Professional Documents
Culture Documents
Identifying Predictors To Pare
Identifying Predictors To Pare
Doctor of Philosophy
Indiana University
August 2019
ProQuest Number: 22618195
All rights reserved
INFORMATION TO ALL USERS
The quality of this reproduction is dependent upon the quality of the copy submitted.
In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if material had to be removed,
a note will indicate the deletion.
ProQuest 22618195
Published by ProQuest LLC (2019 ). Copyright of the Dissertation is held by the Author.
All rights reserved.
This work is protected against unauthorized copying under Title 17, United States Code
Microform Edition © ProQuest LLC.
ProQuest LLC.
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, MI 48106 - 1346
Accepted by the Graduate Faculty, Indiana University, in partial fulfillment of the
Doctoral Committee
_________________________________________________
Nicole Martins, Ph.D., Chair
_________________________________________________
Jonathon Beckmeyer, Ph.D.
_________________________________________________
Bryant Paul, Ph.D.
_________________________________________________
Paul Wright, Ph.D.
August 6, 2019
ii
Copyright © 2019
iii
Acknowledgements
This dissertation is the culmination of a long journey that I completed with the help
of many people. Graduate school is not something one goes through alone and I am so
blessed to have had a stable support system through these years. To begin, I would first
like to thank with the deepest and most sincere gratitude my committee chair, Dr. Nicole
Martins. Since I walked into your office on my visit to Indiana University in 2012, you have
been my advisor and guided me along a trajectory of success. Over the years, you have been
a model of integrity, a mentor, a cheerleader, and a friend. Thank you for your endless
encouragement and advice, and for teaching me how to do good science. I am truly grateful
for all of the time, opportunities, and energy you have invested into my education and
growth.
With equal appreciation and gratefulness, I thank my other Media School committee
members who have been integral to my graduate career and dissertation. Dr. Bryant Paul, I
sincerely appreciate all of the advice and valuable life lessons you have shared with me in
the years that I have known you. I am so very grateful for the opportunities you brought
into my life and all of the time and energy you invested into my education. From The
Summit to helping me through phone interviews, you have included me in every occasion
that offered me employment, growth, and experience. You were not only my mentor and
professor, but you also welcomed me (and Matt) into your home and beautiful family. I will
Dr. Paul Wright, I am so thankful for the email I sent to your advisor years ago when
I was deciding on graduate schools and he pointed me in your direction. In that email, he
expressed his confidence that you would be an “outstanding advisor”’ and, of course, you
iv
were. No matter how busy you were, you always provided thorough, detailed, and
constructive feedback in meetings and on paper. I will never forget the time that you
unfortunately lost all of the edits you made on a manuscript of mine in a Microsoft Word
glitch and, instead of saving yourself time and work by just summarizing your thoughts in
an email, you called me and discussed your feedback over the phone, page by page. Your
level of detail and shrewd advice has, without a doubt, made me a better writer. Thank you
so very much for the time and effort you devoted to my education and for pushing me to be
Dr. Jonathon Beckmeyer, though you were my “minor” advisor, you have played a
major role in my PhD career. The classes that I have taken with you have been instrumental
development. I also owe a substantial amount of my knowledge of statistics to you and the
time and energy you have spent teaching me. Over countless hours and with an abundance
of patience, you have trained me to listen to data and interpret with unbiased eyes and
mind. I have learned so much from you outside of the classroom as well, through the
projects and publications we have co-authored. You are an excellent scientist and
thoughtful, dedicated mentor. You have truly shaped me into a well-rounded and open-
I thank all of the anonymous parents on Mechanical Turk who participated in this
study, and the parents who voluntarily filled out my pilot survey. I also extend my thanks to
Dr. JD Seo at the Indiana Statistical Consulting Center and his class of future statistical
consultants for their feedback on my analysis plan, as well as Hugo Hu at the ISCC for his
v
To the brilliant friends I have made along this academic journey – Glenna, Ashley,
Teresa, Nic, Rachelle, Irene, Cole, Yanyan, Jingjing, Niki, Lucia, and Laura. Adversity has a
way of bringing people together in a special and unique way and I feel so lucky that the
trials of grad school brought us closer. Thank you for being a community of love,
encouragement, advice, and celebration throughout the highs and lows of life.
To my friends in the “real world” – Kristie, Katie E., Morgan, Kaley, Joanna, Rebecca,
Katie C., Tori, India, Caroline, Marina, Anna, and Cheyenne. You all helped maintain my
sanity and supplied never-ending encouragement. Thank you for keeping my spirits up and
I would be lost without the support and love of my mother, father, and sister. You
have always encouraged me, supported me, and believed in me, especially when I didn’t
believe in myself. Mom, you are my role model of strength, dedication, and passion for life;
Dad, you are a fountain of optimism, advice, and knowledge; Aleya, you are my biggest
cheerleader, not only during graduate school but throughout our lives. Your support is
worth more than I can put on paper and I will always look up to you. Special thanks to my
parents and siblings-in-law—Sheryl, David, Ryan, Katie, Joshua, and Viviane—who are the
most wonderful and caring extended family I could wish for. Y’all will probably never read
And finally, to my amazing husband, Matthew. Words cannot fully express how
much you mean to me. Without question or second thought, you have always supported my
goals, empowered me to be my best, and sacrificed so much so that I could succeed. This
achievement would not have been possible without your love and encouragement. You are
vi
Mona Malacane Clay
monitor, and restrict their children’s media activities. A great deal of research has been
conducted on parental mediation of television, movies, and video games; however, few
addition, little research has examined predictors to parental mediation during the teenage
years, which tends to decline as children get older. To address these gaps in the literature,
this study employed the theory of planned behavior (TPB) as a framework for identifying
Parents with a child between the ages of 13 and 18 who own a personal mobile
device were recruited through Amazon’s online labor workforce, Mechanical Turk (N =
592), and responded to a cross-sectional survey assessing their attitudes, subjective norms,
mediation, restrictive mediation, and monitoring of mobile devices. Aspects of the family
system were also evaluated as moderators between the TPB variables and parental
mediation. The results showed that certain attitudinal and control beliefs were associated
with parental mediation behaviors, whereas subjective norms did not impact mediation. No
moderating relationships were found for the family system variables but barriers to
communication did interact with some TPB constructs and active mediation. The findings
are discussed with a focus on practical implications and recommendations for parents.
vii
_________________________________________________
Nicole Martins, Ph.D., Chair
_________________________________________________
Jonathon Beckmeyer, Ph.D.
_________________________________________________
Bryant Paul, Ph.D.
_________________________________________________
Paul Wright, Ph.D.
viii
Table of Contents
ix
Data Collection …………………………………………………………………………..……………. 57
Participants and Sampling ……………………………………………………..………………… 57
Data Screening …………………………………………………………………………...…………… 58
Measures ……………………………………………………………………………………………….… 61
Parental Mediation (DVs) ……………………………………………………………… 61
Theory of Planned Behavior Constructs (IVs) …………………………………. 67
Family System Theory Moderating Variables …………………….…….……… 72
Control Variables …………………………………………………..……………………… 75
Data Analysis ………………………………………………………………………..…..……………… 78
Missing Values …………………………………………………………….………………… 78
Analysis Plan ……………………………………………………………..………….……… 80
CHAPTER 4: Results ……………………………………………………………………..…………………... 86
Descriptive Statistics of the Sample …………………………………..……………………… 86
Model Assessment …………………………………………………………………………………… 98
Measurement Model ……………………………………………………………………… 98
Structural Models …………………………………………………………………………. 106
Hypothesis Testing ………………………………………………………………………………… 112
Active Mediation ……………………………………………………………………………………… 114
Active Mediation Moderation Analyses ………………………………...………… 116
Summed Restrictive Mediation ………………………………………………………………… 120
Summed Restrictive Mediation Moderation Analyses ……………………… 121
Average Restrictive Mediation ……………………………………………………..…………... 123
Average Restrictive Mediation Moderation Analyses ……………………..... 124
Summed Monitoring ………………………………………………………………………………… 127
Summed Monitoring Moderation Analyses ………………………..…………… 128
Average Monitoring …………………………………………………………………………………. 130
Average Monitoring Moderation Analyses ……………………………………… 131
CHAPTER 5: Discussion ……………………………………………………………………………………. 133
Theory of Planned Behavior …………………………………………………..………………… 133
x
Attitudes (H1, H2, and RQ1) …………………………………………..……………… 135
Subjective Norms (H3, H4) ………………………………………………..………...… 140
Perceived Behavioral Control/Self-Efficacy (H5, H6, H7) ………………… 143
Moderation Analyses (RQ2, RQ3, RQ4) ……………………………………………………… 147
Demographic Control Variables ……………………………………………………..………… 150
Limitations ………………………………………………………………………………………….…… 152
Conclusions and Recommendations ………………………………………………………….. 154
Directions for Future Research …………………………………………………………………. 156
References ……………………………………………………………………………...………………………….. 161
Appendices …………………………………………………………………………………………………………
Appendix A: List of Hypotheses, Research Questions, and Summarized
Findings ………………………………………………………………………………………..………… 190
Appendix B: Questionnaire ………………………………………….…………………………… 193
Appendix C: Questionnaire Pre-test Results ……………………………………………… 205
Appendix D: Correlation Results ……………………………………………………………….. 212
Curriculum Vitae ………………………………………………………………………………………..…….…
xi
List of Tables
Table Page
1 Reliability Coefficients and Descriptive Statistics for DVs
and IVs for Sub-samples of Responses that Took
>10, >8, & >5 Minutes to Complete Survey…………………………………….……… 60
2 Reliability Coefficients and Descriptive Statistics
for IVs, DVs and Moderators………………………………………………………………… 63
3 Principal Component Analysis of
Summed Monitoring Questions……………………………………………………….…… 67
4 Means and Standard Deviations for
Barriers to Parent Mediation Items…………………………………….………………… 75
5 Descriptive Statistics for Demographic
and Control Variables ……………………………..…………………………..………………. 90
6 Pearson, Point Biserial and Spearman
Correlations Among All Variables…………………………………………..…….……… 95
7 Pearson Correlations for IVs and DVs…………………………………………………… 97
8 Hierarchical Regression Analysis Predicting
Active Mediation ………………………………………………………………………………… 119
9 Hierarchical Regression Analysis Predicting
Summed Restrictive Mediation …………………………………………………………… 122
10 Hierarchical Regression Analysis Predicting
Average Restrictive Mediation …………………………………………………………… 126
11 Hierarchical Regression Analysis Predicting
Summed Monitoring …...……………………………………………………………………… 129
12 Hierarchical Regression Analysis Predicting
Average Monitoring …………………………………………………………………………… 132
xii
List of Figures
Figure Page
xiii
List of Appendices
Appendix Page
A List of Hypotheses, Research Questions,
and Summarized Findings ………………………………………….………………………… 190
B Questionnaire …………………………………………………………………………………...… 193
C Questionnaire Pre-test Results ……………………………………………………………… 205
D Correlation Results ………………………………………………………………………………. 212
xiv
List of Abbreviations
AM Active Mediation
ATT_AMO Attitudes About Mediation Outcomes
ATT_NA Negative Attitudes about Media Effects and Use
ATT_PA Positive Attitudes about Media Effects and Use
BAR Barriers to Parent Mediation
CFI Comparative Fit Index
CI Confidence Interval
FCQ Family Communication Quality
FST Family System Theory
GFI Goodness of Fit Index
KMA Knowledge of Teen Media Activities
MAR Missing at Random
MCAR Missing Completely at Random
ML Maximum Likelihood
MON_AVG Average Monitoring
MON_SUM Summed Monitoring
NMAR Not Missing at Random
PBC Perceived Behavioral Control
PMU Parent Mobile Device Use
PS_A Autonomy Granting (Parenting Behavior)
PB_C Connection (Parenting Behavior)
PB_R Regulation (Parenting Behavior)
RM_AVG Average Restrictive Mediation
RM_SUM Summed Restrictive Mediation
RMSEA Root Mean Square Error of Approximation
SE Self-Efficacy
SEM Structural Equation Modeling
SN_DN Descriptive Norms
SN_IN Injunctive Norms
SRMR Standardized Root Mean Square Residual
TLI Tucker-Lewis Index
TPB Theory of Planned Behavior
xv
CHAPTER 1: Introduction
behaviors are more prevalent for younger children and steadily decline during the teenage
years. Between the ages of 6 and 12, both parents and children report having rules about
television viewing, video game playing, access to devices, and online activities (Chen &
Chng, 2016; Jordan, Hersey, McDivitt, & Heitzler, 2006; Nikken & Jansz, 2013). These
actions include content restrictions, supervised use, and active discussions about media
content.
However, as children progress into their teen years (age 13+), rules and restrictions
for television, video games, and the Internet decrease (Khurana, Bleakley, Jordan & Romer,
2015; Livingstone & Helsper, 2008; Martins, Matthews, & Ratan, 2017). For instance, a
recent nationally representative survey found that parents of children ages 8-to-12
reported significantly more content-based rules across different technologies than parents
of children ages 13-to-18 (Lauricella et al., 2016). Parents of younger adolescents were
more likely to impose and enforce rules about time spent with media, and these parents
also had more frequent active discussions with their children about TV content, different
There are several reasons for this decline. During the transition to adolescence,
young people are developing into more autonomous, independent, and self-governing
individuals (Spear & Kulbok, 2004). Concurrently, adolescents begin to spend more time
with their peers and construct an individual identity outside of their parental umbrella
(Buhrmester, 1992). As a consequence, their activities become less and less supervised
(Steinberg, 1999) and monitoring simply becomes more challenging for parents as their
1
children spend more time away from them. Teens also have more experience with media,
increasingly own personal mobile devices (Rideout, 2015), and may therefore require less
about media use and potential media effects remain high despite this reduction in media
monitoring (Lauricella et al., 2016). Parents report worrying about Internet-related risks
social media, as well as off-line activities like exposure to mature content in TV and movies
(boyd & Hargiatti, 2013; Madden, Cortesi, Gasser, Lenhart, & Duggan, 2012). Potential
negative effects as a result of teen media consumption are also a concern among parents
(Sandstig, Johansson, & Ringsberg, 2013; for a review of effects, see Wright, Malamuth, &
Donnerstein, 2012).
Put simply, there is a disconnect between parental concerns about teen media use
and their reported behaviors that could address these concerns. The purpose of this
dissertation is to investigate this topic using the theory of planned behavior (Ajzen, 1991)
as a framework for identifying parent factors that predict mediation during the teen years.
To be clear, it is not the purpose of this paper to imply that decreasing mediation is
neglectful, or to condemn parents for not monitoring teen media use to the extent that they
did when they were younger. Rather, this study seeks to understand if and how attitudes,
norms, and perceived behavioral control beliefs relate to media monitoring during teen
years, and integrate these into a theoretical model along with family system factors.
2
Background of this Study
media content with children and can occur before, during, or after media use. Restrictive
mediation includes placing limits on access, time, and/or content. Finally, co-viewing (or
co-using) refers to a parental presence during media use without any discussion (as this
television, movies, and video games and is beginning to encompass the Internet, social
media, and mobile platforms (Collier et al., 2016). These mediation behaviors have been
associated with both positive and negative outcomes in children. Active mediation is
typically associated with better well-being for adolescents. When parents engage their
teens in discussions about media content, they have lower intentions to engage in sexual
activity, exhibit lower levels of aggression and externalizing behavior, and engage in less
risky online behavior (Fisher et al., 2009; Lwin, Stanaland, & Miyazaki, 2008; Padilla-
Walker, Coyne, & Collier, 2016). Restrictive mediation has been shown to be successful at
activity among teens, decreased risk of sharing private information online, and lower
exposure to pornography (Fisher et al., 2009; Lee, 2013). However, restrictive mediation
has also been associated with reactance, or boomerang effects, among teens leading to
behavior (Nikken & de Graaf, 2013). The effects of co-viewing are mixed. Some studies have
found that it is related to increased aggressive tendencies after viewing violent TV shows,
3
indicating that parental co-viewing without discussion can signal tacit approval of such
messages (Nathanson, 1999; Rasmussen, Coyne, Martins, & Densley, 2018). However, other
studies have shown that co-viewing is negatively associated with aggression, positively
associated with prosocial behavior, has no effect on sexual activity, and is related to greater
family connectedness, (Coyne, Padilla-Walker, Stockdale, & Day, 2011; Fisher et al., 2009;
players/iPod touches, hereafter called mobile devices. In this context, I refer to parental
mediation as encompassing overall use of mobile devices, including what, when, and where
adolescents can access. Mobile devices were chosen because recent reports of U.S. teen’s
daily media consumption demonstrate that by the time kids reach the teen years, many
have their own personal smartphone or other portable device (or have access to one) and
these devices occupy a large amount of their daily screen time (Lauricella, Cingel,
Blackwell, Wartella, & Conway, 2014; Rideout, 2015). In addition, teens report that a
plurality of their Internet use is on a smartphone or other mobile device (Lenhart, 2015).
mobile devices and adolescent Internet use in general, finding that parents engage in
similar active, restrictive, and co-using mediation behaviors (e.g. Livingstone & Helsper,
2008); however, this dissertation will focus on active and restrictive mediation. Co-
using/co-viewing will not be assessed because it is less commonly reported among parents
2012), and the nature of mobile devices (i.e. size, portability) make them challenging for
4
parents and teens to use together on a consistent basis. Moreover, studies that have
find that it is less frequent than active and restrictive mediation (e.g. Warren, 2017).
Instead this study will consider parent monitoring 1 of mobile devices − defined as
mediation behavior that has emerged specifically within the context of Internet and mobile
device mediation (Livingstone & Helsper, 2008; Vaterlaus, Beckert, Tulane, & Bird, 2014).
Like co-viewing, monitoring can be unobtrusive (e.g. checking their device later on to
determine previous activities) and also autonomy-supportive, by asking what the child is
doing on their device in real-time and providing guidance (if needed) or freedom (if not)
(Shin & Li, 2017). Given the nascence of this body of research, studies that have focused on
historically older forms of mass media (i.e. television, movies, and video games) along with
mediation literature has been largely atheoretical (Chakroff & Nathanson, 2008). That is,
considered more of a descriptive approach rather than a parsimonious theory from which
we can derive predictions and test hypotheses. Given the ubiquity of media in the lives of
associated with mediation would be quite useful to future researchers as well as advocacy
organizations dedicated to media education and literacy for parents and teens. To address
this gap in the literature, I employed the theory of planned behavior (Ajzen, 1991) as a
1Throughout this paper, the term “parental mediation” is used to refer collectively to active
mediation, restrictive mediation, and monitoring.
5
framework to investigate if and how social and psychological factors predict parental
mediation during the teen years. Additionally, I considered how aspects of the family
system can interact with these social-psychological factors to impact parental mediation.
This dissertation is organized into five chapters. The first chapter consisted of the
introduction and provided background on the purpose of this study and the gap it aims to
fill. The next and second chapter is the literature review. This section discusses teen mobile
device use, reviews research on effects and outcomes of parental mediation, and
summarizes the purpose of this study. The theory of planned behavior and its core
constructs are defined next, along with the independent variables that are the main focus of
this study. I also discuss family system theory and the potential moderating role of the
family environment on parental mediation. The third chapter is the Method section, where
I explain the study design, the survey instrument and measures, and detail my analysis
plan. The fourth chapter is the Results section, which begins with descriptive information,
then explains the evaluation of the measurement and structural models, and findings of the
statistical analyses. Chapter five is the Discussion section and reviews the main findings of
the hypotheses and research questions with a focus on practical implications for parents.
Limitations, conclusions, and directions for future research are also discussed.
6
CHAPTER 2: Literature Review
This chapter has three main sections. First, I begin with a discussion of teen mobile
device ownership and general use. I provide conceptual definitions for the three parental
mediation styles that are investigated in this study, briefly review research on the effects of
parental mediation of mobile devices, and explain the purpose and contribution of this
study. The second section focuses on the theory of planned behavior (TPB) and its core
constructs. This section reviews literature on this study’s independent variables, which are
the hypothesized predictors of parental mediation. The third section reviews family system
theory (FST) and the impact of the family environment on parental mediation behaviors.
The last section details the control variables I employed. My hypotheses and research
questions are embedded throughout the literature review and also listed in Appendix A.
Teens in the United States engage in a variety of media activities, including watching
television (TV) content and movies, playing games, surfing social media, creating their own
content, communicating with others, reading and writing, and listening to music, to name a
few. On an average day, teens spend almost seven hours (6 hours and 40 minutes) 2 using
screen media, namely tablets, smartphones, iPods/MP3 players, TV, computers, and video
games (Rideout, 2015). Listening to music and watching entertainment TV content are the
most common daily activities among teens, followed by using social media and watching
videos (Rideout, 2015). A plurality of this use is done on mobile devices, mainly
2 This does not include screen time in school or for homework. Non-screen media activities include
listening to music and reading print material (Rideout, 2015).
7
Teens also own personal mobile devices or have relatively easy access them.
literacy education and research for parents, teachers, and policy makers (Common Sense
Media, n.d.) − commissioned two recent nationally representative “censuses” of media use
(Census reports were authored by Lauricella et al., 2016 & Rideout, 2015). Approximately
67% of teens in the adolescent study self-reported owning a smartphone, 37% had a tablet,
45% had a laptop, and 20% had an iPod Touch (Rideout, 2015). Parent’s reports of teen
media ownership in the parent study were higher, such that 79% of teens had a
smartphone, 45% had a tablet, 43% had a laptop, and 28% had a portable game player
(Lauricella et al., 2016). Another nationally representative study of 1,637 teens (ages 13-
to- 17) conducted by the PEW Internet and American Life Project reported 73% have or
have access to a smartphone, 58% have a tablet, and 87% have a laptop (Lenhart, 2015).
Put simply, teens are frequent users and many are owners of mobile devices.
Parental Mediation
Scholars from different fields have studied parental mediation, including psychology
(Vaterlaus et al., 2014), communication (Valkenburg, Krcmar, Peeters, & Marseille, 1999),
family science (Padilla-Walker et al., 2016), public health and pediatrics (Radesky et al.,
2016), and education (Valke, Bonte, Wever, & Rots, 2010). This research has accumulated
over several decades and variations in conceptual definitions of what parental mediation is
and how parents enact it have evolved over this time (for a thorough review, see
Nathanson, 2001a). In the most general sense, parental mediation refers to interactions
8
communicate about media (Nathanson, 2001a). The three primary methods of parental
mediation are active, restrictive, and co-using/co-viewing, which have been conceptualized
from research focused on traditional forms of media (e.g. TV and movies viewed on TV sets,
video game consoles; Valkenburg, Piotrowski, Hermanns, & de Leeuw, 2013). This study
focuses on active mediation, restrictive mediation, and monitoring of mobile devices and
content between parents and children that take place before, during, or after viewing/use
(Nathanson, 2001a). Although these conversations can occur between a non-parent and
child (e.g. a sibling; Domoff et al., 2019), the majority of research has focused on parents
encompass topics such as critical evaluation of media and interpretation of content (Nikken
& Jansz, 2013). Researchers have also distinguished different types of active mediation.
Negative active mediation refers to evaluative discussions that would deter imitation and
discussions that would encourage imitation and vicarious learning of desirable behaviors;
asking the child for their knowledge or thoughts (Austin, Bolls, Fujioka, & Engelbertson,
While the bulk of parental mediation research has focused on active mediation of
television content, its definition and application extends to mobile devices and the Internet
9
as well. For example, Shin and Li (2017) assessed active mediation of digital technologies
positive and negative consequences of using mobile devices, providing assistance when
using mobile devices and recommending content. Active mediation has also been measured
as conversations surrounding the importance of rules for mobile device use and why too
much time spent using devices can be detrimental (Hwang et al., 2017). In this study, I
content teen’s access, behaviors they engage in, and applications they use on mobile
devices.
surrounding media content and use (Nathanson, 2001a). It includes restrictions on the
amount of time children can spend using media, the time frames during which children can
or cannot use media, access to devices, and the types of content children can or cannot view
(Nathanson, 2001b; Valkenburg et al., 1999). The consistency with which parents enforce
their rules and limitations is another aspect of restrictive mediation. That is, a parent may
report having specific rules for media use but not strictly enforce them (Valkenburg et al.,
2013). Inter-parental disagreement over certain rules can also impede consistent
For mobile devices and the Internet, restrictive mediation behaviors can include
limitations to access and time. For example, Hwang and colleagues (2017) asked parents if
they have restrictions for smartphone use and prohibit their children “from using
smartphones longer than a certain amount of time” (p. 364). It also includes limitations on
activities that kids can engage in, like not being allowed to use apps that connect a child to
10
strangers or share personal information online (Livingstone & Helsper, 2008). These
restrictions can be enforced by using software that blocks or limits access to certain
content or alert parents when content is downloaded or viewed (Livingstone et al., 2017).
In this study, I conceptually and operationally measured restrictive mediation as rules for
the amount of time and when teens can use media, applications that can/cannot be used,
and the use of parent controls or third-party apps that serve to block or limit access to
specified content. I also asked about the consistency and strictness with which parents
(Dishion & McMahon, 1998). It encompasses attention paid to learning and gaining
knowledge about behavior and includes supervisory actions that are enacted to gain this
information (Kerr & Stattin, 2000). Applied to media use, monitoring includes covert and
overt actions that serve to oversee one’s media use and activities. Monitoring has an
element of co-use, in that a parent can engage in the same activity as their child with the
goal of gaining knowledge of their activities (e.g. playing the same online game,
friending/following a child on their social media page); however, it does not require
parents and children to be in the same shared space while co-use is occurring.
seeing a child’s activities (but not necessarily blocking them; Livingstone & Hesper, 2008;
Livingstone et al, 2017), asking what the child is doing on a device, watching the screen of
the device (Shin & Li, 2017), asking for a child’s password to their device or online
accounts, and following a child’s online accounts to (in)conspicuously check their posts
11
(PEW, 2016). These behaviors can be done openly and overtly so that the child knows they
are being supervised, covertly without the child’s knowledge, or a combination of both. For
instance, a parent can publicly “friend” (or follow) their child’s social media accounts but
randomly look at their account to view their posts, with or without their child’s knowledge.
In this study, I conceptually and operationally measured monitoring as covert and overt
First, co-using is less commonly reported for mobile devices compared to traditional media
(Connell, Lauricella, & Wartella, 2015; Warren, 2017). In a recent study, Warren (2017)
surveyed 531 parents with a child between the ages of 10 and 17 about their parental
mediation practices for four media: television, video games, the Internet, and cellphones.
Parental mediation was reported on a four-point Likert scale and responses were summed
to create total score. Mothers and fathers respectively reported shared TV use of 16.83 and
make them more cumbersome for parents to co-use or co-view with their child (Hwang et
al., 2017). Mobile devices that are owned by the child may be more likely to be considered
personal devices compared to media that are in shared places like TVs in a living room
(Vaterlaus, Beckert, & Bird, 2015). Due to this personal nature, it is unlikely that co-viewing
and co-using is a consistently occurring parent behavior for teens (or children in general
for that matter). Indeed, co-viewing is less commonly reported among teens compared to
younger adolescents (Nielsen, 2015). This is not to imply that parents do not ever co-
use/view mobile devices with their teens; rather, these reasons explain why co-use is less
12
common among teens and parents and therefore less relevant to this study of parental
How does parental mediation impact teen media use and the effects of use? The
growing body of literature concerning parental mediation of mobile devices and the
Internet suggests that different mediation behaviors produce different outcomes (Chen &
Shi, 2019). The mechanisms through which these behaviors impact teens are also varied
and can differ based on parenting behavior (Valkenburg et al., 2013). The next section
details the different ways active, restrictive and monitoring mediation impact teen media
Active mediation. Active mediation has historically been associated with more
positive outcomes than restrictive mediation and co-viewing and co-use (Collier et al.,
disapproval with content or media use, they are encouraging and teaching critical thinking
and evaluation, which can mitigate negative effects of exposure and use in adolescents
accommodating the growing need for independent thought-processes and decision making.
thoughts on media content and seeks to elicit their opinions and input on different topics
which is more focused on parent opinions and ideas, and less likely to consider a child’s
thoughts on matters. Not surprisingly, controlling active mediation has been negatively
13
associated with prosocial behavior and positively related to antisocial behaviors, while
styles of active communication still reports beneficial effects of this parental mediation
behavior. For example, active mediation is associated with the development of self-
regulation and self-regulation is positively associated with prosocial behavior and less
externalizing behaviors over time (Padilla-Walker et al., 2016). Children of parents who
practice active mediation are less likely to share sensitive private information online (Lwin
et al., 2008); less likely to use to the Internet and mobile devices compulsively or
excessively (Hefner, Knop, Schmitt, & Vorderer, 2019; Kalmus, Blinka, & Ólafsson, 2015);
and less likely to engage with strangers online (Shin & Ismail, 2014). Other research has
shown that active mediation of smartphones does not directly protect against the
excessive use of the Internet (van Deursen, Bolle, Hegner, & Kommers, 2015).
Restrictive mediation. Having rules and restrictions for time, content, and access
of media is an effective strategy for promoting healthy media habits among adolescents.
Research has shown that restrictive mediation is associated with lower general media use
and time online (Lee, 2013; Padilla-Walker, Coyne, Kroff, & Memmott-Elison, 2018), as well
as lower odds of developing “smartphone addiction” (Chang et al., 2019). It also negatively
predicts problematic Internet use among tweens and teens (Kalmus et al., 2015), disclosure
of private information online (Lwin et al., 2008), and lower contact with online risks, like
14
accidental exposure to violent and pornographic content, online communication with
strangers, and disclosing private information about one’s parents online (Lee, 2013).
Previous research has also demonstrated the ways in which restrictive mediation is
less effective at preventing undesired behaviors and can actually produce unintended
effects, particularly among teens (e.g. Byrne & Lee, 2011). Reactance—also referred to as
the forbidden fruit effect, or boomerang effect—is the increased motivation to engage in a
behavior (or an increased interest in the behavior) in response to regulations or rules that
threaten freedom or autonomy (Brehm & Brehm, 1981). When parents impose strict rules
on behavior and activities, in some cases, it causes an increase in the undesired behavior.
For example, some research has shown that restrictive mediation increases (Sasson &
Mesch, 2014) or has no effect on risky Internet behaviors (Vaala & Bleakley, 2015).
Restrictive mediation that incorporates technical software to track behavior and block
content can increase risky behaviors like sharing private and personally identifying
information (Lwin et al., 2008; Shin & Ismail, 2014) and is also negatively related to the
more likely to occur in the absence of open conversations between parents and children.
When rules are developed with input from the child and communicated in an autonomy-
supportive manner, rules and restrictions do not appear to instigate reactance and
unintended effects (Byrne & Lee, 2011; Fikkers, Piotrowski, & Valkenburg, 2017;
15
Monitoring. Similar to restrictive mediation, monitoring can produce mixed effects.
For example, some studies have found that monitoring is negatively related to adolescent
social media use (Khurana et al., 2014; Vaala & Bleakley, 2015), experiencing online
harassment (Khurana et al., 2014), and problematic or excessive mobile device use
(Bleakley, Ellithorpe, & Romer, 2016). In other cases, monitoring has been positively
associated with risky online behaviors (Ghosh et al., 2018; Sasson & Mesch, 2014) and
There are a few reasons that can explain why these mixed findings occur. First, it is
possible that, like restrictive mediation, monitoring can be inconsistently practiced and is
like depression, delinquency, and lower academic achievement (Simons & Conger, 2007).
between parents and children (e.g. checking activities after use) and this can signal tacit
Wisniewski, Jia, Xu, Rosson, & Carroll, 2015). Nevertheless, because monitoring practices
are less intrusive than active conversations and rules, it may be a preferable strategy for
parents of teens to avoid reactance while also keeping an eye on their activities.
Looking at the body of parental mediation literature in total, there are areas that
have accumulated quite a bit of research and others that are still nascent. The purpose of
parental mediation of mobile devices, which have proliferated in ownership among teens
16
rather quickly over the past decade and have clearly and unequivocally become a large part
of daily life (Lenhart, 2015; Reid Chassiakos, Radesky, Christakis, Moreno, & Cross, 2016).
In comparison to historically older forms of media like TV and video games, mobile device
technology and penetration has advanced rather exponentially (Twenge, Martin, &
Spitzberg, 2018) yet published research on these devices has understandably not increased
parental mediation of young children (Beyens & Eggermont, 2014), tweens (Shin, Huh, &
Faber, 2012), early to middle adolescents (Beyens, Valkenburg, & Piotrowski, 2019), and
across tween to teens (Rasmussen et al., 2018), but fewer have focused on just the teen
years (e.g. Wisniewski et al., 2015). This is also a dearth of research that has focused on
antecedents of parental mediation (Hwang & Jeong, 2015). So, while studies consistently
find that parental mediation decreases as children get older (Chen & Chng, 2016; Padilla-
Walker, Coyne, Fraser, Dyer, & Yorgason, 2012b), we know less about factors that predict
parental mediation during the teen years, which are also a period of increased media use
(PEW, 2016). Thus, this study seeks to contribute to this area and extend this line of
inquiry.
A third, and related, gap in parental mediation literature is the lack of theoretical
communication scholars have incorporated parenting styles (Katz, Lee & Byrne, 2015), self-
determination theory (Valkenburg et al., 2013), and protection motivation theory (Hwang
et al., 2017) to understand to how specific parent cognitions like perceived risks and
autonomy development motivate parental mediation, fewer efforts have been made to
17
incorporate a single parsimonious theory (e.g. Shin & Kim, 2019; Tsai, Wei, & Tsai, 2014).
As a step in this direction, I employed the theory of planned behavior (TPB; Ajzen, 1991) as
a framework to see if and how media related social and psychological factors predict
parental mediation. The broad question I sought to answer is: Do the constructs of the
if parents will engage in mediation and if so, which behaviors they are likely to enact? A
particular benefit to the TPB is the incorporation of antecedent beliefs that compose the
primary constructs. Identifying and investigating these underlying beliefs and their role as
Lastly, this study responds to calls for more research that incorporates family
variables into our understanding of parental mediation (Collier et al, 2016; Nathanson,
2015). Parental mediation of media does not occur in isolation; rather it is one aspect of
many that are comorbid to parenting and family life as a whole. Toward this end, I assessed
how family communication, parenting behaviors, and other demographic aspects of the
explains that human behavior is a function of their intentions—if someone plans to enact a
behavior is more likely to follow through than someone with a weaker intention. Intentions
are determined by one’s attitudes toward behaviors, subjective norms, and perceived
behavioral control. Attitudes toward behaviors are the positive and negative ideas one holds
about the outcome of a behavior. Subjective norms are the social expectations surrounding
18
a behavior. Lastly, perceived behavioral control encompasses one’s perception that they can
effectively control and/or carry out a behavior. These three constructs collectively
influence one’s intentions to engage in a behavior, which indicate how hard a person is
willing to try, and how much effort they will exert, to perform a behavior. A model of the
predictors, that influence parental mediation behaviors. The TPB provides such an
explanatory function as it also includes the beliefs that inform one’s attitude toward a
behavior, subjective norms, and perceived behavioral control. That is, it takes into account
the salient antecedent beliefs that underlie each of these constructs. Behavioral beliefs
influence one’s attitudes toward a behavior, normative beliefs influence subjective norms,
and control beliefs influence perceived behavioral control (see Figure 1).
This theory and its constructs can be applied to parental mediation research (e.g.
Krcmar & Cingel, 2016). Specifically, attitudes toward behaviors refer to positive and
negative ideas about mediation its resulting outcomes; subjective norms encompass
mobile devices (e.g. parents should/should not be mediating); and perceived behavioral
control refers to one’s capacity to successfully enact mediation behaviors. These constructs
can be used to test predictions regarding when and why parental mediation would occur
19
20
Given the potential utility of the TPB in understanding predictors of behavior, this
study utilized it as a framework to identify and synthesize variables that have been
“framework” rather than a formal test of the theory because this is one of the few studies
that have used the TPB in parental mediation research (Krcmar & Cingel, 2016; Shin & Kim,
2019; Tsai et al., 2014) and, as such, our understanding of if and how it applies is still
developing. I also did not assess intentions—the precursor to behavior—to mediate; rather
constructs are determined by their antecedent behavioral, normative, and control beliefs.
The following sections review research concerning variables that comprise the antecedent
beliefs.
Azjen (1991) explains that attitudes toward behaviors are the positive and negative
ideas associated with a behavior and its outcomes. One’s attitude toward a behavior is
determined by their behavioral beliefs, which are shaped by past experiences and the
expected consequences associated with a given behavior. Applied to this study, the
behavioral beliefs hypothesized to influence attitudes toward parental mediation are (a)
negative ideas parents have about media effects and use of media, (b) positive ideas
parents have about media effects and use of media, and (c) parents attitudes toward the
21
Previous research has found that parent attitudes and beliefs about media effects
are related to their mediation behaviors. That is, when parents have a general positive or
negative attitude about specific media activities that their teen engages in, and the
potential effects of these activities, they are more likely to enact parental mediation
behaviors that reflect these attitudes. An early study by Nathanson (2001b) demonstrated
this effect with parents of younger children and attitudes toward violent entertainment
television content. Individuals were asked about the perceived utility and the perceived
harm of their child watching aggressive programming for children and their self-reported
restrictive and active mediation behaviors. Parents who perceived violent content as
potentially harmful and of low social utility were more likely to use restrictive mediation
Similar findings have been replicated in a more recent study with parents of
adolescents ages 8-to-18 and their negative attitudes toward media portrayals of relational
aggression (Rasmussen et al., 2018). The results showed that attitudes impact mediation in
different ways but tended to positively predict them. Specifically, parents who had negative
attitudes about relational aggression in the media in general used more restrictive
mediation, while parents who perceived that the media their child consumed had relational
Negative attitudes also predict parental mediation of video games. Shin and Huh
(2011) found that parents of teens who believed that video games in general would have a
negative influence on their teen were more likely to practice restrictive mediation (i.e. did
not allow game play). Similarly, Nikken and Jansz (2006) surveyed dyads of children and
22
parents and found that parents who believed that playing video games would negatively
impact their child’s behaviors and attitude practiced more restrictive and active mediation.
Parents also report negative concerns about their teen’s use of the Internet and
mobile devices. Worries and concerns include detrimental impacts on their sleep and
the possibility of over-sharing personal information online (Lauricella et al., 2016). These
For example, Warren (2017) found that negative attitudes predict parental
mediation of cell phones and the Internet in a sample of American parents with a child
between the ages of 10 and 17. In this study, parents were asked about their beliefs that
from media use (p. 491) and reported their mediation practices for the Internet and cell
phones. Negative attitudes positively predicted restrictive mediation of the Internet, and
In a study of South Korean parents with a younger adolescent in the 4th, 5th, or 6th
grade, Hwang and colleagues (2017) investigated the role of perceived threat of
smartphone addiction). Their hypothesis that perceived severity and susceptibility would
be positively related to restrictive and active mediation was partially supported, such that
severity positively predicted both styles mediation of smartphones but susceptibility did
23
not. Thus, the potential consequences of addiction motivated parents to actively and
There is also evidence that parent concerns predict monitoring. Researchers in the
Netherlands surveyed parents of children between 9 and 16 years old about their
mediation practices for the Internet and their views of the Internet (Sonck, Nikken, & de
Haan, 2013). Parents who were concerned about their child interacting with strangers and
seeing inappropriate content online used more monitoring strategies than parents who did
not have these concerns. Based on these findings collectively, the following hypothesis is
predicted:
H1: Negative attitudes about media effects and use will be positively associated
with active, restrictive, and monitoring behaviors.
The focus of the previous studies was on potential negative effects of teen media use
and activities; however, there are benefits to media use as well. For example, some
research has found that teens who play strategic video games report higher problem
solving skills, which indirectly improves academic grades (Adachi & Willoughby, 2013).
Social media use can increase teens’ cognitive and affective empathy over time (Vossen &
(Valkenburg & Peter, 2007). Parents also perceive and report that media use can have
benefits for their teen-aged children. For instance, parents believe that social media can
help maintain relationships with peers and help with academics and school work
(Lauricella et al., 2016). Parents who text and call their children, watch TV and movies
together, and play video games with their children also report higher family connection
24
The ways in which these positive attitudes about media use and activities impact
parental mediation vary by media platform and sometimes do not impact them at all. For
example, Nikken & Jansz (2006) surveyed 536 parents and children (between 8 and 18
years old) about parental mediation of video games (active, restrictive, and co-playing) and
different positive and negative perceptions of the effects of playing video games. Positive
effects of game playing included cognitive benefits (e.g. improved concentration), learning
curiosity”), and social-emotional benefits (e.g. relaxation, “improving social skills, learning
to control emotions”) (p.190). Negative effects included physical impacts (e.g. joint pain,
dizziness, and nausea) and behavioral and attitudinal impacts (e.g. “becoming violent,
imitating rude language, considering violence as a normal way of solving problems”) (p.
190). The results did not show significant predictive relationships between perceptions of
cognitive benefits of video game playing and active or restrictive mediation; however,
In a more recent study, Schaan and Melzer (2015) assessed parents’ attitudes
concerning media effects as predictors of parental mediation for television and video
games, using items from Nikken and Jansz’s (2006) measure of positive and negative media
effects on children. The target children in this study were between 9 and 12 years old.
Their findings showed associations between perceived positive effects and some parental
mediation behaviors. Specifically, positive effects of video game playing predicted active
of video games. Perceptions of positive effects did not impact parental mediation of
television.
25
There are mixed findings regarding whether or not positive attitudes impact
parental mediation of mobile devices and the Internet. For example, Shin and Li (2017)
asked parents of adolescent-aged children to rate their attitudes toward digital technology
using semantic differential scales. Examples of attitudes toward using digital technology
Their results did not find evidence that positive attitudes significantly impact active
Warren (2017) adapted Nikken and Jansz’s (2006) measure and assessed the
impact of positive attitudes on parental mediation of cellphones and the Internet. Positive
effects of media use were unrelated to mediation of cell phones; however, they did
positively predict active mediation of the Internet. That is, parents who thought that using
the Internet would benefit their child were more likely to talk to their children about things
like appropriate online behaviors, safe and unsafe online behaviors, and why certain
A recent study surveying Dutch and American parents also found mixed results
regarding parental mediation of social media use and attitudes. Krcmar and Cingel (2016)
surveyed parents with a child between 10 and 18 years old about their attitudes toward
their child’s use of social media, specifically asking if they “approved of social media” for
their child and if they thought social media were “good” for their child (p. 253). For Dutch
parents, these attitudes negatively predicted restrictive mediation of social media and
positively predicted co-use; whereas for American parents, attitudes positively predicted
26
both restrictive mediation and co-use. Positive attitudes were unrelated to active
Based on this limited research and mixed findings, the impact of positive attitudes
about media effects and use on parental mediation of mobile devices is inconclusive. It
could be the case that positive attitudes about mobile devices instigate active mediation of
some activities, as Warren (2017) found, or not impact it at all, per Shin and Li (2017).
Conversely, positive attitudes could negatively predict restrictive mediation, as Krcmar and
Cingel (2016) found among Dutch parents. This relationship does have face validity such
that, in theory, parents who approve of a behavior would be less inclined to restrict it;
however, the opposite was found among American parents and positive attitudes predicted
more restrictive mediation. Given these mixed findings, I posed a research question in lieu
of a hypothesis regarding the impact of positive attitudes of media effects and use on
parental mediation:
RQ1: Are parents’ positive attitudes about media effects and use associated with
active, restrictive, and monitoring behaviors?
Lastly, the beliefs parents hold concerning the outcomes of their mediation can also
influence their behaviors. That is, if parents believe their mediation actions could prevent
undesired effects or behaviors, or reduce the potential for their child to experience these
outcomes, parents could be more likely to continue mediation during the teenage years.
effects or behaviors, or would possibly instigate negative behaviors in their child, could
prevent parents from engaging in these actions. To be clear, the focus of this attitude is the
on the perceived outcome of parental mediation; whereas, the focus of positive or negative
attitudes of media effects and use is on the perceived impact of mobile device use on teens
27
themselves. Put differently, attitudes about outcomes refers to parent perceptions about
the resulting effect of their mediation behaviors, and positive (and negative) attitudes of
media effects and use refers to the perceived effects on teens as a result of their personal
Studies on the extent to which parent expectations about the outcomes of mediation
influence mediation behaviors are limited but qualitative research indicates that children’s
responses to mediation may play a role. Evans and colleagues (2011) conducted qualitative
interviews with 180 adolescents and their parents asking about obstacles they experience
when trying to restrict their child’s TV time and reasons they choose not to restrict. In this
study, several parents of teens commented that their children would likely try to evade any
rules/restrictions if they were imposed and instead parents thought they would have to
personally oversee their child’s media use to enforce such rules. The children in this study
responded similarly, indicating that they would seek to change their parents mind through
any means possible (e.g. begging, getting upset, “[beating] my head against the wall”; p.
1231). Parents also reported benefits to their child’s TV viewing as a reason for not
wanting to restrict or limit too much. Specifically, co-viewing increased feelings of family
closeness and also helped facilitate conversations about sensitive topics with children.
child between the ages of 3 and 16 to investigate how attitudes, norms, and self-efficacy
predict parent’s intentions to restrict their child’s TV viewing. Of particular interest to this
section are the results for attitudes, which were assessed by asking parents to rate the
likelihood of several outcomes if they were to limit TV viewing. Attitudes were the
28
researchers noted four specific attitudes that differentiated parents who intended to limit
TV from those who did not: believing that restrictions would “help [their child] get more
exercise, do better in school, [make them] more likely to talk to the family, and decrease
2013, p. 528). Thus, parents who felt their actions could help their child, and decrease the
likelihood they would encounter undesired content, had higher intentions of engaging in
restrictive behaviors versus those who did not believe these actions would make a
difference.
Similar findings have been observed for active and restrictive mediation of digital
technologies, social media and the Internet. Parents who think they can influence their
child’s use of digital technologies are more likely to use active and restrictive mediation
(Shin & Li, 2017), and those who believe that having some level of control over their child’s
social media would be good for them are more likely to engage in active mediation (Krcmar
& Cingel, 2016). There is also evidence that indicates these beliefs impact monitoring, such
that parents who think involvement in their child’s use of the Internet positively affects
To date, there are few studies that have investigated parental mediation of mobile
devices and fewer have assessed the specific role of parent attitudes toward mediation
outcomes. There is, however, evidence that positive attitudes about the outcomes of
mediation can positively predict active and restrictive mediation of television, the Internet,
and social media (Bleakley et al., 2013; Krcmar & Cingel, 2016; Sonck et al., 2013). Based on
this research, I expect similar findings for parental mediation of mobile devices and predict
29
H2: Parents with positive attitudes toward mediation outcomes will be more
likely to engage in active, restrictive, and monitoring behaviors.
Subjective Norms
(or not perform) a particular behavior (Ajzen, 1991). These are influenced by one’s
normative beliefs, the perception that other important people (e.g. friends, family members)
above definition for subjective norms is, more specifically, an injunctive norm—a
perception that important others approve of a behavior and support it or vice versa
(Sheeran & Orbell, 1999). Another type is descriptive norms—the perceived or actual
behaviors of others. According to the TPB, both norms influence behavior; however, the
former refers to perceptions of others’ social approval for a behavior (i.e. what people think
you should or should not do) and the latter refers to one’s perceptions of how others
behave (i.e. what is the “normal” thing to do; Sheeran & Orbell, 1999). Injunctive norms
were part of the original conception of subjective norms in the TPB and more recently, the
author has recommended that the distinction be made between the two norms and
Applied to this study, injunctive norms concern the perceived social pressure and
approval to monitor teen media, namely what parents should be doing. That is, other
people believe that parents should be actively or restrictively mediating and monitoring
their teen’s mobile devices, or that they should not be doing so. This perceived social
expectation can come from a variety of sources, one of which may be the media that
parents consume. For instance, popular press reports of teens and social media often focus
on risks, danger, and negative consequences (Draper, 2012; Stern & Odland, 2017). The
30
tone of these stories imply that most teens are, or will, engage in risky behaviors and that
online dangers are common and/or increasing in frequency. Reports typically end with the
advice that parents should vigilantly “keep tabs” on their children’s online behaviors (Stern
& Odland, 2017). Frequent exposure to these types of news stories could cultivate the
recently released an updated policy statement reviewing both risks and benefits to
and Media, 2016). Included in this statement are general media monitoring suggestions for
parents of adolescents, again implying that parents should be taking these actions. These
Few studies have investigated the extent to which injunctive norms specifically
impact parental mediation behaviors, but those that have do find relationships between
norms and parent behaviors. For example, Krcmar and Cingel (2016) asked parents with a
child between the ages of 10 and 18 about their normative perceptions towards allowing
their child to use social media. Specifically, parents were asked to rate their agreement
with statements like, “Most of my friends and family with kids the age of mine approve of
social media use for their children” and “Most of my friends and family with kids the age of
mine think social media is just fun for their children” (p. 254). These ideas represent
injunctive norms because they refer to a general perception of others approval for a certain
behavior. Parents who endorsed these norms towards allowing their child to use social
31
Bleakley and colleagues (2013) found similar results. In this study, parents were
asked about subjective norms in a few ways. There were direct self-report statements for
injunctive norms (i.e. “[Do] most people who are important to you think you should not or
you should limit [your] child’s television viewing; p. 527), as well as questions asking about
the underlying beliefs that inform injunctive norms. Specifically, parents rated their
perceptions that other people thought they should or should not limit TV, including
“teachers, other parents you know, your [spouse/partner], and your child’s/children’s
doctor” (p. 527). The findings were the same for both types of questions: parents who
reported stronger normative pressure were significantly more likely to report higher
injunctive norms and parental mediation, there are studies outside of this field that
demonstrate associations between injunctive norms and other parent behaviors. For
example, injunctive norms have been found to positively predict parent-initiated sexual
Cabriales, Ronis, & Zhou, 2008) and parent communication about alcohol (Napper,
Hummer, Lac, & LaBrie, 2015). Although sexual activities and alcohol use are, of course,
quite different from mobile device use and activities, parents do report concerns that their
children are engaging in risky behaviors on them (Lauricella et al., 2016). Based on this
research collectively and the theoretical predictions of the TPB, I expect that injunctive
norms will positively impact parental mediation in this study with the following
hypothesis:
H3: Injunctive norms will be positively associated with active, restrictive, and
monitoring behaviors.
32
Descriptive norms of parental mediation are the perceived mediation behaviors
performed by others. That is, whether or not other people (e.g. friends, family members)
inferred through media reports as well, especially if the reports include parent interviews
or actual media monitoring data (e.g. PEW, 2016). Additionally, parents may participate in
first-hand conversations about media monitoring customs with other parents, which could
inform these norms. Information could also come second-hand from one’s own children,
who might relay the practices of their friends’ parents (e.g. “Mona’s parents do/do not
descriptive norms and parental mediation but the evidence that does exist suggests a
connection. In the Krcmar and Cingel (2016) study reviewed above, parents were also
asked about their perceived norms to control their child’s social media, rating statements
such as “Most of my friends and family with kids the age of mine control their children’s use
of social media” and “My friends and family with kids the age of mine think it’s smart to
keep track of what kids are doing on social media” (p. 254). The first statement is
others; however, the second statement might be considered more of an injunctive norm, as
showed that parents who endorsed these norms towards controlling social media engaged
3It should be noted that while there is a distinction between the normative beliefs themselves, the
source of information (e.g. media, friends, family) may overlap for injunctive and descriptive norms.
33
Likewise, in the Bleakley et al (2013) study, parents were asked about their
descriptive normative beliefs for limiting TV (i.e. “Most people like you have not or have
limited their child’s television viewing”) and rated their ideas that other people—“close
friends, other parents you know, and your family members”—have limited their child’s TV
viewing (p. 527). Once again, the findings showed that stronger descriptive normative
perceptions across different social referents positively predicted intentions to limit their
child’s television.
Research outside the field of parental mediation has also found significant
relationships between descriptive norms and other parenting behaviors such as the use of
car safety seats (Jeffrey, Whelan, Pirouz, & Snowdon, 2016), intentions to vaccinate (Wang,
Lam, Wu, & Fielding, 2015), and allowing adolescent drinking (Gilligan, Thompson, Bourke,
Kypri, & Stockwell, 2014). Taking the research within and outside of the parental
mediation studies into consideration, I expect that descriptive norms will positively impact
H4: Descriptive norms will be positively associated with active, restrictive, and
monitoring behaviors.
Perceived behavioral control (PBC) is defined as the “extent to which people believe
that they are capable of performing a given behavior, that they have control over its
performance” (Fishbein & Ajzen, 2010, p. 154). The focus of this construct is perceived
control over one’s behavior and actions, not their control over the outcome, which is a
outcomes”; Ajzen, 2002a). Applied to this study, PBC is conceptualized as the belief that
refer to the thoughts an individual holds about factors that facilitate or hinder their control
information from others and the environment, and external factors that can help or impede
performance of the behavior (Ajzen, 1991). These control beliefs then inform one’s
In fact, PBC was derived from Albert Bandura’s conceptualization of self-efficacy (SE),
which he defined as “beliefs in one's capabilities to organize and execute the courses of
action required to produce given levels of attainments” (Ajzen, 1991; Bandura, 1998, p.
625). Ajzen (2002a) acknowledges that they reflect a similar underlying concept—which is
one’s perceived confidence that a behavior is within their control and capacity to
perform—but are operationally measured in different ways. The term self-efficacy is far
more common in parental mediation research than perceived behavioral control (e.g.
Nathanson, Eveland, Park, & Paul, 2002; Shin, 2018); however, since SE and, in a few cases,
PBC have been shown in previous research to impact parental mediation behaviors (e.g.
Krcmar & Cingel, 2016), they will be discussed together but measured separately in this
study.
several have looked specifically at mobile devices. One recent study of parents examined
(Hwang et al, 2017). The researchers defined self-efficacy as one’s perceived capability to
engage in parental mediation and measured it by asking parents to rate their endorsement
35
of statements such as, “I am confident that I can engage in mediation” and “I am capable of
engaging in mediation if I wanted to” (p. 364). The results showed that self-efficacy
positively predicted active and restrictive mediation of children’s smartphones. That is,
parents who felt confident in their ability to engage in parental mediation were more likely
parent self-efficacy on active, restrictive, and monitoring of mobile devices (Chang et al.,
2019). To measure self-efficacy, Chang and colleagues (2019) asked parents to rate their
confidence to implement active, restrictive, and monitoring behaviors for their child, like
limiting the time they can use the Internet and teaching them how to use devices safely.
Parent self-efficacy positively predicted engaging in all three parental mediation behaviors.
In a study with parents in the U.S., Shin (2018) assessed active and restrictive
control to engage in conversations about safety and privacy on smartphones and their
smartphones. This type of self-efficacy positively predicted both active and restrictive
mediation.
According to the TPB, perceived behavioral control can directly impact one’s
behaviors, whereas attitudes and subjective norms indirectly influence behavior through
intentions (see Figure 1; Ajzen, 1991). PBC can, of course, also indirectly influence
behaviors through intentions; however, Ajzen (1991) explains that control directly predicts
36
behavior when a “person can decide at will to perform or not perform the behavior” (p.
182). If there are no substantial impediments, parental mediation can be such a behavior
because parental mediation is inherently a parent-enacted activity and has been shown in
previous research to positively predict parental mediation (e.g. Chang et al., 2019). Thus,
consistent with the theoretical predictions of the TPB and previous research, I predict there
will be a positive and direct relationship between PBC/SE and parental mediation
behaviors:
What are the control beliefs underlying one’s PBC and SE that can influence parental
mediation? The TPB explains that control beliefs are shaped by one’s previous experiences
and information from others and one’s environment (Ajzen, 1991). Parent media use,
especially with mobile devices, could be a factor impacting their confidence and perceived
control to mediate. Parents who feel comfortable using mobile devices themselves could be
more likely to media their children’s mobile devices. There is some support for this idea
from a handful of studies that have assessed parental mediation of mobile devices and the
Internet.
Shin and Li (2017) asked parents about two aspects of their digital technology use—
which referred to smartphones, tablets, and computers—and their active, restrictive, and
monitoring behaviors. The first focused on the amount of time parents spent using digital
technologies per day and was measured with hour-based ordinal categories (e.g. 1 hour per
day, 2-3 hours per day). The second asked if parents use digital technology for specific
activities, such as “household tasks, social media, and personal communication” (p. 10). The
researchers found that digital technology activities significantly and positively predicted all
37
three parental mediation behaviors but did not observe significant relationships for
cross-sectional survey, Lee (2013) asked mothers of adolescents about their self-reported
Internet skills and practices for restrictive mediation. Specifically, mothers were asked
whether they had rules about sharing private information online, online gaming,
skills were measured by rating one’s ability to do several activities such as uploading
content, using a blog, searching for specific information, and paying online games. The
findings showed that mothers who reported greater Internet skills were more likely to
restrictively mediate their child’s Internet use. Though the study did not measure parent
Internet use, it was argued that skills are likely acquired through frequent use (Lee, 2013).
monitoring mediation of mobile devices, Chang and colleagues (2019) asked parents about
their mobile device skills. These skills were measured by asking parents about their ability
network, take pictures and/or video and upload them to social media, and blocking app
Mascheroni, and Ólafsson, 2014). Parent weekly use of TV, smartphones, tablets, and
computers was measured but not analyzed as predictors of mediation. Nevertheless, the
researchers found that parent mobile device skills positively predicted active mediation
and monitoring.
38
Although there are few studies to date that have investigated the impact of parent
mobile device use on their mobile device mediation practices, research on other media
platforms can provide context. For instance, parents who play video games often are more
likely to co-play and discuss video games with their children (Martins et al., 2017; Nikken &
Jansz, 2006), but those who play less are more likely to restrictively mediate (Martins et al.,
2017). Some research has also shown that parents who use the Internet daily are more
likely to restrictively mediate and monitor (Ghosh et al., 2018; Livingstone & Helsper,
2008; Sonck et al., 2013). These findings collectively suggest that parents use of the
platforms they intend to mediate does influence their mediation behaviors. Accordingly, I
hypothesize that parent mobile device use will have a positive relationship on their
mediation practices:
H6: Parent mobile device use will be positively associated with active,
restrictive, and monitoring behaviors.
Control beliefs could also be influenced by parent awareness of their child’s media
activities. That is, the extent to which parents know what media their child uses frequently
can impact their ability to communicate, monitor, or restrict their activities. If one does not
know what their child watches on their mobile devices, what social media profiles they
maintain, and/or how much time they spend on their devices then they may feel less
confident in their ability to discuss these activities or concerns with their child.
Findings from qualitative research suggest that the inverse of this idea occurs, such
that lower parental knowledge of what teens do while using and interacting on mobile
devices hinders their ability to mediate (Clark, 2009; Vaterlaus et al., 2014). If parents
cannot see what their child is doing they may, as a result, feel at a loss to talk to their child
about media (Clark, 2009). Another barrier is that parents feel less knowledgeable and
39
skilled than their children about technology and therefore less able to gain knowledge
about their media activities, which can obstruct engagement in mediation (Vaterlaus et al,
2014; Yardi & Bruckman, 2011). It is also the case that parents who report lower
awareness of their child’s social media and Internet activities have more concerns about
negative media effects (Rosen, Cheever, and Carrier, 2008). And, as discussed in a previous
section of this literature review, concerns about negative media effects and use are
positively associated with parental mediation (Hwang et al, 2017; Sonck et al, 2013).
Few quantitative studies have investigated the role of parent knowledge on mobile
device mediation behaviors but there is evidence to suggest a relationship. Using the theory
of planned behavior, Krcmar and Cingel (2016) asked parents about their perceived
behavioral control to allow their child to use social media and to control or mediate it.
Though the researchers did not explicitly assess parent’s knowledge of their child’s social
media use, some of their item-level measures asked about parent familiarity with what
their child does on social media and their ability to find out. Example statements they asked
included, “I can easily find out what my child is doing on social media” and “It’s hard to
know how much time my child is spending on social media (reverse coded)” (p. 254). The
results showed that greater endorsement of these statements positively predicted active
mediation and co-use of social media. Thus, the more parents felt they could familiarize
themselves with their child’s social media activities, the more likely they were to have
active conversations.
General parent involvement—i.e. being familiar with and involved in your child’s
life and activities—has been shown to positively predict active and restrictive mediation of
digital technology (Shin & Li, 2017). The same has been found for parent involvement and
40
greater co-playing, restrictive, and active mediation of video games (Martins et al., 2017).
Also, parental engagement with children in their general leisure activities has been shown
to positively predict active and restrictive mediation of television among younger children,
and engaging with children during their chores and duties is positively related to active,
From these findings overall, the literature suggests that the more parents interact
with their children, the greater their likelihood of engaging parental mediation. I believe
the mechanism that encourages this behavior is parent knowledge of teen media use and
activities. That is, parental knowledge of teen media activities can open pathways of
communication between parent and child (active communication) and/or familiarity with
the activities that could then be monitored or limited (restrictive and monitoring). Based
relate to motivations for mediation is a central focus of this study; however, it is also
important to recognize that parental mediation does not happen in a vacuum and it is likely
that factors other than parental attitudes, norms, and control beliefs influence their
these factors because it positions media use as occurring within the broader environment
of family processes.
Family system theory (FST; Cox & Paley, 1997) explains that a family is a dynamic
unit of interlinking parts that play an essential role in shaping each member’s behavior.
41
There are three properties of systems that inform the family system perspective. First, the
family unit is considered a “complex, integrated whole” in which individual members are
interdependently organized and reciprocally influential to the whole system (Cox & Paley,
1997). A common quote to describe this property is, “the whole is greater than the sum of
its parts.” Each family member has important and unique characteristics which have the
potential to impact the whole family system. To fully understand human behavior, FST
contends that one must examine the ways in which these individual characteristics interact
together. Families are not simply the combined characteristics of the individuals within it;
rather it is these characteristics and their interaction together that are key to
understanding behavior. For example, media behaviors in families of children with large
age gaps (e.g. a 16 year old and a 6 year old) are different than those with narrower age
gaps or single children, as these families adjust their activities to balance the
developmental needs of the different children and the family as a whole (Davies & Gentile,
2012).
intangible boundaries that symbolically guide family interactions. Parents are typically at
the top of the family, establishing family rules, routines, and expectations for children’s
behaviors (Steinberg, 1999). For example, family media rules and expectations could be no
phones at the dinner table and no television until homework has been completed (Hiniker,
Schoenebeck, Kientz, 2016). Despite a hierarchical nature, individuals at any level have
bidirectional influence on others. For instance, conflict within a sibling subsystem can
create discord within the parent dyad over how best to negotiate an issue (or conflict
42
within the marital dyad if there is disagreement over conflict resolution) (Cox & Paley,
2003).
response to changes and challenges within their environment. Families can experience
normative transitions (e.g. new school, new job) or unexpected events (e.g. trauma, death),
and during these transitions and/or events, families may unconsciously or purposefully
renegotiate family rules, patterns of behavior, and boundaries to adapt to these new
changes and to re-establish equilibrium within the family. Take the example of a child
moving from middle school to high school. This graduation is a significant social transition
for an adolescent and is accompanied by less parent and teacher oversight, more
adjustments that the adolescent him-or- herself must grow into, but parents must also
modify their behaviors to allow their child more freedom and independence.
Applied to the family media environment, FST recognizes that family relationships,
communication, and interactions can shape parent behaviors. This is in contrast to a more
(Jordan, 2002). By looking beyond the media platform, this perspective expands the
possible field of predictors that explain why or why not parents choose to engage in
mediation.
Family system theory moderating variables. The TPB predicts that parents with
positive attitudes toward mediation behaviors and their outcomes (attitudes), believe that
others endorse and perform these behaviors themselves (subjective norms), and believe
43
mediation. When the opportunity arises, these parents are likely to engage in parental
mediation (Ajzen, 1991). FST, however, contends that elements other than attitudes,
norms, and self-efficacy influence parent behaviors. In order to fully understand a parent’s
behavior, FST emphasizes taking into account broader aspects of the family system and
parenting because these provide an important context for parent behaviors (Jordan, 2002).
sections, this study also assessed how aspects of the family system interacted with the TPB
constructs to affect parental mediation behaviors. Two aspects of the family system were
mediation: parenting behaviors and family communication (Nathanson, 2015). The next
sections describe these concepts and their potential moderating roles between the TPB
respond to and socialize their child and are characterized along two dimensions of
by being attuned, supportive, and acquiescent to children’s special needs and demands”;
Demandingness refers to “the claims parents make on children to become integrated into
the family whole, by their maturity demands, supervision, disciplinary efforts, and
willingness to confront the child who disobeys” (Baumrind, 1991, pp. 61-62). Over the
years, others have used different terms to refer to these constructs such as warmth,
support, firmness, and control (Baumrind, 2005; Coyne et al., 2011; Wisenblit, Priluck, &
Pirog, 2013). For consistency between this literature review discussion of the concepts and
44
the measures I used to asses them, I will use the terms connection and regulation to refer to
responsiveness and demandingness from this point forward (Robinson, Mandleco, Olsen, &
Hart, 2001).
Parent ratings for connection and regulation determine their placement into one of
four parenting style typologies: neglecting parents are low on both dimensions; permissive
parents are characterized by high connection but low regulation; authoritarian parents are
low on connection and high on regulation; lastly, authoritative parents are high on both
boundaries and expectations with a nurturing orientation and is typically regarded as the
“best” parenting style, associated with several positive youth outcomes (Baumrind, 1991;
for review see Spera, 2005). This study focuses on the core parenting behavior constructs
indicates that parents who exhibit these characteristics are typically proactive media
behaviors on parental mediation among parents of 11-to-15 year olds. In a two year study,
they asked parents to respond to measures that assessed parenting connection and
regulation as well as their general (i.e. non-platform specific) active, restrictive, and co-
viewing mediation behaviors. For both mothers and fathers, connection and regulation
positively predicted active mediation one year later, while regulation positively predicted
connection and regulation again predicted active and restrictive mediation, while
45
connection predicted restrictive mediation (Padilla-Walker, Coyne, Fraser, Dyer, &
Yorgason, 2012b).
Warren and Aloia (2019) assessed parenting styles and mediation of cell phones
among parents with a child between 12 and 17 years old. Their study did not measure the
mediation of cellphones.
Among parents with older adolescents (ages 13-to-17), Sanders et al (2016) asked
platforms (TV, video games, tablets, and computers). Their results showed that connection
and regulation both positively predicted technology related parenting strategies. Eastin
and colleagues (2006) also found that authoritative parenting among parents with a child
between 13 and 18 was associated with active (called “evaluative” in their study) and
and emotional development of the child,” is a parenting behavior construct that has
developed over time to account for psychological discipline that parents use to impart
within the child (Barber, 1996, p. 3296; Baumrind, 1991). Prior research demonstrates that
46
considered a time in which young people develop and seek out autonomy (Spear & Kulbok,
2004). Consequently, parents are likely to adapt their mediation practices to accommodate
the development of autonomy in their child while also continuing to monitor as they see fit
Several studies have looked at the role of autonomy granting in parental mediation.
A recent investigation assessed parenting behaviors and parent use of technical monitoring
of mobile devices for their child between the age of 13 and 17 (Ghosh, et al., 2018).
Technical monitoring was loosely defined as the use of parent control applications that
allowed parents visibility into their child’s text messages and what apps their child
downloaded and used on their smartphone. These researchers did not observe significant
relationships between connection and regulation; however, they did find that autonomy
granting negatively predicted the use of technical monitoring. Similarly, the longitudinal
studies conducted by Padilla-Walker and colleagues (2011, 2012b) found that mothers
who practice autonomy granting demonstrate more co-viewing and less restrictive
mediation over time. Research has also shown that adolescent-reported autonomous
important component of the family system that directly impact parental mediation. It also
provides evidence for the idea that parenting behaviors may play a moderating role
between the TPB constructs and parental mediation. That is, it could be the case that
parents have attitudes, social norms, and PCB that predict mediation, but their connection,
regulation, and autonomy granting practices interact with these concepts to influence their
47
actual behavior. For example, as I discussed previously, parents who have negative
attitudes toward media effects and use are more likely to practice active and restrictive
mediation (Warren, 2017); however, if they also value autonomy granting, parents could
practice less active and restrictive mediation. Or, parents who report strong injunctive
norms (i.e. the belief that others think parents should mediate) towards mediation but
exhibit low connection and regulation may engage in less mediation. Additionally, parents
with low smartphone self-efficacy are less likely to engage in parental mediation, but low
smartphone self-efficacy combined with high parenting efficacy predicts active mediation
of smartphones (Shin, 2018). Given that little research has been conducted on the
is, parents must communicate with their children to convey rules, discuss content, and
know what media their child is using. Open and comfortable general family communication
is correlated with active, restrictive, and co-viewing mediation of television and Internet
use (An & Lee, 2010; Chng, Li, Liau, & Khoo, 2015). Frequent parent-child communication is
positively associated with parent active and restrictive mediation of the Internet, and
active, restrictive, and co-use of cellphones (Warren, 2017). Conversely, when adolescents
report that it is difficult to talk to their parents about risky online experiences, their
parents are more likely to underestimate the occurrence of risks and disagree with their
48
children over risk-prevention practices (Byrne, Katz, Lee, Linz, & McIlrath, 2014; Byrne &
Lee, 2011). Put simply, a comfortable communication environment is fertile ground for
mediation is hindered.
mediation, it stands to reason that it may play a moderating role in the TPB–parental
mediation behavior relationship. It could be that parents have attitudes, norms, and PCB
that predict mediation, but family communication quality interacts with these concepts to
influence their actual behavior. For instance, parent negative attitudes toward media use
and effects is predictive of active parental mediation, but low communication quality
between parents and children can act as a barrier to negatively predict active mediation.
Alternatively, low SE combined with open family communication could positively predict
active mediation and monitoring. The potential moderating relationships between family
communication and the TPB constructs will be examined with the following research
question:
RQ3: Does family communication quality moderate the relationship between the
TPB constructs and active, restrictive, and monitoring behaviors?
Perceived parent barriers. Other aspects of the parent-child relationship could also
explain why parents choose not to engage in mediation during the teen years. Child age is a
consistent negative predictor of parental mediation. Both younger adolescents and their
parents report that communication about media occurs more frequently before the teen
years, rather than during them (Rideout, 2015; Shin & Kim, 2019). Parents provide more
active and restrictive mediation during early adolescent years and then taper these
evident explanation is that children become “too old” for such monitoring. Research to
support this idea is sparse, but a recent qualitative study with parents of emerging adults
(average age of target child was 19.9 years) hints that child maturity plays a role (Vaterlaus
et al., 2015). Parents and their teen-aged children were asked to report their mediation
habits for cellphones and the Internet, how they mediated these technologies, and, if they
reported no monitoring, why they choose not to. The authors stated, “Common among all
participants, an emerging adult’s age was presented as justification for not mediating
Two additional reasons were mentioned as reasons for not mediating. The first was
child trust. For teens, this meant parent’s felt they had demonstrated appropriate behavior
on these technologies when they were younger which gained their parent’s trust. Among
parents, they reasoned that their actions during the tween years laid the foundation for
trust. A quote from one mother in the study said, “Matter of trust at this point. If we have
not set principles for him by this point, then we have already gone wrong” (p. 355). The
second reason was respect for child privacy. As children develop into young adults, it is a
natural and common transition to distance themselves from their parents as they develop
their own identity. Parents who perceived their children to be more adult-like, or growing
into adults, were less likely to monitor as this felt like an invasion of privacy. Child trust
and parent respect for teen privacy have also been reported in other qualitative and
50
To summarize these points so far, (1) parental mediation prior to the teen years, (2)
perceptions of child maturity, (3) child trust, and (4) respect for child’s privacy are all
potential reasons for why parents decrease mediation during the teen years. Collectively
these reasons reflect a parent’s growing recognition of child autonomy during the teen
years and, like the other family system variables, have the potential to interact with the
TPB constructs to impact parental mediation. For example, parents with positive attitudes,
social norms, and PCB to mediate may choose not to because they feel they have already
parental mediation, parents also report functional barriers to mediation. For example, the
amount of time conversations can take and the energy it takes to enforce rules and
restrictions (Beyens, Eggermont, & Nathanson, 2016). At times, parents report a lack of
knowledge of how these conversations should go or tools they can use to help them
monitor their child’s activities on mobile devices (Vaterlaus et al., 2015). Their child may be
resistant to rules and conversations (Jordan et al., 2006) or simply not interested in using
mobile devices or are light users. These reasons also have the potential to interact with the
TPB variables to affect parental mediation. Given the nascence of this research, the
moderating role of these collective barriers will be explored with the following research
question:
Control Variables
It is important control for variables in statistical analyses that are known to co-vary
or impact the relationships of interest in this study (Field, 2013). The inclusion of control
51
variables enables a researcher the ability to parse out the bias they exert on the
independent and dependent variables in a model. As a result, the relationships between the
independent and moderating variables of interest on the dependent variable can be more
accurately assessed. Two family system-related variables will be controlled for in this
study.
The first control variable is parent accessibility, which is the amount of time that
parents are available to spend time with their children. In other words, it is the total
waking hours that parents are available to their children. Though parents do not
necessarily need to be physically present to have rules about media or covertly monitor,
active mediation and the enforcement of rules are inherently shared activities between a
child and parent and previous research suggests that accessibility impacts parent behavior.
In two studies, Warren (2001, 2002) found that parent accessibility significantly influenced
parental mediation, such that high-access parents reported more co-viewing (2001) and
active and restrictive mediation of television viewing (2002) than low-access parents.
Applied to this study, it could be that parents who are not as accessible to their children
have less knowledge of their media habits, lower efficacy to engage in conversations about
media, have less time to conduct these conversations and so on; thus, parent accessibility
could potentially impact the independent and dependent variables, therefore, it will be
Family structure is another variable to control for. Family structure refers to the
relationship between parents and children (e.g. biological, step) and their marital
status/history (e.g. (re)married, separated, divorced; Cox & Paley, 1997). Regarding marital
status, some research has shown that single parents use less restrictive mediation (Gentile,
52
Nathanson, Rasmussen, Reimer, & Walsh, 2012) but more positive active mediation
compared to married parents (Austin, Knaus, & Meneguelli, 1997). Less empirical research
has examined the impact of parent-child relationship on parental mediation (e.g. biological,
step); however, according to FST, a non-biological relationship between parent and child
may impact family interactions such that a non-biological parent would feel less confident
parent. These aspects of family structure will be examined and controlled for if they
Several demographic variables were also used as controls, including parent and
child age, gender, ethnicity, parent education, household income, and parent employment.
These variables have been shown in previous research to impact the dependent and
independent variables of interest in this study. For instance, some studies have found that
younger parents use less restrictive and active mediation (Krcmar, & Cingel, 2016; Shin &
Kim, 2019; Lauricella et al., 2016); mothers tend to practice more parental mediation than
fathers (Chang et al., 2019; Lauricella et al., 2016; Nikken & Opree, 2018); and non-White
parents practice more active and restrictive mediation (Lauricella et al., 2016; Top, 2016).
Previous research has consistently demonstrated that child age and parental mediation
have a negative relationship (Chen & Chng, 2016; Rideout, 2015; Khurana et al., 2015). The
effect of parent education, income, and employment is inconsistent. Some research has
found that parents who reach higher levels of education practice more active and
restrictive mediation (Warren & Aloia, 2019), while others have found that parents who
complete fewer years of education use more monitoring and active mediation (Beyens et
al., 2019; Chang et al., 2019). Some studies find that lower to mid-level household income is
53
associated with more active mediation (Lauricella et al., 2016) and higher income is
associated with less active and restrictive mediation (Top, 2016), while others have found
that higher income is positively associated with restrictive mediation and using technical
monitoring (Nikken & Opree, 2018). Parent employment can impact restrictive mediation
but have no effect on other methods (Warren, 2017). In some cases, parent education,
employment and household income have no impact on parental mediation (Bleakley et al.,
2013; Shin & Kim, 2019). Given the mixed findings for these demographic characteristics,
their effect on the primary independent and dependent variables were examined and
controlled for when significant associations impacted the main relationships of the study.
54
CHAPTER 3: Method
Study Design
survey of parents with a child between the age of 13 and 18 who owns a personal mobile
device. Data was collected through a self-administered online survey hosted on Qualtrics.
The survey instrument included measures for the TPB constructs, FST variables, parental
mediation and monitoring behaviors, and demographic variables. To review, the dependent
variables in this study were active, restrictive and monitoring parental mediation of
Internet-enabled mobile devices. The independent variables were the TPB constructs—
efficacy. The moderators in this study included parenting behaviors, family communication
quality, and parent barriers to mediation. Control variables included parent accessibility,
family structure, and parent and child demographics (age, gender, household income,
parent education, ethnicity, and employment status). Figure 2 illustrates these variables
55
56
Data Collection
Participants were recruited through Mechanical Turk between March and April of
2018. There were two orderings of the survey such that half of the sample viewed the
parental mediation questions first and half viewed the TPB questions first. No effects of
order were observed. Participants received $1.00 for completing the survey. The full
survey (including screening and attention check questions) is in Appendix B. The survey
was pilot-tested with a convenience sample of 61 parents before collecting the final sample.
This was done to test the survey flow, assess the clarity of questions, and solicit feedback
regarding parenting and mobile devices. No major issues were reported or detected but
small changes to the survey were made to improve question clarity. The results of the pilot-
online workforce/labor market that has become popular in social scientific research as a
source of diverse and easily accessible population samples (Buhrmester, Kwang, & Gosling,
2011), including several parental mediation studies (e.g. Martins et al., 2017; Rasmussen,
White, King, Holiday, & Densley, 2016). Some critical evaluations of data gathered from
individuals on MTurk have noted that the reliability of measures on MTurk can be lower
than those from community samples (Rouse, 2015). Attentiveness during data collection is
a commonly mentioned concern, and the threat of artificial intelligence systems (i.e. “bots”)
has also been an issue noted in the literature (Casler, Bickel, & Hackett, 2014; Goodman,
Cryder, & Cheema, 2013; Litman, 2018). Several steps were taken to heed these cautions
57
First, participants completed six pre-survey screening questions to determine their
eligibility to participate in this study (Rouse, 2015). These questions asked the parents
gender, if they had a child within the target age range, if this child owned a personal mobile
device, if the parent spoke to the child regularly (at least four times per week), and if the
parent owned a personal device. Second, this study used three attention/accuracy check
questions spaced throughout the survey to subvert artificial intelligence systems and
individuals who had more than 50 approved HITs and were located in the United States.
Data Screening
Six hundred and thirty five responses were downloaded from Qualtrics. As noted
above, there were several attention check questions evenly spaced throughout the survey.
Six individuals incorrectly answered all 3 of these attention questions and were not
included in the sample. Fifteen responses were missing 17% or more data for the TPB
behavior measures, parental mediation measures, or FST questions and were excluded
from the sample. Six responses were removed for completing the survey twice (as this
violates the assumption that the data are independent). Seven responses reported within
the survey that their child did not own a mobile device and were not retained for the final
sample. In all, a total of 34 responses were excluded from the final sample during this first
The average time to complete the survey was 21.47 minutes (SD = 40.43). The
skewness (11.63) and kurtosis (157.61) of this distribution indicated that the data were
positively skewed and leptokurtic (highly peaked; Tabachnick & Fidell, 2007). The time it
took to complete the survey ranged from 3.37 minutes to 651.25 minutes with a median of
58
15.10 minutes. The upper end of this range can be explained by Mechanical Turk’s
assignment acceptance system. MTurk workers can add an assignment to their “assignment
queue” to work on until the assignment expires (Amazon Developer Guide, 2017; Amazon
Mechanical Turk, 2018). I set the assignment to expire after 4 days, so in theory, Mturk
workers had 96 hours to complete this survey after adding it to their queue before it would
close automatically.
To investigate the lower end of the range, I analyzed the means, standard deviations,
and internal reliability coefficients for all variables while excluding individuals who took
less than 10 minutes to complete the survey (n=490); less than 8 minutes to complete the
survey (n = 536); and less than 5 minutes to complete the survey (n=592). Table 1
compares the full sample (n = 601) to each of these sub-samples. There are slight
differences among the reliability coefficients but none exceed 0.02. The survey was
approximately 127 questions 4 and while the data do not substantially differ in Table 1, it is
unlikely that individuals who took less than 5 minutes to complete the survey filled it out to
fidelity and with care; therefore, the 9 responses that took less than 5 minutes to finish the
survey were excluded as well. After removing all of these responses, the final sample size
retained was 592 (N). The average time to complete the survey was 22.03 minutes (SD =
40.58).
4Including the filter questions at the beginning of the survey, all of the potential follow up
questions, and open-ended responses, the survey was a total of 136 questions.
59
Table 1. Reliability Coefficients and Descriptive Statistics for DVs and IVs for Sub-samples of
Responses that Took >10, >8, & >5 Minutes to Complete Survey
All >10 Minutes >8 Minutes >5 Minutes
n = 601 n = 490 n = 536 n = 592
Mean (SD) 21.47 (40.43) 25.01 (44.28) 23.38 (42.45) 22.03 (40.58)
Median 15.10 17.02 16.09 15.14
Range 3.37 – 651.25 10.00 – 651.25 8.01 – 651.25 5.08 – 651.25
M SD α M SD α M SD α M SD α
Parental
Mediation
AM 3.00 0.74 .89 2.98 0.75 .90 3.00 0.75 .90 3.00 0.75 .90
RM_SUM 3.16 1.55 .72 a 3.16 1.51 .70 a 3.19 1.52 .70 a 3.19 1.53 .71 a
RM_AVG 2.84 0.72 .82 2.86 0.72 .82 2.86 0.73 .82 2.84 0.72 .82
MON_SUM 2.30 1.38 .66 a 2.38 1.36 .64 a 2.35 1.37 .65 a 2.33 1.37 .65 a
MON_AVG 3.07 1.00 .81 3.06 1.01 .81 3.09 1.02 .82 3.08 1.01 .82
Attitudes
Toward
Behaviors
ATT_NA 3.13 0.99 .90 3.16 0.99 .90 3.16 1.00 .90 3.14 1.00 .90
ATT_PA 3.59 0.76 .89 3.62 0.75 .89 3.62 0.75 .89 3.59 0.76 .89
ATT_AMO 3.33 0.81 .74 3.54 0.82 .75 3.35 0.82 .75 3.33 0.81 .75
Subjective
Norms
SN_IN 4.37 1.34 .87 4.45 1.36 .88 4.44 1.35 .87 4.38 1.34 .87
SN_DN 4.08 1.43 .89 4.10 1.46 .90 4.10 1.46 .90 4.08 1.44 .90
Perceived Behavioral
Control/Self-Efficacy
PBC 3.75 0.93 .86 3.81 0.92 .86 3.80 0.93 .87 3.76 0.93 .86
SE 3.83 0.95 .87 3.88 0.94 .87 3.87 0.94 .87 3.84 0.95 .87
PBC & SE 3.79 0.90 .92 3.84 0.88 .92 3.83 0.89 .92 3.80 0.90 .92
PMU 3.89 0.70 .69 3.93 0.69 .67 3.93 0.69 .68 3.90 0.69 .68
KMA 3.69 0.82 .88 3.72 0.81 .87 3.72 0.81 .88 3.70 0.82 .88
60
Family System
Theory
PB_C 4.31 0.77 .91 4.41 0.70 .91 4.40 0.70 .91 4.33 0.75 .92
PB_R 4.15 0.78 .91 4.23 0.75 .91 4.22 0.75 .91 4.17 0.77 .91
PB_AG 3.95 0.74 .86 4.00 0.72 .86 4.00 0.73 .87 3.97 0.73 .86
FCQ 3.81 0.73 .82 3.87 0.70 .82 3.87 0.70 .82 3.83 0.72 .82
BAR 2.27 0.81 .84 2.20 0.79 .83 2.23 0.81 .84 2.27 0.81 .84
Note. Actual minutes, not decimal minutes, are reported in this table; a Kuder-Richardson
20 coefficient for binary variables.
Measures
The survey was comprised of the measures described in the next sections. The
dependent variables are discussed first followed by the independent, moderating, and
control variables. Cronbach’s alpha (α) is reported for scaled (Likert) measures and Kuder-
same manner but used specifically to refer to reliability for binary items (Cortina, 1993;
Cronbach, 1951). The commonly used cutoff of α > .70 was employed in this study (Field,
2013). Note that some of the measures described next were created or adapted for this
study; however, when possible, comparable alphas from previous studies are noted. The
reliability coefficients and descriptive statistics for IVs, DVs, and moderators are in Table 2,
and descriptive statistics for demographic and control variables are in Table 5.
Parents were asked about their active and restrictive mediation and monitoring
behaviors. These questions were modeled after traditional parental mediation measures
(Gentile et al., 2012; Nathanson, 1999, Nathanson & Botta, 2003; Valkenburg et al., 1999)
61
and measures adapted for parental mediation of mobile devices and the Internet (Hiniker
et al., 2016; PEW, 2016; Rideout, 2015; Shin & Li, 2017). Collectively these questions
concerned parental mediation of what, when, and where adolescents can access on their
mobile devices. Table 2 contains the means and standard deviations for these measures
62
Table 2. Reliability Coefficients and Descriptive Statistics for IVs, DVs and Moderators
Variable M (SD) α
Parental Mediation
Active mediation 3.00 (0.75) .90
Summed restrictive mediation 3.19 (1.53) .71 a
Average restrictive mediation 2.84 (0.72) .82
Summed monitoring 2.33 (1.37) .65 a
Average monitoring 3.08 (1.01) .82
Attitudes Toward Behaviors
Negative attitudes about media use and effects 3.14 (1.00) .90
Positive attitudes about media use and effects 3.59 (0.76) .89
Attitudes about mediation outcomes 3.33 (0.81) .82
Subjective Norms
Injunctive 4.38 (1.34) .87
Descriptive 4.08 (1.44) .90
Perceived Behavioral Control/Self-Efficacy
Perceived behavioral control 3.76 (0.93) .86
Self-efficacy 3.84 (0.95) .87
Parent mobile device use 3.90 (0.69) .68
Parent knowledge of teen media activities 3.70 (0.82) .88
Family System Theory
Parenting behavior: Connection 4.33 (0.75) .92
Parenting behavior: Regulation 4.17 (0.77) .91
Parenting behavior: Autonomy granting 3.97 (0.73) .86
Family communication quality 3.83 (0.72) .82
Barriers to parent mediation 2.27 (0.81) .84
Note. N=592. a Kuder-Richardson 20 coefficient for binary variables.
63
Active mediation. Seven questions for active mediation asked about parent
guidance and conversations relating to the use and content of mobile devices. Examples
include “How often do you discuss what content is appropriate or inappropriate to view on
mobile devices? How often do you discuss what personal information should or should not
Never, 2 = Sometimes but not often, 3 = Several times a month, 4 = Several times a week, 5 =
Everyday). The alpha for these questions was α= .90, which is comparable to previous
Restrictive mediation. Five questions for restrictive mediation asked about time
and content rules. Examples include “Do you have rules about when your child can use
their mobile device? Do you have rules about what apps your child can or cannot use on
their mobile device?” These questions had binary response options (yes/no) and the
mediation questions, they were asked a follow-up question about how strict or consistent
they are with enforcing rules on a 4-point scale (1 = Not at all, 2 = A little, 3 = Moderately, 4
= Very). For example, if a parent responded “yes” to the question, “Do you have rules about
when your child can use their mobile device?” they were asked, “How would you describe
how strict or consistent you are with enforcing rules about when your child can be on their
mobile device?” If parents responded affirmatively to all five binary response questions,
affirmatively to four out of five binary response questions, then they answered a total of
eight questions for restrictive mediation (and so on). Cronbach’s alpha for the strictness
64
questions was high (α= .82) and comparable to previous research (Warren, 2017, α = .84).
From this point forward, the binary response questions will be referred to as “summed
Monitoring. Seven questions for monitoring asked about overt and covert means of
supervising mobile devices. Four of these questions had binary response options (yes/no)
and, for some questions, if parents responded affirmatively they were asked a follow-up
question for more information. For example, if a parent responded “yes” to the question,
“Are you friends with your child on their social media profiles (e.g. Facebook, Twitter) for
the purpose of monitoring their posts or activity?” they were also asked, “On which social
media sites are you friends with your child?” (Facebook, Twitter, Instagram, Snapchat,
Pinterest, YouTube, Tumblr, Email, and Other) and “How often do you check your child’s
mobile device after they have used it to monitor what they are doing (both immediately
after or some time later)?” The remaining three questions asked about the frequency of
monitoring and were reported on a 5-point Likert scale (1 = Never, 2 = Sometimes but not
Cronbach’s alpha for the Likert response questions was high (α = .81) and
comparable to similar questions in a previous study (Sonck et al., 2013, α = .79) but KR20
for the four binary response questions was low (KR20= .65). The low KR20 for this
measure may be due to the low number of items (Cortina, 1993; Tavakol & Dennick, 2011).
Relatedly, it could also be due to the different types of tasks that were assessed in these
questions. Measures of internal consistency assess the degree to which “all items in a test
measure the same concept or construct and hence it is connected to the inter-relatedness
65
of the items within the test” (Tavakol & Dennick, 2011, p. 52). To investigate this issue, I
conducted a principal component analysis with varimax (orthogonal) rotation on the four
items. The analysis yielded two factors with two questions loading onto each factor
explaining a total of 77.68% of the variance (Table 3). Factor 1 explained 49.09% of the
variance and comprised two questions relating to passwords: “Do you know your child’s
passwords for any of his/her social media and/or online accounts? Do you know your
child’s passwords for any of his/her mobile devices?” Factor 2 explained 28.60% of the
variance and comprised two questions relating to friending on social media: “Are you
friends with your child on their social media profiles (e.g. Facebook, Twitter) for
fun/entertainment? Are you friends with your child on their social media profiles (e.g.
Facebook, Twitter) for the purpose of monitoring their posts or activity?” Given that these
items loaded onto factors in an expected manner, they were retained for further analyses.
From this point forward, the four binary response questions will be referred to as “summed
monitoring” and the three Likert response questions asking about frequency will be
66
Table 3. Principal Component Analysis of Summed Monitoring Questions
Loadings
Item Factor 1 Factor 2 Communalities
Do you know your child’s passwords for any of
.852 .768
his/her social media and/or online accounts?
Do you know your child’s passwords for any of
.891 .794
his/her mobile devices?
Are you friends with your child on their social
media profiles e.g. Facebook, Twitter for .891 .794
fun/entertainment?
Are you friends with your child on their social
media profiles e.g. Facebook, Twitter for the .834 .751
purpose of monitoring their posts or activity?
Eigenvalue 1.963 1.144
% of Variance 49.086 28.598
Total Variance 77.684
Ajzen’s guidelines on how to construct a formal TPB questionnaire include the use
of elicitation research methods, self-report and direct observation of the target population,
latent and manifest measures, and 7-point semantic differential response options (Ajzen,
2002b; Fishbein & Ajzen, 2010). Given that this dissertation is employing the TPB as a
guiding theoretical framework rather than formally testing its application, exact TPB
measures were not utilized. The goal here instead was to investigate the extent to which
the hypothesized antecedent beliefs outlined in the literature review underlie each of the
TPB constructs and ultimately discern if these beliefs predict parental mediation. To this
end, existing measures from previous studies were employed, adapted, or created for this
study.
67
TPB: Attitudes
To review, attitudes toward behaviors are the positive and negative ideas a parent
holds about active, restrictive, and monitoring mediation and teen mobile device usage.
one’s attitudes included (a) the negative and positive beliefs parents have about media
effects and use of media and (b) the attitudes parents have about the expected outcome of
mediation behaviors.
Negative attitudes about media effects and use. Negative attitudes about media
effects and use were assessed with nine items from the most recent Common Sense census
of parents with teens and tweens (Lauricella et al., 2016). These questions comprised
concerns about content (e.g. “How worried are you about your child being exposed to
images of drugs, alcohol, and tobacco use?”), emotional well-being (e.g. “How worried are
you that your child’s ability to communicate well with other people will decrease because
of the amount of time they spend online or mobile devices?”), and behaviors (e.g. “How
worried are you about your child receiving or sending sexual images or videos?”).
Responses were reported on a 5-point scale (1 = Not at all worried, 2 = Slightly worried, 3 =
these questions was high (α = .90) but lower than similar questions from a previous study
Positive attitudes about media effects and use. Positive attitudes about media
effects and use were assessed with eight items from the Common Sense census (Lauricella
et al., 2016). These questions comprised benefits such as creative expression (e.g. “How
much do you agree or disagree that using the Internet and mobile devices supports my
68
child’s creativity?”), supporting education (e.g. “How much do you agree or disagree that
using the Internet and mobile devices helps with schoolwork or education?”), and personal
development (e.g. “How much do you agree or disagree that using the Internet and mobile
devices supports my child’s social skills?”). Responses were reported on a 5-point scale (1 =
agree). (α = .89). Cronbach’s alpha for these items was high (α = .89) and comparable to
similar questions from a previous study (Warren, 2017, α = .87). Parents were also given
space to report any other concerns, benefits, and potential effects of media use in open-
ended responses.
Positive attitudes about mediation outcomes. Parents were asked five questions
regarding their attitudes about the outcomes of active, restrictive, and monitoring
behaviors (also adapted from the Common Sense census; Lauricella et al., 2016). Example
questions include, “How effective do you think rules about mobile devices and the Internet
would be at preventing your child from engaging in behaviors you do not want them to?”;
“How effective do you think conversations about content and apps would be at preventing
undesired effects or behaviors in your child?”; and “How likely do you think rules about
mobile devices and the Internet would cause conflict or disagreements with your child?
(reverse coded).” Responses were reported on a 5-point scale (1 = Not at all effective, 2 =
higher values indicated more positive beliefs that mediation outcomes would be effective
(α = .75).
69
TPB: Subjective Norms
Injunctive norms. Five items asked about injunctive norms of parental mediation
behaviors. These questions were created for this dissertation. To review, injunctive norms
(Fishbein & Ajzen, 2010). This set of questions was introduced with the statement, “In
general, people think that parents of teenagers should…” followed by referent mediation
behaviors (e.g. “Have rules about when their child can use their mobile device… Have
conversations with their children about media content and apps they view on their mobile
devices”). Responses were reported on an 8-point scale (0 = The majority of people DO NOT
think this way – 7 = The majority of people DO think this way) (α= .87).
mediation behaviors. These questions were also created for this dissertation. To review,
descriptive norms are the perceived or actual behaviors of others (Fishbein & Ajzen, 2010).
This set of questions was introduced with the statement, “The next questions ask you to
think about how many parents with a kid your child’s age do the following behaviors,”
followed by the same referent mediation behaviors asked for injunctive norms (e.g. “Have
rules about when their child can use their mobile device… Have conversations with their
children about media content and apps they view on their mobile devices”). Responses
To review, PBC and SE are conceptually similar constructs in that both refer to one’s
perceived confidence that they can perform a behavior; however, they differ slightly in
70
behavior in terms of what they can/cannot do (i.e. capability) under varying degrees of
obstacles (Bandura, 2006), while measures of PBC ask individuals to think about behaviors
in terms of perceived controllability (Fishbein & Ajzen, 2010). Ajzen (2002a) has
acknowledged the overlap between the two concepts and even stated that measures of PBC
“should contain self-efficacy as well as controllability items” (Ajzen, 2002b, p. 7). To heed
this recommendation while also maintaining the distinction between PBC and SE, a two-
part question was designed as a direct measure of PBC and SE for this dissertation.
The PBC set of questions was introduced with the following instructions: “Please
rate how much you believe each statement is something that you feel you are capable of, or
is a behavior within your control to perform.” The statements referred to each mediation
behavior (e.g. “Limit when my child can use their mobile device… Discuss media content
and apps my child views on their mobile device(s)) and responses were reported on a 5-
point scale (1 = Not at all capable – 5 = Very capable). Cronbach’s alpha for these items was
good (α = .86) and comparable to similar questions in a previous study (Krcmar & Cingel,
2016, α = .81).
The SE set of questions was introduced with the following instructions: “Now
please rate your confidence that you could do these behaviors even if they are difficult.” The
statements referred to each mediation behavior (e.g. “Limit when my child can use their
mobile device… Discuss media content and apps my child views on their mobile device(s))
and responses were reported on a 5-point scale (1 = Not at all confident – 5 = Very
Parent mobile device use. Parent mobile device use was measured with a mobile
technology activities roster developed by Shin and Li (2017). Parents were asked to rate
71
how frequently they use a mobile device for six different tasks including, “job-related tasks,
household tasks, social media, educational purposes, leisure, and personal communication”
(Shin & Li, 2017, p. 10). Responses were reported on a 5-point scale (1 = Never, 2 =
Sometimes but not often, 3 = Several times per month, 4 = Several times per week, 5 = Every
day). Cronbach’s alpha for this measure was low (α= .68) and examined further with a
The analysis yielded two factors. Factor 1 explained 39.72% of the variance and
comprised three items relating to personal use (personal communication, leisure, and
social media), with factor loadings from .706 to .819. Factor 2 explained 18.30% of the
variance and comprised three items relating to work tasks (job-related tasks, household
tasks, and educational purposes), with factor loadings from .694 to .749. Given these
findings, the three items relating to personal use and three items relating to work use were
measured using eight items from the Common Sense census (Lauricella et al., 2016). These
questions asked parents to report how often they are aware of their child’s media activities
such as what they watch on TV, which online videos they view, and what they post about
themselves online. Responses were reported on a 5-point scale (0=My child doesn’t do this
activity, 1 = Never, 2 = Once in a while, 3 = Some of the time, 4 = Most of the time, 5 = All of the
Version (PSDQ-S; Robinson et al., 2001) was used to measure connection (5 items),
72
regulation (5 items), and autonomy granting (5 items). Example statements included “I
have warm and intimate moments with my child” (connection), “I emphasize the reasons
for rules” (regulation), and “I allow my child to give input into family rules” (psychological
Rarely, 3 = Sometimes, 4 = Often, 5 = Always). Cronbach’s alphas for these items were:
connection (α= .92), regulation (α= .91) and autonomy granting (α= .86), which are higher
than previous studies (Padilla-Walker et al., 2012b, connection (α= .78), regulation (α= .82)
Apprehension Scale (C-PCAS; Lucchetti, Powers, & Love, 2002) were adapted and used to
measure quality of family communication. The C-PCAS was originally developed for use
with young adults and measures the degree of anxiety and openness toward dyadic
about things that happened during the day” (adapted to, “My child appears relaxed when
we are talking about things that happened during the day”) and “When in casual
conversations with my father/mother I don’t feel I have to guard what I say” (adapted to,
“When in casual conversations with my child, they do not appear guarded”). Responses
= Agree, 5 = Strongly agree). Cronbach’s alpha for these items was good (α= .82) but lower
than the full 25-item scale (Lucchetti et al., 2002, α = .92 for mothers, α = .92 for fathers).
Parent barriers to mediation. Parents were asked nine questions about potential
barriers to media conversations developed from several sources (Rideout, 2007; Vaterlaus
et al., 2015), including the impact of previous conversations/rules (“I had these
73
conversations when my child was younger”), their child’s maturity (“My child being mature
enough that I don’t think these conversations are necessary”), lack of knowledge (“Not
knowing what to say to my child”), time constraints (“How long the conversations would
take”), trust (“Trusting my child to act and behave responsibly while using their mobile
device”), and privacy (“Respect for my child’s privacy”). This set of questions was
There are many reasons why parents may not have conversations about the
Internet or mobile devices with their teenager. “Conversations” can include
discussions about online behavior or activities and when, how long, and for what
purposes teens can use their mobile device. Listed below are some common reasons
that other parents of teenagers have said they don’t discuss the Internet or mobile
devices with their children. To what extent do each of the following statements
affect having rules or conversations with your child about using mobile devices and
the Internet? If you don’t feel any barriers to talking with your child about mobile
devices, please select “not at all”. (Appendix B)
Moderately, 5 = A great deal) (α= .84). Descriptive statistics for these items are in Table 4.
Parents were also given space to report any other barriers they perceived to mediation in
an open-ended response.
74
Table 4. Means and Standard Deviations for Barriers to Parent Mediation Items
Items M SD
How long the conversations would take 1.96 (1.17)
My child being mature enough that I don’t think
2.61 (1.32)
these conversations are necessary
Respect for my child’s privacy 2.70 (1.29)
Trusting my child to act and behave responsibly
3.04 (1.26)
while using their mobile device
I don’t think these conversations will make any
1.93 (1.17)
difference
My child being unreceptive to these conversations 2.02 (1.19)
My child not being interested in mobile devices 1.67 (1.05)
Not knowing what to say to my child 1.82 (1.14)
I had these conversations when my child was
2.68 (1.34)
younger
M (SD) of all items 2.27 (0.81)
Control Variables
available to spend with their child each week (Warren, 2001, 2017). Accessibility was
measured by asking parents to report about how many waking hours they are home or
other places with their family during a typical week day (Monday – Friday) and on a typical
weekend day (Saturday and Sunday). Responses were reported on a 6-point scale (1 = One
hour, 2 = Two hours, 3 = Three hours, 4 = Four hours, 5 = Five hours, 6 = Six hours, 7 = If more
than 6 hours, please specify). Parents were given the option to report more than 6 hours in
an open-ended response.
75
Ninety-four parents wrote in responses higher than 6 hours for weekday hours they
are home with their family 5. These responses ranged from “7” hours – “40+” hours, and 9
individuals wrote in responses that reflected being home “all day.” To account for this
range of answers, I kept responses that wrote in an estimate of 14 hours and under because
these responses can reflect parents who wake up with their children, home school, and
then remain available to their children throughout the day. For example, a household that
wakes up at 7am and goes to bed at 9pm reflects a 14 hour potential for parent
Two hundred and eighty eight parents wrote in responses higher than 6 hours for
weekend hours. Here again, I kept responses that wrote in an estimate of up to 14 hours
and under and multiplied this value by 2 (M = 13.75, SD = 7.17). The weekday and weekend
hours were then summed to create an overall score of parent accessibility (M = 39.16, SD =
14.53).
Family structure. Two questions asked about the parent composition within the
household as a measure of family structure (Austin et al., 1997; Cox & Paley, 1997). The
first question asked parents to report their relationship to the child of interest. The options
were (1) Birth/biological parent, (2) Step-parent, (3) Adoptive parent, (4) Grandparent, (5)
Foster parent, and (6) Other [open –ended response]. The second question asked, “If there
are other adults in your household, what is their relationship to your child? [Select all
options that apply]”. Response options were (1) Birth/biological parent, (2) Step-parent, (3)
Adoptive parent, (4) Grandparent, (5) Foster parent, and (6) There are no other adults living
5Ninety six parents selected “More than 6 hours” but 2 did not write in an estimate of their waking
hours, therefore 94 responses are discussed.
76
in my house. For statistical analyses, this variable was dichotomized as both biological-
Marital status. Parents were asked to report their current marital status. Response
options were (1) Single, (2) In a relationship, (3) Married, (4) Divorced, and (5) Prefer not to
answer. For statistical analyses, this variable was dichotomized as married (=1) and non-
married (=0).
Child demographics. Parents were asked to report their child’s age, grade, gender,
ethnicity, and the number and type of child-owned media devices that are in the home. For
statistical analyses, gender was dichotomized as female (=1) and male (=0), and ethnicity
Adult demographics. Parents were asked to report their age, gender, household
income, education, ethnicity, employment status, and the number and types of devices in
the home. Options for household income were (1) under $30,000, (2) $30,000-$49,999, (3)
$50,000-$69,999, (4) $70,000-99,999, and (5) $100,000 or more. Options for education were
(1) Less than 8th grade, (2) Some high school, (3) High school degree or GED, (4) Some college
(2 or 4 year), (5) College graduate (e.g. B.A., B.S.), (6) Master’s degree (e.g. M.A., M.S., M.B.A.),
(7) Professional degree (e.g. J.D., M.D.), and (8) Doctorate degree (e.g. PhD, EdD). Options for
ethnicity were (1) American Indian/Alaskan Native, (2) Asian or Pacific Islander, (3)
White/Non-Hispanic, (7) Other (open-ended response), and (8) Prefer not to answer.
Options for employment status included (1) Employed full-time, (2) Employed part-time, (3)
Self-employed, (4) Student, (5) Retired, and (6) Unemployed, (7) Disabled, (8)
Homemaker/family manager, (9) Other, and (10) Prefer not to answer. For statistical
77
analyses, gender was dichotomized as female (=1) and male (=0), and ethnicity was
Data Analysis
conducted in IBM SPSS Statistics 25. Prior to any inferential analyses, several steps were
taken to prepare the data. First, the data were checked for inconsistent responses (e.g.
discrepant answers from the pre-screening questions that were repeated within the
survey) and missing data. Second, internal reliability was checked for measures used in the
Missing Values
The Missing Value Analysis function in SPSS was used to investigate any patterns to
missing data. Little’s MCAR test is a chi-square statistic used to indicate whether data
values are in a missing completely at random (MCAR) pattern. A significant p value for
Little’s test indicates that the data are not MCAR. The other possible patterns of missing
data are missing at random (MAR) or not missing at random (NMAR) (Byrne, 2010). Values
that are MAR can be “linked to the observed values of other variables in the data,” while
NMAR values vary systematically and may be related to the dependent variable (Byrne,
2010, p. 354; Tabachnick & Fidell, 2007). The results of Little’s MCAR test were statistically
significant indicating that the data were not missing completely at random (MCAR), χ2
The greatest amount of missing data (15%; 89 missing values) was observed in the
Total Parent Accessibility variable. This is understandable given the amount of open-ended
responses that were recorded for this variable and how the Total Accessibility variable was
78
calculated (only values of 14 hours and under were retained for calculation; see Parent
Accessibility in Control Variables section for more information). I considered this pattern of
missing data to be MAR because the pattern can be predicted or determined by another
variable in the data (Tabachnick & Fidell, 2007). There are several options for missing data
techniques. Listwise deletion removes the entire case that contains any missing values and
can substantially reduce the sample size. Pairwise deletion excludes cases relating to a pair
of variables with missing data but this can cause the sample size to vary across variables.
Mean substitution replaces missing values with the variable’s mean and can reduce
maximization (EM) calculates predicted scores using a series of regressions and then
imputes the entire data set with maximum likelihood (ML) estimations, which assumes the
variable is normally distributed (Byrne, 2010; Tabachnick & Fidell, 2007). I used mean
substitution to impute missing values because this variable was not normally distributed
The second greatest amount of missing data (7.8%) was within the average
five binary response restrictive mediation questions (summed restrictive mediation), they
were asked a follow up question about how strict or consistent they are with enforcing
rules on a 4-point scale (average restrictive mediation). Thus, if parents responded “no” to
pattern can be predicted or determined by another variable in the data (Tabachnick &
Fidell, 2007). The skewness (-0.35) and kurtosis (-0.12) of this variable indicated a slight
79
negative skew and a slightly platykurtic (flat) distribution; however, neither value
suggested the data were particularly non-normal (Tabachnick & Fidell, 2007). Because
these data were approximately normally distributed and percentage of missing data was
low, I used the ML method within AMOS to estimate missing data for average restrictive
mediation.
Analysis Plan
between the TPB constructs and active, restrictive, and monitoring behaviors. Structural
equation modeling (SEM) was used to investigate these predictions. SEM offers some
advantages over other statistical methods such as hierarchical regressions or path analysis
(Kline, 2011). First, unlike path analysis, SEM distinguishes between latent and observed
variables. Observed variables are directly collected from the data (e.g. self-reported
measure of communication quality) and latent variables are composed of one or more
directly measured variables, called indicators. The proposed TPB constructs in this study
were all latent variables. For example, subjective norms were a latent variable composed
the of descriptive and injunctive norm subscales. SEM also factor analyzes the indicators of
Second, SEM calculates and accounts for measurement error for each observed
variable in a model. This is a special advantage to using SEM over multiple regression,
which assumes that IVs are measured without error (Tabachnick & Fidell, 2007). Though
care was taken to avoid issues that contribute to non-random measurement errors (e.g.
counter balancing the survey, attention check questions), it is unrealistic to eliminate all
80
error associated with self-report data. Indicators with large error terms were identified
Third, SEM analyzes multiple complex relationships across latent and observed
variables simultaneously in one model. While multiple and hierarchical regressions and
path analyses can ultimately provide these same mathematical results, SEM is a more
parsimonious and powerful multivariate method to test the relationships proposed in this
The remaining research questions (2, 3, and 4) asked if family system variables and
barriers to parental mediation moderated the relationships between the TPB constructs
calculated for each respective moderator and independent variable, and then tested as a
analyses (Kline, 2011). The first step is specification, which involves drawing the
hypothesized model using SEM graphical symbols. Latent variables are denoted with
circles or ellipses and observed variables are denoted with squares or rectangles.
Directional relationships are drawn with straight arrows and correlations or covariances
are denoted with a curved, double-headed arrow line. Two important terms to also be
aware of are exogenous and endogenous variables. These refer to independent and
The second step is to identify the model (Kline, 2011). Model identification is
achieved if “a unique solution for the values of the structural parameters can be found”
(Byrne, 2010, p. 33). The model must have at least 0 degrees of freedom (i.e. the number of
81
free parameters subtracted from the number of known parameters), latent variables must
be scaled, and a unique value must be estimable for each parameter. An under-identified
(or unidentified) model occurs when the number of parameters is greater than the data
variances and covariances. It is impossible to estimate a unique value for each parameter
with an under-identified model because there are an infinite number of solutions (Byrne,
2010). A just-identified model occurs when the number of parameters is equal to the
number of data variances and covariances. It is possible to estimate a unique value for each
parameter with a just-identified model but it has 0 degrees of freedom and cannot be
rejected. An over-identified model occurs when there are more observations (i.e. data
variances and covariances) than free parameters, resulting in positive degrees of freedom.
The third step is to select measures that assess the constructs of interest, followed by
data collection and data preparation for analysis (i.e. data screening and cleaning) (Kline,
2011).
The fourth step is model estimation (Kline, 2011). This step, model estimation, is the
substantive aspect of SEM. It involves evaluating the model fit and the parameter estimates.
A measurement model must be assessed first. The measurement model represents the
relationship between the observed variables that compose the latent constructs (Byrne,
2010). The measurement model is essentially a confirmatory factor analysis (CFA) model
that specifies how well the observed indicator variables load onto the unobserved latent
constructs. The structural models can be assessed after a satisfactory measurement model
is fit to the data. The structural model specifies the direct or indirect relationships between
82
latent or manifest variables (i.e. tests the hypotheses and answers proposed research
questions) 6.
Model fit is evaluated by different fit indices. Fit indices refer to how well the sample
data approximate a specified model and there are several that should be consulted for SEM,
as no single indicator can suggest a good model fit (Kline, 2011). The most basic of fit
indices is the model chi-square (χ2) and it signifies the discrepancy between the sample
0.05) is desirable and indicates a good fit of the model to the data. Specifically, “the higher
the probability associated with χ2, the closer the fit between the hypothesized model and
the perfect fit” (Byrne, 2010, p. 76). However, scholars caution that large sample sizes will
likely result in a significant χ2 and it is unrealistic to assume that hypothesized models will
perfectly fit the population (only approximate it) (Byrne, 2010; Kline 2010); therefore one
Five other indices were consulted to evaluate model fit (Byrne, 2010). The Goodness
of Fit Index (GFI) indicates, in lay terms, how well the model fits the data compared to no
model at all (Kline, 2010). This index ranges from 0 to 1.00 and values close to 1.00 indicate
a better fit. The general rule of thumb is that values greater than or equal to .90 are
acceptable and values greater than or equal to .95 are excellent. The Comparative Fit Index
(CFI) is an incremental fit index. It compares the fit of the hypothesized model to the
independence model, which would hypothetically be a poor fit to the data (Brosseau-Laird
& Savalei, 2014). The CFI measures the improvement in fit of the hypothesized model to the
independence model. It ranges from 0 to 1.00 and values greater than 0.95 indicate a good
6 Kline (2011) describes two more steps to SEM. The fifth step is re-specification and is necessary if
initial model fit is poor. The sixth step is to report results.
83
fit (Brosseau-Laird & Savalei, 2014; Byrne, 2010). The Tucker-Lewis Index (TLI) is another
incremental fit index that measures fit and parsimony, such that “it rewards hypothesized
models that result in greater improvement in fit relative to the baseline model and have
few model parameters” (Brosseau-Laird & Savalei, 2014, p. 461). It also ranges from 0 to
1.00 and values greater than 0.95 indicate a good fit (Byrne, 2010).
measure of model fit (Byrne, 2010). A small RMSEA value indicates a better model fit, such
that values of 0.05 indicate “good” model fit and those between 0.05 and 0.08 are
considered reasonable (Byrne, 2010; Kline, 2011). AMOS also reports a 90% confidence
interval for RMSEA, with a narrow interval that includes .05 reflecting a more suitable
model fit.
The Root Mean Square Residual (RMR) “represents the average residual value
derived from the fitting of the variance-covariance matrix of the hypothesized model to the
variance-covariance matrix of the sample data” (Byrne, 2010, p. 77). These residuals are
unstandardized and depend on the scale of the reported variable; so, if scales between
variables are different, it is recommended that the standardized RMR (SRMR) is reported.
The SRMR is the average of these residuals and is based on the variance-covariance
matrices of the sample and predicted model transformed into correlation matrices. The
SRMR, then, is the overall average difference between the predicted and observed
correlations. Like RMSEA, smaller values are desirable and values of 0.05 indicate good
model fit and those between 0.05 and 0.08 are acceptable (Byrne, 2010).
standardized parameter estimates were also evaluated. The standard errors of parameter
84
estimates should not be too large or too small, though there is no designated cut-off
85
CHAPTER 4: Results
The first section of results details the descriptive characteristics of the sample.
Results for correlation analyses between the demographic variables and independent and
dependent variables are discussed in Appendix D and also viewable in Tables 6 and 7. The
measurement model evaluation process is described next, followed by the structural model
evaluation. The last section presents findings for my hypotheses and research questions.
Descriptive statistics for all demographic variables are in Table 5. The majority of
parents in the sample were female (74.2%) and the average age was 38.3 years old (SD =
(1.7%), Other (0.7%), and 1.4% preferred not to answer. This distribution is similar to
recent national population estimates from the U.S. Census Bureau (76.6% White, 13.4%
Black/African-American, 5.8% Asian, 0.2% Pacific Islander, 18.1% Hispanic or Latino, 1.3%
American Indian/Alaskan Native; U.S. Census Bureau, 2010-2018). The majority of parents
unemployed (0.8%), disabled (0.8%), retired (0.3%), other type of employment (0.2%),
and a small percentage preferred not to answer this question (0.3%). A plurality of the
sample had a college degree (41.7%), completed some college education (2 or 4 year;
86
31.4%), had a Master’s degree (12.3%), a high school degree or GED (9.5%), a professional
degree (2.9%), doctorate degree (1.2%), and 0.3% reported completing some of their high
school education. The educational background of this sample is higher than recent national
population estimates (30.9% of people aged 25+ holding a bachelor’s degree or higher; U.S.
Eleven percent of the sample reported a household income of less than $30,000;
25% reported income between $30,000 and $49,999; 20.8% reported income between
$50,000 and $69,999; 25.3% reported income between $70,000 and $99,999; and 17.6%
reported income of $100,000 or more. The mode income distributions from this sample are
comparable to recent national population estimates (median household income (in 2017
The majority of parents were the birth/biological parent of the child they were
grandparent (1.3%), foster parent (0.5%), and other/open-ended (0.5%). The majority of
people reported one other adult in their household (84.3%), followed by no other adults
(11.9%), and two adults (3.5%) 7 (M number of other adults in home = 0.91, SD = 0.39).
Other adults in the household were mainly birth/biological parents (71.1%), followed by
step-parents (12.1%), grandparents (6.3%), adoptive parents (1.7%), foster parents (0.2%)
and 11.5% reported there were no other adults in their home. The average weekday and
weekend parent accessibility was 26.45 hours (SD = 10.76) and 13.75 hours (SD = 7.17)
and total average weekly parent accessibility was 39.16 hours (SD = 14.53).
7Three individuals selected each “Other adult” option (e.g. birth, step, biological, foster etc.) and
were excluded from the descriptive analyses for this variable.
87
The majority of children reported by parents were male (55.4%) and the average
child age was 14.5 years old (SD = 1.46). Approximately 29.7% were 13 years old; 28%
were 14 years old; 17.9% were 15 years old; 10.5%% were 16 years old; 9.1% were 17
years old; and 4.2% were 18. About 71.8% of children were White/Non-Hispanic, followed
American (1.7%), and 1.4% preferred not to answer this question. Approximately 21.5% of
children were in 7th grade, 23.5% were in 8th grade, 22% were in 9th grade, 15.4% were in
10th grade, 7.9% were in 11th grade, 8.3% were in 12th grade, and 1.4% reported “Other”
Parents reported an average of 6 devices in their home (SD = 1.94). The most
computer (86.5%), tablet (84.1%), video game console (77.5%), desktop computer
(58.1%), streaming device (56.4%), portable game player (38.5%), MP3 player (32.9%),
and other devices (0.8%). Parents reported an average of 4.13 (SD = 1.89) child-owned
devices—devices that were not shared or temporarily owned with limited functionality.
The most commonly reported child-owned device was a smartphone (88%), followed by
tablets (66.4%), video game console (58.6%), laptop computer (57.3%), TV (55.2%),
portable game player (24.7%), MP3 player (22%), desktop computer (20.8%), and
8There were seven OPEN-ENDED responses to this question. Two parents wrote that their child
had graduated, four parents reported their child was a college freshman, and one parent said their
child was in 6th grade.
88
Sixty one percent of parents reported knowing the passcode to their child’s mobile
device(s) and 12.8% said there was no lock on their child’s device(s). Approximately 60%
of parents knew a password(s) to their child’s social media or online accounts, reporting an
average of 3.01 known passwords (SD = 1.60). The most commonly reported known
passwords were for Facebook (41.6%), Email (34.3%), Instagram (30.7%), YouTube
(28.5%), Snapchat (21.6%), Twitter (12%), Pinterest (6.6%), other sites (2.7%), and
Tumblr (1%). Sixty percent of parents reported being friends with their child on social
media for fun/entertainment and 52.4% reported being friends with their child on social
media for the purpose of monitoring their posts or activity. Parents were friends with their
child on an average of 2.36 social media sites (SD = 1.33), the most common being
Twitter (13.7%), Email (13%), Pinterest (5.1%), Tumblr (1%), and other sites (0.3%).
Pearson correlations for all variables are reported in Table 6. For visual clarity,
Table 7 contains correlations for the independent and dependent variables only.
Spearman’s rho (rs) correlations were calculated for ranked variables, including child
grade, household income, parent education, and employment (Field, 2013). Point biserial
correlations (rpb) were calculated for dichotomous variables, including child and parent
gender, child and parent ethnicity, marital status, and family structure (Field, 2013). The
89
Table 5. Descriptive Statistics for Demographic and Control Variables
Variable n % M (SD)
Parent Characteristics
Age 38.34 (7.87)
Gender
Female 437 74.2
Male 150 25.3
Prefer not to answer 2 0.3
Missing values 3 0.5
Race/Ethnicity
White/Non-Hispanic 438 74.0
Black/African American/Non- 58 9.8
Hispanic
Asian or Pacific Islander 38 6.4
Latino/Latina 20 3.4
Mexican American 11 1.9
American Indian/Alaskan Native 10 1.7
Prefer not to answer 8 1.4
Other 4 0.7
Missing values 5 0.8
Marital Status
Married 431 72.8
In a relationship 75 12.7
Single 57 9.6
Divorced 22 3.7
Prefer not to answer 7 1.2
Employment Status
Employed full-time 398 67.2
Employed part-time 73 12.3
Self-employed 56 9.5
Homemaker/family manager 39 6.6
Student 10 1.7
Unemployed 5 0.8
Disabled 5 0.8
Retired 2 0.3
Prefer not to answer 2 0.3
Other 1 0.2
Missing values 1 0.2
90
Education
College graduate (e.g. B.A., B.S.) 247 41.7
Some college (2 or 4 year) 186 31.4
Master’s degree (e.g. M.A., M.S., 73 12.3
M.B.A.)
High school degree or GED 56 9.5
Professional degree (e.g. J.D., 17 2.9
M.D.)
Doctorate degree (e.g. PhD, EdD) 7 1.2
Some high school 2 0.3
Less than 8th grade 0 0
Missing values 4 0.7
Income
Under $30,000 66 11.1
$30,000-$49,999 148 25.0
$50,000-$69,999 123 20.8
$70,000-$99,999 150 25.3
$100,000 or more 104 17.6
Missing values 1 0.2
Parent relationship to child
Birth/biological parent 530 89.5
Step-parent 34 5.7
Adoptive Parent 15 2.5
Grandparent 8 1.3
Foster Parent 3 0.5
Other 2 0.3
Average number of other adults in
0.91 (0.39)
home a
Other adults’ relation to child
Birth/biological parent 419 71.1
Step-parent 71 12.1
No other adults in home 68 11.5
Grandparent 37 6.3
Adoptive Parent 10 1.7
Foster Parent 1 0.2
Parent accessibility
26.45 (10.76)
Weekday
13.75 (7.17)
Weekend
39.16 (14.53)
Total
Child Characteristics n % M (SD)
Gender
Male 328 55.4
Female 263 44.4
Prefer to self-describe 0 0
91
Missing values 1 0.2
Age
13 176 29.7
14 166 28.0
15 106 17.9
14.54 (1.46)
16 62 10.5
17 54 9.1
18 25 4.2
Missing values 3 0.5
Race/Ethnicity
White/Non-Hispanic 425 71.8
Black/African American/Non- 63 10.6
Hispanic
Asian or Pacific Islander 36 6.1
Latino/Latina 25 4.2
Other 13 2.2
American Indian/Alaskan Native 11 1.9
Mexican American 10 1.7
Prefer not to answer 8 1.4
Missing values 1 0.2
Grade
7th grade 127 21.5
8th grade 139 23.5
9th grade 130 22.0
10th grade 91 15.4
11th grade 47 7.9
12th grade 49 8.3
Other 8 1.4
Missing values 1 0.2
Media Devices in Home n % M (SD)
92
Smartphone 521 88.0
Tablet 393 66.4
Video game console 347 58.6
Laptop computer 339 57.3
TV 327 55.2
Portable game player 146 24.7
MP3 player 130 22.0
Desktop computer 123 20.8
Streaming device 120 20.3
Does parent know the passcode for
child’s mobile devices?
Yes 361 61.0
No 155 26.2
There is no lock on devices 76 12.8
Does parent know any passwords for
child’s social media (SM) or online
accounts?
Yes 352 59.5
No 193 32.6
Child doesn’t use SM or online 43 7.3
accounts
Missing values 4 0.7
Parent-known passwords 3.01 (1.60)
Facebook 246 41.6
Email 203 34.3
Instagram 182 30.7
YouTube 169 28.5
Snapchat 128 21.6
Twitter 71 12.0
Pinterest 39 6.6
Other 16 2.7
Tumblr 6 1.0
Is parent friends with child on social
media SM for fun/entertainment?
Yes 355 60.0
No 146 24.7
Child doesn’t use SM 88 14.9
Missing values 3 0.5
Is parent friends with child on social
media SM to monitor activity/posts?
93
Yes 310 52.4
No 185 31.3
Child doesn’t use SM 94 15.9
Missing values 3 0.5
Social media sites that parents are
2.36 (1.33)
friends with their child on
Facebook 334 56.4
Instagram 205 34.6
Snapchat 135 22.8
YouTube 83 14.0
Twitter 81 13.7
Email 77 13.0
Pinterest 30 5.1
Tumblr 6 1.0
Other 2 0.3
a There were 2 missing values for this variable and 3 responses were excluded for selecting
each adult option (e.g. birth, step, and adoptive parent); therefore the n = 487.
94
95
96
97
Model Assessment
proposed indicators reliably reflected underlying latent constructs in this dataset. The
measurement model estimates factor loadings of the observed measures onto the
unobserved latent variables using confirmatory factor analysis (Byrne, 2010). Three
subscales were used as indicators for attitudes (positive attitudes about media effects and
use = 8 items; negative attitudes about media effects and use = 9 items; attitudes about
mediation outcomes = 5 items); two subscales were used as indicators for subjective norms
(injunctive norms = 5 items; descriptive norms = 5 items); and five subscales were used as
indicators for PBC/SE (personal–parent mobile device use = 3 items; work–parent mobile
device use = 3 items; knowledge of teen media activities = 8 items; direct measure of PBC =
5 items; direct measure of SE = 5 items). One observed indicator of each latent variable had
a fixed loading of 1 and the latent constructs were free to correlate with one another
(Byrne, 2010). The error terms for indicators were uncorrelated. This measurement model
98
Figure 3. Hypothesized Measurement Model
Note. ATT_NA = Negative attitudes about media effects and use; ATT_PA = Positive
attitudes about media effects and use; ATT_AMO = Attitudes about mediation outcomes; IN
= Injunctive norms; DN = Descriptive norms; PMU Personal= Parent mobile device use for
personal tasks; PMU Work= Parent mobile device use for work related tasks; KMA =
Knowledge of teen media activities; PBC = Direct measure of perceived behavioral control
to mediate; SE = Direct measure of self-efficacy to mediate. χ2(32) = 250.66, p < .001, GFI =
.922, CFI = .868, RMSEA = .108 (90% CI = .095, .120), SRMR = .083, TLI = .814.
99
In Figure 3, latent constructs are represented in ellipses and observed variables are
represented in rectangles. Fit indices for this model were poor, χ2(32) = 250.66, p < .001,
GFI = .922, CFI = .868, RMSEA = .108 (90% CI = .095, .120), SRMR = .083, TLI = .814.
Specifically, the CFI and TLI were lower than desirable and the RMSEA and SRMR were
higher than desirable. The RMSEA is a parsimony-adjusted index with lower values
between .05 and .08 considered ideal (Kline, 2011). In general, RMSEA will decrease with
more degrees of freedom; however, Kline (2011) explains, “RMSEA does not necessarily
favor models with more degrees of freedom. This is because the effect of correction for
parsimony diminishes as the sample size becomes increasingly large” (p. 206). The null
hypothesis for the chi-square is that attitudes is a latent construct with a three factor
structure; subjective norms is a latent construct with a two factor structure; and perceived
behavioral control is a latent construct with a five factor structure (Byrne, 2010). This
hypothesis was rejected with a significant p-value, which was expected given the large
sample size; however, taken together with low CFI and TLI values, these indices could
indicate an issue of parsimony such that there are more than necessary parameters in the
To address this issue, I focused on the PBC/SE latent construct because it had the
most observed indicators (five) and therefore was the most ostensible culprit affecting
model parsimony. To simplify this construct I combined the separated indicators for
parent mobile device use, and perceived behavioral control and self-efficacy. Parent mobile
device use was initially conceptualized as a single measure but entered as two indicators
based on a principal component analysis which revealed two factors (personal tasks and
work tasks); thus, combining them is essentially reverting to my a priori analysis plan.
100
Combining the measures of PBC and SE has theoretical support. Ajzen (2002a)
acknowledges the similarity between perceived behavioral control and self-efficacy, noting
that these concepts are not always independent of each other and measures of PBC “should
contain items that assess self-efficacy as well as controllability” (p. 680) 9. I also conducted a
principle component analysis on the 10 items to determine if there was statistical support
for combining these measures. They did indeed load onto one factor which explained
59.57% of the variance and also demonstrated high internal consistency as a single
measure (α = .92).
After these steps were taken, the model was re-ran and the fit of this measurement
model was an improvement over the previous but still a poor fit to the data, χ2(17)= 74.97,
p < .001, GFI = .970, CFI = .934, RMSEA = .076 (90% CI = .059, .094), SRMR = .057, TLI =
.891. The RMSEA decreased but still indicated unacceptable fit, and the CFI and TLI were
below the desired cut off of .95. This model is displayed in Figure 4.
9 Ajzen (2002a) also states that, “Depending on the purpose of the investigation, a decision can be
made to aggregate over all items, treating perceived behavioral control as a unitary factor, or to
distinguish between self-efficacy and controllability by entering separate indices into the prediction
equation” (p. 680).
101
Figure 4. Re-specified Measurement Model
.11*
.17*
.88*
.27**
.80**
.88**
.86**
.42**
.20**
.57**
.66**
Note. Standardized estimates are within straight arrows, correlations are within curved
arrows; ATT_NA = Negative attitudes about media effects and use; ATT_PA = Positive
attitudes about media effects and use; ATT_AMO = Positive and negative attitudes about
mediation outcomes; IN = Injunctive norms; DN = Descriptive norms; PMU = Parent mobile
device use; KMA = Knowledge of teen media activities; PBC&SE = Combined direct
measures of perceived behavioral control and self-efficacy to mediate; χ2(17)= 74.97, p <
.001, GFI = .970, CFI = .934, RMSEA = .076 (90% CI = .059, .094), SRMR = .057, TLI = .891.
* p < .05 ** p < .001
102
The standardized parameter estimates (β) are shown in the arrows between the
latent constructs and their indicators. These values represent the amount of change in the
latent construct that is attributed to a standard deviation change of the indicator. All of the
βs were significant; however, the betas for negative attitudes about media use and effects
(β = .11, p = .05) and positive attitudes about media use and effects (β = .17, p = .003) were
quite low, suggesting that they did not indicate, or load onto, the Attitudes latent construct.
Consequently, the model was re-specified with the three attitudes measures entered as
over the previous; however, some fit indices were still not within the recommended range,
χ2(13)= 56.03, p < .001, GFI = .978, CFI = .951, RMSEA = .075 (90% CI = .055, .095), SRMR =
.045, TLI = .894. The RMSEA was still undesirably high and the TLI was below the
recommended cut off of .95. There were no other theoretical, data-driven, or a priori
modifications to impose on this model; therefore it was retained as the final, albeit
untenable, measurement model. This model was over-identified with 36 distinct sample
standardized parameter estimates (β) are shown in the arrows between the latent
constructs and their indicators and correlations are shown within the curved arrows
(Figure 5). All of the βs were significant (and their standard errors were all below .09) and
7 of the 10 correlations between manifest and latent constructs were significant. The non-
significant correlations were between negative attitudes about media effects and use and
attitudes toward mediation outcomes (r = .08, p = .06); negative attitudes about media
103
effects and use and positive attitudes about media effects and use (r = -.05, p = .24); and
positive attitudes about media effects and use and subjective norms (r = .07, p = .11).
104
Figure 5. Retained Measurement Model
-.05
.08
.18**
.16**
.07
.23**
.80** .16*
.10
.89**
.76**
.43**
.19**
.58**
.66**
Note. Standardized estimates are within straight arrows, correlations are within curved
arrows; ATT_NA = Negative attitudes about media effects and use; ATT_PA = Positive
attitudes about media effects and use; ATT_AMO = Positive and negative attitudes about
mediation outcomes; IN = Injunctive norms; DN = Descriptive norms; PMU = Parent mobile
device use; KMA = Knowledge of teen media activities; PBC&SE = Combined direct
measures of perceived behavioral control and self-efficacy to mediate; χ2(13)= 56.03, p <
.001, GFI = .978, CFI = .951, RMSEA = .075 (90% CI = .055, .095), SRMR = .045, TLI = .894.
* p < .05 ** p < .001
105
Summary. The initial hypothesized model consisted of 10 indicators for 3 latent
constructs (see Figure 3). This measurement model failed to reach an acceptable level of fit
and was re-specified with a priori and data-driven reasoning to improve the parsimony of
the model by combining two couples of indicators (Figure 4). The re-specified model was
an improvement over the first but still exhibited low factor loadings within the attitudes
latent construct. The three indicators for this construct were then re-specified as manifest
variables and this measurement model was improved over the previous two (Figure 5).
Nevertheless, some fit indices were still above or below recommended cutoff values and
indicated that the hypothesized model had mis-fit with the sample data. With no other
hypothesized relationships (i.e. paths) among the exogenous and endogenous variables.
Three structural models were fit for the three endogenous variables of interest in this
study (active mediation, restrictive mediation, and monitoring). As with the measurement
model, the structural models were evaluated by their fit indices, the magnitude and
statistical significance of the standardized parameter estimates, and their standard errors.
Rather than rely on the correlational results alone, all demographic variables were
included as covariates in the first specified model for each endogenous variable and non-
Missing data was handled using the Maximum Likelihood method in AMOS
(Arbuckle, 2017). When data are incomplete, some fit measures cannot be computed.
Specifically GFI and SRMR cannot be computed because these indices compare the fit of the
106
hypothesized model to the independence model (i.e. a poor fitting model to the data) or the
saturated model (i.e. a perfectly fitting model to the data) (Arbuckle, 2017). Computing
these models with incomplete data is problematic given the number of parameters AMOS
would need to estimate without sufficient data to use (Arbuckle, 2017); therefore, these fit
The structural model for active mediation with all control variables exhibited poor
model fit, χ2(178)= 1698.71, p < .001, CFI = .409, RMSEA = .120 (90% CI = .115, .125), TLI =
.233. The non-significant demographic variables trimmed from this model were income (β
= -.037, p = .26), child gender (β = -.022, p = .51), family structure (β = -.002, p = .53), parent
employment (β = -.001, p = .98), marital status (β = .003, p = .92), and parent education (β =
-.028, p = .39). Removing these variables and re-specifying the model decreased the model
chi-square but did not improve the fit, χ2(79)= 1320.61, p < .001, CFI = .461, RMSEA = .163
(90% CI = .155, .171), TLI = .181 (Figure 6). A model containing no control variables had a
better but still poor fit to the data, χ2(16)= 127.88, p < .001, CFI = .902, RMSEA = .109 (90%
CI = .092, .127), TLI = .779. Given this poor model fit, the calculated path estimates should
The structural model for restrictive mediation was specified with restrictive
mediation as a latent endogenous variable and the summed and frequency measures as
indicators. The restrictive mediation structural model with all control variables could not
converge within 500 iterations, nor could a model with no control variables (Figure 7).
The structural model for monitoring was specified with monitoring as a latent
endogenous variable and the summed and frequency measures as indicators. The
107
monitoring structural model with all control variables could not converge within 500
Summary. The active mediation structural model fit was poor and the restrictive
mediation and monitoring structural models would not converge to a solution and could
not be evaluated. These results were not entirely unexpected given that some fit indices for
the measurement model were not within the recommended ranges. Ultimately the results
implied that the model was mispecified and the hypothesized variance/covariance matrix
did not reach an acceptable indication of fit to the data variance/covariance matrix. Put
another way, the series of equations arranged in paths predicted by the theory of planned
behavior did not estimate or approximate the data. This did not mean that the constructs
and measures themselves were nonsignificant. On the contrary, the significant paths noted
in Figure 4 between indicators and their latent constructs indicated that there were
108
Figure 6. Structural Model Predicting Active Mediation with Control Variables
Note. ATT_NA = Negative attitudes about media effects and use; ATT_PA = Positive
attitudes about media effects and use; ATT_AMO = Positive and negative attitudes about
mediation outcomes; IN = Injunctive norms; DN = Descriptive norms; PMU = Parent mobile
device use; KMA = Knowledge of teen media activities; PBC&SE = Combined direct
measures of perceived behavioral control and self-efficacy to mediate; CAge = Child age;
CEth = Child ethnicity (White/Non-Hispanic = 1, Non-white = 0); TAccess = Total parent
accessibility; PAge = Parent age; PGen = Parent gender (Male = 0, Female = 1); PEth =
Parent ethnicity (White/Non-Hispanic = 1, Non-white = 0).
109
Figure 7. Structural Model Predicting Restrictive Mediation
Note. ATT_NA = Negative attitudes about media effects and use; ATT_PA = Positive
attitudes about media effects and use; ATT_AMO = Positive and negative attitudes about
mediation outcomes; IN = Injunctive norms; DN = Descriptive norms; PMU = Parent mobile
device use; KMA = Knowledge of teen media activities; PBC&SE = Combined direct
measures of perceived behavioral control and self-efficacy to mediate; ResMedAvg =
Averaged restrictive mediation (frequency); and ResMedSum = Summed restrictive
mediation (yes/no).
110
Figure 8. Structural Model Predicting Monitoring
Note. ATT_NA = Negative attitudes about media effects and use; ATT_PA = Positive
attitudes about media effects and use; ATT_AMO = Positive and negative attitudes about
mediation outcomes; IN = Injunctive norms; DN = Descriptive norms; PMU = Parent mobile
device use; KMA = Knowledge of teen media activities; PBC&SE = Combined direct
measures of perceived behavioral control and self-efficacy to mediate; MonitorSum =
Summed monitoring measure (yes/no); and MonitorAvg = Averaged monitoring measure
(frequency).
111
Hypothesis Testing
Hierarchical regressions were used to test the hypotheses and research questions in
lieu of poor fitting structural equation models. Demographic variables that were
significantly correlated with dependent variables were entered into the first block (see
Table 6 for all correlations). Specifically, child age, child grade, child ethnicity, parent age,
parent gender, parent ethnicity, and income were entered in the first block of the
regression predicting active mediation; child age, child grade, and parent age were entered
in the first block of the regression predicting summed restrictive mediation; child age, child
grade, parent age, parent gender, parent accessibility, and parent employment were
entered in the first block of the regression predicting average restrictive mediation; child
gender and parent gender were entered in the first block of the regression predicting
summed monitoring; and child age, child grade, parent age, and parent gender were
entered in the first block of the regression predicting average monitoring. The TPB
variables were entered into the second block. Parenting behaviors (connection, regulation,
were entered into the third block. The last block consisted of interactions for the proposed
parental mediation. Moderation was analyzed using model 1 of Hayes PROCESS macro for
Scatterplots of the residuals for the outcome and predictor variables were evaluated
Multicollinearity was checked by reviewing the correlation matrix for values > .70, as well
as the tolerance and variance inflation factors (VIF). Tolerance below 0.2, VIF values above
112
10, and an average VIF of “substantially greater than 1” would indicate issues of
multicollinearity (Field, 2013). There were several instances of high correlations that were
examined further. First, child age and grade were strongly correlated (rs = .87, p < .001), as
were parent and child ethnicity (rpb = .88, p < .001) (Table 6). Therefore, child grade and
child ethnicity were excluded from the first block of demographic predictors and child age
Second, the correlation between injunctive norms and descriptive norms was high
(r = .71) (Table 7). To control for this collinear effect, I ran separate regressions alternating
entering injunctive norms in one block and entering descriptive norms in the subsequent
block thereby controlling the effect of one norm to examine the individual effect of the
other. I ran these analyses for all of the outcome variables and injunctive and descriptive
norms were never significant; therefore, they were entered together into the same block
The last correlation of note was between connection and regulation (r = .73, p <
.001) (Table 7). To control for this collinearity, I entered autonomy granting and regulation
into one block and connection into the subsequent block. In summary, there were 5 blocks
for each regression: block one contained demographic variables, block two contained the
theory of planned behavior variables, block three contained the family system variables,
block four contained parenting connection, and block five contained the interactions.
Cooks distance and Mahalanobis distance were evaluated to check for outliers.
Cook’s distances greater than 1 indicate cases that have a significant impact on the model
as whole (Field, 2013). Mahalanobis distances measure the distance between a case and
the multivariate mean(s) of the predictor variables (Field, 2013). Outlier cases were
113
determined using the critical value of chi-square for p = .05 and degrees of freedom
equaling the number of predictors for each regression. The Durbin-Watson statistic is
reported to test the assumption that errors are independent, with values around 2 being
desirable and indicating that residual errors are uncorrelated (Field, 2013).
independent variable; however, the results below are reported by dependent variable so
that similar analyses are grouped together and straightforward to follow. Appendix A is a
summarized list of the hypotheses, research questions and their findings. Note that the
coefficients and p-values reported below are at each block they were entered and the
coefficients and p-values in Tables 8 through 12 are from the fourth block of the regression
model.
Active mediation. Results of the regression analysis for active mediation are
displayed in Table 8. Missing value cases were excluded listwise and cases with a
Mahalanobis distance of 28.87 or greater (df = 18) were excluded from the analysis. There
The first block of demographic variables was significant and explained 10% of the
variance, F(5, 509) = 11.21, p < .001, Durbin-Watson = 1.95. In this block, child age (β = -
.20, p < .001), parent gender (coded as female =1, male = 0; β = .15, p < .001), and parent
significant. The second block of TPB variables was significant and explained an additional
24% of variance, F(13, 501) = 19.94, p < .001. In this block, negative attitudes about media
effects and use (β = .15, p = .001) (H1), positive attitudes about media effects and use (β =
.09, p = .02) (RQ1), parent mobile device use (β = .08, p = .05) (H6), and knowledge of teen
114
media activities (β = .43, p < .001) (H7) all positively predicted active mediation. The third
block with family communication quality, parenting regulation, autonomy granting, and
F(17, 497) = 18.14, p < .001. In this block, autonomy granting (β = .16, p = .001) and
With the addition of this third block of variables, attitudes about mediation outcomes
became significant in this block (β = .11, p = .02) (H2), and positive attitudes about media
effects and use (β = .05, p = .19) (RQ1) and parent mobile device use (β = .05, p = .27) (H6)
were no longer significant. The last block with parenting connection was significant, F(18,
496) = 17.16, p < .001, but the change in R2 was not significant (∆R2 = .001, p = .42) 10.
supported. Specifically, negative attitudes about media effects and use (H1) and attitudes
toward mediation outcomes (H2) positively predicted active mediation. Parent mobile
device use (H6) and knowledge of teen media activities (H7) did predict active mediation at
their block of entry; however, when all variables were entered into the model, these
Specifically, injunctive norms (H3), descriptive norms (H4), and PBC/SE (H5) did not
attitudes about media effects and use would be related to active mediation and the results
showed that they positively predict this behavior; however, with the addition of the FST
10The significant F-ratio for this step indicates that the overall model significantly predicted the
outcome variable; however, parenting connection did not explain any additional variance nor did it
significantly improve the model’s ability to predict the outcome (Field, 2013).
115
Active mediation moderation analyses. The remaining research questions (2, 3,
and 4) asked whether parenting behaviors, family communication quality, and barriers to
parental mediation moderate relationships between the TPB constructs and active
mediation. Family communication quality (RQ3) was not a significant predictor of active
mediation; therefore, there was no evidence that it would moderate relations between the
TPB variables and active mediation. The parenting behavior of autonomy granting (RQ2)
and barriers to parental mediation (RQ4) were significant and subsequently tested as
moderators with the significant TPB variables: negative attitudes about media effects and
use, positive attitudes about media effects and use, attitudes about mediation outcomes,
parent mobile device use, and knowledge of teen media activities. Interaction terms were
created for these variables and the significant moderators: barriers to parental mediation
and autonomy granting. Predictor variables were mean centered prior to moderation
analyses. To reduce the potential for multicollinearity, each interaction was estimated in a
separate regression model. Model 1 of Hayes PROCESS macro for SPSS was used to test the
moderating variables.
The interaction between negative attitudes about media effects and use and
autonomy granting was not significant (b = -.07, p = .08), F(1, 497) = 3.12, p = .08, R2 = .38.
The interaction between positive attitudes about media effects and use and
autonomy granting was not significant (b = .07, p = .17), F(1, 497) = 1.91, p = .17, R2 = .38.
granting was not significant (b = .04, p = .36), F(1, 497) = 0.83, p = .36, R2 = .38.
The interaction between parent mobile device use and autonomy granting was not
116
The interaction between knowledge of teen media use and autonomy granting was
The interaction between negative attitudes about media effects and use and barriers
variance in the model, F(1, 497) = 4.80, p = .03, R2 = .39. To examine this interaction effect,
simple slopes were estimated at three levels of perceived barriers: 1 SD below the mean, at
the mean, and 1 SD above the mean. When perceived barriers were low there was a non-
significant positive relationship between negative attitudes and active mediation (b = .04, p
= .37); at the mean of perceived barriers there was a positive relationship between
negative attitudes and active mediation (b = .09, p = .002); and when perceived barriers
were high there was a stronger positive relationship between negative attitudes and
The interaction between positive attitudes about media effects and use and barriers
variance in the model, F(1, 497) = 5.45, p = .02, R2 = .39. To examine this interaction effect,
simple slopes were estimated at three levels of perceived barriers: 1 SD below the mean, at
the mean, and 1 SD above the mean. When perceived barriers were low there was a
significant positive relationship between positive attitudes and active mediation (b = .13, p
= .01); at the mean of perceived barriers there was a non-significant positive relationship
between positive attitudes and active mediation (b = .05, p = .20); and when perceived
barriers were high there was a non-significant negative relationship between positive
117
The interaction between attitudes about mediation outcomes and barriers to
parental mediation was not significant (b = .01, p = .87), F(1, 497) = 0.03, p = .87, R2 = .38.
The interaction between parent mobile device use and barriers to parental
the model, F(1, 497) = 5.04, p = .03, total R2 = .39. To examine this interaction effect, simple
slopes were estimated at three levels of perceived barriers: 1 SD below the mean, at the
mean, and 1 SD above the mean. When perceived barriers were low there was a significant
positive relationship between parent mobile device use and active mediation (b = .15, p =
.01); at the mean of perceived barriers there was a non-significant positive relationship
between parent mobile device use and active mediation (b = .06, p = .16); and when
perceived barriers were high there was a non-significant negative relationship between
The interaction between knowledge of teen media use and barriers to parental
mediation was not significant (b = .08, p = .06), F(1, 497) = 3.64, p = .06, R2 = .38.
Per these results, to answer research question 2, autonomy granting did not
moderate any relationships between the TPB variables and active mediation. To answer
active mediation and negative attitudes about media effects, positive attitudes about media
effects, and parent mobile device use. When parents reported average or above average
barriers to parental mediation, negative attitudes about media effects and use positively
predicted active mediation; and when parents reported low perceived barriers to parental
mediation, positive attitudes about media effects and use and parent mobile device use
118
Table 8. Hierarchical Regression Analysis Predicting Active Mediation (N = 515)
M (SD) B SE β p R2 ∆ R2 Tol. VIF
AM
3.01 (0.68)
(Range 1-5)
Block 1 .10**
Child age 14.45 (1.39) -.05 .02 -.09 .03 .709 1.410
Parent age 37.94 (7.39) -.01 .00 -.08 .07 .730 1.369
Parent gender
0.75 (0.43) .16 .06 .10 .01 .921 1.086
(female =1)
Parent ethnicity
(White/Non- 0.78 (0.41) -.09 .06 -.05 .16 .902 1.108
Hispanic =1)
Incomea 3.18 (1.26) -.01 .02 -.02 .61 .867 1.153
Block 2 .34** .24**
ATT_NA 3.14 (0.94) .08 .03 .11 .00 .798 1.253
ATT_PA 3.61 (0.69) .05 .04 .05 .18 .790 1.266
ATT_AMO 3.33 (0.76) .09 .04 .10 .02 .640 1.563
SN_IN 4.40 (1.28) -.05 .03 -.09 .11 .408 2.449
SN_DN 4.11 (1.36) .03 .03 .06 .32 .394 2.539
PMU 3.93 (0.65) .05 .04 .05 .23 .725 1.380
KMA 3.73 (0.76) .34 .04 .38 .00 .685 1.459
PBC & SE 3.84 (0.84) -.02 .04 -.03 .58 .496 2.017
Block 3 .38** .04**
FCQ 3.83 (0.68) -.06 .05 -.06 .21 .484 2.065
PB_R 4.17 (0.73) .09 .06 .10 .11 .343 2.918
PB_AG 3.96 (0.70) .17 .05 .17 .00 .497 2.011
BAR 2.24 (0.80) .11 .04 .13 .00 .614 1.628
Block 4 .38** .00
PB_C 4.34 (0.72) -.05 .06 -.05 .42 .302 3.308
Note. Coefficients are reported from the fourth block of the model; Tol. = Tolerance;
Average VIF = 1.57.
a Under $30k = 1; $30-$49,9k = 2; $50-$69,9k = 3; $70-$99,9k = 4; $100k or more = 5.
119
Summed restrictive mediation. Results of the regression analysis for summed
restrictive mediation are in Table 9. Summed restrictive mediation refers to the binary
response questions regarding time and content rules (e.g. “Do you have rules about when
your child can use their mobile device?”). Missing value cases were excluded listwise and
cases with a Mahalanobis distance of 25 or greater (df = 15) were excluded from the
The first block of demographic variables was significant and explained 11% of the
variance, F(2, 522) = 31.21, p < .001, Durbin-Watson = 1.93. Specifically, child age (β = -.27,
p < .001) and parent age (β = -.10, p = .03) were significant. The second block of TPB
variables was significant and explained an additional 20% of variance, F(10, 514) = 22.73, p
< .001. Negative attitudes about media effects and use (β = .21, p < .001) (H1), attitudes
about mediation outcomes (β = .25, p < .001) (H2), and knowledge of teen media activities
(β = .20, p < .001) (H7) all positively predicted summed restrictive mediation. The third
block with family communication quality, parenting regulation, autonomy granting, and
barriers to parental mediation was significant, F(14, 510) =16.45, p < .001; however, none
of the coefficients in this block were significant and the change in R2 was not significant
(∆R2 = .001, p = .51). The last block with parenting connection was significant, F(15, 509)
=15.44, p < .001, but the change in R2 was not significant (∆R2 = .002, p = .28).
attitudes about media effects and use (H1), attitudes about mediation outcomes (H2), and
parent knowledge of teen media activities (H7) all positively predicted summed restrictive
(H3), descriptive norms (H4), PBC/SE (H5), and parent mobile device use (H6) did not
120
significantly impact summed restrictive mediation. Research question 1 asked if parent’s
positive attitudes about media effects and use would be related to summed restrictive
mediation and the results showed that they did not impact this behavior.
questions (2, 3, and 4) asked whether parenting behaviors, family communication quality,
and barriers to parental mediation moderate the relationship between the TPB constructs
and summed restrictive mediation. None of these variables were significant predictors of
summed restrictive mediation; therefore, there was no evidence that they moderated
121
Table 9. Hierarchical Regression Analysis Predicting Summed Restrictive Mediation (N = 525)
M (SD) B SE β p R2 ∆ R2 Tol. VIF
RM_SUM
3.25 (1.47)
(Range 1-5)
Block 1 .11**
Child age 14.46 (1.39) -.15 .05 -.14 .00 .741 1.349
Parent age 37.91 (7.30) -.02 .01 -.08 .05 .760 1.316
Block 2 .31** 20**
ATT_NA 3.16 (0.95) .31 .06 .20 .00 .834 1.200
ATT_PA 3.61 (0.69) .00 .09 .00 .98 .839 1.192
ATT_AMO 3.31 (0.75) .51 .09 .26 .00 .644 1.553
SN_IN 4.39 (1.26) -.10 .07 -.09 .14 .391 2.560
SN_DN 4.13 (1.35) .07 .07 .06 .29 .374 2.674
PMU 3.93 (0.66) -.06 .09 -.03 .54 .769 1.301
KMA 3.73 (0.76) .37 .09 .19 .00 .700 1.428
PBC & SE 3.82 (0.84) .12 .09 .07 .18 .501 1.995
Block 3 .31** .01
FCQ 3.81 (0.68) -.09 .11 -.04 .43 .508 1.970
PB_R 4.15 (0.75) .10 .13 .05 .43 .332 3.010
PB_AG 3.94 (0.70) -.06 .11 -.03 .60 .492 2.032
BAR 2.25 (0.79) .01 .09 .00 .95 .612 1.633
Block 4 .31** .00
PB_C 4.31 (0.74) -.15 .14 -.07 .28 .289 3.460
Note. Coefficients are reported from the fourth block of the model; Tol. = Tolerance;
Average VIF = 1.72.
*p < .05 **p < .001
122
Average restrictive mediation. Results of the regression analysis for average
restrictive mediation are in Table 10. Average restrictive mediation refers to the strictness
and consistency that parents reported in enforcing their time and content restrictions (e.g.
“How would you describe how strict or consistent you are with enforcing rules about when
your child can be on their mobile device?”). Missing value cases were excluded listwise and
cases with a Mahalanobis distance of 28.87 or greater (df = 18) were excluded from the
The first block of demographic variables was significant and explained 9% of the
variance, F(5, 476) = 9.59, p < .001, Durbin-Watson = 2.12. In this block, child age (β = -.11,
p = .02), parent gender (β = .11, p = .01), parent accessibility (β = .19, p < .001) and parent
employment (β = -.14, p = .002) were significant. The second block of TPB variables was
significant and explained an additional 15% of variance, F(13, 468) = 11.51, p < .001. In this
block, negative attitudes about media effects and use (β = .13, p = .002) (H1), attitudes
about mediation outcomes (β = .21, p < .001) (H2), and PBC/SE (β = .19, p < .001) (H5)
positively predicted average restrictive mediation. The third block with family
mediation was significant and explained an additional 2% of variance, F(17, 464) = 9.60, p
< .001. In this block, regulation positively predicted average restrictive mediation (β = .15,
p = .02). The last block with parenting connection (β = .16, p = .04) was significant and
explained an additional 1% of variance, F(18, 463) = 9.37, p < .001. With the addition of
connection to the regression model, regulation was no longer significant (β = .08, p = .22).
attitudes about media effects and use (H1), attitudes toward mediation outcomes (H2), and
123
PBC/SE (H5) positively predicted average restrictive mediation. Hypotheses 3, 4, 6, and 7
were not supported. Specifically, injunctive norms (H3), descriptive norms (H4), parent
mobile device use (H6), and knowledge of teen media activities (H7) did not significantly
attitudes about media effects and use would be related to average restrictive mediation and
the results showed that they did not impact this behavior.
and 4) asked whether parenting behaviors, family communication quality, and barriers to
parental mediation moderate the relationship between the TPB constructs and average
mediation (RQ4) were not significant predictors of average restrictive mediation, therefore,
there was no evidence that they moderated relations between the TPB variables and
average restrictive mediation. The connection and regulation parenting behaviors (RQ2)
were significant and subsequently tested as moderators with the significant TPB variables:
negative attitudes about media effects and use, attitudes about mediation outcomes, and
PBC/SE. Interaction terms were created for these variables and connection and regulation.
Predictor variables were mean centered prior to moderation analyses. Model 1 of Hayes
PROCESS macro for SPSS was used to test the moderating variables.
The interaction between negative attitudes about media effects and use and
connection was not significant (b = -.03, p = .47), F(1, 462) = 0.52, p = .47, R2 = .27.
The interaction between attitudes about mediation outcomes and connection was
124
The interaction between PBC/SE and connection was not significant (b = -.04, p =
The interaction between negative attitudes about media effects and use and
regulation was not significant (b = -.03, p = .41), F(1, 462) = 0.68, p = .41 R2 = .27.
The interaction between attitudes about mediation outcomes and regulation was
The interaction between PBC/SE and regulation was not significant (b = -.06, p =
Per these results, to answer research question 2, connection and regulation did not
moderate any relationships between the TPB variables and average restrictive mediation.
125
Table 10. Hierarchical Regression Analysis Predicting Average Restrictive Mediation (N =
482)
M (SD) B SE β p R2 ∆ R2 Tol. VIF
RM_AVG
2.85 (0.69)
(Range 1-4)
Block 1 .09**
Child age 14.40 (1.38) -.01 .02 -.02 .69 .746 1.340
Parent age 37.93 (7.23) .00 .00 -.05 .26 .762 1.313
Parent gender 0.74 (0.44) .17 .07 .11 .01 .889 1.125
Parent
39.29 (11.50) .00 .00 .05 .22 .819 1.221
accessibility
Parent
7.43 (1.01) -.06 .03 -.08 .05 .943 1.060
employmenta
Block 2 .24** .15**
ATT_NA 3.19 (0.96) .09 .03 .12 .01 .817 1.224
ATT_PA 3.61 (0.70) .00 .04 .00 .98 .795 1.258
ATT_AMO 3.36 (0.76) .20 .05 .22 .00 .633 1.579
SN_IN 4.41 (1.28) .00 .03 -.01 .93 .396 2.526
SN_DN 4.16 (1.37) -.02 .03 -.05 .47 .389 2.569
PMU 3.93 (0.65) -.05 .05 -.05 .29 .757 1.321
KMA 3.78 (0.75) .07 .04 .08 .10 .716 1.397
PBC & SE 3.87 (0.82) .10 .05 .11 .04 .513 1.949
Block 3 .26* .02*
FCQ 3.85 (0.68) -.05 .06 -.05 .36 .496 2.018
PB_R 4.19 (0.72) .08 .07 .08 .22 .335 2.981
PB_AG 3.98 (0.70) -.02 .06 -.02 .72 .496 2.016
BAR 2.22 (0.81) -.02 .04 -.03 .58 .592 1.690
Block 4 .27* .01*
PB_C 4.36 (0.72) .15 .07 .16 .04 .275 3.636
Note. Coefficients are reported from the fourth block of the model; Tol. = Tolerance;
Average VIF = 1.58.
a Prefer not to answer = 1; Retired or unemployed = 2; Disabled or Other = 3; Student = 4;
126
* p < .05 **p < .001
Summed monitoring. Results of the regression analysis for summed monitoring
are in Table 11. Summed monitoring refers to overt and covert means of viewing
adolescent behaviors on mobile devices (e.g. “Are you friends with your child on their
social media profiles (e.g. Facebook, Twitter) for the purpose of monitoring their posts or
activity?”). Missing value cases were excluded listwise and cases with a Mahalanobis
distance of 25 or greater (df = 15) were excluded from the analysis. There were no Cook’s
The first block of demographic variables was significant and explained 2% of the
variance, F(2, 534) = 4.13, p = .02, Durbin-Watson = 2.00. Only parent gender was
significant in this block (β = .10, p = .03). The second block of TPB variables was significant
and explained an additional 11% of the variance F(10, 526) = 8.02, p < .001. In this block,
attitudes about mediation outcomes (β = .14, p = .007) (H2), PBC/SE (β = .13, p = .01) (H5),
parent mobile device use (β = .09, p = .04) (H6), and knowledge of teen media activities (β =
.09, p = .04) (H7) positively predicted summed monitoring. The third block with family
mediation was significant, F(14, 522) =6.36, p < .001; however, none of the coefficients in
this block were significant and the change in R2 was not significant (∆R2 = .02, p = .09). With
the addition of this third block of variables, knowledge of teen media activities (H7) was no
longer significant (β = .09, p = .09). The last block with parenting connection was
significant, F(15, 521) =5.94, p < .001, but the change in R2 was not significant (∆R2 = .00, p =
.70).
supported. Specifically, attitudes about mediation outcomes (H2), PBC/SE (H5), and parent
127
mobile device use (H6) all positively predicted summed monitoring. Knowledge of teen
media activities (H7) did significantly predict summed monitoring at its block of entry;
however, when all variables were entered into the model, knowledge of teen media
activities was no longer significant. Hypotheses 1, 3, and 4 were not supported. Specifically,
negative attitudes about media effects and use (H1), injunctive norms (H3), and descriptive
norms (H4) did not significantly impact summed monitoring. Research question 1 asked if
parent’s positive attitudes about media effects and use would be related to summed
monitoring and the results showed that they did not affect this behavior.
3, and 4) asked whether parenting behaviors, family communication quality, and barriers
to parental mediation moderate the relationship between the TPB constructs and summed
therefore, there was no evidence that they moderated relations between the TPB variables
128
Table 11. Hierarchical Regression Analysis Predicting Summed Monitoring (N = 537)
M (SD) B SE β p R2 ∆ R2 Tol. VIF
MON_SUM
2.37 (1.36)
(Range 1-4)
Block 1 .02*
Child gender
0.45 (0.50) .14 .12 .05 .21 .929 1.077
(female =1)
Parent gender 0.74 (0.44) .20 .13 .07 .12 .912 1.097
Block 2 .13** .11**
ATT_NA 3.13 (0.97) .03 .06 .02 .64 .827 1.210
ATT_PA 3.62 (0.69) -.02 .09 -.01 .81 .828 1.208
ATT_AMO 3.33 (0.76) .27 .09 .15 .00 .639 1.565
SN_IN 4.42 (1.27) .05 .07 .05 .46 .431 2.320
SN_DN 4.15 (1.37) .02 .06 .02 .71 .421 2.374
PMU 3.92 (0.67) .22 .10 .11 .02 .757 1.321
KMA 3.74 (0.76) .15 .09 .08 .08 .735 1.360
PBC & SE 3.82 (0.84) .30 .09 .19 .00 .511 1.959
Block 3 .15** .02
FCQ 3.82 (0.68) -.02 .12 -.01 .87 .495 2.019
PB_R 4.17 (0.75) -.13 .13 -.07 .31 .335 2.984
PB_AG 3.96 (0.71) .18 .11 .09 .11 .480 2.083
BAR 2.26 (0.80) .16 .09 .09 .08 .629 1.591
Block 4 .15** .00
PB_C 4.33 (0.74) -.05 .14 -.03 .70 .290 3.444
Note. Coefficients are reported from the fourth block of the model; Tol. = Tolerance;
Average VIF = 1.64.
*p < .05 **p < .001
129
Average monitoring. Results of the regression analysis for average monitoring are
in Table 12. Average monitoring refers to the frequency with which parents would engage
in monitoring behaviors (e.g. “How often do you ask your child what they are doing on their
mobile device when they are using it?”). Missing value cases were excluded listwise and
cases with a Mahalanobis distance of 26.30 or greater (df = 16) were excluded from the
The first block of demographic variables was significant and explained 10% of the
variance, F(3, 538) = 20.33, p < .001, Durbin-Watson = 1.99. Specifically, child age (β = -.26,
p < .001) and parent gender (β = .14, p < .001) were significant. The second block of TPB
variables was significant and explained an additional 23% of the variance, F(11, 530) =
23.61, p < .001. In this block, negative attitudes about media effects and use (β = .21, p <
.001) (H1), attitudes about mediation outcomes (β = .14, p = .001) (H2), PBC/SE (β = .09, p
= .04) (H5), and parent knowledge of teen media activities (β = .27, p < .001) (H7) were
significant. The third block with family communication quality, parenting regulation,
autonomy granting, and barriers to parental mediation was significant, F(15, 526) = 17.53,
p < .001; however, none of the coefficients in this block were significant and the change in
R2 was not significant (∆R2 = .00, p = .48). With the addition of this third block of variables,
PBC/SE (H5) was no longer significant (β = .09, p = .06). The last block with parenting
connection was significant, F(16, 525) = 16.45, p < .001, but the change in R2 was not
supported. Specifically, negative attitudes about media effects and use (H1), attitudes
toward mediation outcomes (H2), and knowledge of teen media activities (H7) positively
130
predicted average monitoring. PBC/SE (H5) did significantly predict average monitoring at
its block of entry; however, when all variables were entered into the model, PBC/SE was no
longer significant. Hypotheses 3, 4, and 6 were not supported. Specifically, injunctive norms
(H3), descriptive norms (H4), and parent mobile device use (H6) did not significantly
impact average monitoring. Research question 1 asked if parent’s positive attitudes about
media effects and use would be related to average monitoring and the results showed that
and 4) asked whether parenting behaviors, family communication quality, and barriers to
parental mediation moderate the relationships between the TPB constructs and average
therefore, there was no evidence that they moderated relations between the TPB variables
131
Table 12. Hierarchical Regression Analysis Predicting Average Monitoring (N = 542)
M (SD) B SE β p R2 ∆ R2 Tol. VIF
MON_AVG
3.08 (0.98)
(Range 1-5)
Block 1 .10**
Child age 14.50 (1.44) -.07 .03 -.11 .01 .746 1.341
Parent age 38.26 (7.72) -.01 .01 -.07 .07 .786 1.273
Parent gender 0.74 (0.44) .23 .08 .10 .01 .929 1.076
Block 2 .33** .23**
ATT_NA 3.14 (0.97) .20 .04 .19 .00 .830 1.205
ATT_PA 3.61 (0.70) -.06 .05 -.05 .25 .822 1.216
ATT_AMO 3.32 (0.77) .18 .06 .14 .00 .640 1.563
SN_IN 4.39 (1.29) -.06 .04 -.08 .14 .416 2.402
SN_DN 4.13 (1.39) .08 .04 .11 .06 .400 2.503
PMU 3.92 (0.67) .05 .06 .03 .40 .767 1.304
KMA 3.73 (0.77) .32 .06 .25 .00 .690 1.450
PBC & SE 3.81 (0.85) .11 .06 .09 .06 .517 1.936
Block 3 .33** .00
FCQ 3.83 (0.68) -.09 .07 -.07 .20 .503 1.988
PB_R 4.16 (0.75) .05 .08 .04 .50 .339 2.950
PB_AG 3.96 (0.71) .09 .07 .07 .19 .494 2.023
BAR 2.25 (0.80) -.03 .06 -.02 .60 .633 1.580
Block 4 .33** .00
PB_C 4.33 (0.74) -.06 .09 -.05 .46 .297 3.372
Note. Coefficients are reported from the fourth block of the model; Tol. = Tolerance;
Average VIF = 1.63.
*p < .05 **p < .001
132
CHAPTER 5: Discussion
The overarching goal of the present study was to identify factors associated with
mobile devices. Toward this end, the theory of planned behavior (TPB) was employed as a
framework to investigate if and how attitudes, norms, and perceived behavioral control
beliefs relate to parental mediation of mobile devices during teen years, and integrate these
into a theoretical model along with family system factors. The results showed that
behavioral and control beliefs are stronger predictors than norms of parental mediation.
Family system variables did not influence or moderate relationships between the TPB
variables and parental mediation, but parent barriers did interact with some of the TPB
This chapter first discusses the application of the TPB to parental mediation and the
limitations that were uncovered during the statistical analyses. The findings and
implications for each TPB construct are summarized next, followed by the family system
moderators and demographic variables that were examined. The last section addresses the
limitations of this study, practical recommendations for parents, and future directions for
research.
toward a behavior, subjective norms, and perceived behavioral control. The underlying
beliefs that inform these constructs are behavioral beliefs, normative beliefs, and control
parental mediation was the crux of this study. The initial data analysis plan attempted to
133
analyze them using structural equation modeling. Although there were significant paths in
the measurement model, neither the measurement nor the structural models exhibited
From a statistical perspective, these findings demonstrated that the paths arranged
and specified to model the TPB did not approximate the data. The measurement model fit
statistics 11 indicated that although the hypothesized model as a whole was better than no
model at all (i.e. Goodness of Fit Index, GFI), it was a passable improvement to the poor-
fitting null model (i.e. Comparative Fit Index (CFI) and Tucker-Lewis Index (TLI)).
Essentially, the constructs, and measures I chose to assess them, did not work together in a
I believe the model fit issues occurred because I strictly adhered to the structure of
the TPB. While I was primarily using the TPB as a framework to identify constructs that
predict parental mediation, I was also interested in testing the predictive validity of the
model as a whole. During the model specification stage, I did inspect the modification
indices and observed that allowing covariance between some of the error terms would
have improved the model fit; however, there was no theoretical basis for allowing this
modification (and not allowing others) per the TPB, and it would have deviated the
structure of the model from the TPB. Eliminating some constructs all together would also
have improved the model but, again, this would have strayed from the structure of the TPB.
The real goal is to test a theory by specifying a model that represents predictions of
that theory among plausible constructs measured with the appropriate indicators. If
such a model does not ultimately fit the data, then this outcome is interesting
11The retained measurement model fit statistics were, χ2(13)= 56.03, p < .001, GFI = .978, CFI =
.951, RMSEA = .075 (90% CI = .055, .095), SRMR = .045, TLI = .894.
134
because there is value in reporting models that challenge or debunk theories. (p.
189)
The goal of all empirical research is to uncover answers to questions. Ultimately, the
answer to the question asked in this study of whether or not the TPB performs well as a
structural model predicting parental mediation is that it did not. While the statistical model
was not useful, there were significant relationships between the constructs of TPB and
parental mediation. Indeed, the SRMR—standardized root mean square residual, a fit
index that indicates the overall average difference between predicted and observed
values—was good, which suggested that there were noteworthy paths to be explored
despite the model as a whole not approximating the sample data. The next sections discuss
For the majority of parental mediation behaviors, negative attitudes about media
effects and use (H1) was positively associated with, and the second strongest TPB
coefficient for, active mediation, summed restrictive mediation, and average monitoring.
That is, parents who were concerned about their teen being exposed to undesirable
behaviors (e.g. cyberbullying or sexting), were more likely to have discussions about
mobile devices (active mediation); set and frequently apply restrictions for mobile devices
(summed and average restrictive mediation); and practice frequent overt monitoring
(average monitoring). These findings are consistent with parental mediation on other
platforms, like video games (Shin & Huh, 2011) and television (Rasmussen et al., 2018),
and also mirror findings from research on active and restrictive mediation of cell phones
(Warren, 2017), restrictive mediation of the internet (Warren, 2017), and monitoring of
135
the Internet (Sonck et al., 2013). Thus, concern about potential harm is a rather consistent
motivator for parents to be involved with their child’s media activities and this finding
The exception to this finding was for summed monitoring, where negative attitudes
did not significantly impact parent behavior. This result could be due to the items that
on social media. The act of knowing passwords and passcodes can be a rather intrusive
behavior, especially for parents of teens who are trying to build trust and respect privacy
boundaries as children get older (Blackwell, Gardiner, & Schoenebeck, 2016; Vaterlaus et
al., 2015). Respect for privacy could be a stronger motive than negative attitudes about
mobile media use and activities when it comes to these types of monitoring behaviors.
About 60% of parents reported being friends with their child on social media for fun
and entertainment and about half (52%) reported being friends with their child on social
media for the purpose of monitoring their posts or activity. Being friends with one’s child
for fun and entertainment is a positively valenced activity and less likely to be motivated by
negative attitudes. By this logic though, negative attitudes should theoretically motivate
parents to friend their child on social media for the purpose of monitoring activity, perhaps
out of concern for the content their child is posting or who they are talking to (Vaterlaus et
al., 2014). It could be that there is a variable not measured in this study that acts as a
barrier to this behavior. For example, some research has reported that children create
multiple accounts or have different, more restrictive privacy settings for their parents than
other followers (Redden & Way, 2017). If this is the case, then regardless of their attitude
136
(positive or negative) parents may be less aware of their child’s authentic online activities
Positive attitudes about media effects and use (RQ1) positively predicted active
mediation, however, the coefficient was small (β = .08, p = .04) and no longer significant
when all variables were added to this model. Positive attitudes did not predict any other
mediation behaviors. For active mediation, it makes sense that when parents have a
positive attitude toward mobile devices and what kids can do with/on them, they could
have more topics to talk about with their child, hence more active mediation. It would also
follow that for restrictive mediation, having a positive outlook on the uses, activities, and
effects of mobile devices could lead to fewer restrictions, which has indeed been observed
among Dutch parents and restrictive mediation of social media use (Krcmar & Cingel,
2016). There is also some evidence for this explanation from research on television. Beyens
and Eggermont (2014) found that stronger positive parental attitudes toward TV predicted
using TV as a babysitter, which then predicted higher TV viewing among young children. I
did not measure teen mobile device use in this study, but it stands to reason that positive
attitudes toward mobile devices is associated with a more laid-back approach from parents
and, hence, null relationships with parental mediation. The same argument can be said for
monitoring behaviors: although the coefficients for this variable were not significant in this
study, they did trend negative for both summed and average monitoring, indicating that
It should also be noted that parents did not report strong positive attitudes toward
the use of mobile devices, which is another possible explanation for my null findings. The
overall average of positive attitudes was 3.59 (SD = 0.76; Table 2), which corresponds to a
137
tepid score with respect to the Likert response scale used for this measure 12. This relatively
weak attitude could have lowered the chances of observing a significant relationship
An alternative but parallel explanation is that positive attitudes are more important
for predicting mediation of specific activities rather than the platform of mobile devices. Put
another way, positive attitudes motivate activity-specific mediation but this relationship
does not hold when these activities are conducted on a mobile device. To this point, the
study conducted by Krcmar and Cingel (2016) found that positive attitudes toward
children’s use of social media positively predicted restrictive mediation and co-using;
however, Shin and Li (2017) did not find a relationship between positive attitudes and
Likewise, Warren (2017) found that positive attitudes had no impact on active mediation of
cell phones but they did significantly predict active mediation of the Internet. So, while
there is some evidence that shows positive attitudes influence mediation of specific
activities, I did not find that they were as influential or motivating for mediation of mobile
often included the “use of the Internet” but this was in reference to use on a mobile device.
The most obvious explanation for this difference is that mobile devices are simply
more difficult for parents to mediate than other platforms due to the mobility, privacy, and
freedom they afford users. Difficulty combined with a positive (albeit lukewarm) attitude
on their use and effects could explain a hands-off approach by parents. Future research
12Appendix B contains the items for this measure. The Likert response scale used was 1 = Strongly
disagree, 2 = Disagree, 3 = Somewhat agree, 4 = Moderately agree, 5 = Strongly agree.
138
should continue investigating differences in device-specific mediation for the same activity
Lastly, positive attitudes about mediation outcomes (H2) positively predicted all
mediation behaviors. In fact, this was the only antecedent belief that was significant for
every parental mediation dependent variable. Believing that one’s efforts to intervene in
teen’s mobile device use would be effective at preventing negative effects or behaviors is
clearly an important motivating factor for parents. This is an aspect of parental mediation
research which has been observed in previous parental research. Bleakley et al (2013)
strongest predictor for engaging in this behavior. Shin and Li (2017) also found that the
more parents believed they can influence their child’s use of mobile devices the more likely
they were to practice active and restrictive mediation. Moreover, this finding is consistent
with other TPB studies that have examined the attitude–behavior relationship for different
aspects of parenting behaviors like spanking (Taylor, Hamvas, Rice, Newman, & Dejong,
It is encouraging to find that this construct was a rather strong predictor and that it
was associated with all of the parental mediation behaviors. Reinforcing positive outcomes
of mediation should be echoed by other researchers and the relationship between these
attitudes and child outcomes ought to be further explored. Increasing positive attitudes
toward the outcome of mediation can also be incorporated into messages that directly
139
Subjective Norms (H3, H4)
Neither injunctive norms (H3) nor descriptive norms (H4) were significantly related
to any parental mediation behaviors. This finding is somewhat unexpected given that past
research has found associations between these norms and parental mediation. For
descriptive norms and intention to limit TV viewing (i.e. restrictive mediation); though, it
was weaker than the relationship between attitudes and intention to limit TV viewing.
Krcmar and Cingel (2016) found that parents who endorsed norms related to controlling
social media were more likely to use restrictive mediation with their children. There is also
digital games in the classroom, such that parents who believe teachers and experts endorse
the use of games in the classroom are more likely to support these ideas themselves and
believe they provide learning opportunities (Bourgonjon, Valcke, Soetaert, de Wever, &
Schellens, 2011).
A difference between these studies and the current one is the platform of interest,
which could be impacting perceived norms and thus the relationship between norms and
parental mediation. The average of injunctive norms (i.e. the extent to which parents
thought that other people believe they should be mediating) in my study was 4.39 (SD =
1.34, range = 1-7). This value is lower but comparable to the Bleakley et al (2013) study
which reported 4.91, 5.54, and 4.47 for spouse, teacher, and “other parents you know”
injunctive norms to limit TV viewing. Bleakley et al (2013) reported average values of 2.37,
2.38, and 2.25 for other parents, close friends, and family member’s descriptive norms (i.e.
the perceived behaviors of others) for limiting TV viewing. These values are considerably
140
lower than what I observed in this study, which was 4.07 (SD = 1.44). My results compare
more closely to a recently published investigation conducted by Shin and Kim (2019),
which looked at predictive relationships between TPB constructs and intentions to engage
values of 4.48 (SD = 0.72) and 3.59 (SD = 1.12) for injunctive norms from one’s spouse and
friends and descriptive norms of other parents. In their study, Shin and Kim (2019) found
that injunctive norms were significantly associated with active mediation and descriptive
norms weakly but significantly predicted both active and restrictive mediation of
smartphones.
One key difference between my study and the studies conducted by Shin and Kim
(2019) and Bleakley et al (2013) is that they measured intentions to mediate, whereas I
asked for parents to self-report their mediation behaviors. The TPB positions intentions as
the precursor to behavior; however, intentions do not always lead to behavioral action
(Ajzen, Brown, & Carvajal, 2004). In simple terms, people do not always end up doing
things that they intend to do, which would explain why Shin and Kim (2019) and Bleakley
et al (2013) found significant associations between norms and intentions, yet I did not find
significance between norms and behaviors. It could also be the case that an intervening
variable that was not measured in this study mediates the relationship between norms and
behavior, namely social pressure to conform (or motivation to comply) to norms (Ajzen,
1991). That is, parents in my study reported that others believe they should mediate
(injunctive norms) and that other parents with kids their child’s age mediate as well
(descriptive norms) but they may not have felt the need or motivation to conform to these
141
Another alternative explanation and difference between my study and others is the
proximity and specificity of groups that were asked for parents to keep in mind while
answering these items. My questions asked parents to think about (a) people in the U.S. in
general for injunctive norms and (b) parents with a kid your child’s age for descriptive
norms. Shin and Kim (2019) asked about parent’s “child, spouse/partner, and close
friends” for injunctive norms, and “typical parent[s] of an individual your child’s age” (p.
150). Bleakley et al (2013) asked about the parent’s spouse, their child’s teacher, and
“other parents you know” for injunctive norms and other parents, close friends, and family
member’s for descriptive norms. Perhaps if I specified a more distinctive (and potentially
more important) group to parents in this study, their motivation to comply with this group
could have been higher and consequently could have led to significant relations between
Ultimately, within the context of the TPB variables that I measured and analyzed, my
results demonstrated that other behavioral and control beliefs mattered more to parents
than perceptions of what others do and think. As Ajzen (1991) explains, the predictive
power of different aspects of the TPB will vary depending on the behavior, the sample of
study, and other contextual factors. In this study, it appeared that norms did not impact
142
Perceived Behavioral Control/Self-Efficacy (H5, H6, H7)
restrictive mediation, summed monitoring, and average monitoring 13. Thus, parents who
felt confident/in control of parental mediation strategies were more likely to consistently
enforce rules and restrictions for mobile device use (average restrictive mediation), kept
track of passwords and activities on their devices (summed monitoring), and consistently
monitored their activities on mobile devices (average monitoring). These findings are
smartphones (Hwang et al, 2017) and parent monitoring of mobile devices and the Internet
(Chang et al., 2019). PBC/SE was not associated with active mediation or summed
devices), which is in contrast to findings from previous research (Krcmar & Cingel, 2016;
Why did PBC/SE not predict active and summed restrictive mediation in this study?
First, it is important to note that PBC/SE did impact one aspect of restrictive mediation: for
parents who had rules and restrictions for mobile devices, PBC/SE to mediate positively
predicted how consistently parents reported enforcing their rules (i.e. average restrictive
mediation); however, PBC/SE was not an important factor for determining whether or not
A possible explanation for these null relationships is that keeping the two measures
of PBC and SE combined confounded their separate effect on the dependent variables. To
13PBC/SE was significant when entered into step two of the average monitoring regression (β = .09,
p = .04) but became marginally significant in the final step of this model after all other independent
variables were entered (β = .09, p = .06; Table 12).
143
test this explanation, I re-ran the regression predicting active mediation and entered the
individual PBC and SE variables (rather than the combined measure, which was merged to
improve measurement model parsimony). Neither PBC (β = .02, p = .76) nor SE (β = -.09, p
= .20) were significant predictors of active mediation. I also re-ran the regression
predicting summed restrictive mediation, entering the individual PBC and SE variables
rather than the combined measure, and again PBC (β = .09, p = .19) and SE (β = -.05, p = .43)
were not significant predictors of summed restrictive mediation. Separating the PBC and SE
variables did however reveal a ceiling effect for these measures, which is also a likely
contributing cause for my null findings. Although they were both reported on 5-point
Likert scales, the mean of the overall sample for PBC was 3.76 (SD = 0.93) and the mean for
SE was 3.84 (SD = 0.95). These values are already quite high and after missing data were
excluded listwise and cases with Mahalanobis distances were omitted, the means for these
variables were higher and their standard deviations were lower, further narrowing their
variance: for active mediation, the mean of PBC was 3.80 (SD = 0.87) and the mean of SE
was 3.89 (SD = 0.88); for summed restrictive mediation, the mean of PBC was 3.79 (SD =
0.87) and the mean of SE was 3.85 (SD = 0.89). Thus, because parents in this sample
reported rather high perceived control and high self-efficacy, the impact of these variables
Parent mobile device use (H6) significantly predicted active mediation 14 and
summed monitoring, indicating that the more parents use mobile devices themselves, the
more likely they were to have conversations with their teens and monitor aspects of their
14 Parent mobile device use was significant when entered in step two of the active mediation
regression (β = .08, p = .05) but was no longer significant in the final step of this model after all
other independent variables were entered (β = .05, p = .23; Table 8).
144
mobile devices. The null relationships between parent mobile device use and restrictive
relationship between these two constructs. The lack of significant findings for this variable
could be related to the behavior I assessed. In my study, parents were asked on a 5-point
Likert scale (1 = Never to 5 = Every day) about how often they used their mobile device for
six activities: job-related tasks, household tasks, social media, educational purposes,
leisure, and personal communication. I adapted this measure from Shin and Li (2017), who
asked parents to specify if they used devices for seven different tasks on a binary response
(yes/no) scale. In a separate set of questions, these researchers asked parents to estimate
the amount of time they spend on devices using a Likert scale from “no usage” to “4 hours
and above” (p. 10). They found that the amount of time spent using devices did not impact
parental mediation, but the more skills parents reported, the more likely they were to use
In a similar vein, Chang and colleagues (2019) asked parents to indicate if they were
able to do 11 tasks related to the Internet and mobile devices using a binary response
(yes/no) (e.g. “Do you know how to download apps? Do you know how to connect to a Wi-
Fi network from a smartphone?”; p. 27). Their findings also demonstrated that the more
mobile device skills that parents reported, the more likely they were to actively mediate
and monitor mobile devices and the Internet. The ages of children in Shin and Li (2017;
grades 1-6) and Chang et al (2019; 5th grade) studies were lower than in this study, which
is important to consider when comparing these findings; however, the broader takeaway
from these studies and mine could be that frequency of mobile device use (as I measured it)
145
matters less for parental mediation than simply being able to use mobile devices
summed restrictive mediation, summed monitoring 15, and average monitoring. In fact, this
variable was the strongest TPB predictor for active mediation and average monitoring and
significance when predicting average restrictive mediation in the step that it was added (β
= .09, p = .07) and in the final step of the regression model (β = .08, p = .10) but did not
reach p < .05. Thus, knowledge of teen media activities appears to impact the presence of
rules and restrictions parents have for mobile devices but does not influence the extent to
motivating factor for the majority of parental mediation behaviors. This variable is similar
to general parental involvement —knowing and doing activities with one’s child—which
mediation of video games (Martins et al., 2017), television (Warren, 2001; Warren et al.,
2002), and digital technologies (Shin & Li, 2017). However, to my knowledge, knowledge of
media activities itself has not been assessed as an antecedent to parental mediation of
mobile devices for teens. Efforts to increase parental mediation during the teen years
should incorporate messages that encourage parents to be (or become) aware of their
15Knowledge of teen media activities was significant when entered in step two of the summed
monitoring regression (β = .09, p = .04) but was no longer significant in the final step of this model
after all other independent variables were entered (β = .08, p = .08; Table 11).
146
Moderation Analyses (RQ2, RQ3, RQ4)
restrictive, and monitoring behaviors. Some direct associations were observed, such that
autonomy granting positively predicted active mediation, while regulation and connection
positively predicted average restrictive mediation. These findings are consistent with
previous research which has also shown direct relationships between these different
somewhat unexpected that parenting behaviors were not significantly associated with
more dependent variables because previous research has found that connection predicts
active and restrictive mediation of digital technologies (Shin & Li, 2017), regulation is
positively associated with active and restrictive mediation of media generally (Padilla-
Walker & Coyne, 2011; Padilla-Walker et al., 2012b), and autonomy granting has been
shown to negatively predict monitoring (Ghosh et al., 2018). On the other hand, several
studies have also observed null findings between parenting behaviors and active,
(Hwang et al., 2017; Shin & Li, 2017; Warren & Aloia, 2019).
Given that this is one of the first research endeavors to test the predictive nature of
the TPB and parenting behaviors for parental mediation of mobile devices, it is
understandable that there will be differences in research findings across studies. In fact,
these null findings are a novel contribution to this body of literature, as my results suggest
that the TPB constructs are more important in explaining parental mediation of mobile
devices than parenting behaviors. For all dependent variables, the block containing the TPB
147
constructs explained more variance than the block(s) containing parenting behaviors,
indicating that parenting behaviors did not account for additional variance in the
dependent variables above and beyond the TPB constructs and control variables. Put
another way, parenting behaviors did not enhance or diminish parent’s likelihood of
mediation is a goal for parent interventions, as parenting behaviors can be relatively stable
across the parenting years (Carrasco, Rodriguez, del Barrio, & Holgado, 2011; Loeber et al.,
relationships between the TPB constructs and active, restrictive, and monitoring behaviors.
Family communication quality did not significantly predict any of the parental mediation
behaviors and therefore did not demonstrate evidence of moderation. Similar to parenting
behaviors, it is likely the case that the different TPB constructs were more instrumental
outcomes, however, communication quality has been shown to be an influential factor. For
with greater disclosure of negative or risky online experiences (Wisniewski, Xu, Rosson, &
of children’s risky online experiences (Byrne et al., 2014). It has also been found that
associated with lower likelihood of problematic Internet use (Chng et al., 2015). Thus,
148
research indicates that communication quality appears to be a greater factor for children to
The last research question asked if barriers to parental mediation (RQ4) moderated
relationships between the TPB constructs and active, restrictive, and monitoring behaviors.
moderating relationships between negative attitudes about media effects and use, positive
attitudes about media effects and use, and parent mobile device use and active mediation.
When barriers were average or above average, negative attitudes were more strongly
related to active mediation. These results replicate findings from video game and television
research which have also demonstrated that negative attitudes galvanize parents to
When barriers were low, the association between positive attitudes about media
effects and use and active mediation was higher; and when barriers were low, the
relationship between parent mobile device use and active mediation was also higher. Put
simply, when parents do not feel hindered by certain functional (e.g. not knowing what to
say) and practical barriers (e.g. how long the conversations would take) they actively
mediate more.
Barriers were not significant in the models predicting restrictive mediation and
monitoring and thus did not moderate relationships between the TPB variables and these
mediation strategies. Though this variable significantly moderated some relationships for
active mediation, it is important to note that its mean was quite low (M = 2.27, SD = 0.81),
which corresponded to the barriers collectively having a “slight” impact on their ability to
enforce rules or have conversations with their children about mobile devices and the
149
Internet, per the Likert scale I used. The significant negative correlations between barriers
.20, p < .001), and average monitoring (r = -.11, p = .01) suggest that there is a relationship
occurring; however, the TPB variables were ultimately more influential and accounted for
more variance than these barriers. Future research should continue to investigate how
Beyond the TPB and family system variables, certain parent and child
characteristics were associated with parental mediation. Control variables that were
measured in this study included parent accessibility, family structure, parent and child age,
parent and child gender, parent and child ethnicity, parent education, household income,
and parent employment. Child age and parent age were significantly and negatively
associated with all dependent variables except for summed monitoring, which is consistent
with previous research (Shin, 2018; Shin & Kim, 2019). Parent gender significantly
predicted all dependent variables except for summed restrictive mediation, indicating that
mothers practiced more parental mediation than fathers. It is important to note that this
sample predominantly consisted of women (N women = 437, 74%; N men = 150, 26%) 16,
which surely impacted these findings; however, previous studies have shown that mothers
mediate more than fathers so this finding is not completely spurious (Chang et al., 2019;
Lauricella et al., 2016; Padilla-Walker & Coyne, 2011; Warren, 2017). Child gender did not
significantly predict any dependent variables, as other studies on parental mediation of TV,
digital technologies, and the Internet have also found (Shin, 2018; Shin & Li, 2017; Warren,
16 There were five missing values for this variable, therefore the total N adds to 587.
150
2017). Parent ethnicity was associated with active mediation, such that Non-White parents
parents, teens, and digital devices also observed this specific relationship; however, the
greater majority of research has not found associations between parent ethnicity and
mediation (Bleakley et al., 2013; Shin & Kim, 2019; Warren & Aloia, 2019). Parent
who were spent more waking hours around their children also reinforced their rules and
restrictions more frequently for mobile devices. While previous research has demonstrated
television (Warren, 2001, 2002), this is likely the first study which has found such
relationships for parental mediation of mobile devices. Relatedly, parent employment was
negatively associated with average restrictive mediation, demonstrating that higher levels
of employment (and potentially lower accessibility) were associated with less frequent
and income—did not influence any of the dependent variables. The nonsignificance
observed for these variables is likely due to the homogeneity of this sample. Specifically,
89.4% were the biological parents of the target child, 65% of households consisted of two
biological parents and 72.8% of individuals reported being married. A sample with more
diversity, or which specifically targets non-biological and/or divorced parents, could shed
There are mixed findings regarding the impact of parent education on parental
mediation of cell phones and smartphones. Some studies have found no relationship
151
between parent education and any form of mediation (Shin & Kim, 2019; Warren, 2017),
while others have shown that parents who have completed more education practice more
active and restrictive mediation of cell phones (Warren & Aloia, 2019) and less educated
parents practice more monitoring (Chang et al., 2019). Similarly, some research has
with lower income households practicing more monitoring of mobile devices (Lauricella et
al., 2016), while others have not observed this relationship (Shin, 2018; Warren, 2017). It
should be noted that the educational background of parents in this sample was higher than
recent national estimates. Specifically, 30.9% of individuals in the U.S. over the age of 25
have a bachelor’s degree or higher, while 58.4% of people in the same age bracket in this
study reported holding bachelor’s degree or higher (U.S. Census Bureau, 2010-2018). My
results add to the mixed findings for these demographic characteristics and more research
is necessary to understand if and how parent income and education impact their mediation
Limitations
There are several limitations to this study that should be acknowledged. The first
limitation concerns the sample of parents who responded to this survey on Mechanical
Turk. Although previous research has praised Mechanical Turk for its diversity and
generalizability (Buhrmester et al., 2011), parents who responded to this study were
mostly women and fit some aspects of a WEIRD population (White, Educated,
Industrialized, Rich, and Democratic) (Henrich, Heine, Norenzayan, 2010; Jordan &
also have higher digital literacy and skills than parents who do not work on MTurk or use
152
mobile devices less frequently (Stewart, Chandler, & Paolacci, 2017; Huang, Li, Chen, &
The use of a single reporter (i.e. one parent) is a second limitation in this study. That
is, parents in the same household may not engage in the same mediation behaviors or
agree on media rules for their children (e.g. Mares et al., 2018). Disagreements could
impact the consistency with which parental mediation is practiced in the home and how it
was self-reported in the survey. The target child’s perspective was also not taken into
account. Prior work has shown that parents and children differ in their reporting of
parental mediation, with parents reporting more frequent mediation than their children
(e.g. Nikken & Jansz, 2006; Sonck et al., 2013) and children reporting communication is
more difficult than parents (Byrne et al., 2014). Communication is inherently bi-directional
and the quality of it is a subjective judgement. The personal nature of some questions
relating to parenting, communication, and involvement in one’s child media use could have
garnered more socially desirable responses from parents. Additionally, family system
theory explains that parents and children have a bi-directional influence on each other and,
as such, child behaviors and characteristics that were not measured in this study have the
ability to impact parenting practices (Cox & Paley, 1997). For instance, adolescent self-
with parental mediation over time, such that higher adolescent self-regulation is associated
with sharper declines in restrictive mediation into the teen years (Padilla-Walker et al.,
2012b).
153
Third, the cross-sectional nature of this survey limits the discussion of causal and
directional inferences drawn from the results. For instance, my findings show that there is
a significant and strong relationship between knowledge of teen media activities and active
mediation but I cannot conclusively say that this knowledge causes or precedes active
mediation. It could very well be that conversations about media lead to more knowledge.
A final limitation to this study was the length of the survey, which was
approximately 127 questions, or 136 questions including the filter questions at the
beginning of the survey and all of the follow-up questions. Steps were taken to decrease the
likelihood of low quality data, such as pre-testing the survey, interspersing attention and
respondents, and counter-balancing the survey; however, these steps are not guaranteed to
produce perfect data. Realistically, it is likely that fatigue and repetition had some effect on
participant attentiveness.
Despite these limitations, several novel and valuable contributions can be gained
from this study. As the focus of this research investigation was on social and psychological
these findings. For interventions or media literacy campaigns aimed at increasing parental
mediation during the teen years, educating parents about positive outcomes of mediation
would be an excellent starting point. The findings from this study suggest that reinforcing
the idea that parent’s efforts are worthwhile and effective at preventing undesired effects
154
Encouraging knowledge of teen media activities is another behavior to target. This
type of knowledge was a significant and strong predictor of all parental mediation
behaviors except for average restrictive mediation. For parents of teens, increasing
knowledge of media activities may be challenging as kids at this age could reject questions
and parent actions (Livingstone, 2008; Madden et al., 2012). Indeed, high involvement in
teen media use is incongruous with autonomy seeking (Vaterlaus et al., 2015) and can
result in the opposite outcome of what was intended, namely further distance from parents
and increased engagement in the unwanted behavior (e.g. Nikken & de Graaf, 2013). An
agreement developed by parents and children is a possible compromise. The Family Media
easy to use website to do just this (HealthyChildren.org, n.d.; Reid Chassiakos et al., 2016).
The tool was designed to create an individualized plan for household media use and reflect
goals, values, and boundaries parents and children agree upon (Korioth, 2016). Wise words
to keep in mind for parents of teens while creating this agreement or any media rules are,
A third consistent and strong predictor to most parental mediation behaviors was
negative attitudes about media effects and use. This finding indicates that parents are
conscious of the potential negative consequences of mobile device use, worry about the
impact of media use on their own children (e.g. Robb, 2019), and these worries play an
important motivating role for parental mediation. Of the nine topics assessed in this
measure, the most concerning for parents in this study were teens spending too much time
online or on their mobile device (M = 3.61, SD = 1.17), over-sharing personal details of their
155
plethora of research-based and openly-accessible information directed at parents
regarding these topics and these publications offer simple recommendations to address
these concerns. For instance, Gabrielli and colleagues (2018) developed the acronym TECH
parenting—Talk, Educate, Co-view, and House Rules—to offer behavioral strategies and
conversation starters for managing media (Crist, 2018; Gabrielli, Marsch, & Tanski, 2018).
The American Academy of Pediatrics also recommends that parents converse, engage in
shared activities with their kids, and moreover recognize and accept that the occasionally
risky activities teens engage in are, simply put, “part of adolescent development” (American
Academy of Pediatrics, 2015). Recent research from Common Sense Media also highlights
that teens share these concerns, not only for their use but also for their parents’ (Robb,
2019; Timberg, 2019). To this point, parents should be encouraged to reflect on their own
device use and consider whether they are modeling behaviors they would want their
children to do.
While positive attitudes about media effects, norms and, to some extent, perceived
behavioral control were less vital in determining parental mediation in this study, future
research should continue to investigate their potential roles with different and more
diverse samples of parents. This limitation aside, the findings from this study show that
some of the TPB constructs are important factors in understanding predictors to parental
The findings from this study present several avenues for future research. First and
foremost, more research is needed on mobile devices and parental mediation given their
widespread, almost omnipresent, role in the daily life of adults and teenagers (Rideout,
156
2015; Rideout & Robb, 2018). Compared to the bodies of literature investigating parental
mediation of TV, video games, and movies (Nathanson, 2001a), research on mobile devices
is in its infancy. Indeed, some results in this study did not replicate findings from previous
research on other platforms (i.e. TV and video games) suggesting that mediation of mobile
opportunities) for parents. For example, previous research has demonstrated that
al., 2013) and positive attitudes toward video game playing predict active mediation
(Schaan & Melzer, 2015); however, I did not observe these same effects in this study. The
interactive, portable, private and personal nature of mobile devices present a new media
environment that has surely caused parents to adjust their mediation strategies. More
work is needed to understand these changes and future researchers should continue to
mediation outcomes and knowledge of teen media activities. These variables emerged as
strong positive predictors for almost all of the parental mediation outcomes and were
clearly important factors that impact parents. Replicating these results in different and
more diverse samples will increase the generalizability of the findings. Future research
should also look at the determinants of these variables, such as parent media literacy, past
behavior, and/or demographic characteristics. In turn, these could become target variables
Future research should also examine different areas of PBC/SE and how they impact
parental mediation of mobile devices. For example, Shin (2018) assessed different domains
157
of self-efficacy, including smartphone self-efficacy, parenting self-efficacy, and parental
mediation self-efficacy. The results showed that all three types of self-efficacy predicted
interaction was observed such that parents who were high in parenting self-efficacy
engaged in active mediation even when they reported low smartphone self-efficacy. This
between parenting behaviors, the TPB constructs and parental mediation. More research is
needed to clarify the relationships between PBC/SE, parenting, the TPB constructs, and
Different types of self-efficacy could also be explored in future studies. For example,
response efficacy has been demonstrated in previous research to predict active mediation
(e.g. Nathanson et al., 2002). Response efficacy refers to the perceived impact or
effectiveness of one’s behaviors (Hwang et al, 2017) and is similar to the measure of
attitudes about mediation outcomes used in this study, which significantly predicted active
mediation and summed restrictive mediation. The more parents perceived that their
child (i.e. positive attitudes about mediation outcomes) the more likely they were to engage
in active and restrictive mediation. Thus, it appears that active mediation of mobile devices
is more a function of parent’s attitudes that their efforts will make an impact on their child,
158
It was surprising to find so few direct relationships between parenting behaviors
and parental mediation, given that several other investigations have demonstrated that
these styles do impact parent behavior (Ghosh et al., 2018; Shin & Li, 2017). More research
behaviors in parental mediation of mobile devices. Future work could continue looking at
in mediation of mobile devices. Another route can look at how parenting behaviors are
combined with mediation and the impact these different styles have on adolescent
controlling, or inconsistent active and restrictive mediation and how these behaviors
impact problematic mobile device use and aggressive behaviors (Fikkers et al., 2017;
Over the course of writing this dissertation, I have had several informal, anecdotal
conversations with parents of teenagers to discuss their attitudes about mobile devices and
mediation. Child trust, respect for privacy, the amount of time kids use their mobile device,
and past mediation behaviors were consistent themes that emerged during these
reduced mediation over time. I did include these topics in my omnibus measure of parent
barriers to mediation; however, close examination of the effect they impose on parental
mediation was beyond the scope, time, and resources of this study but is an area ripe for
future investigations.
Though mobile device use and ownership among teens has grown substantially and
precipitously over the past 5 years (Twenge et al., 2018), research on parental mediation of
159
these technologies is still developing. This study was a step towards filling in this gap and
furthering our understanding of the social and psychological predictors that influence
parental mediation during the teen years. My hope is that these results inform prospective
research directions and media literacy campaigns that educate and empower parents to
160
References
Adachi, P. J. C., & Willoughby, T. (2013). More than just fun and games: The longitudinal
http://doi.org/10.1007/s10964-013-9913-9
Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human
Ajzen, I. (2005). Attitudes, personality, and behavior (2nd ed.). New York, NY: Open
https://ebookcentral.proquest.com/lib/iub-ebooks/detail.action?docID=287791
Ajzen, I. (2002a). Perceived behavioral control, self-efficacy, locus of control, and the theory
http://doi.org/10.1111/J.1559-1816.2002.Tb00236.X
from http://www.unix.oit.umass.edu/~aizen/pdf/tpb.measurement.pdf
Ajzen, I., Brown, T. C., & Carvajal, F. (2004). Explaining the discrepancy between intentions
and actions: The case of hypothetical bias in contingent valuation. Personality and
https://doi.org/10.1177/0146167204264079
Ajzen, I., & Fishbein, M. (2005). The influence of attitudes on behaviour. In D. Albarracın, B.
T.Johnson, & M. P. Zanna (Eds.), The handbook of attitudes (pp. 173–221). Mahwah,
161
Amazon Developer Guide. (2017). Working with HITs: Deadlines and expirations [Amazon
https://docs.aws.amazon.com/AWSMechTurk/latest/AWSMechanicalTurkRequest
er/Concepts_HITsArticle.html
Amazon Mechanical Turk. (2018, January 26). Improvements to the MTurk worker site
worker-site-a27d72ffc7c4
American Academy of Pediatrics. (2015). Children and media: Tips for parents. AAP News.
room/Pages/Children-And-Media-Tips-For-Parents.aspx
http://doi.org/10.1542/peds.2016-2592
An, S.K., & Lee, D. (2010). An integrated model of parental mediation: The effect of family
http://doi.org/10.1080/01292986.2010.496864
ftp://public.dhe.ibm.com/software/analytics/spss/documentation/statistics/25.0
en/amos/Manuals/IBM_SPSS_Amos_User_Guide.pdf
Austin, E. W., Bolls, P., Fujioka, Y., & Engelbertson, J. (1999). How and why parents take on
http://doi.org/10.1080/08838159909364483
162
Austin, E. W., Knaus, C., & Meneguelli, A. (1997). Who talks how to their kids about TV: A
http://doi.org/10.1080/08824099709388685
Bandura, A. (1998). Health promotion from the perspective of social cognitive theory.
http://doi.org/10.1080/08870449808407422
Bandura, A. (2006). Guide for constructing self-efficacy scales. In F. Pajares & T. Urdan
8624.1996.tb01915.x
http://doi.org/10.1177/0272431691111004
Beyens, I., & Eggermont, S. (2014). Putting young children in front of the television:
http://doi.org/10.1080/01463373.2013.860904
163
Beyens, I. ., Eggermont, S., & Nathanson, A. I. (2016). Understanding the relationship
longitudinal study of reciprocal patterns and the moderating role of maternal stress.
Blackwell, L., Gardiner, E., & Schoenebeck, S. (2016). Managing expectations: Technology
tensions among parents and teens. In Proceedings of the 19th ACM Conference on
Bleakley, A., Ellithorpe, M., & Romer, D. (2016). The role of parents in problematic Internet
http://doi.org/10.17645/mac.v4i3.523
Bleakley, A., Piotrowski, J. T., Hennessy, M., & Jordan, A. (2013). Predictors of parents’
intention to limit children’s television viewing. Journal of Public Health, 35(4), 525-
532. http://doi.org/10.1093/pubmed/fds104
Bourgonjon, J., Valcke, M., Soetaert, R., de Wever, B., & Schellens, T. (2011). Parental
1444. http://doi.org/10.1016/j.compedu.2010.12.012
boyd, D., & Hargittai, E. (2013). Connected and concerned: Variation in parents’ online
2866.POI332
164
Brehm, S. S., & Brehm, J. W. (1981). Psychological reactance: A theory of freedom and
Brosseau-Liard, P. E., & Savalei, V. (2014). Adjusting incremental fit indices for
https://doi.org/10.1080/00273171.2014.933697
Boer & J. Dunn (Eds.), Children’s sibling relationships: Developmental and clinical
Buhrmester, M., Kwang, T., & Gosling, S. D. (2011). Amazon’s Mechanical Turk: A new
Byrne, B. M. (2010). Structural equation modeling with AMOS: Basic concepts, applications,
and programming (2nd ed.). New York, NY: Taylor & Francis.
Byrne, S., Katz, S. J., Lee, T., Linz, D., & McIlrath, M. (2014). Peers, predators, and porn:
http://doi.org/10.1111/jcc4.12040
Byrne, S., & Lee, T. (2011). Toward predicting youth resistance to Internet risk prevention
http://doi.org/10.1080/08838151.2011.546255
Carrasco, M. A., Rodríguez, M. A., del Barrio, M. V., & Holgado, F. P. (2011). Relative and
165
children and adolescents. Psychological Reports, 108(1), 149–166.
https://doi.org/10.2466/02.09.10.21.pr0.108.1.149-166
Casler, K., Bickel, L., & Hackett, E. (2014). Separate but equal? A comparison of participants
and data gathered via Amazon’s MTurk, social media, and face-to-face behavioral
http://doi.org/10.1016/j.chb.2013.05.009
Chakroff, J. L., & Nathanson, A. I. (2008). Parent and school interventions: Mediation and
Ltd. http://doi.org/10.1002/9781444302752.ch24
Chang, F., Chiu, C., Chen, P., Chiang, J., Miao, N., Chuang, H., & Liu, S. (2019). Children’s use of
https://doi.org/10.1016/j.chb.2018.11.048
Chen, V. H. H., & Chng, G. S. (2016). Active and restrictive parental mediation over time:
Chen, L., & Shi, J. (2019). Reducing harm from media: A meta-analysis of parental
http://doi.org/10.1177/1077699018754908
Chng, G. S., Li, D., Liau, A. K., & Khoo, A. (2015). Moderating effects of the family
166
Cyberpsychology, Behavior, and Social Networking, 18(1), 30–36.
http://doi.org/10.1089/cyber.2014.0368
Clark, L. S. (2009). Digital media and the generation gap. Information, Communication &
Collier, K. M., Coyne, S. M., Rasmussen, E. E., Hawkins, A. J., Padilla-Walker, L. M., Erickson, S.
http://doi.org/10.1037/dev0000108
Common Sense Media. (n.d.). Common Sense Media: Our mission. Retrieved March 6, 2019
from https://www.commonsensemedia.org/about-us/our-mission
Connell, S. L., Lauricella, A. R., & Wartella, E. (2015). Parental co-use of media technology
with their young children in the USA. Journal of Children and Media, 9(1), 5–21.
http://doi.org/10.1080/17482798.2015.997440
9010.78.1.98
Cox, M. J., & Paley, B. (1997). Families as systems. Annual Review of Psychology, 48, 243–67.
Cox, M. J., & Paley, B. (2003). Understanding families as systems. Current Directions in
Coyne, S. M., Padilla-Walker, L. M., Stockdale, L., & Day, R. D. (2011). Game on... girls:
Associations between co-playing video games and adolescent behavioral and family
167
outcomes. Journal of Adolescent Health, 49(2), 160–165.
http://doi.org/10.1016/j.jadohealth.2010.11.249
Crist, C. (2018, June 28). TECH parenting style could help kids manage their media use.
kids-technology/tech-parenting-style-could-help-kids-manage-their-media-use-
idUSKBN1JO2RZ
Cronbach L.J. (1951). Coefficient alpha and the internal structure of tests. Psychometrika,
Davies, J. J., & Gentile, D. A. (2012). Responses to children’s media use in families with and
425. http://doi.org/10.1111/j.1741-3729.2012.00703.x
Dishion, T. J., & McMahon, R. J. (1998). Parental monitoring and the prevention of child and
https://doi.org/10.1023/A:1021800432380
Domoff, S. E., Radesky, J. S., Harrison, K., Riley, H., Lumeng, J. C., & Miller, A. L. (2019). A
naturalistic study of child and family screen media and mobile device use. Journal of
1275-1
http://doi.org/10.1080/17482798.2011.587147
168
Eastin, M. S., Greenberg, B. S., & Hosfchire, L. (2006). Parenting the Internet. Journal of
Erickson, L. B., Wisniewski, P., Xu, H., Carroll, J. M., Rosson, M. B., & Perkins, D. F. (2016).
http://doi.org/10.1002/asi
Evans, C. A., Jordan, A. B., & Horner, J. (2011). Only two hours?: A qualitative study of the
Field, A. (2013). Discovering statistics using IMB SPSS Statistics. Thousand Oaks, CA: SAGE
Publications.
Fikkers, K. M., Piotrowski, J. T., & Valkenburg, P. M. (2017). A matter of style? The
http://doi.org/10.1016/j.chb.2017.01.029
Fishbein, M., & Ajzen, I. (2010). Predicting and changing behavior: The reasoned action
Fisher, D., Hill, D. L., Grube, J. W., Bersamin, M. M., Walker, S., & Gruber, E. L. (2009).
Gabrielli, J., Marsch, L., & Tanski, S. (2018). TECH parenting to promote effective media
3718
169
Gentile, D. A., Nathanson, A. I., Rasmussen, E. E., Reimer, R. A., & Walsh, D. A. (2012). Do you
see what I see? Parent and child reports of parental monitoring of media. Family
Ghosh, A. K., Badillo-Urquiola, K. A., Xu, H., Rosson, M. B., Carroll, J. M., & Wisniewski, P.
Ghosh, A. K., Badillo-Urquiola, K., Rosson, M. B., Xu, H., Carroll, J., & Wisniewski, P. J. (2018,
apps on teens' mobile devices. Presented at ACM Conference on Human Factors (CHI)
Gilligan, C., Thompson, K., Bourke, J., Kypri, K., & Stockwell, T. I. M. (2014). “Everybody else
org.proxyiub.uits.iu.edu/10.15288/jsad.2014.75.908
Gold, J. (2015). Screen-smart parenting: How to find balance and benefit in your child’s use of
social media, apps, and digital devices. The Guilford Press: New York, NY.
Goodman, J. K., Cryder, C. E., & Cheema, A. (2013). Data collection in a flat world: The
Guilamo-Ramos, V., Jaccard, J., Dittus, P., & Collins, S. (2008). Parent-adolescent
170
communicate. Health Psychology, 27(6), 760–769.
http://doi.org/10.1037/a0013833
Hayduk, L., Cummings, G., Boadu, K., Pazderka-Robinson, H., & Boulianna, S. (2007).
Testing! testing! One, two, three – Testing the theory in structural equation models!
http://doi.org/10.1016/j.paid.2006.10.001
A regression based approach. New York, NY: The Guilford Press. Retrieved from
https://ebookcentral.proquest.com/lib/iub-ebooks/detail.action?docID=1186800
https://www.healthychildren.org/English/media/Pages/default.aspx
Hefner, D., Knop, K., Schmitt, S., & Vorderer, P. (2019). Rules? Role model? Relationship?
https://doi.org/10.1080/152132692018.1433544%0ARules?
Henrich, J., Heine, S. J., & Norenzayan, A. (2010). The weirdest people in the world?
Hiniker, A., Schoenebeck, S. Y., & Kientz, J. A. (2016, February). Not at the dinner table:
Huang, G., Li, X., Chen, W., & Straubhaar, J. D. (2018). Fall-behind parents? The influential
171
Behavioral Scientist, 62(9), 1186–1206.
https://doi.org/10.1177/0002764218773820
Hwang, Y., Choi, I., Yum, J.-Y., & Jeong, S.-H. (2017). Parental mediation regarding children’s
http://doi.org/10.1089/cyber.2016.0555
Hwang, Y., & Jeong, S.-H. (2015). Predictors of parental mediation regarding children’s
743. http://doi.org/10.1089/cyber.2015.0286
Retrieved from
https://www.ibm.com/support/knowledgecenter/en/SSLVMB_23.0.0/spss/base/s
yn_reliability_missing.html
Jeffrey, J., Whelan, J., Pirouz, D. M., & Snowdon, A. W. (2016). Boosting safety behaviour:
Descriptive norms encourage child booster seat usage amongst low involvement
http://doi.org/10.1016/j.aap.2016.03.006
Jordan, A. B. (2002). A family systems approach to examining the role of Internet in the
home. In J.Turow & A. L. Kavanaugh (Eds.), The wired homestead: An MIT press
sourcebook on the Internet and the family (pp. 141–160). Cambridge, MA: MIT Press.
Jordan, A. B., Hersey, J. C., McDivitt, J. A., & Heitzler, C. D. (2006). Reducing children’s
television viewing time: A qualitative study of parents and their children. Pediatrics,
172
Jordan, A., & Prendella, K. (2019). The invisible children of media research. Journal of
Kalmus, V., Blinka, L., & Ólafsson, K. (2015). Does it matter what mama says: Evaluating the
Katz, S. J., Lee, T., & Byrne, S. (2015). Predicting parent-child differences in perceptions of
how children use the internet for help with homework, identity development, and
http://doi.org/10.1080/08838151.2015.1093479
Kerr, M., & Stattin, H. (2000). What parents know, how they know it, and several forms of
1649.36.3.366
Khurana, A., Bleakley, A., Jordan, A. B., & Romer, D. (2015). The protective effects of
http://doi.org/10.1007/s10964-014-0242-4
Kline, R. B. (2011). Principles and practice of structural equation modeling (3rd ed). New
http://proxyiub.uits.iu.edu/login?url=https://search-ebscohost-
com.proxyiub.uits.iu.edu/login.aspx?direct=true&db=nlebk&AN=332476&site=ehos
t-live&scope=site
173
Korioth, T. (2016, October 21). Family Media Plan helps parents set boundaries for kids.
https://www.aappublications.org/news/2016/10/21/MediaParents102116
Krcmar, M., & Cingel, D. P. (2016). Examining two theoretical models predicting American
and Dutch parents’ mediation of adolescent social media use. Journal of Family
Lauricella, A. R., Cingel, D. P., Beaudoin-Ryan, L., Robb, M. B., Saphir, M., & Wartella, E. A.
(2016). The Common Sense census: Plugged in parents of teens and tweens. San
https://www.commonsensemedia.org/research/the-common- sense-census-
plugged-in-parents-of-tweens-and-teens-2016
Lauricella, A., Cingel, D., Blackwell, C., Wartella, E., & Conway, A. (2014). The mobile
generation: Youth and adolescent ownership and use of new media. Communication
Lee, S. J. (2013). Parental restrictive mediation of children’s Internet use: Effective for what
http://doi.org/10.1177/1461444812452412
Lenhart, A. (2015, April). Teens, social media, and technology overview. Washington, DC:
Pew Research Center’s Internet & American Life Project. Retrieved from
http://www.pewInternet.org/2015/04/09/teens-social-media-technology-2015/
Litman, L. (2018, August 10). Concerns about bots on Mechanical Turk: Problems and
mechanical-turk-problems-and-solutions
174
Livingstone, S. (2008). Taking risky opportunities in youthful content creation: Teenagers’
use of social networking sites for intimacy, privacy and self-expression. New Media
Livingstone, S., Haddon, L., Vincent, J., Mascheroni, G., & Ólafsson, K. (2014). Net children go
mobile: The UK report. London, UK: London School of Economics and Political
Livingstone, S., & Helsper, E. J. (2008). Parental mediation of children’s Internet use. Journal
http://doi.org/10.1080/08838150802437396
Livingstone, S., Ólafsson, K., Helsper, E. J., Lupiáñez-Villanueva, F., Veltri, G. A., & Folkvord, F.
(2017). Maximizing opportunities and minimizing risks for children online: The role
Loeber, R., Drinkwater, M., Yin, Y., Anderson, S. J., Schmidt, L. C., & Crawford, A. (2000).
Lucchetti, A. E., Powers, W. G., & Love, D. E. (2002). The empirical development of the Child
Parent Communication Apprehension Scale for use with young adults. Journal of
http://doi.org/10.1207/S15327698JFC0203_1
Lwin, M. O., Stanaland, A. J. S., & Miyazaki, A. D. (2008). Protecting children’s privacy online:
175
Madden, M., Cortesi, S., Gasser, U., Lenhart, A., & Duggan, M. (2012). Parents, teens, and
online privacy. Washington, DC: Pew Internet & American Life Project. Retrieved
from: http://pewInternet.org/Reports/2012/Teens-and-Privacy.aspx.
Manning, M. (2009). The effects of subjective norms on behaviour in the theory of planned
http://doi.org/10.1348/014466608X393136
Mares, M. L., Stephenson, L., Martins, N., & Nathanson, A. I. (2018). A house divided:
Parental disparity and conflict over media rules predict children’s outcomes.
http://doi.org/10.1016/j.chb.2017.12.009
Martins, N., Matthews, N. L., & Ratan, R. A. (2017). Playing by the rules: Parental mediation
http://doi.org/10.1177/0192513X15613822
Meeus, A., Eggermont, S., & Beullens, K. (2018). Constantly connected: The role of parental
https://doi.org/10.1093/hcr/hqy015
Napper, L. E., Hummer, J. F., Lac, A., & LaBrie, J. W. (2015). What are other parents saying?
176
Nathanson, A. (1999). Identifying and explaining the relationship between parental
http://doi.org/0803973233
Associates.
Nathanson, A. I. (2001b). Parent and child perspectives on the presence and meaning of
201–220. http://doi.org/10.1207/s15506878jobem4502
Nathanson, A. I. (2015). Media and the family: Reflections and future directions. Journal of
Nathanson, A. I., & Botta, R. A. (2003). Shaping the effects of television on adolescents’ body
304–331. http://doi.org/10.1177/0093650203252409
Nathanson, A. I., Eveland Jr., W. P., Park, H.-S., & Paul, B. (2002). Perceived media influence
http://doi.org/10.1207/s15506878jobem4603
Nielsen (2015). Grow and tell: As children age from toddlers to teens, their media palate
https://www.nielsen.com/us/en/insights/news/2015/grow-and-tell-as-
children-age-from-toddlers-to-teens-their-media-palate-changes.html
177
Nikken, P., & de Graaf, H. (2013). Reciprocal relationships between friends’ and parental
mediation of adolescents’ media use and their sexual attitudes and behavior. Journal
-9873-5
Nikken, P., & Jansz, J. (2006). Parental mediation of children’s video game playing: A
comparison of the reports by parents and children. Learning, Media and Technology,
Nikken, P., & Jansz, J. (2013). Developing scales to measure parental mediation of young
http://doi.org/10.1080/17439884.2013.782038
Nikken, P., & Opree, S. J. (2018). Guiding young children’s digital media use: SES-differences
in mediation concerns and competence. Journal of Child and Family Studies, 27(6),
1844–1857. http://doi.org/10.1007/s10826-018-1018-3
Opgenhaffen, M., Vandenbosch, L., Eggermont, S., & Frison, E. (2012). Parental mediation of
A latent growth curve analysis in early and middle adolescence. Journal of Children
Padilla-Walker, L. M., & Coyne, S. M. (2011). “Turn that thing off!” Parent and adolescent
http://doi.org/10.1016/j.adolescence.2010.09.002
Padilla-Walker, L. M., Coyne, S. M., & Collier, K. M. (2016). Longitudinal relations between
178
externalizing problems. Journal of Adolescence, 46, 86–97.
http://doi.org/10.1016/j.adolescence.2015.11.002
Padilla-Walker, L. M., Coyne, S. M., & Fraser, A. M. (2012a). Getting a high-speed family
connection: Associations between family media use and family connection. Family
Padilla-Walker, L. M., Coyne, S. M., Fraser, A. M., Dyer, W. J., & Yorgason, J. B. (2012b).
Parents and adolescents growing up in the digital age: Latent growth curve analysis
http://doi.org/10.1016/j.adolescence.2012.03.005
Padilla-Walker, L. M., Coyne, S. M., Kroff, S. L., & Memmott-Elison, M. K. (2018). The
protective role of parental media monitoring style from early to late adolescence.
017-0722-4
PEW. (2016). Parents, teens and digital monitoring. Washington, DC: Pew Internet and
http://www.pewInternet.org/2016/01/07/how-parents-monitor-their-teens-
digital-behavior/
Radesky, J. S., Eisenberg, S., Kistin, C. J., Gross, J., Block, G., Zuckerman, B., & Silverstein, M.
about child mobile technology use. Annals of Family Medicine, 14(6), 503–508.
http://doi.org/10.1370/afm.1976
179
Rasmussen, E. E., Coyne, S. M., Martins, N., & Densley, R. L. (2018). Parental mediation of US
Rasmussen, E. C., White, S. R., King, A. J., Holiday, S., & Densley, R. L. (2016). Predicting
parental mediation behaviors: The direct and indirect influence of parents’ critical
thinking about media and attitudes about parent-child interactions. Journal of Media
Redden, M. S., & Way, A. K. (2017). ‘Adults don’t understand’: Exploring how teens use
http://doi.org/10.1080/00909882.2016.1248465
Reid Chassiakos, Y., Radesky, J., Christakis, D., Moreno, M. A., & Cross, C. (2016). Children
http://doi.org/10.1542/peds.2016-2593
Rideout, V. (2007). Parents, children, & media. A Kaiser Family Foundation Survey. Menlo
https://www.kff.org/other/poll-finding/parents-children-media-a-kaiser-family-
foundation/
Rideout, V. (2015). The Common Sense census: Media use by tweens and teens. San Francisco,
https://www.commonsensemedia.org/research/the-common-sense-census- media-
use-by-tweens-and-teens
180
Rideout, V., & Robb, M. B. (2018). Social media, social life: Teens reveal their experiences. San
https://www.commonsensemedia.org/research/social-media-social-life-2018
Robb, M. B. (2019). The new normal: Parents, teens, screens, and sleep in the United States.
https://www.commonsensemedia.org/research/The-New-Normal-Parents-Teens-
and-Devices-Around-the-World
Robinson, C. C., Mandleco, B., Olsen, S. F., & Hart, C. H. (2001). The parenting styles and
Rodríguez-de-Dios, I., van Oosten, J. M. F., & Igartua, J. J. (2018). A study of the relationship
between parental mediation and adolescents’ digital skills, online risks and online
http://doi.org/10.1016/j.chb.2018.01.012
Rosen, L. D., Cheever, N. A., & Carrier, L. M. (2008). The association of parenting style and
child age with parental limit setting and adolescent MySpace behavior. Journal of
http://doi.org/10.1016/j.appdev.2008.07.005
181
Sanders, W., Parent, J., Forehand, R., & Breslend, N. L. (2016). The roles of general and
Sandstig, G., Johansson, B., & Ringsberg, K. (2013). The behavioral consequences of parents
6579/2013.07.007
Sasson, H., & Mesch, G. (2014). Parental mediation, peer norms and risky online behavior
http://doi.org/10.1016/j.chb.2013.12.025
Schaan, V. K., & Melzer, A. (2015). Parental mediation of children’s television and video
game use in Germany: Active and embedded in family processes. Journal of Children
Sheeran, P., & Orbell, S. (1999). Augmenting the theory of planned behavior: Roles for
children’s smartphone use in the United States. Journal of Children and Media, 12(4),
465–477. http://doi.org/10.1080/17482798.2018.1486331
Shin, W., & Huh, J. (2011). Parental mediation of teenagers’ video game playing:
http://doi.org/10.1177/1461444810388025
182
Shin, W., Huh, J., & Faber, R. J. (2012). Tweens’ online privacy risks and the role of parental
http://doi.org/10.1080/08838151.2012.732135
Shin, W., & Ismail, N. (2014). Exploring the role of parents and peers in young adolescents’
Shin, W., & Kim, H. K. (2019). What motivates parents to mediate children’s use of
https://doi.org/10.1080/08838151.2019.1576263
Shin, W., & Li, B. (2017). Parental mediation of children’s digital technology use in
http://doi.org/10.1080/17482798.2016.1203807
Simons, L. G., & Conger, R. D. (2007). Linking mother – father differences in parenting to a
Sonck, N., Nikken, P., & de Haan, J. (2013). Determinants of Internet mediation: A
comparison of the reports by Dutch parents and children. Journal of Children and
Spear, H. J., & Kulbok, P. (2004). Autonomy and adolescence: A concept analysis. Public
1209.2004.021208.x
183
Spera, C. (2005). A review of the relationship among parenting practices, parenting styles,
http://doi.org/10.1007/s10648-005-3950-1
Stern, S. R., & Odland, S. B. (2017). Constructing dysfunction: News coverage of teenagers
http://doi.org/10.1080/15205436.2016.1274765
Stewart, N., Chandler, J., & Paolacci, G. (2017). Crowdsourcing samples in cognitive science.
https://doi.org/10.1016/j.tics.2017.06.007
Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics (5th ed.). Boston, MA:
Pearson Education.
Tavakol, M., & Dennick, R. (2011). Making sense of Cronbach’s alpha. International Journal
Taylor, C. A., Hamvas, L., Rice, J., Newman, D. L., & Dejong, W. (2011). Perceived social
community sample of parents. Journal of Urban Health: Bulletin of the NeW York
Timberg, C. (2019, May 29). Many teens sleep with their phones, survey finds – just like
https://www.washingtonpost.com/business/technology/mimicking-their-parents-
many-teens-sleep-with-their-phones-survey-finds/2019/05/28/1bf2ee68-8188-
11e9-9a67-a687ca99fb3d_story.html?noredirect=on&utm_term=.2515165dae5c
184
Top, N. (2016). Socio-demographic differences in parental monitoring of children in late
Tsai, T. H., Wei, C. H., & Tsai, C. Y. (2014). Investigating parental intention of using internet
012-9750-z
Twenge, J. M., Martin, G. N., & Spitzberg, B. H. (2018). Trends in U.S. adolescents’ media use,
1976-2016: The rise of digital media, the decline of TV, and the (near) demise of
U.S. Census Bureau. (2010-2018). Quick Facts: National population estimates, July 1, 2018
https://www.census.gov/quickfacts/fact/table/US#
Vaala, S. E., & Bleakley, A. (2015). Monitoring, mediating, and modeling: Parental influence
on adolescent computer and Internet use in the United States. Journal of Children
Vahedi, Z., Sibalis, A., & Sutherland, J. E. (2018). Are media literacy interventions effective at
https://doi.org/10.1016/j.adolescence.2018.06.007
Valcke, M., Bonte, S., De Wever, B., & Rots, I. (2010). Internet parenting styles and the
impact on Internet use of primary school children. Computers & Education, 55(2),
454–464. http://doi.org/10.1016/j.compedu.2010.02.009
185
Valkenburg, P., Krcmar, M., Peeters, A. L., & Marseille, N. M. (1999). Developing a scale to
43(1), 52–66.http://doi.org/10.1080/08838159909364474
Valkenburg, P. M., & Peter, J. (2007). Internet communication and its relation to well-being:
http://doi.org/10.1080/15213260709336802
Valkenburg, P. M., Piotrowski, J. T., Hermanns, J., & de Leeuw, R. (2013). Developing and
http://doi.org/10.1111/hcre.12010
van Deursen, A. J. A. M., Bolle, C. L., Hegner, S. M., & Kommers, P. A. M. (2015). Modeling
habitual and addictive smartphone behavior: The role of smartphone usage types,
Vaterlaus, J. M., Beckert, T. E., & Bird, C. V. (2015). “‘At a certain age it’s not appropriate to
http://doi.org/10.1177/2167696815581277
Vaterlaus, J. M., Beckert, T. E., Tulane, S., & Bird, C. V. (2014). “They always ask what I’m
doing and who I’m talking to”: Parental mediation of adolescent interactive
http://doi.org/10.1080/01494929.2014.938795
186
Villarruel, A. M., Cherry, C. L., Cabriales, E. G., Ronis, D. L., & Zhou, Y. (2008). A parent
http://doi.org/10.1521/aeap.2008.20.5.371.A
Vossen, H. G. M., & Valkenburg, P. M. (2016). Do social media foster or curtail adolescents’
http://doi.org/10.1016/j.chb.2016.05.040
Wang, L. D-L., Lam, W. W. T., Wu, J., & Fielding, R. (2015). Psychosocial determinants of
Chinese parental HPV vaccination intention for adolescent girls: Preventing cervical
Warren, R. (2001). In words and deeds: Parental involvement and mediation of children’s
http://doi.org/10.1207/S15327698JFC0104_01
http://doi.org/10.1207/s15506878jobem4703
Warren, R. (2017). Multi-platform mediation: U.S. mothers’ and fathers’ mediation of teens’
http://doi.org/10.1080/17482798.2017.1349685
Warren, R., & Aloia, L. (2019). Parenting style, parental stress, and mediation of children’s
http://doi.org/10.1080/10570314.2019.1582087
187
Warren, R., Gerke, P., & Kelly, M. A. (2002). Is there enough time on the clock ? Parental
https://doi.org/10.1207/s15506878jobem4601_6
Wisenblit, J. Z., Priluck, R., & Pirog, S. F. (2013). The influence of parental styles on
http://doi.org/10.1108/JCM-02-2013-0465
Wisniewski, P., Jia, H., Xu, H., Rosson, M. B., & Carroll, J. M. (2015, March). “Preventative” vs.
“reactive”: How parental mediation influences teens’ social media privacy behaviors.
http://doi.org/10.1145/2675133.2675293
Wisniewski, P., Xu, H., Rosson, M. B., & Carroll, J. M. (2017). Parents just don’t understand:
Why teens don’t talk to parents about their online risk experiences. In Proceedings
of the 2017 ACM Conference on Computer Supported Cooperative Work and Social
https://doi.org/10.1145/2998181.2998236
Wright, P. J., Malamuth, N. M., & Donnerstein, E. (2012). Research on sex in the media and
Singer & J. L. Singer (Eds.), Handbook of children and the media (2nd ed., pp. 273-
188
Yardi, S., & Bruckman, A. (2011). Social and technical challenges in parenting teens’ social
media use. Presented at the ACM Conference on Human Factors (CHI) in Computing
189
APPENDIX A
H1: Negative attitudes about media effects and use will be positively associated with active,
H2: Parents with positive attitudes toward mediation outcomes will be more likely to
H3: Injunctive norms will be positively associated with active, restrictive, and monitoring
behaviors.
H4: Descriptive norms will be positively associated with active, restrictive, and monitoring
behaviors.
190
H5: Perceived behavioral control and self-efficacy will be positively associated with active,
H6: Parent mobile device use will be positively associated with active, restrictive, and
monitoring behaviors.
H7: Parental knowledge of teen media activities will be positively associated with active,
RQ1: Are parents’ positive attitudes about media effects and use associated with active,
Active mediation: Positive attitudes positively predicted active mediation but was
no longer significant after all variables were entered into the model.
Summed restrictive mediation: Positive attitudes did not predict summed
restrictive mediation.
Average restrictive mediation: Positive attitudes did not predict average
restrictive mediation.
Summed monitoring: Positive attitudes did not predict summed monitoring.
Average monitoring: Positive attitudes did not predict average monitoring.
191
RQ2: Do parenting behaviors (connection, regulation, autonomy granting) moderate the
relationship between the TPB constructs and active, restrictive, and monitoring behaviors?
RQ3: Does family communication quality moderate the relationship between the TPB
RQ4: Do perceived barriers to parent mediation moderate the relationship between the
Active mediation: Yes. Average and high perceived barriers combined with
negative attitudes about media effects and use is associated with higher active
mediation; low perceived barriers combined with positive attitudes about media
effects and use is associated with higher active mediation; and low perceived
barriers combined with parent mobile device use is associated with higher active
mediation.
Summed restrictive mediation: No.
Average restrictive mediation: No.
Summed monitoring: No.
Average monitoring: No.
192
APPENDIX B
Filter questions
The following questions will help determine if you are eligible to participate in this study.
Please note that answering NO to some of these questions does not necessarily disqualify
you from participating in the survey.
2. Do you own a personal mobile device (smartphone, tablet, laptop, Internet enabled MP3
player/iPod Touch)? YES/NO
(3a) IF YES, does your child have their own personal mobile device (smartphone,
tablet, laptop, Internet enabled MP3 player/iPod Touch)? YES/NO
(3b) Do you speak to your child between the ages of 13-18 at least four times per
week? YES/NO
Consent Message
We are interested in learning more about the role that parents play in their child’s use of
media. You will be asked to complete a questionnaire about your child’s access to and use
of media and your behaviors related to your child’s media use. You will also be asked a few
questions about your background, including education, income, and ethnicity. This
information is collected to determine the diversity of the sample and will be kept
completely confidential.
This survey will take about 20 minutes to complete. Would you like to participate in this
survey? YES/NO
193
Child Demographics
For the remainder of this survey, please think about the child in your family
between the ages 13 and 18 who has had the most recent birthday. Even if you have
multiple children between these ages, please think about this particular child when
answering these questions.
9. Which of the following devices does your child personally have - i.e. devices that are
not shared, belong to the family, or temporarily owned with limited functionality
(e.g. a tablet provided by child’s school)? …CHECK ALL THAT APPLY
Television set
Smartphone
Tablet (such as iPad, Galaxy Tab, Nexus, Kindle or similar product)
Video game console with TV-connectivity (such as X-Box, Wii, Playstation)
Streaming device (such as Google Chromecast, Roku, Amazon Fire Stick)
Desktop computer
Laptop computer
Internet enabled iPod Touch or MP3 player (besides a phone or tablet)
Portable game player (e.g. Nintendo DS)
194
Parental Mediation
The following questions ask about ways parents may act regarding their child’s use of
mobile devices. Mobile devices include smartphones, tablets, laptops, and Internet-enabled
MP3 players (e.g. iPod Touch, Zune). These questions represent a wide range of behaviors that
some parents do, and some do not do. For each question, please rate the extent to which you
do these things during a typical week with your child.
Active mediation
How often do you… NEVER, SOMETIMES BUT NOT OFTEN, SEVERAL TIMES A MONTH,
SEVERAL TIMES A WEEK, EVERY DAY
10. Talk to your child about the content (e.g. videos, apps, games) that he/she watches,
hears, or plays while using their mobile device?
11. Discuss what content is appropriate and inappropriate to view on mobile devices?
(e.g. violent or sexually explicit content)
12. Discuss what behaviors are appropriate and inappropriate while using their mobile
device? (e.g. talking to strangers, cyberbullying…)
13. Talk about the apps that your child uses on their mobile device?
14. Recommend apps or content to your child that you like or use?
15. Talk about the apps or content that your child likes and/or uses?
16. What personal information should or should not be shared online or on a mobile
device?
Restrictive mediation
17. Do you have rules about how much time your child can spend on their mobile
device? YES/NO
18. IF YES, How would you describe how strict or consistent you are with
enforcing rules about how much time your child can spend on their mobile
device? NOT AT ALL, A LITTLE, MODERATELY, VERY
19. Do you have rules about when your child can use their mobile device? (e.g. no
devices during dinner time, after they are done with homework) YES/NO
20. IF YES, How would you describe how strict or consistent you are with
enforcing rules about when your child can spend on their mobile device? NOT
AT ALL, A LITTLE, MODERATELY, VERY
21. Do you have rules about what apps your child can or cannot use on their mobile device?
YES/NO
22. IF YES, How would you describe how strict or consistent you are with
enforcing rules about what apps your child can or cannot use on their mobile
device? NOT AT ALL, A LITTLE, MODERATELY, VERY
195
23. Do you approve your child’s app purchases or downloads on their mobile devices?
YES/NO
24. IF YES, How would you describe how strict or consistent are you are with
enforcing approval of app purchases and downloads on your child’s mobile device?
NOT AT ALL, A LITTLE, MODERATELY, VERY
25. Do you use any parental controls or third-party apps (e.g. Net Nanny) on your child’s
mobile device that allow you to block them from using certain sites or apps? YES/NO
If YES, How often do you use these parental controls or third party apps? NEVER,
RARELY, SOMETIMES, OFTEN, ALWAYS
[AC1: Many people have unique hobbies. For example, crocheting, urban bike riding,
gardening, collecting antique coins, and building miniature ships. Please select exercising.
Are any of the options below a hobby that you do?]
Monitoring
Do you know your child’s passwords for…
26. Any of his/her social media and/or online accounts? YES, NO, MY CHILD DOESN’T HAVE
SOCIAL MEDIA
27. IF YES, which media accounts do you have their password for? FB, TWITTER,
INSTAGRAM, SNAPCHAT, PINTEREST, YOUTUBE, TUMBLR, EMAIL ACCOUNT,
OTHER
28. His/her mobile devices? YES, NO, THERE IS NO LOCK ON ANY OF THEIR DEVICES
29. Are you friends with your child on their social media profiles (e.g. Facebook, Twitter)
for fun/entertainment? YES, NO, MY CHILD DOESN’T HAVE SOCIAL MEDIA
(30) IF YES, on which social media sites are you friends with your child? FB,
TWITTER, INSTAGRAM, SNAPCHAT, PINTEREST, YOUTUBE, TUMBLR, EMAIL
ACCOUNT, OTHER
(31) IF YES, How often do you look at their social media profiles for
fun/entertainment? NEVER, SOMETIMES BUT NOT OFTEN, SEVERAL TIMES A
MONTH, SEVERAL TIMES A WEEK, EVERY DAY
32. Are you friends with your child on their social media profiles (e.g. Facebook, Twitter)
for the purpose of monitoring their posts or activity? YES, NO, MY CHILD DOESN’T HAVE
SOCIAL MEDIA
(33) IF YES, How often do you look at their social media profiles for the purpose of
monitoring their activity? NEVER, SOMETIMES BUT NOT OFTEN, SEVERAL TIMES A
MONTH, SEVERAL TIMES A WEEK, EVERY DAY
196
How often do you… NEVER, SOMETIMES BUT NOT OFTEN, SEVERAL TIMES A MONTH,
SEVERAL TIMES A WEEK, EVERY DAY
34. Ask your child what they are doing on their mobile device when they are with
you?
35. Keep an eye on the screen of your child’s mobile device when they are using it?
36. Check your child’s mobile device after they have used it to monitor what they are
doing (both immediately after or sometime later)?
Listed below are some common reasons that other parents of teenagers have said they don’t
discuss the Internet or mobile devices with their children.
To what extent do each of the following statements affect having rules or conversations with
your child about using mobile devices and the Internet? If you don’t feel any barriers to
talking with your child about mobile devices, please select “not at all”.
NOT AT ALL, SLIGHTLY, SOMEWHAT, MODERATELY, A GREAT DEAL
67. Covert monitoring of your child’s mobile devices (e.g. checking their web/app
history)
68. Overt monitoring of your child’s mobile devices (e.g. asking your child to show
you what they are doing on their device)
69. How likely do you think rules about mobile devices and the Internet would cause
conflict or disagreements with your child? (REVERSE)
[AC2: How are you feeling right now? Although we would like to know how you are
feeling, please select slightly pleased below so we know you are paying attention.]
198
Injunctive normative beliefs:
The next questions ask you to keep in mind people in United States in general.
Please move the slider to the point on the scale that corresponds closest to the way you feel
about the statement.
In general, people think that parents of teenagers should…
70. Have rules about when their child can use their mobile device
71. Have rules about what their child can/cannot do on their mobile device
72. Have conversations with their children about media content and apps they view
on their mobile devices
73. Monitor their child’s mobile device and Internet use in a covert manner (e.g.
checking their web/app history)
74. Monitor their child’s mobile device and Internet use in an overt manner (e.g. ask
child to show parent what they are doing on their device)
199
83. Monitor what my child does on the Internet and their mobile device in a covert
manner (e.g. checking their web/app history)
84. Monitor what my child does on the Internet and their mobile device in an overt
manner (e.g. ask child to show parent what they are doing on their device)
200
Family System Theory
Parenting Behaviors
Please answer the following questions by selecting the response that best describes
how you are with your child. NEVER, RARELY, SOMETIMES, OFTEN, ALWAYS
Connection (Responsiveness)
104. I encourage my child to talk about their troubles.
105. I am responsive to my child’s feelings and needs.
106. I give my child comfort and understanding when he or she is upset.
107. I praise my child when he or she is good.
108. I have warm and emotionally connected moments with my child.
Regulation (Demandingness)
109. I give my child reasons why rules should be obeyed.
110. I help my child to understand the impact of their behavior by encouraging them
to talk about the consequences of his/her own actions.
111. I explain the consequences of my child’s behavior.
112. I emphasize the reasons for rules.
113. I explain to my child how I feel about their good and bad behavior.
Autonomy granting
114. I show respect for my child’s opinions by encouraging them to express
themselves.
115. I encourage my child to freely express him/herself even if they are disagreeing
with me.
116. I allow my child to give input into our family rules.
117. I take my child’s desires into account before asking him/her to do something.
118. I take into account my child’s preferences when making plans for the family.
[AC3: Please select the option below that is not an animal. Thank you for doing this task.]
201
Parent accessibility
125. During a typical weekday, about how many waking hours per day are you at home or
other places with your family?
Less than an hour
1 hour
2 hours
3
4… (continued) More than 6
126. During a typical weekend day, about how many waking hours per day are you at home
or other places with your family?
Less than an hour
1 hour
2 hours
3
4… (continued) More than 6
Family structure
127. Which option below best describes your relationship with your child?
Birth/biological Parent
Step-parent
Adoptive Parent
Grandparent
Foster Parent
Other
128. If there are other adults in your household, what is their relationship to your child?
[select all options that apply]
Birth/biological Parent
Step-parent
Adoptive Parent
Grandparent
Foster Parent
There are no other adults living in my household
Parent Demographics
This information is being asked just to make sure we are talking to a representative group of
people.
202
130. What is your age?_________
133. What is the last grade or level of education that you have completed?
Less than 8th grade
Some high school
High school degree or GED
Some college (2 or 4 year)
College graduate (e.g. B.A., B.S.)
Master’s degree (e.g. M.A., M.S., M.B.A.)
Professional degree (e.g. J.D., M.D.)
Doctorate degree (e.g. PhD, EdD)
203
136. Which of the following devices do you have in your home? (select all that apply)
Television
Smartphone
Tablet
Video game console with TV-connectivity (e.g. Xbox, Wii, Playstation)
Streaming device (e.g. Google Chromecast, Roku, Amazon Fire Stick)
Desktop computer
Laptop computer
MP3 player
Portable game player (e.g. Nintendo DS)
Other (OPEN-ENDED)
204
APPENDIX C
Prior to data collection, the research instrument was pre-tested with a convenience
sample of parents. Demographic information was not collected for this convenience
sample. Pre-test data was collected from January 4, 2018 to February 13, 2018.
Excluding outliers 17 and incomplete responses, the average time to complete the pre-test
survey was 19.49 minutes (SD = 8.53). Cronbach’s alpha was computed for continuous
responses and Kuder-Richardson 20 (KR20) was computed for binary responses. With two
exceptions, Cronbach’s alphas for all measures ranged from .76 - .92 (Table A). Two
variables—parent mobile device use and average restrictive mediation—had alphas of .50
and 0.58, respectively. The parent mobile device use measure (6 items) asked how
frequently parents use their own personal mobile device for the following six activities:
job-related tasks, household tasks, social media, educational purposes, leisure, and
personal communication. The low alpha for this measure may be due to the low number of
items in this measure (Cortina, 1993; Tavakol & Dennick, 2011). It could also be due to the
different types of tasks that were assessed in these questions (e.g. job-related versus
scale, which is the degree to which “all items in a test measure the same concept or
construct and hence it is connected to the inter-relatedness of the items within the test”
(Tavakol & Dennick, 2011, p. 52). The items were moderately correlated in the expected
17To identify outliers, the duration to complete the survey (in minutes) was converted to z-scores.
Scores of 2 or more were considered outliers, resulting in the exclusion of four responses for the
calculation of this variable.
205
directions (e.g. job-related tasks and social media were negatively correlated); therefore,
questions to the initial questions about parent’s time and content restrictions. For example,
parents were asked if they “had rules about when your child can use their mobile device”
affirmatively, they were asked a follow-up question about how strict or consistent they are
with enforcing their rules or restrictions (average restrictive mediation). Given the nature
of this question, there were a number of missing values (i.e. parents answering “no” to the
first question did not answer the follow-up strictness question). The reliability analysis
command in SPSS uses listwise deletion for cases that have missing data and there is no
option to change this feature (IBM, n.d.). Consequently, SPSS omitted 59 cases with partial
data and analyzed reliability for the remaining 8 cases (parents who answered each follow-
up question). It is unlikely that this issue will be present with a larger data sample,
The questionnaire was counter-balanced such that parents either saw the parental
mediation (PM) questions first or the theory of planned behavior (TPB) questions first.
Thirty seven parents saw the PM questions first and 30 parents saw the TPB first. An order
effect for one variable was observed: Parents who saw the PM questions first (M = 4.70, SD
= .43) reported significantly higher regulation scores (parenting behavior) than parents
who saw the TPB questions first (M = 4.33, SD = .72), t(41) = 2.33, p = .03. Levene’s test
indicated unequal variance (F = 9.00, p= .02) so degrees of freedom were adjusted from 56
to 41.
206
Lastly, open-ended feedback was solicited on the last page of the pre-test survey
with the following message: “Thank you for completing this survey. If you have any
feedback about the survey flow, question wording, or anything else you noted, please
describe it below.” Responses to this question were used to adjust the research instrument
207
My child not being
interested in mobile Likert 1.21 (0.55)
devices
Not knowing what to say
Likert 1.28 (0.64)
to my child
Attitudes Toward
Behaviors
ATT_NA Likert Averaged 9 2.68 (0.96) 0.89
ATT_PA Likert Averaged 9 2.57 (0.64) 0.81
ATT_AMO Likert Averaged 5 2.68 (0.91) 0.80
Subjective Norms
SN_IN Likert Averaged 5 6.00 (2.30) 0.90
SN_DN Likert Averaged 5 4.91 (2.08) 0.92
PBC/SE
208
Video game 36
console 10
Streaming device 9
Desktop computer 58
Laptop computer 15
MP3 player 21
Portable game
player
Does parent know any
passwords for child’s
%
social media or online
accounts?
Yes
No 73
Child doesn’t use 22
SM or online 5
accounts
Average number of
known passwords for
1-7 1.82 (1.64)
child’s social media or
online accounts
Known account password %
Facebook 28
Twitter 8
Instagram 34
Snapchat 24
Pinterest 3
YouTube 15
Tumblr 0
Email 58
Other 9
Demographics n
Child gender b
Male 29
Female 37
Child grade
7th grade 12
8th 13
9th 12
10th 14
11th 6
12th 8
Other 2
209
Note. TPB = Theory of Planned Behavior; PM = Parental Mediation; a Kuder-Richardson 20
coefficient for binary variables; b One parent did not report their child’s gender; C Eight
parents who took the pre-test reported that their child did not have a personal mobile
device; however, this was a requirement for parents to participate in the full sample.
210
“The question about control was a bit
confusing. The spectrum includes "I have
control" and "I do not control." These
These instructions were
seem like different things to me. I feel like Self-efficacy
edited for clarity.
I have control, but I don't actively control.
I'm just not a very controlling parent (at
least in my mind, lol).”
211
APPENDIX D
Correlation Results
Pearson correlations for all variables are reported in Table 6. For visual clarity,
Table 7 contains correlations for the independent and dependent variables only.
Spearman’s rho (rs) correlations were calculated for ranked variables, including child
grade, household income, parent education, and employment (Field, 2013). Point biserial
correlations (rpb) were calculated for dichotomous variables, including child and parent
gender, child and parent ethnicity, marital status, and family structure (Field, 2013).
(yes/no questions), r(461) = .19, p < .001, positive attitudes about media effects and use,
r(501) = .13, p = .004, attitudes about mediation outcomes, r(501) = .21, p < .001, perceived
behavioral control, r(501) = .26, p < .001, self-efficacy, r(501) = .27, p < .001, parent mobile
device use, r(501) = .18, p < .001, connection, r(500) = .26, p < .001, regulation, r(501) =
.26, p < .001, autonomy granting, r(501) = .14, p < .001, family communication, r(501) =
.23, p < .001, and parent barriers to mediation, r(501) = -.27, p < .001.
associated with family communication quality, rpb(590) = -.10, p = .01. Household income
was significantly correlated with active mediation, rs(589) = -.09, p = .03, parent mobile
device use, rs(589) = .17, p < .001, and parent knowledge of teen media activities, rs(589) = -
.12, p = .003. Parent education was significantly associated with positive attitudes about
media effects and use, rs(586) = -.10, p = .02, attitudes about mediation outcomes, rs(586) =
-.08, p = .04, self-efficacy, rs(586) = -.12, p = .003, connection, rs(585) = -.17, p < .001,
regulation, rs(586) = -.14, p = .001, autonomy granting, rs(586) = -.09, p = .04, family
212
communication quality, rs(586) = -.13, p = .002, and barriers to parental mediation, rs (586)
= .13, p = .002.
Parent age was significantly correlated with active mediation, r(589) = -.21, p < .001,
summed restrictive mediation (yes/no questions), r(589) = -.22, p < .001, averaged
restrictive mediation (Likert response questions), r(53) = -.10, p = .02, averaged monitoring
scale (Likert response questions), r(589) = -.19, p < .001, descriptive normative beliefs,
r(587) = -.13, p = .002, parent mobile device use, r(589) = .10, p = .01, knowledge of teen
media activities, r(589) = -.20, p < .001, connection, r(588) = .20, p < .001, regulation,
r(589) = .19, p < .001, autonomy granting, r(589) = .10, p = .01, and family communication
quality, r(589) = .11, p = .01. Parent gender (coded as female = 1, male = 0) was
significantly associated with active mediation, rpb (585) = .10, p = .01, averaged restrictive
mediation, rpb (539) = .13, p = .003, summed monitoring scale, rpb (585) = .11, p = .01,
averaged monitoring scale, rpb (585) = .11, p = .01, injunctive normative beliefs, rpb (583) =
.11, p = .01, parent mobile device use, rpb (585) = .09, p = .03, parent knowledge of teen
media activities, rpb (585) = .09, p = .03, connection, rpb (584) = .10, p = .01, autonomy
granting, rpb (585) = .08, p = .05, family communication quality, rpb (585) = .11, p = .01.
significantly correlated with active mediation, rpb (585) = -.12, p = .004, negative attitudes
about media effects and use, rpb (585) = -.16, p < .001, positive attitudes about media effects
and use, rpb (585) = -.10, p = .02, and barriers to parental mediation, rpb (585) = -.13, p <
.001.
Child age was significantly associated with active mediation, r (587) = -.26, p < .001,
summed restrictive mediation, r (587) = -.35, p < .001, averaged restrictive mediation,
213
r(541) = -.13, p = .002, averaged monitoring scale, r(587) = -.29, p < .001, negative attitudes
about media effects and use, r(587) = -.09, p = .03, attitudes about mediation outcomes, r
(587) = -.21, p < .001, injunctive normative beliefs, r (585) = -.10, p =.02, descriptive
normative beliefs, r (585) = -.21, p < .001, perceived behavior control, r (587) = -.15, p <
.001, self-efficacy, r (587) = -.18, p < .001, parent knowledge of teen media activities, r (587)
= -.30, p < .001, and barriers to parental mediation, r (587) = .15, p < .001. Child gender
(coded as female = 1, male = 0) was significantly correlated with summed monitoring scale,
rpb (589) = .08, p = .05, perceived behavioral control, rpb (589) = .09, p = .03, self-efficacy, rpb
(589) = .12, p < .01, connection, rpb (588) = .15, p < .001, and family communication quality,
rpb (589) = .12, p = .003. Child grade was significantly correlated with active mediation,
rs(589) = -.24, p < .001, summed restrictive mediation, rs(589) = -.28, p < .001, averaged
restrictive mediation, rs(543) = -.12, p = .004, averaged monitoring, rs(589) = -.27, p < .001,
negative attitudes about media effects and use, rs(589) = -.09, p = .03, attitudes about
mediation outcomes, rs(589) = -.24, p < .001, descriptive normative beliefs, rs(587) = -.17, p
< .001, perceived behavioral control, rs(589) = -.19, p < .001, self-efficacy, rs(589) = -.20, p <
.001, knowledge of teen media activities, rs(589) = -.29, p < .001, and barriers to parental
significantly associated with active mediation, rpb (589) = -.09, p = .03, negative attitudes
about media effects and use, rpb (589) = -.11, p = .01, positive attitudes about media effects
and use, rpb (589) = -.09, p = .03, and barriers to parental mediation, rpb (589) = -.10, p = .02.
Parent employment was negatively correlated with averaged restrictive mediation, rs(543)
214
= -.14, p = .001. And lastly, family structure (coded as biological family = 1, non-biological
215
Mona Malacane Clay
EDUCATION
Martins, N., Mares, M.L., Malacane, M., & Peebles, A. (2016). Liked characters get a moral
pass: Young viewers’ evaluations of social and physical aggression in tween
sitcoms. Communication Research. https://doi.org/10.1177/0093650216644016
Martins, N., Malacane, M., Kraus, A., & Lewis, N. (2016). A content analysis of teen
parenthood in “Teen Mom” reality programming. Health Communication, 31(12),
1548-1556. https://doi.org/10.1080/10410236.2015.1089465
RESEARCH ASSISTANTSHIPS
WORK EXPERIENCE
Graduate Work-Study • Created the Media School research participant pool and
Indiana University facilitated recruitment for research studies.
August – January 2019 • Controlled the online sign-up system for research studies
~20 hours/week (Acuity Scheduling) and handled logistics for participants
and researchers.
• Managed SnapStream (TV monitoring and recording
software) requests for all IU Media School faculty.
• Attended weekly research meetings, recorded meeting
minutes, and distributed to listserv.
• Created informational digital “posters” for research being
conducted in the IU Media School.
• Inventoried and ordered supplies for psychophysiological
equipment.
Research Associate • Organized and oversaw recruitment of four focus groups
Dr. Robert Potter & (between 4 and 14 people/group) as commissioned by
WTIU Public Television WTIU Public Television and Broadcasting.
May – September 2017 • Assisted Principal Investigator (Robert Potter) in
~15 hours/week moderating focus group interviews.
• Managed data collection, entry, and analysis.
• Helped create and present final deliverables to WTIU.
Interim Program • Oversaw and facilitated 8-week summer camp for girls
Director ages 5-13.
Girls Incorporated of • Responsible for 60 girls and 10 staff members.
Monroe County • Implemented gender specific programming reflecting the
May – August 2018 Girls Inc. mission to inspire girls to be strong, smart, and
~55 hours/week bold.
• Coordinated developmentally appropriate programming
for the 2018 Fall semester.
Independent • Researched and developed pre- and post- evaluation
Consulting materials for an intervention program aimed at
The Harnisch strengthening young girls’ leadership skills.
Foundation • Target outcomes that were assessed in the evaluation
August – November included learning agility, empathy, collaboration, and self-
2016 awareness.
~15 hours/week • Advised on best practices for data collection and
evaluation.
REFEREED
Beckmeyer, J., & Malacane, M. (2018, April). Patterns of adolescent romantic involvement:
Associations with psychosocial adjustment. Poster presented at the Society for
Research on Adolescence (SRA) Biennial Meeting. Minneapolis, MN.
Malacane, M. (2016, November). Sexy and they know it: Messages about physical
appearance and reinforcement in adolescent TV programs. Paper presented at the
102nd Annual Meeting of the National Communication Association (NCA).
Philadelphia, PA.
Malacane, M., & Martins, N. (2015, November). A content analysis of sexual socialization
messages in popular programs produced for adolescents. Paper presented at the
101st Annual Meeting of the National Communication Association (NCA). Las Vegas,
NV.
Martins, N., Malacane, M., Kraus, A., & Lewis, N. (2015, May). Gritty or glamorous?: A
content analysis of teen parenthood in teen mom reality programming. Paper
presented at the 65th Annual International Communication Association (ICA)
Conference: San Juan, Puerto Rico.
Martins, N., Mares, M.L., Malacane, M., & Peebles, A. (2014, May). Liked characters get a
moral pass: Young viewers’ evaluations of social and physical aggression in tween
sitcoms. Paper presented at the 64th Annual International Communication
Association (ICA) Conference: Seattle, WA.
Funk, M., Malacane, M., Connolley, S., & Sherman, R. (2013, November). Does ‘woman’ mean
‘sex’ in the language of action movie theatrical trailers? Paper presented at the 99th
Annual Meeting of the National Communication Association (NCA): Washington, DC.
NON-REFEREED
Malacane, M. (2016, April). Cognitive barriers and parent mediation. Panel presentation
(Topic: Sex in the Media) at the 1st annual Media School Graduate Association
(MSGA) Conference: Bloomington, IN.
Martins, N., Lynch, T.M, Read, G., Kraus, A., Malacane, M., & Tompkins, J. (2015, April).
Social media, body image, and identity. Panel presentation on the Representation of
Girls in the Media and Toys sponsored by The White House Council on Women and
Girls and the U.S. Department of Education: Washington, DC.
Malacane, M., Goren, M. J., & Treadway, J.R. (2011, November). The influence of popular
adolescent television programs. Poster presented at the 32nd annual conference for
the Society of Southeastern Social Psychologists (SSSP): Johnson City, TN.
Ratti, H., Malacane, M., Treadway, J.R., & Goren, M. J. (2011, March). The increase in
sexualized messages in adolescent programs. Poster presented at the 34th annual
University of Georgia Convention of the Behavioral Sciences: Athens, GA.
• Fall 2015, invited guest lecture discussing Media and the Sexual Double Standard
(Course: Social Scientific Perspectives of Gender)
• Spring 2015, Media Arts and Sciences Speakers Series, The Media School, Women’s
role in action movie trailers: A content analysis examining sexual and agentic
portrayals, 1982-2013.
• Spring 2014, invited guest lecture discussing Sex in the Media (Course: Children and
Media)
• Spring 2013, invited guest lecture discussing Children as a Special Audience (Course:
Children and Media)
• Fall 2012, invited guest lecture discussing Sex and the Media (Course: Media &
Society)
• Fall 2012, invited guest lecture discussing Children and Media (Course: Media &
Society)
Indiana University
• $500, College of Arts and Sciences, Travel Award Grant (2016)
• $250, Top Student Paper Award (first author), National Communication Association
- Mass Communication Division (2016)
• $275, Student Travel Grant Recipient, National Communication Association -
Women’s Caucus (2015)
• Top Paper Award (co-author), International Communication Association (ICA),
Children, Adolescents, and Media Division (2014)
• $250, Student Travel Grant Recipient, National Communication Association -
Women’s Caucus (2013)
University of Georgia
• Best Symposium Presentation, 2011
• UGA Presidential Scholar, 2010-2011
• Dean’s List, 2007-2011
• HOPE Scholar, 2007-2011
• Multicultural Achievement Program Award Recipient, 2007-2009
SERVICE
Indiana University
• M.A. representative for the Media School Graduate Student Advisory Committee,
Spring 2014
• Writer for Telecommunications GradBlog, iutelecomgrad.wordpress.com, Fall
2013-Spring 2015
University of Georgia
• Active Minds, Research Liaison
• Psi Chi Honor Society for Psychology, Secretary
• Center for Undergraduate Research (CURO), Honors Researcher
• Forgotten Friends, Volunteer
PROFESSIONAL AFFILIATIONS
Reproduced with permission of copyright owner. Further reproduction prohibited without permission.