Children and Youth Services Review 108 (2020) 104584

Contents lists available at ScienceDirect

Children and Youth Services Review

journal homepage: www.elsevier.com/locate/childyouth

Pre-college STEM camps as developmental context: Mediational relations T

between gender, career decidedness, socioemotional development, and
engagement
⁎
Ryan J. Gagnona, , Alexandra Sandovalb
a
Clemson University, 276A Lehotsky Hall, Clemson, SC 29634, United States
b
Clemson University, 263 Lehotsky Hall, Clemson, SC 29634, United States

A R T I C LE I N FO A B S T R A C T

Keywords: One potential mechanism to mitigate the underrepresentation of females within STEM is the use of Out-of-
STEM School-Time programs and camps with supportive staff, instructors, and parallel orientation towards socio-
Adolescence emotional development. This study examines factors which may mediate the relation between gender and STEM
Career decidedness career aspiration within a pre-collegiate week-long STEM camp experience among a sample of 365 adolescent
Gender
campers. Campers were an average of 15.58 years old, primarily white (69.8%), and male (51.8%). The results of
Camp
Pre collegiate
the mediational structural equation model (SEM) suggested socioemotional development did not mediate the
relation between gender and career decidedness, and there was no statistical difference between male and fe-
male STEM career decidedness. The lack of effect could be a result of the systematic and ongoing emphasis
towards the development of STEM skills and aspiration within minority and underrepresented groups, specifi-
cally females.

1. Introduction 1.1. The gender gap within STEM

An aggregate of research examining the career fields of science,
technology, engineering, and math (STEM), indicates a longstanding
marginalization and absence of women from the field (Hughes,
Nzekew, & Molyneaux, 2013; Wang & Degol, 2017). While the causes
and consequences of the lack of representation of these groups within
STEM are seemingly well documented (Dasgupta & Stout, 2014; Wang
& Degol, 2013), exploration to establish strategies to address these
deficiencies remain ongoing. Specifically, one potential mechanism is
the use of informal science programs and semi-structured STEM courses
to encourage the continuous participation of unrepresented groups
across STEM career paths, with a specific focus on youth STEM pro-
grams to encourage them in their future careers (Beier et al., 2018;
Holmes, Redmond, Thomas, & High, 2012). However, the factors within
these programs that promote STEM career decidedness are less under-
stood (Beier et al., 2018; Liben & Coyle, 2014). Consequently, the
present study examines factors which may enhance STEM career de-
cidedness through the attendance of pre-collegiate STEM programs.
Below research exploring the gender gap within STEM careers is ex-
plored, literature examining potential contributors to STEM program
success are shared, and the current study is presented.
The gender gap within some aspects of the STEM career field is
potentially declining; in some areas, girls and young women are per-
forming equivalent to or better than their male counterparts in STEM
competencies (Klahr, Triona, & Williams, 2007). Indeed, Stoet and
Geary (2018) illustrated adolescent females tended to perform as good
or better than their male counterparts in science, mathematics, and
reading competencies. However, the overall representation of women
within actual STEM remains limited compared to men (Eccles, 2011).
For instance, about 50% of all science and engineering (S&E) degrees
have been awarded to women since the late 1990s, but approximately
64% of scientists or engineers working in the for-profit S&E sector are
male (National Science Foundation, 2017). While there are other po-
tential outlets for STEM employment (e.g., education, government), the
for-profit sector is the largest STEM employer, “regardless of gender,
race, ethnicity, or disability status” (p. 12, National Science
Foundation, 2017). The lack of balance between degrees awarded (50%
male: 50% female) and those actually employed in their degreed field
(64% male: 43% female) represents a challenge within STEM, pointing
to a potential issue with pipeline for young women and girls interested
in a STEM career (Stoet & Geary, 2018). Put differently, it appears the

⁎
Corresponding author.
E-mail addresses: rjgagno@clemson.edu (R.J. Gagnon), asandov@clemson.edu (A. Sandoval).

https://doi.org/10.1016/j.childyouth.2019.104584
Received 25 May 2019; Received in revised form 27 October 2019; Accepted 28 October 2019
Available online 20 November 2019
0190-7409/ © 2019 Elsevier Ltd. All rights reserved.
R.J. Gagnon and A. Sandoval
early competencies and requisite skills sets are present in both adoles-
cent females and males, but females tend to engage in STEM careers at
lower rates and in many cases forgo participating in a STEM centered
college major, representing a potential “exodus of talent among girls
and women who could otherwise become the next generation of sci-
entists, engineers, and creators of technology” (Dasgupta & Stout, 2014,
p. 22).
1.2. Out-of-school-time programming to enhance STEM career decidedness
The motivation to pursue (or not pursue) a STEM career may begin
in adolescence (Lubinski, 2010), where a variety of macro- and micro-
level factors can inhibit later gender equity across STEM fields (Sinclair,
Nilsson, & Cederskar, 2019; Stoet & Geary, 2018). More specifically,
evidence suggests gender disparities within STEM appear early in
adolescence, illustrating the need for “intervention” at phases earlier
than emerging adulthood (e.g., 18–24 years old) where STEM equity
programs are more frequent (Kang, Hense, Scheersoi, & Keinonen,
2019). For instance, out-of-school-time STEM programs that target
early to middle adolescents have been associated with the development
of STEM competencies and corresponding confidence in one’s “fit” or
desire towards a STEM career, in both mixed gender settings and female
only settings (Holmes et al., 2012; Hughes et al., 2013). In a review of
research examining the gender gap within the STEM field, Wang and
Degol (2017) noted career paths and preferences are often established
in adolescence, and gender frequently plays a meaningful role in this
decision-making process. Programs that cultivate aspiration towards
STEM-centered learning in adolescence could lead to meaningful de-
creases in the gender gap within STEM (Wegemer & Eccles, 2019).
Both within-school and out-of-school-time (OST) STEM programs
serving adolescents are generally orientated to enhance engagement
and motivation in STEM centered learning (Kong, Dabney, & Tai,
2014). In many cases these programs are designed to foster youth
motivations to engage in advanced STEM learning (i.e., college major
selection) and careers (Dasgupta & Stout, 2014; Jungert, Hubbart,
Dedic, & Rosenfield, 2019), especially for cohorts historically under-
represented in STEM (e.g., some ethnic/racial minorities, persons with
disabilities, females) (Eccles & Wang, 2016; Gasiewski, Eagan, Garcia,
Hurtado, & Chang, 2012). For example, an emerging body of evidence
suggests participation in these STEM centered experiences may enhance
female motivation to engage in further STEM education, conceptualized
by an improvement or shift in decidedness on a STEM centered career
(Mohr-Schroeder et al., 2014; Wang & Degol, 2017). However, research
in this area is relatively scant and based primarily upon college student
(i.e., 18–24 years old) samples rather than adolescents, illustrating a
gap in our understanding of how these programs may decrease gender
effects on STEM career decidedness at earlier phases (Kang et al., 2019).
In seeming parallel with the developing research exploring con-
tributors to the gender gap within STEM, there is an escalation in re-
search regarding the role of OST STEM programs on addressing this
gap. More specifically, research examining OST STEM programs has
shifted focus from solely on STEM program attendance and outcomes,
to an additional focus towards understanding the qualities of STEM
programs which best promote continuous STEM-centered learning and
career development (Kong et al., 2014; Wang & Degol, 2013). This shift
towards a more complete story of the “how and what” of OST STEM
programs and camps has increased understanding of mechanisms which
may magnify desirable program effects. For example, when a re-
sponsive and supportive instructor is present within a STEM course,
students tend to become more motivated and perform at higher effort
levels in proceeding STEM contexts and courses (Kramer et al., 2018;
Leon, Nunez, & Liew, 2015). Similarly, in the context of summer STEM
experiences, camp counselors can play a unique role in the improve-
ment of aspirations towards a STEM career (i.e., career decidedness).
This focus towards motivations was demonstrated by Fields (2009) who
indicated caring and supportive counselors enhanced STEM learning,
2
Children and Youth Services Review 108 (2020) 104584
appreciation, and career decidedness among adolescent campers. In a
broader study of OST programs, Tiffany, Exner-Cortens, and Eckenrode
(2012) illustrated similar findings, where the presence of caring and
supportive staff tended to enhance programmatic quality. Explicating
on the unique contexts within STEM programs, some research also
suggests the intentional exposure of youth to active scientists, may
narrow the gender gap within STEM, where female participants become
more likely to shift to a STEM career path after their participation in
these programs, due in part to role modelling behaviors demonstrated
by STEM instructors and scientists (Kang et al., 2019; Mohr-Schroeder
et al., 2014; Mostache, Matloff-Nieves, Kekelis, & Lawner, 2013).
Importantly, simple attendance of a STEM program alone is likely
insufficient to improve an adolescents STEM career decidedness levels,
rather quality of participation and engagement with a particular program
or setting tend to also represent necessary conditions for outcome
achievement (Roth, Malone, & Brooks-Gunn, 2010; Simpkins, Little, &
Weiss, 2004). Further, there is a breadth of literature linking caring and
supportive staff behaviors with greater rates of participant engagement
and corresponding achievement of desirable program outcomes in the
broader sphere of adolescent program research (Simpkins, 2015;
Tiffany, Exner-Cortens, & Eckenrode, 2013) and to a lesser degree
within STEM centered programs (Beier et al., 2018; Kramer et al., 2016;
Leon et al., 2015). The links between supportive staff behaviors, camper
engagement, and camper outcomes illustrate how quality STEM
summer programs (e.g., science camps) can lead to greater engagement
with STEM coursework, greater aspirations towards STEM college ma-
jors, career interests, and improved self-confidence among males and
females (Dieker, Grillo, & Ramlakhan, 2012; Fields, 2009; Kong et al.,
2014).
1.3. Socioemotional skill development within OST STEM programs
Although underexamined within the context of STEM program re-
search, the development of socioemotional skills and self-determinate
behaviors (e.g., autonomy, relatedness, competence) in parallel with
STEM career decidedness represents an additional conduit to under-
stand how to best serve and cultivate the next generation of STEM
professionals (Beier et al., 2018; Jungert et al., 2019; Kramer et al.,
2016; Wilson et al., 2015). Indeed, some research suggests programs
that also target and foster socioemotional skills are more likely to en-
hance STEM career decidedness and motivations to engage in current
and future STEM education (Falk et al., 2016; Guay, Senecal, Gauthier,
& Fernet, 2003; Leon et al., 2015). For instance, programs emphasizing
the development and/or enhancement of friendship skills, self-esteem,
and communication abilities can facilitate a sense of belonging to a
community interested in STEM and thus enhance motivation to more
readily involve oneself in later STEM-centric opportunities (Eccles,
2011; Sinclair et al., 2019). This engagement and satisfaction with
course materials and activities has been associated with greater rates of
aspiration towards STEM and academic performance (Vedder-Weiss &
Fortus, 2011; Wilson et al., 2015). For instance, in an examination of
the relationship between gender and motivation within STEM, Solanki
and Xu (2018) demonstrated when female undergraduate students de-
velop greater levels of motivation, they are generally more engaged
with thier STEM course work.
Pre-collegiate STEM summer camps may offer one potential context
to develop these socioemotional and motivational skills in parallel with
STEM career aspirations and career decidedness (Bevan, Petric, &
Wilkinson, 2014; Kang et al., 2019). Indeed, research examining career
path development programs among adolescents suggests levels of ca-
reer decidedness are often intertwined with positive youth development
and wellbeing (Hirschi, 2009; Lubinski and Persson-Benbow, 2006). At
a broader level, research examining camp as a context for positive
youth development suggests engaged participation can improve youth
communication skills, attitude, self-regulation, exploration and curi-
osity, cooperation skills, and personal responsibility (Garst & Gagnon,
R.J. Gagnon and A. Sandoval
2016; Nugent, Barker, Grandgennett, & Adamchuk, 2010; Sibthorp,
Bialeschki, Morgan, & Browne, 2013), and the development of these
skills within quality STEM programming may lead to greater aspiration
towards a STEM career and thus career decidedness (London, Pastor,
Servon, Rosner, & Wallace, 2010; Pace-Marshall, McGee, McLaren, &
Veal, 2011), regardless of participant gender (Myers, Jahn, Gailliard, &
Stoltzfus, 2011).
2. The present study
Given the ongoing and systematic allocation of resources towards
enhancing representation of females in STEM careers, the examination
of potential factors which may enhance or explain a female deciding on
a STEM career remains critical (Krishnamurthi, Bevan, Rinehart, &
Coulon, 2013; National Science Foundation, 2017). As such, this study
examines contextual factors which may explain the relations between
gender and career decidedness resulting from a pre-collegiate STEM
camp experience. Specifically, the mediational influences of camp
counselor behaviors, STEM course instructor behaviors, camp structure
engagement, and five dimensions of youth reported socioemotional skill
development (YRSESD) were examined, to determine how they may
account for the relationship between gender, career decidedness and
career undecidedness (see also Fig. 1). It is hypothesized that suppor-
tive course instructor behaviors (H1), supportive camp counselor be-
haviors (H2), camp structure engagement (H3), and YRSESD (H4) will
mediate the relationship between camper gender and STEM career
decidedness. Further, it is hypothesized that supportive course in-
structor behaviors (H5), supportive camp counselor behaviors (H6),
camp structure engagement (H7), and YRSESD (H8) will mediate the
relationship between camper gender and STEM career undecidedness.
Fig. 1. Conceptual Model of Potential Mediational Processes. Paths reduced for parsimony of presentation.
3
Children and Youth Services Review 108 (2020) 104584
3. Method
3.1. Participants
The 365 study participants were an average of 15.58 years old
(SD = 1.12 years) and primarily identified as male (n = 189, 51.8%);
with 47.1% (n = 172) identifying as female, and one respondent
identifying as non-binary. Participants primarily identified as white
(n = 254, 69.8%), with the remainder identifying as either African-
American (n = 68, 18.7%), Asian origin (n = 19; 5.2%), Hispanic or
Latino origin (n = 12, 3.3%), or multiple race (n = 11, 3%). Post-hoc
power analyses to determine the minimum necessary sample for testing
the study measurement model and hypotheses indicate the sample of
365 respondents was sufficient for testing of the measurement and
structural models [(average between factor r = 0.01, α = 0.05,
λ = 39.22) (R2 ≥ 0.08, p ≤ 0.05, λ = 30.78, α = 0.05, 10 predictor
variables)] (Cohen, Cohen, West, & Aiken, 2003).
3.2. Procedures
As part of a larger study exploring the potential benefits of re-
sidential summer camps and out-of-school-time experiences for ado-
lescents, the study data were collected from adolescents attending a
pre-collegiate STEM camp program. Specifically, a purposive sampling
technique was utilized, where the population of interest was asked to
participate in the study (Shadish, Cook, & Campbell, 2002). At the end
of their one-week experience, campers at the study site were asked to
respond to a paper questionnaire provided by a member of the camp
administration team, who was trained in the constructs underpinning
the items within the questionnaire and how to address common ques-
tions which may arise. Of a possible 424 respondents, 390 opted to
participate in the study, indicating a 91.98% response rate. As noted in
greater detail in the data processing and results sections below, 25
R.J. Gagnon and A. Sandoval
respondents were removed from the data set after data processing,
leading to a final overall sample of 365 (post data processing response
rate = 86.08%).
The camp is marketed as a pre-collegiate STEM experience, where
campers (e.g., aged 13–17) participate in an experiential STEM course,
with active college professors. Campers engage with professors in au-
thentic experiments and coding sessions tailored to their interests, fa-
cilitating closer relationships between camper and professor; im-
portantly, courses are designed with low instructor to camper ratios
(i.e., 1:12–15), to reflect this aim. Each week long experience has
specific tracks for a camper, including: bioengineering (i.e., cell culture,
bioinstrumentation), veterinary medicine (livestock anatomy, parasite
identification), civil engineering (i.e., structural design, building failure
testing), autonomous vehicle engineering (i.e., robotics development
and coding), audio engineering (developing and testing music record-
ings), and built environment design (i.e., energy efficient building de-
velopment, computer animated design). Across the one-week experi-
ence, campers stay on-site in campus residence halls, with current
university students acting as counselors to guide and mentor campers as
they potentially progress towards college and a STEM career.
To assess how/if campers were oriented towards a STEM career,
campers were asked to identify their desired college major, following
the format of Wegemer and Eccles (2019). Of total respondents, 35
indicated “don’t know” or left the question blank. Of valid responses,
305 indicated a STEM-centered major (e.g., engineering, biochemistry,
zoology, astrophysics, computer science) and 25 indicated either hu-
manities (e.g., history, creative writing, journalism, political science) or
a business centered major (e.g., entrepreneurship, sports management).
3.3. Measures
3.3.1. Youth reported socioemotional skill development
To assess potential changes in socioemotional skills resulting from
the camp experience, a modified version of the Parental Perceptions of
Developmental Outcomes (PPDO) scale was utilized (Gagnon & Garst,
2019; Garst & Gagnon, 2016). Specifically, the items comprising the
PPDO were adapted from a parental perspective (e.g., As a result of
camp…my child participates in new learning experiences) to a child per-
spective (e.g., As a result of my camp experience…I participate in new
learning experiences). Within the present study the YRSESD (Youth Re-
ported Socioemotional Skill Development) is comprised of five factors
and 21-items for the present investigation: (1) communication skills
(e.g., I participate more in discussions; 4-items), (2) responsibility (e.g., I
take responsibility for my own actions; 5-items), (3) self-regulation (e.g., I
manage disappointment well; 4-items), (4) attitude (e.g., I am more
hopeful about the future; 4-items), and (5) exploration (e.g., I seek chal-
lenges beyond my comfort zone; 4-items). Respondents were asked to
reflect on how their experience at the camp changed them rating the
items on a 1 (strongly disagree) to 5 (strongly agree) Likert format, where
higher scores indicate improvement in the corresponding factor. The
YRSESD scale has demonstrated acceptable levels of internal con-
sistency across the five factors in past research (α = 0.855–0.938) and
in the present study (α = 0.884–0.918).
3.3.2. Career (Un)Decidedness
To assess aspirations towards a STEM career, a modified version of
the goal decidedness/commitment scale (Hirschi, 2009) was adapted to
reflect two potential dimensions. First, career decidedness assessed the
degree to which adolescent beliefs about a potential career path match
with their academic interests (e.g., I see how my interests can become a
career). Second, career undecidedness, assessed how adolescents’ con-
fidence about a potential career may have changed or declined (e.g., I
am confused about a future career). When responding to these items,
campers were asked to reflect on how their experience at camp changed
their desired career path (i.e., college major). The items were measured
on a 1 (strongly disagree) to 5 (strongly agree) Likert style scale, where
4
Children and Youth Services Review 108 (2020) 104584
higher scores indicate better understanding of what career path the
respondent would like to pursue (i.e., career decidedness) or greater
confusion about a potential career path (i.e., career undecidedness).
The scale exhibited acceptable levels of internal consistency in past
research (α = 0.86) and the present study (α = 0.771; 0.637).
3.3.3. Engagement and support
To measure engagement, support, and involvement with the camp,
three factors were assessed, based in part upon the Tiffany-Eckenrode
Program Participation Scale (TEPPS) (Tiffany et al., 2012). Specifically,
the personal development subscale from the TEPPS was adapted for the
present study to assess two potential mechanisms of personal devel-
opment: supportive camp counselor behaviors (e.g., My counselor took a
personal interest in me, 5-items) supportive STEM course instructor be-
haviors (e.g., My instructor took personal interest in me; 5-items). In past
research (α = 0.82) and the present study (α = 0.849; 0.836) the
subscales have exhibited acceptable levels of internal consistency. In
addition to these measures, three generic items historically utilized at
the program site to assess participant engagement with camp activities
and structure were employed, (e.g., The camp activities were fun; 3-
items), which also demonstrated an acceptable level of internal con-
sistency within the present study (α = 0.814). All items were measured
on a 1 (strongly disagree) to 5 (strongly agree) Likert style scale, where
higher scores indicate greater rates of satisfaction and comfort.
3.4. Data processing and analyses
Prior to testing of the measurement model and examination of the
study hypotheses, the data were screened for outliers and normality
within EQS 6.3 software. First, the data were examined for multivariate
kurtosis (i.e., multivariate kurtosis estimate 500 greater than other
cases in data set; Byrne, 2006). The results of this examination in-
dicated 15 cases were contributing to nonnormality within the data set
and were correspondingly removed from additional analyses, leading to
a sample of 375. Second, the data were explored for normality; the
results of which indicated the data were non-normal in their dis-
tributive properties (e.g., normalized estimate = 54.606). As such,
robust estimation techniques were utilized to account for skewness in
the data set in proceeding analyses and in identification of fit indices
[e.g., Satorra-Bentler chi-square (S/Bχ2); Bentler, 2006]. Next the data
were examined for missing data issues. Due to evidence of non-nor-
mality, a robust version of Little (1988) test of MCAR (Missing Com-
pletely at Random) was utilized to examine the data for systematic
causes of missingness with a p > .05 criterion (Bentler, 2006). The
results of this analysis, as part of testing of the measurement model, the
MCAR criterion was not achieved [mean & variance adjusted
χ2(1 3 0) = 221.303, p < .001], suggesting the data were not MCAR.
However, given the low rate of missing data (95.73% of cases had no
missing data; no variable/item had more than 0.53% missing values), a
Full Information Maximum Likelihood (FIML) technique was utilized to
simulate missing values in measurement model testing and an Ex-
pectation Maximization (EM) technique was utilized to impute missing
values in the structural models (Bentler, 2006; Byrne, 2006).
After data processing, the analyses in the present study consisted of
two phases, first a confirmatory factor analysis (CFA) was conducted to
examine the measurement properties of the scale, and two structural
equation models (SEM) were utilized to test the study hypotheses. To
assess if acceptable model fit was present within the measurement and
structural models, the Comparative Fit Indices (CFI) and Non-Normed
Fit Indices (N-NFI) were utilized, where levels closer to 1.0 indicate
“better” model fit (e.g., CFI = 0.900). Additionally, the Root Mean
Squared Error Approximation (RMSEA) and its 90% confidence inter-
vals were utilized to assess model parsimony (e.g., unnecessary paths),
where levels closer to zero indicate better RMSEA scores [e.g.,
RMSEA = 0.060 (90%, CI 0.049 to 0.069)] (Byrne, 2006). As part of
the CFA, a combination of goodness-of-fit indices were utilized.
R.J. Gagnon and A. Sandoval
Specifically, Cronbach’s Alphas (α) were examined to determine the
internal consistency of each factor, with levels above 0.600 typically
indicating acceptability. Further, the factor loadings (λ), for each item
were assessed, with loadings greater than 0.500 typically suggesting
acceptability (Brown, 2015). Finally, to assess that each factor was
accounting for more unique variance than error, Average Variance
Extracted (AVE) scores were utilized, where values greater than 0.500
generally suggest acceptability (Kline, 2016). The discriminant validity
of the scale was assessed utilizing between factor correlations levels,
where levels greater than 0.800 suggest the factors may be sharing
more variance than they are uniquely accounting for (Byrne, 2006). To
further substantiate discriminant validity, √AVE was also utilized,
where the √AVE should be greater than the between factor correlation
levels of each corresponding factor (i.e., √AVE > r). Importantly,
measures of acceptable fit and validity were also conditioned upon past
studies utilizing the factors and items within the current study, and fit
indices were not “rule-bound” by arbitrary cutoff levels (e.g., “Golden
Rules,” see also Marsh, Hau, & Wen, 2004).
To test the study hypotheses, two mediational models were ex-
amined. First, a partially mediated model with the direct paths from the
study independent variable (IV), camper gender, to the dependent
variables (DV), career decidedness and undecidedness. Second, a fully
mediated model with no direct paths between the IV (camper gender)
and DVs (career decidedness and career undecidedness). In both
models, camper gender was dummy coded (Male = 0; Female = 1) for
analysis of direct and mediational effects, where when a score (re-
gression coefficient, β) is higher, this higher β indicates a positive effect
of female gender on the outcome of interest. To determine if media-
tional effects were possible, the S/Bχ2 levels for the partially and fully
mediated models were tested utilizing the S/Bχ2 difference test, where
a nonsignificant result (p > .05) in combination with a nonsignificant
path from the IV to DVs indicates the potential for mediation (Byrne,
2006). Further, to test the significance and magnitude of potential
mediational effects (i.e., indirect effects), the Sobel test was utilized
(Hayes, 2013).
4. Results
As part of the measurement analysis, the data were again screened
for highly non-normal cases, which indicated an additional ten outliers
within the data set and were removed from additional analyses, leading
to a final sample of 365. Correspondingly, the CFA results indicated
acceptable model fit: [S/Bχ2(6 9 5) = 1185.227, p ≤ 0.001, N-
NFI = 0.926, CFI = 0.934, RMSEA = 0.043 (90%, CI 0.039 to 0.047)].
The scale exhibited acceptable convergent validity as illustrated in
Table 1 with all factor loadings greater than 0.500, all internal con-
sistency levels greater than 0.700, and all AVE levels greater than
0.500, with the exception of the career undecidedness factor. While this
factor exhibited acceptable internal consistency, one-item (I want to
study different subjects) indicated a poor loading (λ = 0.452). Inspection
of the covariance matrices and LaGrange Multiplier (LM) test results to
realign or remove this factor did not indicate meaningful improvement
to model fit (Byrne, 2006). As such both the item and corresponding
factor was retained. Additionally, the evidence presented in Table 2
demonstrates the discriminant validity of the scale, where the majority
of √AVE levels were greater than corresponding between factor corre-
lations, with the exception of the exploration and attitude factors.
Specifically, their √AVE levels (0.814; 0.835) were slightly lower than
the exploration/attitude correlation levels (r = 0.846) and self-reg-
ulation/attitude levels (r = 0.829).
However, these differences were not meaningful enough to suggest
the need for model modification (e.g., √AVE minus r < 0.100), so the
model was retained (Byrne, 2006; Farrell, 2010). Furthermore, six of
the ten correlations comprising the socioemotional measures (i.e., F6 to
F10 in Table 2) exhibited higher than desirable correlation levels (e.g.,
r ≥ 0.800). However, similar high shared variance has been
5
Children and Youth Services Review 108 (2020) 104584
demonstrated in past studies utilizing these subscales, with differential
findings and effect sizes across factors within the same overall struc-
tural models, supporting their unshared/unique variance and suitability
for SEM (e.g., Gagnon & Garst, 2019).
Given the evidence of acceptable measurement properties, two
models were produced to examine the study hypotheses: (1) a partially
mediated model, with the direct path from the IV (gender) to the DVs
(career decidedness and undecidedness) included, and (2) a fully
mediated model, with the direct path from gender (IV) to the career
decidedness (DV) and undecidedness removed (DV). As part of the
examination of the SEM model fit, the LaGrange Multiplier (LM) fit
indices were examined (Byrne, 2006). The LM test statistics suggested
an item within the self-regulation factor was sharing error variance
with an item in the attitude factor (denoted by *** in Table 1), in ad-
dition to two items within the supportive camp counselor behaviors
factor (denoted by * in Table 1), and two items within the responsibility
factor (denoted by ** by in Table 1). As such the covariance matrices
and LM respecification indices were explored, which did not provide
sufficient evidence to re specify or remove the items without engaging
in data-driven (rather than theory-driven) modelling. Thus, given the
relative strength of factor loadings and acceptable levels of model fit,
the items were retained in their prespecified factors with their errors
covaried.
Both the partially mediated [S/Bχ2(7 4 1) = 1368.092, p ≤ 0.001,
N-NFI = 0.905, CFI = 0.914, RMSEA = 0.048 (90%, CI 0.044 to
0.052)] and fully mediated [S/Bχ2(7 4 3) = 1369.646, p ≤ 0.001, N-
NFI = 0.905, CFI = 0.914, RMSEA = 0.048 (90%, CI 0.044 to 0.052)]
models exhibited no substantive differences between their respective fit
indices. As noted earlier, the factors comprising the five socioemotional
factors exhibited high between factor correlations (e.g., r < 0.700) in
the CFA. To mitigate the high between factor correlations the dis-
turbance terms across the five factors were covaried for both the partial
and full mediation models. Within the partially mediated model, gender
had no significant direct effect on career decidedness (β = −0.055,
SE = 0.062, p = .284) or career undecidedness (β = 0.040,
SE = 0.052, p = .532). Further, the results of the S/Bχ2 difference test
indicated no statistically significant difference between the partially
and fully mediated models [ΔS/Bχ2(2) = 1.385, p = .501]. Taken to-
gether, the results of these analyses indicated the possibility of med-
iation, and thus appropriateness of testing for mediational effects. As
such the hypotheses (H1-H8) were tested utilizing the Sobel test in the
fully mediated model to determine if significant mediation was present.
As illustrated in Table 3, the Sobel test results indicate no significant
mediational (i.e., indirect) effects were present within the model, de-
monstrating a lack of support for the study hypotheses (H1 – H8).
However, while tertiary to the study purpose, there were significant
direct effects of gender on exploration score (β = 0.106, SE = 0.078,
p = .049), suggesting females in the sample were more curious about
new experiences than their male peers as a result of the STEM camp.
Further, the results indicated supportive camp counselor behaviors
(β = 0.182, SE = 0.050, p = 0.002), supportive course instructor be-
haviors (β = 0.338, SE = 0.055, p ≤ 0.001), and camp structure en-
gagement (β = 0.161, SE = 0.052, p = .006), all positively influenced
career decidedness score. Additionally, attitude both positively influ-
enced career decidedness score (β = 0.433, SE = 0.122, p = .009), and
negatively influenced career undecidedness score (β = −0.464,
SE = 0.217, p = .033).
5. Discussion
This study examined contextual factors which may explain the re-
lation between gender and career decidedness resulting from a pre-
collegiate STEM camp experience. Given the heavy dedication of re-
search dollars and resources towards developing pathways for groups
underrepresented in STEM, understanding how pathways such as
summer camp experiences enhance aspiration and representation of
R.J. Gagnon and A. Sandoval Children and Youth Services Review 108 (2020) 104584

Table 1
Descriptive and Confirmatory Statistics.
Factor/Item M◊ (SD) λ α AVE

Career Decidedness 0.771 0.538

I better understand what I want to do for a job. 3.962 (0.858) 0.775
I better understand of what I can do in the future. 4.144 (0.821) 0.810
I see how my interests can become a career. 4.204 (0.770) 0.599

Career Undecidedness 0.637 0.402

I want to change my career. 2.203 (0.985) 0.509
I am confused about a future career. 2.314 (1.117) 0.862
I want to study different subjects. 3.067 (1.122) 0.452

Supportive Camp Counselor Behaviors 0.849 0.538

My counselor shared their knowledge of college. * 4.265 (0.811) 0.694
My counselor provided guidance about college. * 4.075 (0.896) 0.678
My counselor was enthusiastic. 4.572 (0.624) 0.609
My counselor took personal interest in me. 4.174 (0.873) 0.830
My counselor valued my opinion. 4.242 (0.801) 0.831

Camp Structure Engagement 0.814 0.597

The daily camp structure was appropriate. 4.047 (0.924) 0.696
The program provided a variety of activities. 4.248 (0.826) 0.825
The camp activities were fun. 4.171 (0.834) 0.792

Supportive Instructor Behaviors 0.836 0.511

The instructor was prepared. 4.429 (0.777) 0.708
My instructor provided guidance about college. 4.160 (0.883) 0.826
My instructor took personal interest in me. 3.885 (1.067) 0.716
The instructor allowed me to gain new knowledge. 4.426 (0.861) 0.717
The courses were challenging. 3.626 (1.106) 0.589

Communication 0.903 0.702

I participate more in discussions. 3.568 (0.868) 0.810
I communicate better with others. 3.661 (0.874) 0.876
I better share thoughts and ideas verbally. 3.599 (0.908) 0.888
I ask questions. 3.616 (0.920) 0.774

Responsibility 0.918 0.693

I take responsibility for my own actions. ** 3.892 (0.901) 0.845
I take care of my own things. ** 4.027 (0.902) 0.789
I share in work responsibilities. 3.877 (0.884) 0.813
I follow through when asked to do something. 4.057 (0.878) 0.854
I better follow directions. 3.845 (0.926) 0.861

Self-Regulation 0.884 0.662

I better handle success and failure. 3.755 (0.947) 0.849
I manage disappointment well. 3.624 (0.956) 0.863
I deal effectively with conflict. 3.713 (0.946) 0.861
I don’t get frustrated easily. *** 3.473 (1.032) 0.667

Attitude 0.915 0.731

I have a good mental attitude. *** 3.869 (0.920) 0.871
I have a generally “positive” view on life. 3.948 (0.937) 0.894
I am more hopeful about the future. 4.080 (0.930) 0.816
I show a positive attitude when around others. 3.974 (0.912) 0.837

Exploration 0.901 0.698

I participate in new learning experiences. 4.109 (0.848) 0.854
I am curious about new topics and subjects. 4.226 (0.810) 0.762
I seek challenges beyond my comfort zone. 4.010 (0.898) 0.850
I am willing to try new experiences. 4.237 (0.820) 0.872

Note: ◊ Means (M) are based upon Expectation Maximization (EM) Values; λ: standardized coefficient (factor loading); AVE: Average Variance Extracted; α:
Cronbach’s alpha; *, **, ***: indicates covariance between corresponding items.

such groups remains an important area of inquiry (Krishnamurthi et al.,
2013). Within the present study, the mediational analyses indicated a
lack of support for the eight study hypotheses. Importantly, as de-
monstrated in the partially mediated model, there was no significant
direct effect of camper gender on career decidedness or career un-
decidedness score. The lack of direct effects in this model does indicate
mediational processes could be in place if there is support for differ-
entiation in STEM career decidedness (or undecidedness) due to camper
gender. However, in the current study, there was no evidence of in-
direct (i.e., mediational) effects utilizing the eight potential mediator
factors. Specifically, the results indicated a lack of significant media-
tional effect of supportive course instructor behaviors on the relation-
ship between camper gender, career decidedness (H1), and career
undecidedness (H5). Similarly, there was no significant mediational
effect of supportive camp counselor behaviors on the relationship be-
tween career decidedness (H2), career undecidedness (H6), and camper
gender.
While some in-school (Leon et al., 2015; Solanki & Xu, 2018) and
out-of-school (Fields, 2009) research suggests instructor and camp
counselor behaviors and traits should enhance the effect of STEM
programs on career decidedness, this was not supported in the current
study. The lack of differentiation seems to reflect literature suggesting
negligible to nonexistent differences in STEM career decidedness, per-
sistence, or aspiration due to participant gender regardless of instructor
behaviors (Chow & Salmela-Aro, 2011; Falk et al., 2016; Jungert et al.,
2019; Price, 2010; Solanki & Xu, 2018). There was also a lack of support

6
R.J. Gagnon and A. Sandoval Children and Youth Services Review 108 (2020) 104584

Table 2
Between Factor Correlations.
Factor F1 F2 F3 F4 F5 F6 F7 F8 F9 F10

F1. Career Decidedness 0.733

F2. Career Undecidedness −0.509 0.634
F3. S.C.C.B. 0.449 −0.189 0.733
F4. C.S.E. 0.463 −0.188 0.440 0.772
F5. S.C.I.B. 0.581 −0.231 0.363 0.353 0.715
F6. Communication 0.523 −0.237 0.497 0.475 0.437 0.838
F7. Responsibility 0.566 −0.288 0.392 0.461 0.448 0.822 0.832
F8. Self-Regulation 0.554 −0.232 0.417 0.456 0.426 0.776 0.831 0.814
F9. Attitude 0.620 −0.308 0.376 0.462 0.438 0.786 0.852 0.829 0.855
F10. Exploration 0.606 −0.235 0.386 0.439 0.486 0.720 0.804 0.740 0.846 0.835

Note: All factor correlations p ≤ 0.05; Bold indicates √AVE; S.C.C.B. = Supportive Camp Counselor Behaviors; C.S.E. = Camp Structure Engagement;
S.C.I.B. = Supportive Course Instructor Behaviors.

Table 3 for the mediating effect of camp structure engagement on the re-
Strength and Significance of Direct and Indirect Effects and Sobel Tests. lationship between camper gender, career decidedness (H3), and career
β SE p-value
undecidedness (H7). While some research indicates the relationship
between gender and career decidedness may be explained in part by
Gender to Mediating Variable participant engagement (Vedder-Weiss & Fortus, 2011; Wilson et al.,
Gender → S.C.C.B. 0.075 0.077 0.178 2015), this effect was not demonstrated within the present study. In-
Gender → C.S.E. 0.006 0.077 0.910
Gender → S.C.I.B. 0.102 0.081 0.066
deed, similar results have been illustrated in studies of college students,
Gender → Communication 0.031 0.087 0.573 where the relationship between gender and STEM aspiration was not
Gender → Responsibility 0.032 0.087 0.554 influenced by engagement with the course activities or with the in-
Gender → Self-Regulation 0.037 0.089 0.490 structors themselves (Gasiewski et al., 2012).
Gender → Attitude 0.003 0.088 0.960
The lack of support for hypotheses H4 and H8 suggested these so-
Gender → Exploration 0.106 0.078 0.049
cioemotional skills (i.e., communication, responsibility, self-regulation,
Mediating Variable to Career Decidedness
attitude, and exploration) did not account for the relationship between
S.C.C.B. → C.D. 0.182 0.050 0.001*
C.S.E. → C.D. 0.161 0.052 0.006
gender, career decidedness (H4), and career undecidedness (H8). The
S.C.I.B. → C.D. 0.338 0.055 0.001* development of these skills was a desired outcome of the organization
Communication → C.D. −0.193 0.083 0.079 facilitating the camp in addition to the fostering of STEM career de-
Responsibility → C.D. 0.019 0.120 0.904 cidedness. Examination of the distributions of the item means within
Self-Regulation → C.D. 0.066 0.089 0.585
the five factors suggests greater variation across these items than those
Attitude → C.D. 0.433 0.122 0.009
Exploration → C.D. 0.138 0.108 0.287 comprising the career decidedness factor. Given the high between
factor correlations across these factors, the lack of an effect across all
Sobel Test (Mediating Effect on Career Decidedness)
Gender → S.C.C.B. → C.D. 0.014● 0.015 0.346 five factors was somewhat unsurprising, despite evidence of the sub-
Gender → C.S.E. → C.D. 0.001● 0.012 0.937 scales producing differential effect sizes in previous studies (Gagnon &
Gender → S.C.I.B. → C.D. 0.034● 0.028 0.217 Garst, 2019). However, some research does suggest this lack of effect
Gender → Communication → C.D. −0.006● 0.017 0.724 may be increasingly normative within adolescent samples. For instance,
Gender → Responsibility → C.D. 0.001● 0.004 0.884
the work of Beier et al. (2018) and Falk et al. (2016) both suggested
Gender → Self-Regulation → C.D. 0.002● 0.007 0.716
Gender → Attitude → C.D. 0.001● 0.038 0.972 STEM interest scores did not meaningfully differentiate due to gender
Gender → Exploration → C.D. 0.015● 0.016 0.351 or self-concept in their research.
Mediating Variable to Career Undecidedness There are likely several complementary explanations for the lack of
S.C.C.B. → C.U. −0.057 0.080 0.401 mediational or direct effects evidenced in the present study. First, and
C.S.E. → C.U. −0.039 0.088 0.589 perhaps the most hopeful explanation, the lack of effect could be a
S.C.I.B. → C.U. −0.085 0.077 0.212 result of the systematic and ongoing emphasis towards the development
Communication → C.U. 0.161 0.154 0.283
Responsibility → C.U. −0.250 0.200 0.204
of STEM skills and aspiration within female adolescents (Krishnamurthi
Self-Regulation → C.U. 0.134 0.162 0.409 et al., 2013). Put differently, the programs, resources, and overarching
Attitude → C.U. −0.464 0.217 0.033 cultural shift towards alleviating longstanding suppression of females
Exploration → C.U. 0.258 0.163 0.071 from the STEM professions could be working. Indeed, evidence pre-
Sobel Test (Mediating Effect on Career Undecidedness) sented through the National Science Foundation and others suggests,
Gender → S.C.C.B. → C.U. −0.004● 0.007 0.565 females continue to be represented at greater rates in terms of aspira-
Gender → C.S.E. → C.U. 0.001● 0.003 0.938
tion towards STEM, STEM college major selection, and STEM careers
Gender → S.C.I.B. → C.U. −0.009● 0.010 0.406
Gender → Communication → C.U. 0.005● 0.015 0.736
(National Science Foundation, 2017; Wai, Lubinski, Benbow, & Steiger,
Gender → Responsibility → C.U. −0.008● 0.023 0.724 2010). Second, it is likely the sample examined in the current study
Gender → Self-Regulation → C.U. 0.005● 0.013 0.710 itself was biased. As noted earlier, the participants opted into a STEM
Gender → Attitude → C.U. −0.001● 0.041 0.972 summer camp experience, this may imply the participants were already
Gender → Exploration → C.U. 0.027● 0.027 0.302
“decided” on a STEM career; indeed, the means within career decid-
Note: S.C.C.B. = Supportive Camp Counselor Behaviors; C.S.E. = Camp edness factor suggest high scores in this area (i.e., ceiling effects). As
Structure Engagement; S.C.I.B. = Supportive Course Instructor Behaviors; suggested in research exploring afterschool and summer-based STEM
C.D. = Career Decidedness; C.U. = Career Undecidedness; β = regression experiences, children and adolescents opting into STEM experiences
coefficient; SE = Standard Error; ● = Mediating effect; * = p ≤ 0.001. beyond their regular coursework may already be motivated to engage
in these experiences, and thus able to overcome constraints that would
stymie their less motivated or less equipped peers (Pace-Marshall et al.,
2011). Third, it is also probable there are unmeasured mediators

7
R.J. Gagnon and A. Sandoval
explaining the relationship between gender and career decidedness.
Indeed, some research suggests persistence may account for the re-
lationship between gender and decidedness, where females with greater
rates of intrinsic motivation tended to persist towards a STEM college
major (Griffith, 2010).
5.1. Limitations
Beyond the limitations already described, a few warrant deeper
explanation. The high between factor correlations evidenced in Table 2
among the five socioemotional factors (YRSESD) illustrates potential
multicollinearity, and thus calls into question the discriminant validity
of these factors (Kline, 2016). Further, the high-correlations point to
another potential issue across the measures, where a 2nd order factor
may be acting as a common-cause of the socioemotional factors (Byrne,
2006). Similar issues with poor variation across socioemotional mea-
sures have been demonstrated within camp research (e.g., Garst &
Gagnon, 2016; Sibthorp et al., 2013), and the theoretically appropriate
use of 2nd order factor models has been found to introduce parsimony
to models (i.e., reduction of SEM parameters and paths), especially
when high correlations are present (Gagnon, Garst, & Townsend, 2019).
In addition to the collinearity issues within the model, some items in-
dicated ceiling effects which could harm the inferences that could be
made about the relations and effects within the models (Bandalos,
2018). In future research, the 10-factors comprising the measure in the
current study should be explored with greater ranges (e.g., 1 to 7 versus
1 to 5), to potentially alleviate this issue.
The present study utilized a cross-sectional design, which seriously
limits the conclusions which can be drawn from the findings.
Longitudinal data could potentially provide support for the study
findings, but also could illustrate if shifts (or lack thereof) in career
decidedness scores are moderated by STEM camper gender. More spe-
cifically, it is possible male campers had different slopes (i.e., shifts)
than their female counterparts, and an exploration of this shift con-
textualized by other potential factors (e.g., prior STEM education ex-
periences, socioemotional skill shifts, counselor/instructor behaviors)
could demonstrate which unique combination(s) of camper traits and/
or contextually dependent factors best promote STEM career decided-
ness in both female and male campers. Importantly, there is also a likely
selection bias present within the study sample; specifically, all campers
opted in to a STEM centered experience. Thus, the effects (and/or lack
thereof) demonstrated in the present study may fail to replicate within a
comparison sample not opting in to a STEM experience. Ideally, a study
with a randomized design assigning campers to either a STEM or non-
STEM experience may demonstrate if STEM career decidedness is due to
the context or factors exclusive of the camp.
6. Conclusion
It appears that for this study, with this sample, at this site, gender
did not play a meaningful role in STEM career decidedness or un-
decidedness. Importantly, with larger shifts in STEM education, the lack
of effects may be explained by the developing trend of orientation to-
wards enhancing equity within the STEM field (College Board, 2018).
Indeed, the increasing research and resource allocation to address these
concerns, suggests an ongoing a systematic push of women, women of
color, and many minority youth cohorts towards the STEM field (Eccles
& Wang, 2016; Johnston-Nicholson, Collins, & Holmer, 2004; National
Science Foundation, 2017). However, the lack of representation of
people of color and those with disabilities within the STEM career field
remains problematic, and potentially stifles the next generation of sci-
entists and problem solvers.
Declaration of Competing Interest
None of the authors has a conflict of interest to report. This
8
Children and Youth Services Review 108 (2020) 104584
submission has not been published previously and is not under review
for publication elsewhere. This submission has been approved by all
authors and that, if accepted, it will not be published elsewhere in the
same form, in English or in any other language, including electronically
without the written consent of the copyright-holder.
Appendix A. Supplementary material
Supplementary data to this article can be found online at https://
doi.org/10.1016/j.childyouth.2019.104584.
References
Beier, M. E., Kim, M. H., Saterbak, A., Leautaud, V., Bishnoi, S., & Giberto, J. M. (2018).
The effect of authentic project-based learning on attitudes and career aspirations in
STEM. Journal of Research in Science Teaching, 56(1), 3–23. https://doi.org/10.1002/
tea.21465.
Bentler, P. M. (2006). EQS 6 Structural Equations Program Manual. Encino, CA:
Multivariate Software Inc.
Bevan, B., Petich, M., & Wilkinson, K. (2014). Tinkering is serious play. Educational
Leadership: Journal of the Department of Supervision and Curriculum Development, 72(4),
28–33.
Byrne, B. M. (2006). Structural equation modeling with EQS: Basic concepts, applications and
programming (2nd ed.). New York, NY: Routledge.
Chow, A., & Salmela-Aro, K. (2011). Task-values across subject domains: A gender
comparison using a person-centered approach. International Journal of Behavioral
Development, 35(3), 202–209. https://doi.org/10.1177/0165025411398184.
Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2003). Applied multiple regression/cor-
relation analysis for the behavioral sciences (3rd ed.). New York, NY: Routledge.
College Board (2018). Number of females and underrepresented students taking AP computer
science courses spikes again. New York: The College Board.
Dasgupta, N., & Stout, J. G. (2014). Girls and women in science, technology, engineering,
and mathematics: STEMing the tide and broadening participation in STEM careers.
Policy Insights from the Behavioral and Brain Sciences, 1(1), 21–29. https://doi.org/10.
1177/2372732214549471.
Dieker, L., Grillo, K., & Ramlakhan, N. (2012). The use of virtual and simulated teaching
and learning environments: Inviting gifted students into science, technology, en-
gineering, and mathematics careers (STEM) through summer partnerships. Gifted
Education International, 28(1), 96–106. https://doi.org/10.1177/
0261429411427647.
Eccles, J. (2011). Gendered educational and occupational choices: Applying the Eccles
et al. model of achievement-related choices. International Journal of Behavioral
Development, 35(3), 195–201. https://doi.org/10.1177/0165025411398185.
Eccles, J. S., & Wang, M. (2016). What motivates females and males to pursue careers in
mathematics and science? International Journal of Behavioral Development, 40(2),
100–106. https://doi.org/10.1177/0165025415616201.
Falk, J. H., Staus, N., Dierking, L. D., Penuel, W., Wyld, J., & Bailey, D. (2016).
Understanding youth STEM interest pathways within a single community: The
Synergies project. International Journal of Science Education, Part B, 6(4), 369–384.
https://doi.org/10.1080/21548455.2015.1093670.
Farrell, A. M. (2010). Insufficient discriminant validity: A comment on Bove, Pervan,
Beatty, and Shiu (2009). Journal of Business Research, 63, 324–327. https://doi.org/
10.1016/j.jbusres.2009.05.003.
Fields, D. A. (2009). What do students gain from a week at science camp? Youth per-
ceptions and the design of an immersive, research-oriented astronomy camp.
International Journal of Science Education, 31(2), 151–171. https://doi.org/10.1080/
09500690701648291.
Gagnon, R. J., & Garst, B. A. (2019). Research and practice: An applied examination of the
parental perceptions of developmental outcomes scale. Journal of Outdoor Recreation,
Education, and Leadership, 11(1), 3–20. https://doi.org/10.18666/JOREL-2019-V11-
I1-9070.
Gagnon, R. J., Garst, B. A., & Townsend, J. A. (2019). Tough decisions in medical spe-
cialty camps: Relationships between camp dosage, outcomes, and camper attendance.
Social Sciences and Medicine, 221, 49–57. https://doi.org/10.1016/j.socscimed.2018.
12.014.
Garst, B. A., & Gagnon, R. J. (2016). A structural model of camp director practices and
outcomes: Does intention toward program outcomes matter? Journal of Park and
Recreation Administration, 34(4), 51–65. https://doi.org/10.18666/JPRA-2016-V34-
I4-7293.
Gasiewski, J. A., Eagan, M. K., Garcia, G. A., Hurtado, S., & Chang, M. J. (2012). From
gatekeeping to engagement: A multicontextual, mixed method study of student
academic engagement in introductory stem courses. Research in Higher Education, 53,
229–261. https://doi.org/10.1007/s11162-011-9247-y.
Griffith, A. L. (2010). Persistence of women and minorities in STEM field majors: Is it the
school that matters? Economics of Education Review, 29, 911–922. https://doi.org/10.
1016/j.econedurev.2010.06.010.
Guay, F., Senecal, C., Gauthier, L., & Fernet, C. (2003). Predicting career indecision: A
self- determination theory perspective. Journal of Counseling Psychology, 50(2),
165–177. https://doi.org/10.1037/0022-0167.50.2.165.
Hayes, A. F. (2013). Introduction to mediation, moderation, and conditional process analysis:
A regression-based approach. New York, NY: Guilford Press.
Hirschi, A. (2009). Career adaptability development in adolescence: Multiple predictors
R.J. Gagnon and A. Sandoval
and effect on sense of power and life satisfaction. Journal of Vocational Behavior, 74,
145–155. https://doi.org/10.1016/j.jvb.2009.01.002.
Holmes, S., Redmond, A., Thomas, J., & High, K. (2012). Girls helping girls: Assessing the
influence of college student mentors in an afterschool engineering program.
Mentoring & Tutoring: Partnership in Learning, 20(1), 137–150. https://doi.org/10.
1080/13611267.2012.645604.
Hughes, R. M., Nzekew, B., & Molyneaux, K. J. (2013). The single sex debate for girls in
science: A comparison between two informal science programs on middle school
students’ STEM identity formation. Research in Science Education, 43(5), 1979–2007.
https://doi.org/10.1007/s11165-012-9345-7.
Johnston-Nicholson, H., Collins, C., & Holmer, H. (2004). Youth as people: The protective
aspects of youth development in after-school settings. The ANNALS of the American
Academy of Political and Social Science, 591(1), 55–71. https://doi.org/10.1177/
0002716203260081.
Jungert, T., Hubbart, K., Dedic, H., & Rosenfield, S. (2019). Systemizing and the gender
gap: Examining academic achievement and perseverance in STEM. European Journal
of Psychology of Education, 34(2), 479–500 10.1007%2Fs10212-018-0390-0.
Kang, J., Hense, J., Scheersoi, A., & Keinonen, T. (2019). Gender study on the relation-
ships between science interest and future career perspectives. International Journal of
Science Education, 41(1), 80–101. https://doi.org/10.1080/09500693.2018.
1534021.
Klahr, D., Triona, L. M., & Williams, C. (2007). Hands on what? The relative effectiveness
of physical versus virtual materials in an engineering design project by middle school
children. Journal of Research in Science Teaching, 44(1), 183–203. https://doi.org/10.
1002/tea.20152.
Kline, R. B. (2016). Principles and practice of structural equation modeling (4th ed.). New
York, NY: Guilford.
Kong, X., Dabney, K. P., & Tai, R. H. (2014). The association between science summer
camps and career interest in science and engineering. International Journal of Science
Education, Part B, 4(1), 54–65. https://doi.org/10.1080/21548455.2012.760856.
Kramer, N. C., Kracora, B., Lucas, G., Dehghani, M., Ruther, G., & Gratch, J. (2016).
Closing the gender gap in STEM with friendly male instructors? On the effects of
rapport behavior and gender of a virtual agent in an instructional interaction.
Computers and Education, 99, 1–13. https://doi.org/10.1016/j.compedu.2016.04.
002.
Krishnamurthi, A., Bevan, B., Rinehart, J., & Coulon, V. (2013). What after-school STEM
does best: How stakeholders describe youth learning outcomes. After-school Matters,
42–49.
Leon, J., Nunez, J. L., & Liew, J. (2015). Self-determination and STEM education: Effects
of autonomy, motivation, and self-regulated learning on high school math achieve-
ment. Learning and Individual Differences, 43, 156–163. https://doi.org/10.1016/j.
lindif.2015.08.017.
Liben, L. S., & Coyle, E. F. (2014). Chapter three-developmental interventions to address
the STEM gender gap: Exploring intended and unintended consequences. Advances in
Child Development and Behavior, 47, 77–115. https://doi.org/10.1016/bs.acdb.2014.
06.001.
Little, R. J. A. (1988). A test of missing completely at random for multivariate data with
missing values. Journal of the American Statistical Association, 83(404), 1198–1202.
London, R. A., Pastor, M., Servon, L. J., Rosner, R., & Wallace, A. (2010). The role of
community technology centers in promoting youth development. Youth and Society,
42(2), 199–228. https://doi.org/10.1177/0044118X09351278.
Lubinski, D., & Persson-Benbow, C. (2006). Study of mathematically precocious youth
after 35 years uncovering antecedents for the development of math-science expertise.
Perspectives on Psychological Science, 1(4), 316–345.
Lubinski, D. (2010). Spatial ability and STEM: A sleeping giant for talent identification
and development. Personality and Individual Differences, 49(4), 344–351. https://doi.
org/10.1016/j.paid.2010.03.022.
Marsh, H. W., Hau, K.-T., & Wen, Z. (2004). In search of golden rules: Comment on
hypothesis- testing approaches to setting cutoff values for fit indexes and dangers in
overgeneralizing Hu and Bentler's (1999) findings. Structural Equation Modeling: A
Multidisciplinary Journal, 11(3), 320–341. https://doi.org/10.1207/
s15328007sem1103_2.
Mohr-Schroeder, M. J., Jackson, C., Miller, M., Walcott, B., Little, D. L., Speler, L., et al.
(2014). Developing middle school students' interests in STEM via summer learning
experiences: See blue STEM camp. School Science and Mathematics, 114(6), 291–301.
https://doi.org/10.1111/ssm.12079.
Mostache, H. S., Matloff-Nieves, S., Kekelis, L., & Lawner, E. K. (2013). Effective STEM
programs for adolescent girls: Three approaches and many lessons learned.
Afterschool Matters, 17–25.
Myers, K. K., Jahn, J. L. S., Gailliard, B. M., & Stoltzfus, K. (2010). Vocational Anticipatory
Children and Youth Services Review 108 (2020) 104584
Socialization (VAS): A communicative model of adolescents’ interests in STEM.
Management Communication Quarterly, 25(1), 87–120. https://doi.org/10.1177/
0893318910377068.
National Science Foundation (2017). Women, minorities, and persons with disabilities in
science and engineering: 2017. Arlington, VA.: National Science Foundation.
Nugent, G., Barker, B., Grandgenett, N., & Adamchuk, V. I. (2010). Impact of robotics and
geospatial technology interventions on youth stem learning and attitudes. Journal of
Research on Technology in Education, 42(4), 391–408. https://doi.org/10.1080/
15391523.2010.10782557.
Pace-Marshall, S., McGee, G. W. M., McLaren, E., & Veal, C. C. (2011). Discovering and
developing diverse STEM talent: Enabling academically talented urban youth to
flourish. Gifted Child Today, 34(1), 16–23. https://doi.org/10.1177/
107621751103400107.
Price, J. (2010). The effect of instructor race and gender on student persistence in STEM
fields. Economics of Education Review, 901–910. https://doi.org/10.1016/j.
econedurev.2010.07.009.
Roth, J. L., Malone, L. M., & Brooks-Gunn, J. (2010). Does the amount of participation in
afterschool programs relate to developmental outcomes? A review of the literature.
American Journal of Community Psychology, 45(3–4), 310–324. https://doi.org/10.
1007/s10464- 010-9303-3.
Shadish, W. R., Cook, T. D., & Campbell, D. T. (2002). Experimental and quasi-experimental
designs for causal inference. Belmont, CA: Wadsworth, Cengage Learning.
Sibthorp, J., Bialeschki, M. D., Morgan, C., & Browne, L. (2013). Validating, norming, and
utility of a youth outcomes battery for recreation programs and camps. Journal of
Leisure Research, 45(4), 514–536. https://doi.org/10.18666/jlr-2013-v45-i4-3897.
Simpkins, S. D. (2015). When and how does participating in an organized after-school
activity matter? Applied Developmental Science, 19(3), 121–126. https://doi.org/10.
1080/10888691.2015.1056344.
Simpkins, S. D., Little, P. M., & Weiss, H. B. (2004). Understanding and measuring at-
tendance in out-of-school programs. Issues and Opportunities in Out-of-School Time
Evaluation, 7, 1–12.
Sinclair, S., Nilsson, A., & Cederskar, E. (2019). Explaining gender-typed educational
choice in adolescence: The role of social identity, self-concept, goals, grades, and
interests. Journal of Vocational Behavior, 101(A), 54–71. https://doi.org/10.1016/j.
jvb.2018.11.007.
Stoet, G., & Geary, D. C. (2018). The gender-equality paradox in science, technology,
engineering, and mathematics education. Psychological Science, 29(4), 581–593.
https://doi.org/10.1177/0956797617741719.
Solanki, S. M., & Xu, D. (2018). Looking beyond academic performance: The influence of
instructor gender on student motivation in STEM fields. American Educational
Research Journal, 55(4), 801–835. https://doi.org/10.3102/0002831218759034.
Tiffany, J. S., Exner-Cortens, D., & Eckenrode, J. (2012). A new measure for assessing
youth program participation. Journal of Community Psychology, 40(3), 277–291.
https://doi.org/10.1002/jcop.20508.
Tiffany, J. S., Exner-Cortens, D., & Eckenrode, J. (2013). Longitudinal associations be-
tween HIV risk reduction and out-of-school time program participation. Journal of
Adolescent Health, 52, 795–797. https://doi.org/10.1016/j.jadohealth.2013.01.011.
Vedder-Weiss, D., & Fortus, D. (2011). Adolescents’ declining motivation to learn science:
Inevitable or not? Journal of Research in Science Teaching, 48(2), 199–216. https://doi.
org/10.1002/tea.20398.
Wai, J., Lubinski, D., Benbow, C. P., & Steiger, J. H. (2010). Accomplishment in science,
technology, engineering, and mathematics (STEM) and its relation to STEM educa-
tion dose: A 25-year longitudinal study. Journal of Educational Psychology, 102(4),
860–871. https://doi.org/10.1037/a0019454.
Wang, M., & Degol, J. (2013). Motivational pathways to STEM career choices: Using
expectancy–value perspective to understand individual and gender differences in
STEM fields. Developmental Review, 33, 304–340. https://doi.org/10.1016/j.dr.2013.
08.001.
Wang, M. T., & Degol, J. L. (2017). Gender gap in science, technology, engineering, and
mathematics (stem): Current knowledge, implications for practice, policy, and future
directions. Educational Psychology Review, 29(1), 119–140. https://doi.org/10.1007/
s10648-015- 9355-x.
Wegemer, C. M., & Eccles, J. S. (2019). Gendered STEM career choices: Altruistic values,
beliefs, and identity. Journal of Vocational Behavior, 110(A), 28–42. https://doi.org/
10.1016/j.jvb.2018.10.020.
Wilson, D., Jones, D., Bocell, F., Crawford, J., Kim, M., Veilleux, N., et al. (2015).
Belonging and academic engagement among undergraduate STEM students: A multi-
institutional study. Research in Higher Education, 56, 750–776. https://doi.org/10.
1007/s11162-015-9367-x.