You are on page 1of 15

J Youth Adolescence (2007) 36:912–926

DOI 10.1007/s10964-006-9126-6

Parent Relationships, Emotion Regulation, Psychosocial Maturity


and College Student Alcohol Use Problems
Judith L. Fischer · Larry F. Forthun ·
Boyd W. Pidcock · Duane A. Dowd

Received: 2 March 2006 / Accepted: 2 August 2006 / Published online: 13 September 2006

C Springer Science+Business Media, Inc. 2006

Abstract This study tested associations between problems which the two groups were constrained to be similar. Among
in parent-youth relationships and problems with alcohol use young women, emotion regulation and psychosocial matu-
among college students (N = 1592) using structural equa- rity were partial mediators of the effects of parent problems
tion modeling. Hypotheses were that relationships between on alcohol use problems. Among young men, parent prob-
both substance-specific parenting factors (parental drinking) lems were indirectly related to alcohol use problems through
and non-substance-specific parenting factors (parental intru- emotion regulation. Implications for alcohol use prevention
sive control and lack of support) and college student drinking activities on college campuses are discussed.
behaviors would be mediated by the developmental tasks of
managing difficult emotions and establishing a mature psy- Keywords College students . Alcohol use . Psychosocial
chosocial identity. Sex, ethnicity and age were entered as maturity . Emotion . Parent psychological control . Parent
control variables in the analyses and were tested for moder- support . Parent drinking
ating effects. Results showed that the unconstrained model
for males and females differed significantly from a model in With frequent and excessive drinking, college students are
at a high risk for alcohol related problems (Wechsler et al.,
An earlier version of this paper was presented at the National Council 1994, 2000). Problems associated with alcohol misuse af-
on Family Relations Annual Meeting, November, 2004, Orlando, fect the students themselves as well as the lives of other
Florida. students (Wechsler et al., 1994, 2000). Past research has ex-
J. L. Fischer () amined a variety of familial and developmental antecedents
Department of Human Development & Family Studies, Texas to alcohol use among college students to better understand
Tech University, the complex nature of alcohol misuse and its often risky
Lubbock, TX 79409, Research interests in college student alcohol
consequences (see Ham and Hope, 2003 for a review). The
misuse.
e-mail: judith.fischer@ttu.edu research continues to point to parenting styles and behaviors
as important influences on alcohol use problems even into
L. F. Forthun the college years (Ham and Hope, 2003). At the same time,
Department Human Development & Family Studies, Penn State
University DuBois, College Place,
research continues to show a disparity between those who
DuBois, PA 15801, Research interests in adolescent psychosocial misuse alcohol and those who do not on a variety of de-
maturity. velopmental indicators including psychosocial development
(Bishop et al., 1997; Jones and Hartmann, 1988) and emo-
B. W. Pidcock
Department of Counseling Psychology, Lewis & Clark College,
tion regulation (Windle and Davies, 1999). Given the impact
Portland, OR 97219-7899, Research interests in college student that family relationships have on children, it is important to
alcohol misuse. include both parenting factors and developmental consider-
ations in an examination of alcohol use and misuse among
D. A. Dowd
School of Human Ecology, Louisiana Tech University,
college students.
P.O. Box 3167, Ruston, LA 71272, Research interests in young This study examines a mediational model where parent
adult relationships. factors are identified as specific stressors that contribute

Springer
J Youth Adolescence (2007) 36:912–926 913

Fig. 1 Model–Dashed lines are


control variable paths, solid age
lines are paths of parent,
development and alcohol use
variables

PSYCHO-
SOCIAL
MATURITY
ethnicity
EMOTION
REGULATION

PARENT
PROBLEMS ALCOHOL USE
PROBLEMS

PARENT
ALCOHOL
PROBLEMS

sex
Model

to the alcohol use problems of college students but whose of alcoholism (Pandina and Johnson, 1990). Parents’ prob-
influences are partially mediated by the student’s ability to lematic drinking provides modeling of the use and abuse of
regulate difficult emotions and develop healthy psychosocial substances (Kandel and Andrews, 1987) and generates ex-
strengths (see Fig. 1). Tested as well are the possibilities that pectancies about the effects of alcohol (Zucker et al., 2000).
these influences may be fully mediated (Baron and Kenny, Moreover, exposure to parental alcoholism may have en-
1986) or indirectly related (Holmbeck, 1997; Preacher and during effects. DeLucia et al. (2001) theorized that once
Hayes, 2004). Partial mediation is supported when the inde- offspring have initiated alcohol use as a result of exposure to
pendent and mediating variables are significantly related to parental alcoholism, the child’s continued use is independent
the outcome. Full mediation occurs when the prior significant of changes in the alcoholic parent’s drinking. In this study,
path between the independent variable and the dependent youth perceptions of problematic alcohol use by parents are
variable is nonsignificant after the mediator is introduced. assessed.
Indirect effects occur when there is no prior significant path Two non-substance-specific parenting factors negatively
between the independent and dependent variable but there related to offspring drinking problems are parental attach-
are significant independent to intermediate variable effects ment and behavioral control (Bell et al., 2000; Stone et al.,
and significant intermediate to dependent variable effects. 2002). Parent-child relationships that are characterized by
a combination of supervision and support/attachment have
been related both concurrently and longitudinally to re-
Conceptual model duced substance use among adolescents (Adalbjarnardottir
and Hafsteinsson, 2001). By contrast, parent-child relation-
Parental influences ships characterized by poor supervision or a lack of sup-
port/attachment are associated with increased frequency of
The predictor variables in the model include both substance- substance use and abuse (Barnow et al., 2002). In the case of
specific parenting factors and non-substance-specific par- the latter, the emerging demands for autonomy on the part of
enting factors (Jacob and Johnson, 1997). A substance- adolescents may disrupt parent-child attachments and lead
specific parenting factor includes parent’s own use of alcohol to a change in the behaviors of parents attempting to mon-
(Jacob and Johnson, 1997). Offspring with a family history itor the activities of their children (Dishion and McMahon,
of alcoholism are at increased risk of experiencing problems 1998). If efforts to monitor the child fail, parents may re-
with alcohol compared to offspring with no family history sort to psychological control. They may also routinely use

Springer
914 J Youth Adolescence (2007) 36:912–926

such parenting practices. Control attempts that emphasize (Cooper et al., 1995; Iacono et al., 1999). Figure 1 includes
manipulation, exploitation of the parent-child bond, nega- pathways where parent alcohol and parent problems are re-
tive expressions, criticisms and excessive personal control lated to emotion regulation and these are related to alcohol
(Barber, 1996) are related to poorer psychological well be- use problems.
ing as well as to both internalizing and externalizing prob- Similarly, difficulty in managing negative emotions may
lems in adolescents (Barber and Harmon, 2002; Stone et al., result in arrested psychosocial development. According to
2002). Likewise, disruptions in parental support have neg- Chickering and Reisser (1993) and Erikson (1968), an im-
ative effects on substance use, effects that are mediated by portant developmental task for emerging adults and college
adolescent self-control (Wills et al., 2004). students is developing a healthy sense of identity. Identity
In this study, “parent problems” was a predictor variable is defined as a stable sense of self, of knowing “who one
and was assessed by the offspring’s perceptions of parental is” and “where one is going” (Erikson, 1968). This stable
psychological control and poor connection with parents. In sense of identity is supported by various developmental as-
addition, the model predicts that parent alcohol problems are sets, or strengths, including competence, purpose, and in-
related to these non-alcohol-specific parent problems that tegrity (Chickering and Reisser, 1993; Erikson, 1968). Neg-
are in turn associated with such outcomes as offspring al- ative emotions alert the individual to important events (both
cohol misuse (Bijttebier et al., 2006; Wampler et al., 1993). internal and external) that may impact the evolving sense
Figure 1 reflects this partial mediation model among parental of self. However, if negative emotions are left unmanaged
alcohol problems, parenting problems, and outcomes. or are poorly managed, emotion-based psychosocial goals
become disorganized and unfocused and psychological de-
velopment is arrested. Research by Nurmi et al. (1997) links
Mediating factors: Emotion regulation and psychosocial emotions and psychosocial development. College students
maturity with a less mature identity style had the highest levels of
depressive symptoms and difficulty in delaying gratification
Theory and research support the view that the influences (Nurmi et al., 1997). Furthermore, this less mature profile is
of both substance-specific and non-substance-specific par- linked with alcohol use and abuse (Bishop et al., 1997; Jones
enting factors are partially mediated by the developmental and Hartmann, 1988). Thus, the model depicts connections
tasks of regulating emotions and establishing psychosocial between emotion regulation and psychosocial maturity and
strengths. According to Chickering and Reisser (1993), and between both of these developmental tasks and alcohol use
consistent with other developmental theorists (e.g., Erikson, problems.
1968), the ability to manage and regulate emotions is im-
portant during times of transition or crisis. The transition
to college involves numerous changes, including decisions Control variables and moderating effects
about alcohol use and the management of difficult emotions.
A positive transition is facilitated when individuals are able The literature suggests that the relationships among the vari-
to control which emotions they will experience and how they ables in the model may differ based on sex of the partici-
will express them (Gross, 1998). pant. Although national surveys of alcohol use consistently
From a functionalist perspective, only events that are show that females, as compared to males, are much less
important to the person arouse the emotions and provide likely to consume alcohol and experience alcohol-related
the subjective motivation to pursue goal-directed behavior problems (Clements, 1999), these reflect average between
(Campos et al., 1994). However, emotional arousal can be group differences. In order for sex to be a moderator, it
both positive and negative. Of concern is when the adoles- must alter the nature of the associations among variables. A
cent response to parental stress caused by parent alcohol variety of psychosocial correlates of alcohol use differ by
use, poor connection, or psychological control, results in gender (Nolen-Hoeksema, 2004). Research suggests that fe-
negative emotional states such as depression or frustration males may be more sensitive to family disruption and are
(Stone et al., 2002; Windle and Davies, 1999). Negative more likely to respond with increased alcohol consumption
emotional states often lead to efforts by the individual to (Coffelt et al., 2006). Likewise, there is a stronger rela-
manage or minimize the source of the problem or emotion tionship between psychosocial self-esteem and alcohol con-
(Cooper et al., 1995). When unsuccessful, these efforts may sumption among women than men (Walitzer and Sher, 1996).
result in attempts by the adolescent to artificially alleviate the On the other hand, men are more likely to report drinking as
negative feelings through use of alcohol or other illicit sub- a means to cope with emotional distress (Park and Levenson,
stances (Catanzaro and Laurent, 2004; Kassel et al., 2000). 2002). These differences would suggest that both substance-
In fact, the inability to regulate emotions and behaviors has specific and non-substance specific parenting factors in our
been implicated in the development of alcohol use disorders model may be more strongly related to alcohol problems

Springer
J Youth Adolescence (2007) 36:912–926 915

for females. Additionally, low psychosocial maturity may be chosocially mature (Bishop et al., 1997). Thus, the model
more strongly associated with alcohol problems for females illustrates predictions that parent alcohol problems and par-
whereas emotion regulation should be a stronger correlate ent problems are related to offspring alcohol use problems
for males. through disruption in the developmental tasks of emotion
Ethnicity is another potential moderator of the associa- regulation and psychosocial strengths (see Fig. 1). Within
tions in the proposed model. Research indicates that Euro- this overall model, there are also specific predictions of how
pean American youth report higher consumption and conse- parent variables are related to each other and how devel-
quences of alcohol use with Hispanic Americans next and opmental variables are related to each other. First, parent
African Americans lowest (Ham and Hope, 2003; Johnstone, alcohol problems have direct and mediated paths to other
1994). These differences suggest that ethnicity is at least a variables in the study through parent problems. Second, emo-
correlate of drinking patterns and consequences. It may be tion regulation has direct and mediated paths to alcohol use
that the alcohol use differences among diverse youth are re- problems through psychosocial maturity. Finally, because
lated differentially to the variables in this study, but research demographic variables may have an impact on the levels
(Bray et al., 2003) found that the processes of individuation of the variables of the study as well as on the associations
(involving autonomy, identity, and intimacy) were similar among the variables of the study, sex, ethnicity and age are
across European American, Hispanic American, and African tested for their effects within the model and as moderators
American adolescents. On the one hand, differences in levels of the model effects.
of alcohol use are expected, but whether associations among
variables, such as those that involve psychosocial maturity,
vary by ethnicity is uncertain. With a lack of precision in Method
efforts to measure ethnicity (Johnstone, 1994), and the small
percentages of ethnically diverse youth in colleges such as Participants
those used in this study, there are limitations in this study to
full consideration of the role of ethnic diversity. College students were recruited in their classes on three state-
A final possible moderator in this study is age. Age is assisted campuses in the northeast, the south, and the south-
related to alcohol consumption in that drinking reaches a west. Institutional IRB approval was obtained on all three
peak at 19 or 20 and declines afterwards (Johnstone, 1994) campuses. The anonymous volunteers represented a wide
justifying its inclusion as a control variable. Younger college range of majors, ages, and family incomes. The sample used
students would have had more recent at-home experiences for this study was composed of 1,592 college students, 1,064
with parents than older college students, thus, having possi- (66%) women and 528 men (34%). Their ages ranged from
bly stronger associations of parent variables with other vari- 18–25 years with a mean of 20.2 and a standard deviation
ables. In this case, age would act as a moderating variable. of 1.5. Eighty-six percent were White/Non-Hispanic, with
Thus, sex, ethnicity, and age are included in Fig. 1 as exoge- 5% Hispanic and 6% African American. The remaining par-
nous variables and they are tested as possible moderators. ticipants identified themselves as Asian American/Pacific
Islander (1%), Native American/American Indian (< 1%) or
other (2%). Parents’ household incomes ranged from under
Summary and predictions $15,000 to over $100,000 with a mean in the $60,000 to
$75,000 range. Among the students who ever used alcohol
In this study, the ability to manage difficult emotions and (84%), the average onset was in the middle of the 9th grade.
the attainment of psychosocial maturity were identified as The collection of data from three campuses provided a wider
important partial mediating factors between problems with array of college contexts than would the use of data from only
parents, both substance specific and non-substance spe- one campus. The few participants with missing data (n = 27,
cific, and alcohol problems. Consistent with our model, 1.7%) were deleted from the data set of 1,619. Examination
previous research has demonstrated a link between lack of of differences between the original data set and the non-
parental support and connection with difficulty managing missing data set found negligible differences and supported
emotions (Catanzaro and Laurent, 2004). Likewise, research dropping the participants with missing data.
has demonstrated a link between parent problems and diffi-
culty in acquiring psychosocial strengths (Fullinwider-Bush Measures
and Jacobvitz, 1993; Grotevant and Cooper, 1985; Perosa
et al., 1996). Finally, research suggests that alcohol use is There were five latent variables included in the model. The
greater among those who have difficulty managing emo- predictor variables were (a) Parent Alcohol Problems and
tions (Windle and Davies, 1999) and those who are less psy- (b) Parent Problems. The intermediate variables were (c)

Springer
916 J Youth Adolescence (2007) 36:912–926

Emotion Regulation, and (d) Psychosocial Maturity. The de- Lack of parental connection and support was measured
pendent variable was (e) Alcohol Use Problems. by using the communication subscale of the revised Inven-
tory of Parent and Peer Attachment (IPPA) (Armsden and
Parent alcohol problems Greenberg, 1987). The communication subscale specifically
assesses the affective quality of the relationship between
The Children’s Alcohol Screening Test CAST-6 identi- parents and their children on a 5-point Likert scale. Sam-
fied problems offspring experience with parents that arise ple statements include, “I like to get my mother’s point
because of the parents’ drinking (Hodgins and Shimp, 1995). of view on things I’m concerned about” and “I feel it’s
The CAST-6 consisted of 6 items answered with a yes/no no use letting my feelings show around my father.” Par-
format. This scale is a shorter version of the Children’s ticipants were asked to complete the nine subscale items
Alcoholism Screening Test or CAST (Jones, 1983). It has for both mother and father separately. Items were coded so
been used successfully in research involving college students that higher scores reflected lower connection and support.
(Pidcock et al., 2000) and is a recognized screening tool to Cronbach’s alpha was .91 for both the mother and father
identify children of alcoholics (Bijttebier et al., 2006). The subscales.
CAST-6 was chosen for the alcohol-specific parenting mea- Because there were a number of participants without re-
sure because it specifically assesses the degree of impact sponses on the items relating to fathers, and because these
of parental drinking on the participant. Children may not participants were disproportionately distributed by ethnic-
know how much or how frequently their parent drinks but ity, the parental indicators were collapsed across mother and
they can identify whether they’ve been in an argument with father scores. For those with scores on both parent scales,
a drinking parent or if they wished their parent would stop the average of mother and father scores was used. If there
drinking. A sample item was “Did you ever argue or fight were scores for only one parent, then the scale score for
with a parent when he or she was drunk.” Cronbach’s alpha that parent was used. Thus, the indicators for parent prob-
was .90. High scores indicated more problems with parents’ lems were parent control and parent lack of connection and
drinking (Parent Alcohol Problems) whereas a score of zero support.
indicated no problems either because parents did not drink
or because parental drinking was not problematic to the off- Emotion regulation
spring. In order to provide indicators for a latent variable, the
six items were parceled over three indicators, termed castA, Emotion Regulation consisted of 24 questions from the Iowa
castB, and castC. CastA reflected the belief that the parent Managing Emotions Inventory (Hood, 1997) organized into
has a problem with alcohol and a wish that the parent would two indicators concerning recognition and management of
stop drinking. CastB indicated encouragement to the parent depression (manage depression) and frustration (manage
to quit and the wish to hide alcohol from the parent. Finally, frustration). The items were answered with a 5-point for-
castC reported that parents argued with each other and the mat from “Not descriptive of me” to “Very descriptive of
adolescent when drunk. me.” An example is “When feeling frustrated, I find a solu-
tion and move on to other tasks.” Cronbach’s alpha was .85.
Parent problems A high score reflected better recognition and management of
emotions.
There were two indicators of non-substance-specific par-
enting problems (Parent Problems), parental psychological Psychosocial maturity
control (parent control) and parental lack of support and
connection (parent lack support). The Parental Psychologi- Four scales, (a) identity commitment, (b) purpose, (c) com-
cal Control scale reflected college students’ perceptions of petence, and (d) integrity, were used as indicators of the la-
maternal and paternal insensitive control and manipulation tent variable of Psychosocial Maturity. Identity commitment
(Barber, 1996). The Parental Psychological Control scales was measured using the 16-item commitment subscale of the
consisted of 8 items for each parent, such as, “My mother is Ego Identity Process Questionnaire (Balistreri et al., 1995),
a person who is always trying to change how I feel or think answered on a 6-point format from “Strongly Disagree” to
about things” and “My father is a person who brings up past “Strongly Agree.” Identity commitment measured the extent
mistakes when he criticizes me.” These items were answered to which the person had committed to vocation, friends, val-
on a 5-point Likert type scale from “Almost never or never ues, roles, and politics, (e.g., “I am very confident about what
true” to “Almost always or always true” with higher scale kinds of friends are best for me”). Cronbach’s alpha was .77.
scores reflecting greater perceived problems with parents. The psychosocial strengths of purpose and competence were
The Cronbach’s alpha reliability was .87 for mother control measured using subscales from the Psychosocial Inventory of
and .89 for father control. Ego Strengths–Short Form (Markstrom et al., 1997). Each of

Springer
J Youth Adolescence (2007) 36:912–926 917

these subscales had four items from 1 (Not descriptive of me) cohol) per week ranged from 0 to 75 with a mean of 4.19,
to 5 (Very descriptive of me). Each had a Cronbach’s alpha median of 1, and standard deviation of 7.49. Alcohol prob-
of .62. The psychosocial strength of integrity was measured lems averaged 5.3, median was 2, and standard deviation
using the wisdom subscale from the Psychosocial Inven- was 7.19. As described below, three of the indicators were
tory of Ego Strengths–Short Form (Markstrom et al., 1997). transformed to reduce skew.
The wisdom subscale, integrity, measures the psychosocial
strength acquired during the final stage of Erikson’s eight Control variables
stages of development: integrity vs. despair, precursors of
which appear at earlier developmental stages. The integrity Sex was coded as 1 = female (n = 1064), 2 = male
subscale had a Cronbach’s alpha of .62. (n = 528). Ethnicity was 0 = European American (white,
non-Hispanic, n = 1374) or 1 = Other Ethnicity (n = 218).
Alcohol use problems With so few in any of the ethnic groups other than Eu-
ropean American, this Other Ethnicity category was cre-
Four items/scales constituted indictors of the latent variable ated of necessity. In the Other Ethnicity group, 41% were
Alcohol Use Problems. Experiences of problems in the use African American and 29% were Hispanic/Latino. Alco-
of alcohol and estimates of excessive drinking occasions and hol use problem indicators in this study varied by ethnicity
average consumption were used to reflect the various ways in (Wilks lambda = .83, p < .001) similar to that reported in
which alcohol use problems are discussed and measured in the literature (Johnstone, 1994). Comparisons across all eth-
the literature. Although Patock-Peckham and Morgan-Lopez nic groups with Tukey’s B for multiple mean comparisons
(2006) found that alcohol use and problems stemming from found African Americans scored lower than all others on fre-
alcohol use were significantly correlated for women but not quency of binge drinking and lower than Native Americans
men, when men and women were compared in this study, the on binge drinking in the past two weeks. Groups were equiv-
factor loadings of indicators of the latent variable were not alent on alcohol consumption and alcohol problems. Three
significantly different. Thus, the four indicators of Alcohol other significant mean differences were scattered across the
Use Problems were binge in the past two weeks (binge), groups. This heterogeneity by ethnicity across the variables
frequency of binging on the last 10 occasions (frequency of the study should be kept in mind as the analyses proceed.
binge), alcohol consumption per week (consume alcohol), Age was recorded in years. When age was used as a cate-
and alcohol related problems (alcohol problems). gory in the moderating analyses, 18–20 year olds were in the
Frequency of excessive drinking and degree of con- younger group (n = 994) and 21–25 year olds were in the
sumption are often used to measure alcohol use problems older group (n = 598). This split was based on the median
(Wechsler et al., 2000). The participants were asked to (age 20), the legal drinking age and the literature that sug-
“Think back over the last two weeks. How many times have gested a decline in drinking after age 20 (Johnstone, 1994).
you had five or more drinks at a sitting?” Responses to this The variables of the study were assessed for skew and out-
binge item were written in. “Think about the last 10 occa- liers. Although testing of models with maximum likelihood
sions you’ve drunk alcohol. On how many occasions did estimates (as was done in this study) is robust to nonnormal-
you have five or more drinks?” and participants were asked ity (Lei and Lomax, 2005) efforts were made to reduce skew
to write in the number of times (frequency binge). Students when possible. Three of the alcohol use variables were sig-
were requested to “Please write the average number of drinks nificantly skewed. In order to reduce skew, these were trans-
that you consume a week” in the space provided (consume formed using a log to the base 10 transformation (binge and
alcohol). The 10 items of the Alcohol Use Problems scale alcohol problems) or a square root transformation (consume
(alcohol problems) (Wechsler et al., 1994; Wechsler et al., alcohol). There were no outliers. Table 1 includes informa-
2000) reflected the extent to which the students experienced tion on the means, standard deviations, and intercorrelations
various problems due to their own drinking or drug use in the of variables of the study.
past year. The items were answered on a 6-point scale from
“Never” to “10 or more” times. A sample item was “Been Plan of analyses
in trouble with campus or local police.” Cronbach’s alpha
was .83. Higher scores reflected more problems with alco- The plan of analyses called for several stages. First, the hy-
hol/drug use. Frequency of binge drinking in the past two pothesized full model was analyzed. This full model tested
weeks ranged from 0 to 20 with a mean of 1.46, median of 0, for fit of the paths in Fig. 1. Evaluation of the model involved
and standard deviation of 2.32. Frequency of binge drinking examination of various indices of goodness of fit. Through-
over the last 10 drinking occasions ranged from 0 to 10 with out, goodness of fit is estimated with the incremental fit
a mean of 3.18, median of 2, and standard deviation of 3.56. index (IFI), a nonnormed index that is relatively unaffected
Average number of alcoholic drinks consumed (consume al- by sample size, the comparative fit index (CFI), a normed

Springer
918 J Youth Adolescence (2007) 36:912–926

index, both of which should be .90 or higher with .93 or


higher expected for a good fit (Byrne, 1994). The root mean

3.61
.74
square error of approximation (RMSEA), that compensates


I

for complexity of the model (Hu and Bentler, 2000) and is

.47∗∗∗
3.72
.72
minimally influenced by sample size (Browne and Cudeck,
Cm


1993), should be .08 or lower for an acceptable fit and .05

.56∗∗∗
.59∗∗∗
or lower for a good fit (McDonald and Ho, 2002). Although

3.69
.74
P


Chi-square is used often as an index of goodness of fit, it is
likely to be significant with a large N such as occurs in this

.44∗∗∗
.33∗∗∗
.34∗∗∗
4.29
.64
study (Sivo et al., 2006). Therefore Chi-square is not used
IC


to determine goodness of fit but it is reported and is used to
.19∗∗∗
.32∗∗∗
.30∗∗∗
.24∗∗∗
compare models.

3.53
.54
D

In the first stage of the analyses, the full model was tested
.78∗∗∗
.13∗∗∗
.28∗∗∗
.27∗∗∗
.19∗∗∗
and modified through elimination of nonsignificant paths.
3.43
.54
Comparisons were made between the two models for signif-
F

icant change in Chi-square. Goodness of fit was examined for


−.08∗∗∗

−.08∗∗
−.06∗
−.04
−.05

−.04

.27
.62
the models. The second stage tested for possible moderation
CtC

effects by comparing the model constrained, in turn, for sex,


.65∗∗∗

ethnicity, and age, with corresponding unconstrained mod-


−.06∗
−.06∗
.01
.00
−.04
−.04

.23
.57
CtB

els. Significant moderating effects need to be considered in


order to evaluate threats to the generalizability of the model


.75∗∗∗
.73∗∗∗

−.09∗∗∗

−.07∗∗
−.08∗∗
−.05∗
−.01
−.02

.37
.73

and to move to the third stage of testing, a stage that included


CtA

the examination of mediation and indirect effects. Alterna-


.18∗∗∗
.13∗∗∗
.14∗∗∗
−.22∗∗∗
−.22∗∗∗
−.28∗∗∗
−.23∗∗∗
−.21∗∗∗
−.23∗∗∗

tive models are proposed and examined to provide greater


2.40
.76

understanding of the model.


PL

–.51∗∗∗
.17∗∗∗
.15∗∗∗
.16∗∗∗
−.10∗∗∗
−.16∗∗∗
−.21∗∗∗
−.19∗∗∗
−.17∗∗∗
−.28∗∗∗
2.02
.77
PC

Results
.08∗∗∗

.08∗∗∗

.14∗∗∗
−.12∗∗∗
−.15∗∗∗
−.22∗∗∗
−.14∗∗∗
−.11∗∗∗
−.12∗∗∗
.03

.03

.51
.51

The full and modified models


AP

–.66∗∗∗

.08∗∗∗
−.10∗∗∗
−.14∗∗∗
−.18∗∗∗
−.10∗∗∗

The structural equation model with latent variables was


−.07∗
.05
.02
.04
−.01

−.04
1.34
1.52
CA

tested using Amos 6.0. The correlational analyses supported


the orthogonality of the demographic variables with each


.69∗∗∗
.63∗∗∗

.09∗∗∗
−.13∗∗∗
−.16∗∗∗
−.20∗∗∗
−.13∗∗∗
−.08∗∗∗
−.08∗∗∗
.07∗∗
.05∗
.04

.01

3.18
3.56

other; therefore, no correlated paths among these were en-


FB

tered. To the conceptual model in Fig. 1 were added the


Correlations, means, standard deviations (n = 1592)

.68∗∗∗
.82∗∗∗
.65∗∗∗

.09∗∗∗
−.09∗∗∗
−.14∗∗∗
−.18∗∗∗
−.12∗∗∗
−.09∗∗∗

indicators of the latent variables and associated error terms


.05∗

−.05∗
.03
−.02

.01

.25
.32

(tables of unstandardized and standardized coefficients are


B

available on request). Table 2 provides model fit indices.


− .09∗∗∗

Transformed score; ∗ p < .05; ∗∗ p < .01; ∗∗∗ p < .001.


−.10∗∗∗
−.13∗∗∗
−.10∗∗∗

.10∗∗∗

−.15∗∗∗
.08∗∗

−.07∗∗
.06

.03

.02
.03
−.02
−.04
−.04

.14
.34

The extent to which indicators loaded on their respective


E

factors was examined. All of the indicators were signifi-


cantly related to their respective latent variables, p < .001,
.10∗∗∗

−.11∗∗∗
−.13∗∗∗
.06∗

−.06∗

−.05∗
.03
.03

−.02
−.01
.03
−.04
−.02

.01
.02
.03
1.33
.47

and loaded on their respective latent factors. A first calcula-


G

tion of the full model was evaluated in terms of modification


.10∗∗∗

.09∗∗∗

.07∗∗
.05∗

.06∗
.02
.03

−.01

.02
−.02

.04
.06

.05
−.02
−.02
.01
−.01
20.23
1.52

indices that indicated that a path could be freed for corre-


A

lated error terms between frequency of binge and alcohol use


problems. The second calculation of the full model included
Identity Commitment (IC)
Parent Lack Control (PL)

this modification. This full model demonstrated a good fit.


Consume Alcohol (CA)a
Alcohol Problems (AP)a

Manage Depression (D)


Manage Frustration (F)
Frequency Binge (FB)

Table 2 provides the fit indices for this model with the cor-
Parent Control (PC)

Standard Deviation
Competence (Cm)

related error term. Explained variance was 7.2%.


Cast A (CtA)
Cast B (CtB)
Cast C (CtC)
Ethnicity (E)

Purpose (P)

Integrity (I)
Gender (G)

There were five nonsignificant paths involving parent al-


Binge (B)a
Table 1
Variable
Age (A)

cohol problems that were eliminated in the modified model:


Mean

age, sex, and ethnicity to parent alcohol problems, and parent


a

Springer
J Youth Adolescence (2007) 36:912–926 919

Table 2 Model fit for full and


modified models and tests for Model X2 df IFI CFI RMSEA X 2 df
moderation
Full 577.59∗∗∗ 113 .96 .96 .05
Modified 589.54∗∗∗ 120 .96 .96 .05 11.95 7
Moderation effects: Sex
Unconstrained 657.44∗∗∗ 212 .96 .96 .04
Constrained 720.10∗∗∗ 238 .96 .96 .04 62.66∗∗∗ 26
Moderation effects: Ethnicity
Unconstrained 644.63∗∗∗ 212 .96 .96 .04
Constrained 699.77∗∗∗ 238 .96 .96 .04 55.14∗∗∗ 26
Note. na: not applicable as Moderation effects: Age
neither model is a good fit Unconstrained 1991.08∗∗∗ 206 .84 .84 .07
∗∗∗ Constrained 2104.20∗∗∗ 236 .84 .84 .07 na na
p < .001.

alcohol problems to emotion regulation and to psychosocial was significantly more negative (p < .001) for women than
maturity. In addition, paths from sex to parent problems and men (β = − .23, p < .001, vs. β = − .03, ns). in the uncon-
from ethnicity to emotion regulation were nonsignificant and strained model. The indicators of the latent variables were
were deleted. As seen in Table 2, the modified model was not significantly different by sex of the participant. With sex
also a good fit, and as expected when nonsignificant paths are of the participant limiting the generalizability of the modi-
deleted, was not significantly different from the full model. fied model, these results called for testing for mediation and
Explained variance was 7.3%. Before presenting and ac- indirect paths separately for young women and young men.
cepting this modified model as a final model, threats to its
generalizability are assessed through tests of moderation ef- Ethnicity and age of participant
fects.
The results in Table 2 indicated a moderating effect of eth-
Moderation effects nicity. However, when comparing the paths for the Euro-
pean American versus Other Ethnicity group, there were no
Comparing the results of the models unconstrained versus significant paths to alcohol use problems in the model for
constrained provides an indication of the extent to which a Other Ethnicity participants except for age. Despite the fact
control variable acts as a moderator in the model, altering the that age was the only significant path to alcohol use prob-
strength of associations among variables. The original full lems among the participants in the Other Ethnicity group
model had been modified by the deletion of several paths. (β = .22, p < .01), this path was not significantly different
In all but one of the instances (age as moderator), the same (p > .05) from that of the European American participants
paths that had been deleted in the modified model were non- (β = .09, p < .01). The negative effect of parent problems
significant in the multiple group comparisons and therefore on psychosocial maturity was significantly stronger (p < .05)
deleted. As a follow-up to a significant moderation effect, for Other Ethnicity participants (β = − .52, p < .001) than
analyses were fit to pinpoint where groups differed. Tests for European American youth (β = − .27, p < .001).
for the significance of difference of a path across multiple The failure to find any significant pathways among the
groups compared a model with a constrained path against main variables of the study to alcohol use problems in
the unconstrained model. the Other Ethnicity group suggests that, along with age,
there are other unspecified pathways to alcohol use prob-
Sex of participant lems among diverse youth. Additional analyses within the
European American group supported the model presented
The results in Table 2 indicated that sex was a moderator of in Figs. 2 and 3. Thus, the model represents European
the modified model. Figures 2 and 3 provide the standardized American processes only.
path coefficients for females and males (with error terms The variance of the measures of alcohol use problem indi-
omitted for ease of readability). In the unconstrained model, cators was considered as a possible contributing factor in the
explained variance was 9.4% for young women and 5.9% failure to find significant associations between the variables
for young men. Tests for significance of difference of paths of the study and alcohol use problems within the Other Eth-
found two such significant differences: (a) the age to parent nicity group. This was rejected based on higher variances in
problems path was significantly more negative (p < .001) for the Other Ethnicity group. The inclusion of ethnicity in the
women than men (β = − .14, p < .001 vs. β = − .09, ns); analyses within sex (Figs. 2 and 3) is warranted because eth-
(b) the psychosocial maturity to alcohol use problems path nicity is significantly associated with variables of the study.

Springer
920 J Youth Adolescence (2007) 36:912–926

age
.18***
commitment purpose competence integrity
.54 .86 .66 .69

.14** .08*
-.14*** PSYCHO-
SOCIAL
MATURITY
-.02 .31***
.97
depression
ethnicity .82 EMOTION
frustration
REGULATION -.23***
.11*
-.18***
-.09*
parent .66 -.26*** .90 binge
control
-.31*** ALCOHOL USE .75
PARENT frequency
parent .82 PROBLEMS binge
lack support PROBLEMS .92
-.09*
consume
.72 alcohol

.08* alcohol
.24*** problems
.92
castA
.81 PARENT
castB ALCOHOL
.83 PROBLEMS
castC Young Women
Fig. 2 Model for females–standardized coefficients–dashed lines are nonsignificant paths; p < .05, ∗∗ p < .01; ∗∗∗ p < .001

age
.24***
commitment purpose competence integrity
.57 .87 .79 .78

-.09 .12* .12* PSYCHO-


SOCIAL
MATURITY
-.09* .32***
.93
depression
ethnicity .80 EMOTION
frustration
REGULATION -.03
.10
-.09*
-.20***
-.32*** .91
parent .67 binge
control
-.32*** ALCOHOL USE .76
PARENT frequency
parent .71 PROBLEMS binge
lack support PROBLEMS .91
.00
consume
74 alcohol

.04 alcohol
.26*** problems
.93
castA
.82 PARENT
castB ALCOHOL
.72 PROBLEMS
castC Young Men
Fig. 3 Model for males–standardized coefficients–dashed lines are nonsignificant paths; p < .05, ∗∗ p < .01; ∗∗∗ p < .001

Springer
J Youth Adolescence (2007) 36:912–926 921

Although age was associated with the variables in the Parenting variables related to alcohol use problems
study, tests for moderation found that both the constrained through developmental variables
and unconstrained models provided poor fits, even after
adding a path from sex to parent alcohol problems suggested As a first step, the a priori path between alcohol-specific par-
by the modification indices (see Table 2). Age was retained ent alcohol problems and youth alcohol use problems was
as a predictor in the model, but not as a moderator. established. This was significant (β = .08, p < .05) among
To summarize the tests for moderating effects, the modi- young women but not among young men (β = .05, ns). Thus,
fied model was significantly different for males and females. any effects for young women would be mediated effects and
Furthermore, the processes depicted in Figs. 2 and 3 are those among the young men would be indirect effects. With
applicable for European American youth. Age was associ- no significant path from parent alcohol problems to emotion
ated with the variables of the study but attempts to fit models regulation for either women or men (β = .05, ns; β = .00,
with age as a moderator were unsuccessful. ns, respectively), there was no test for parent alcohol prob-
lems to alcohol use problems through emotion regulation.
In this case, the condition that there be a significant IV to
Mediation effects MV path was not present. Tests for mediation for the effects
of parent alcohol problems to alcohol use problems through
Based on the analyses for moderation effects, the tests for the psychosocial maturity proceeded for young women as there
significance of mediation and indirect effects were carried were significant IV to MV and MV to DV paths. There was
out on males and females separately. In accordance with significant partial mediation (bootstrapped estimate = .019,
recommendations by Baron and Kenny (1986) for testing CI = .007 to .034) on the effects of parent alcohol problems
mediation, first, an association between the IV and DV was on alcohol use problems through psychosocial maturity. A
tested. However, in the absence of a significant IV to DV similar test for young men was not calculated as there was
path, then indirect paths via intermediate variables are still no significant MV to DV path.
testable (Holmbeck, 1997). Preacher and Hayes (2004) have Turning to non-alcohol-specific parent problem effects
argued that testing the significance of indirect effects can and on alcohol use problems through developmental vari-
should occur in the absence of a prior IV to DV significant ables, there was a significant a priori path between the
effect. IV and DV (β = .06, p < .05) among women but not
Following recommendations of Preacher and Hayes among men (β = .05, ns). Thus, significant effects included
(2004), mediation tests were carried out using a bootstrap partial mediation among young women (estimate = .026,
procedure they considered superior in terms of power to CI = .014 to. 041) and indirect effects among young men
the Sobel test when testing for mediating/indirect effects. (estimate = .046, CI = .021 to .080) when emotion regu-
The estimate of mediated/indirect effects in this study was lation was the MV. With respect to psychosocial maturity,
obtained using 5000 resamples and confidence intervals of such partial mediation was significant for young women
95%. In the analyses, the latent variables were recomputed (estimate = .067, CI = .047 to .092) but could not be tested
as measured variables in order to use the Preacher and Hayes among young men in the absence of a significant MV to
formula. Factor scores of the indicators of each latent vari- DV effect. Thus, the prediction that developmental variables
able were employed to provide a model with all measured would be mediators or indirect effects between parent vari-
variables for these tests of significance of mediation/indirect ables and alcohol use problems was found for three of the
effects. four tests among young women, but only for one of these
No omnibus test of mediation was conducted. The for- among young men.
mula for calculating indirect effects identified classes of
mediating/indirect variables, but did not test for the spe-
cific linked paths identified in the model. Instead, more Parent alcohol problems related to developmental variables
limited questions based on the issues that built the model through parent problems
were considered. First, were parenting variables related to
alcohol use problems through the developmental variables? The tests in this section assessed whether parent prob-
Second, were parent alcohol problems related to devel- lems constituted a link between parent alcohol problems
opmental variables through parent problems? Third, were and development. The a priori tests of association be-
parent problems related to psychosocial maturity through tween parent alcohol problems and developmental variables
emotion regulation and was emotion regulation related to indicated that, among young women, there was a non-
alcohol use problems through psychosocial maturity? In significant association between parent alcohol problems and
each analysis, age and ethnicity are entered as control emotion regulation (β = .05, ns) and between parent alco-
variables. hol problems and psychosocial maturity (β = − .04, ns).

Springer
922 J Youth Adolescence (2007) 36:912–926

Indirect effects from parental alcohol problems to emo- Alternative models


tion regulation through parent problems were significant
(estimate = − .037, CI = − .057 to − .023) and from par- With age and ethnicity significantly related to alcohol use
ent alcohol problems to psychosocial maturity through par- problems for both young men and women, these were in-
ent problems (estimate = − .057, CI = − .080 to − .039). cluded in both of the alternative models tested. Alternative
There was no significant a priori association of parent alcohol models to those in Figs. 2 and 3 involved leaving out sets
problems to emotion regulation (β = .00, ns) among young of latent variables. When tested, a model based on Fig. 1
men, however, there was a significant indirect effect from (unconstrained for sex) but without emotion regulation and
parent alcohol problems to emotion regulation through par- psychosocial maturity (Alternative Model 1) found reduced
ent problems (estimate = − .050, CI = − .086 to − .025). explained variance in alcohol use problems for women (from
Similarly, there was no prior association between parent al- 9.4 to 3.7%) and for men (from 5.9 to 2.7%). Such a reduction
cohol problems and psychosocial maturity (β = − .04, ns) in explained variance argues for the importance of including
for the young men but there was a significant indirect ef- emotion regulation and psychosocial maturity in the models.
fect from parent alcohol problems to psychosocial maturity Models without parent alcohol problems and parent prob-
through parent problems (estimate = − .074, CI = − .113 lems could also provide an alternative. The resulting uncon-
to − .042). Thus, in all four tests, parent alcohol problems strained Alternative Model 2 was a comparable fit and had
were indirectly related to developmental variables through explained variance in alcohol use problems that differed by
parent problems. small amounts from the unconstrained model among both
young women (9.4% to 8.4%) and young men (5.9 to 5.7%).
However, among the young women explained variance in
Parent problems related to psychosocial maturity through emotion regulation dropped from 10.1 to 3.2%, among young
emotion regulation and emotion regulation related men it dropped from 12.2 to 2.2%. Explained variance in
to alcohol use problems through psychosocial maturity psychosocial maturity dropped from 32.1 to 22.9% in young
women and from 37.1 to 28.8% in young men. In view of the
Finally, the possible mediating or indirect effects involving significant mediation and indirect effects reported in the pre-
emotion regulation as a MV or as a IV were tested. First, with vious section for both men and women, we favor the model
respect to the role of emotion regulation as a possible me- depicted in Figs. 2 and 3 as a more complete model. It al-
diator between parent problems and psychosocial maturity, lows for understanding how parent variables may be linked
there were significant a priori associations between these to personal characteristics that are, in turn, linked to alcohol
parent problems and psychosocial maturity among young use problems.
women (β = − .22, p < .001) and young men (β = − .31,
p < .001). Emotion regulation was a partial mediator in each Discussion
case (Women: estimate = − .055, CI = − .080 to − .036;
Men: estimate = − .073, CI = − .114 to − .042). Among The model and its variations that were tested in this
young women there was a prior association between emotion study were built upon theory and research that supported
regulation and alcohol use problems (β = − .14, p < .001). interconnected links among parenting and developmental
This association was significantly and partially mediated by variables and alcohol misuse in college youth. In this
psychosocial maturity (estimate = − .062, CI = − .087 to study, partial mediation (women) or indirect effects (men)
− .041). In the absence of an MV to DV significant asso- linked both alcohol-specific and non-alcohol specific factors
ciation among young men, there was no test for mediation. (Jacob and Johnson, 1997) to alcohol use problems through
Thus, among young women, emotion regulation provided a both (women) or one (men) of the developmental factors
link between parent problems, psychosocial maturity, and al- in these students. Managing difficult emotions and attain-
cohol use problems. Among young men, emotion regulation ing psychosocial maturity were not only predicted by parent
was a mediator of the parent problems to psychosocial matu- problems, but were also linked to each other. Sex of the par-
rity association but the absence of a link from psychosocial ticipant along with ethnicity and age were introduced in the
maturity to alcohol use problems suggested that male dif- model based on literature that pointed to mean level differ-
ficulties with psychosocial maturity issues could be related ences between these variables and some or all of the variables
to outcomes not tested in this model. Taken together these of the study, but particularly alcohol use problems of the stu-
tests for mediating and indirect effects revealed a networked dents themselves. Males, European Americans, and older
set of relationships among the variables of the model. The students were higher on alcohol use problems than females,
predictions were especially borne out in the case of young those of Other Ethnicity or younger students, respectively.
women. The finding of higher alcohol use problems on the part of

Springer
J Youth Adolescence (2007) 36:912–926 923

males and European Americans is consistent with the litera- sponse to family factors (Coffelt et al., 2006). Helping young
ture (Johnstone, 1994). However, that younger students were women deal with parent problems and emotion regulation
lower on alcohol use problems was surprising in light of lit- in a positive, constructive manner may facilitate their psy-
erature that suggested that drinking reached a peak at ages chosocial growth. Efforts to strengthen their identity forma-
19 and 20 and declined thereafter (Johnstone, 1994). Perhaps tion could provide an avenue not only for greater personal
students in the 2000’s are postponing drinking curtailment growth but also for fewer problems with alcohol. Examina-
until they are finished with college. In the full model, par- tion of these possibilities should be a focus of future research.
ent alcohol problems were unrelated to sex, ethnicity or age,
suggesting that problems with parents who misuse alcohol Young men
transcend these demographic categories. The testing of mod-
erating effects for sex, ethnicity and age indicated that males Contrary to the model that predicted partial mediation,
and females differed significantly, that the model findings among young men only indirect effects were testable. Parent
were applicable only to European Americans and that age problems were indirectly related to alcohol problems through
was not a moderator. emotion regulation. The literature has been quite convincing
in terms of the central role of emotional and behavioral reg-
Young women ulation in the positive development of youth, including the
avoidance of harmful substance use (Iacono et al., 1999;
Among young women, psychosocial maturity played a cen- Wills and Dishion, 2004). Among the young men, there was
tral role in the prediction of alcohol use problems. It di- an indirect link from parent alcohol problems to parent prob-
rectly predicted young women’s alcohol misuse and it served lems to emotion regulation and there was a connected link
as a partial mediator for the effects of emotion regulation from parent problems to emotion regulation to alcohol use
and parent problems. Explanations for how this vulnera- problems. Emotion regulation was directly related to alcohol
bility through psychosocial maturity to alcohol misuse oc- use problems among young men, consistent with the research
curs among young women but not young men focuses on of Park and Levenson (2002).
the nature of women’s identity development and its possi- It may be that less ability to manage disturbing feelings
ble susceptibility to other influences. Historically, Erikson of depression and frustration are related to attempts to self-
(1968) wrote that a young woman would suspend full iden- medicate with alcohol, leading to excessive and problem-
tity formation until the development of a committed roman- atic use (Catanzaro and Laurent, 2004; Kassel et al., 2000).
tic relationship. If she has not attained a strong sense of Moreover, difficulty in managing emotions could be associ-
psychosocial maturity, then such suspension may leave her ated with a greater behavioral undercontrol that disinhibits a
vulnerable to outside influences and low self-esteem. For ex- youth from caution in using substances. Helping college men
ample, those with less mature psychosocial development are to regulate their frustrations and depression could facilitate
more vulnerable to the pressures of peers to drink excessively fewer alcohol use problems. If young men with problems in
(Adalbjarnardottir, 2002). Although the current analysis did regulating emotions use alcohol to self-medicate, then de-
not include peer pressure, there is an established body of creasing the need for self-medication should diminish this
literature citing the important role of peers in adolescent and avenue to alcohol misuse.
youth drinking (Curran et al., 1997; Kandel and Andrews, Other pathways to alcohol misuse untested in this study
1987). Thus, vulnerability to peer pressure could be one are important to consider. For example, sensation seeking
means by which lower psychosocial maturity is linked to is not so much the undercontrol of behavior as the seeking
alcohol use problems. out of stimulation and excitement and has been related to
Among young women who have not yet gained a strong alcohol use problems (Cloninger et al., 1995). Future re-
sense of psychosocial maturity, it may be that they find search should examine such additional pathways and the
comfort in using alcohol. Such an explanation is consis- interactions of predictors to illuminate for whom and under
tent with Walitzer and Sher’s (1996) research where young what conditions there are processes that link parent variables
women’s greater consumption of alcohol was associated with to development variables and from development to alcohol
lower self-esteem. Alcohol misuse could allow these young misuse. This research identifies important roles for emotion
women to delay or deny dealing with such important issues regulation for men and emotion regulation and psychosocial
as self-definition and self-esteem. The failure to regulate maturity for women.
emotions could lead to drinking to self-medicate and a post-
ponement of psychosocial maturity. The significant associa- Limitations and strengths
tions between alcohol-specific and non-alcohol-specific par-
ent factors is consistent with research that highlights young The study adds to the literature on youth and alcohol mis-
women’s compared to young men’s greater drinking in re- use through use of theory and previous research to derive a

Springer
924 J Youth Adolescence (2007) 36:912–926

particular directional model. Nonetheless, the findings are in through individual-focused strategies (e.g., curriculum infu-
the nature of associations. Contrary to the proposed model, sion or specific skills training) or environmental/contextual
it may be that college students’ problems with alcohol pre- strategies (e.g., campus policies, social norms or other me-
clude forming a healthy psychosocial identity, destabilize dia campaigns) (Larimer and Cronce, 2002; Perkins, 2002).
emotional regulation, and foment more problematic relation- Although such strategies have been found to be somewhat
ships with parents. The cross-sectional design is a limitation effective, they ignore other important developmental pro-
in untangling such directions of effects. However, related cesses that may also contribute to alcohol use problems. It
longitudinal studies support this model’s direction of effects is unlikely that parents spontaneously will become less in-
(Bray et al., 2001; Simmons-Morton and Chen, 2005; Weber trusive in their control efforts or less problematic when they
et al., 1989). drink, but there are other avenues to reduce risk. For exam-
This study assessed only the respondents’ point of view ple, improving the students’ own ways of relating to parents
and focused on just two developmental constructs, emotion could foster stronger emotion regulation in young men and
regulation and psychosocial maturity, as mediating or indi- women and greater psychosocial maturity in young women.
rect variables. Obtaining multiple points of view, including The addition of components designed to strengthen the psy-
parent reports of their own parenting, would be a direction chosocial and emotional well being of students throughout
for future research. In this study, the explained variances their college years would enhance alcohol and drug preven-
in alcohol use problems were 9.4% for young women and tion programs.
5.9% for young men. Adding such variables as peer drinking In sum, the mediational and indirect pathways tested in
and sensation seeking would likely increase the explained this research were moderated by sex of the participant. There
variances. Nonetheless, these current findings highlight the was support for a partial mediation model among young
role of development in college student alcohol misuse. With women. The association between parent problems and fe-
results in this study limited to those of European American male college student alcohol use problems was mediated, in
descent, ethnicity should be examined further. In this study, part, by women’s psychosocial maturity. Psychosocial ma-
there were few participants of other ethnicity compared to turity also provided a partially mediated path for the effects
European American participants, furthermore, those in the of women’s emotion regulation on alcohol use problems.
other ethnicity group were quite heterogeneous. Needed in Among college men, the indirect but central role of emotion
the future is a study that includes greater numbers of di- regulation suggests the possible medicinal use of alcohol to
verse participants along with an expanded set of predictors cope with unmanageable emotions that are associated with
of alcohol use problems. parental problems. More broadly, this study indicated that
Strengths of the study include a theory and research-based problems with parents are linked to developmental issues
model, a large sample size, and results that extend beyond that are, in turn, linked to alcohol use problems.
one region or college campus. In addition, there was simul-
taneous consideration of multiple parenting and develop-
mental variables within the same model along with tests of References
alternative models. The generalizability of the model was
tested with the result that there were important male and Adalbjarnardottir S (2002) Adolescent psychosocial maturity and alco-
female variations detected in the prediction of alcohol use hol use: Quantitative and qualitative analysis of longitudinal data.
Adolesc 37:19–53
problems. Although sex of the participant was used as a Adalbjarnardottir S, Hafsteinsson LG (2001) Adolescents perceived
moderating variable and allowed identification of significant parenting styles and their substance use: Concurrent and longitu-
differences in the pathways to alcohol use problems between dinal analyses. J Res Adolesc 11:401–423
males and females, other moderating variables, such as per- Armsden GC, Greenberg MT (1987) The Inventory of Parent and Peer
Attachment: Individual differences and their relationship to psy-
sonality (cf., Fischer and Wampler, 1994) could be identified chological well-being in adolescence. J Youth Adolesc 16(5):427–
and tested. 454
Balistreri E, Busch-Rossnagel NA, Geisinger KF (1995) Development
and preliminary validation of the ego identity process question-
naire. J Adolesc 18:179–192
Implications and conclusions
Barber BK (1996) Parental psychological control: Revisiting a ne-
glected construct. Child Dev 67:3296–3319
The findings indicate a possible role for college or university Barber BK, Harmon EL (2002) Violating the self: Parental psy-
level interventions to discourage problematic alcohol use. chological control of children and adolescents. In: Barber BK
(ed) Intrusive parenting: How psychological control affects
Most alcohol and drug prevention activities on college cam-
children and adolescents. American Psychological Association,
puses focus on changing students’ specific knowledge, atti- Washington, D C, pp 15–52
tudes, and behaviors in regard to alcohol use (DeJong and Barnow S, Schuckit MA, Lucht M, John U, Freyberger HJ (2002) The
Langford, 2002). These outcomes are accomplished either importance of a positive family history of alcoholism, parental

Springer
J Youth Adolescence (2007) 36:912–926 925

rejection and emotional warmth, behavioral problems and peer Fullinwider-Bush N, Jacobvitz DB (1993) The transition to young
substance use for alcohol problems in teenagers: A path analysis. adulthood: Generational boundary dissolution and female iden-
J Stud Alcohol 63:305–315 tity development. Fam Process 32:87–103
Baron R, Kenny D (1986) The moderator-mediator variable distinc- Gross JJ (1998) The emerging field of emotion regulation: An integra-
tion in social psychological research: conceptual, strategic, and tive review. Rev Gen Psychol 2:271–299
statistical considerations. J Pers Soc Psychol 51:1173–1182 Grotevant HD, Cooper CR (1985) Patterns of interaction in family
Bell NJ, Forthun LF, Sun SW (2000) Attachment, adolescent compe- relationships and the development of identity exploration in ado-
tencies, and substance use: Developmental considerations in the lescence. Child Dev 56:415–428
study of risk behaviors. Subst Use Misuse 35(9):1177–1206 Ham LS, Hope DA (2003) College students and problematic drinking:
Bishop DI, Macy-Lewis JA, Schnekloth CA, Puswella S, Struessel GL A review of the literature. Clin Psychol Rev 23:719–759
(1997) Ego identity status and reported alcohol consumption: A Hodgins DC, Shimp L (1995) Identifying adult children of alcoholics:
study of first-year college students. J Adolesc 20:209–218 Methodological review and a comparison of the CAST-6 with
Bray JH, Adams GJ, Getz JG, Baer PE (2001) Developmental, family, other methods. Addiction 90:255–267
and ethnic influences on adolescent alcohol usage: A growth curve Holmbeck GN (1997) Toward terminological, conceptual, and statis-
approach. J Fam Psychol 15:3001–3314 tical clarity in the study of mediators and moderators: Examples
Bray JH, Adams GJ, Getz JG, McQueen A (2003) Individuation, peers, from the child-clinical and pediatric psychology literatures. J Con-
and adolescent alcohol use: A latent growth analysis. J Consult sult Clin Psychol 65:599–610
Clin Psychol 71:553–564 Hood AB (1997) The Iowa student development inventories (2nd edn).
Bijttebier P, Goethals E, Ansoms S (2006) Parental drinking as a risk HiTech Press, The University of Iowa
factor for children’s maladjustment: The mediating role of family Hu L, Bentler PM (2000) Cutoff criteria for fit indexes in covariance
environment. Psychol Addict Behav 20:126–130 structure analysis: Conventional criteria versus new alternatives.
Browne MW, Cudeck R (1993) Alternative ways of assessing model fit. Struct Eqn Model 6:1–55
In: Bollen KA, Long JS (eds) Testing structural equation models. Iacono WG, Carlson SR, Taylor J, Elkins IJ, McGue M (1999) Behav-
Sage, Newbury Park, CA, pp 136–162 ioral disinhibition and the development of substance-use disorders:
Byrne BM (1994) Testing for factorial validity, replication, and in- Findings from the Minnesota twin family study. Dev Psychopathol
variance of a measuring instrument: A paradigmatic application 11:869–900
based on the Maslach Burnout Inventory. Mult Behav Res 29:289– Jacob T, Johnson S (1997) Parenting influences on the development of
311 alcohol abuse and dependence. Alcohol Health Res World 21:204–
Campos JJ, Mumme DL, Kermoian R, Campos RG (1994) The devel- 210
opment of emotion regulation: Biological and behavioral consid- Johnstone BM (1994) Sociodemographic, environmental, and cultural
erations. Monogr Soc Res Child Dev 59:284–303 influences on adolescent drinking behavior. In: Zucker R, Boyd G,
Catanzaro SJ, Laurent J (2004) Perceived family support, negative Howard J (eds) The development of alcohol problems: Exploring
mood regulation expectancies, coping, and adolescent alcohol the biopsychosocial matrix of risk (NIH Publication No. 94-3495).
use: Evidence of mediation and moderation effects. Addict Be- US Department of Health and Human Services, Rockville, MD,
hav 29:1779–1797 pp 169–203
Chickering AW, Reisser L (1993) Education and identity (2nd edn). Jones JW (1983) The children of alcoholics screening test: A validity
Jossey-Bass Publishers, San Francisco study. Bull Soc Psychol Addict Behav 2:155–163
Clements R (1999) Prevalence of alcohol-use disorders and alcohol- Jones RM, Hartmann BR (1988) Ego identity: Developmental differ-
related problems in a college student sample. Journal of American ences and experimental substance use among adolescents. J Ado-
College Health 48:111–118 lesc 11:347–360
Cloninger CR, Sigvardsson S, Przybeck TR, Svrakic DM (1995) Per- Kandel DB, Andrews K (1987) Processes of adolescent socialization
sonality antecedents of alcoholism in a national area probability by parents and peers. Int J Addict 22:319–342
sample. European Arch Psych Clin Neurosc 245:239–244 Kassel JD, Jackson SI, Unrod M (2000) Generalized expectancies for
Coffelt NL, Forehand R, Olson AL, Jones DJ, Gaffney CA, Zens MS negative mood regulation and problem drinking among college
(2006) A longitudinal examination of the link between parent students. J Stud Alcohol 61:332–340
alcohol problems and youth drinking: The moderating roles of Larimer ME, Cronce JM (2002) Identification, prevention and treat-
parent and child gender. Addict Behav 31:593–605 ment: A review of individually-focused strategies to reduce prob-
Cooper ML, Frone MR, Russell M, Mudar P (1995) Drinking to regulate lematic alcohol consumption by college students. J Stud Alcohol
positive and negative emotions: A motivational model of alcohol 14:148–163
use. J Pers Soc Psychol 69:990–1005 Lei M, Lomax RG (2005) The effect of varying degrees of nonnormality
Curran PJ, Stice E, Chassin L (1997) The relation between adolescent in structural equation modeling. Struct Eqn Model 12:1–27
alcohol use and peer alcohol use: A longitudinal random coeffi- Markstrom CA, Sabino VM, Turnder BJ, Berman RC (1997) The
cients model. J Consult Clin Psychol 65:130–140 psychosocial inventory of ego strengths: Development and val-
DeJong W, Langford LM (2002) A typology for campus-based alcohol idation of a new Eriksonian measure. J Youth Adolesc 26:705–
prevention: Moving toward environmental management strategies. 732
J Stud Alcohol 14:140–147 McDonald RP, Ho M-HR (2002) Principles and practice in reporting
DeLucia C, Belz A, Chassin L (2001) Do adolescent symptomatol- structural equation analyses. Psychol Methods 7:64–82
ogy and family environment vary over time with fluctuations in Nolen-Hoeksema S (2004) Gender differences in risk factors and conse-
paternal alcohol impairment? Dev Psychol 37:207–216 quences for alcohol use and problems. Clin Psychol Rev 24:981–
Dishion TJ, McMahon RJ (1998) Parental monitoring and the preven- 1010
tion of child and adolescent problem behavior: a conceptual and Nurmi J, Berzonsky MD, Tammi K, Kinney A (1997) Identity pro-
empirical formulation. Clin Child Fam Psychol Rev 1:61–75 cessing orientation, cognitive and behavioural strategies and well-
Erikson E (1968) Identity: Youth and crisis. Norton, NY being. Int J Behav Dev 21:555–570
Fischer JL, Wampler RS (1994) Abusive drinking in young adults: Pandina RJ, Johnson V (1990) Serious alcohol and drug problems
Personality type and family role as moderators of family-of-origin among adolescents with a family history of alcoholism. J Stud
influences. J Marr Fam 56:469–479 Alcohol 51:278–282

Springer
926 J Youth Adolescence (2007) 36:912–926

Park CL, Levenson MR (2002) Drinking to cope among college stu- Strayer J (2002) The dynamics of emotions and life cycle identity.
dents: Prevalence, problems, and coping processes. J Stud Alcohol Identity: Int J Theory Res 2:47–79
63:486–497 Walitzer KS, Sher KJ (1996) A prospective study of self-esteem and
Patock-Peckham JA, Morgan-Lopez AA (2006) College drinking be- alcohol use disorders in early adulthood: Evidence for gender
haviors: Mediational links between parenting styles, impulse con- differences. Alcoholism: Clin Exp Res 20:1118–1124
trol, and alcohol-related outcomes. Psychol Addict Behav 20:117– Wampler R, Fischer J, Thomas M, Lyness K (1993) Young adult off-
125 spring and their families of origin: Cohesion, adaptability, and
Perkins HW (2002) Social norms and the prevention of alcohol misuse addiction. J Subst Abuse 5:195–201
in collegiate contexts. J Stud Alcohol 14:164–172 Wechsler H, Davenport A, Dowdall G, Moeykens B, Castillo S (1994)
Perosa LM, Perosa SL, Tam HP (1996) The contribution of family Health and behavioral consequences of binge drinking in college:
structure and differentiation to identity development in females. J A national survey of students at 140 campuses. J Am Med Assoc
Youth Adolesc 25:817–837 272:1672–1677
Pidcock BW, Fischer JL, Forthun LF, West SL (2000) Hispanic and Wechsler H, Lee JE, Kuo M, Lee H (2000) College binge drinking in
Anglo college women’s risk factors for substance abuse and eating the 1990s: A continuing problem. J Am Coll Health 48:199–210
disorders. Addict Behav 25:705–723 Wills TA, Dishion TJ (2004) Temperament and adolescent substance
Preacher KJ, Hayes AF (2004) SPSS and SAS procedures for estimating use: A transactional analysis of emerging self-control. J Clin Child
indirect effects in simple mediation models. Behav Res Methods, Adolesc Psychol 33:69–81
Instrum, Comput 36:717–731 Wills TA, Resko JA, Ainette MG, Mendoza D (2004) Role of parent
Simmons-Morton B, Chen R (2005) Latent growth curve analyses of support and peer support in adolescent substance use: A test of
parent influences on drinking progression among early adoles- mediated effects. Psychol Addict Behav 18:122–134
cents. J Stud Alcohol 66:5–13 Windle M, Davies PT (1999) Depression and heavy alcohol use
Sivo SA, Fan X, Witta EL, Willse JT (2006) The search for “optimal” among adolescents: Concurrent and prospective relations. Dev
cutoff properties: Fit index criteria in structural equation modeling. Psychopathol 11:823–844
J Exp Educ 74:267–288 Zucker RA, Fitzgerald HE, Refior SK, Pallas DM, Ellis DA (2000)
Stone G, Buehler G, Barber BK (2002) Interparental conflict, parental The clinical and social ecology of childhood for children of alco-
psychological control, and youth problem behavior. In: Barber holics: Description of a study and implications for a differentiated
BK (ed) Intrusive parenting: How psychological control affects social policy. In: Fitzgerald HE, Lester BM, Zuckerman BS (eds)
children and adolescents. American Psychological Association, Children of addiction: Research, health, and policy issues. Rout-
Washington, DC, pp 53–95 ledge/Falmer Press, New York, pp 109–141

Springer

You might also like