Professional Documents
Culture Documents
6collins Et Al. (2007)
6collins Et Al. (2007)
This study provided a simultaneous confirmatory test of the theory of planned behavior (TPB) in
predicting heavy episodic drinking (HED) among college students. It was hypothesized that past HED,
drinking attitudes, subjective norms, and drinking refusal self-efficacy would predict intention, which
would, in turn, predict future HED. Participants consisted of 131 college drinkers (63% women) who
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
reported having engaged in HED in the previous 2 weeks. Participants were recruited and completed
This document is copyrighted by the American Psychological Association or one of its allied publishers.
questionnaires within the context of a larger intervention study (see Collins & Carey, 2005). Latent factor
structural equation modeling was used to test the ability of the TPB to predict HED. Chi-square tests and
fit indices indicated good fit for the final structural models. Self-efficacy and attitudes but not subjective
norms significantly predicted baseline intention, and intention and past HED predicted future HED.
Contrary to hypotheses, however, a structural model excluding past HED provided a better fit than a
model including it. Although further studies must be conducted before a definitive conclusion is reached,
a TPB model excluding past behavior, which is arguably more parsimonious and theory driven, may
predict HED among college drinkers better than a model including past behavior.
Keywords: college drinking, alcohol use, theory of planned behavior, heavy episodic drinking, college
students
Ajzen’s (1988) theory of planned behavior (TPB) has been address these issues, we used longitudinal, simultaneous, and
widely used to describe health behaviors over which one has confirmatory model testing to examine the ability of the TPB to
volitional control (Godin & Kok, 1996). Researchers over the past predict intention and HED among college drinkers.
decade have applied this theory to describe the psychological
sequelae involved in college drinking. Most studies to date, how-
TPB
ever, have focused on contributions of individual model compo-
nents instead of evaluating simultaneous model fit, and the results According to the TPB, a certain set of motivational factors leads
of some of these studies have not been entirely consistent with the to an intention to act in a certain way (Ajzen, 1991). Given the
predicted model (e.g., Wall, Hinson, & McKee, 1998). Moreover, right opportunity, people will translate this intention into behavior.
the role of past behavior in the TPB has not been adequately The role of intention in this model is believed to be twofold:
defined and remains somewhat controversial (Ajzen, 2002c). Fi- Attitudes, subjective norms, and perceived behavioral control
nally, although the theory has been applied to college drinking should predict intention, and intention, in turn, should predict
(Conner, Warren, Close, & Sparks, 1999; O’Callaghan, Chant, behavior. Several studies focusing on college drinkers have estab-
Callan, & Baglioni, 1997), it has been used less often to describe lished the prediction of behavior by intention (Armitage, Norman,
the occurrence of heavy episodic drinking (HED) among college & Conner, 2002; Conner et al., 1999; O’Callaghan et al., 1997).
students (see Johnston & White, 2003; Norman, Bennett, & Lewis, Attitudes refer to people’s evaluation of their own behavior.
1998; Wall et al., 1998). Because HED can be particularly harmful Attitudes concerning alcohol use appear to be important in con-
to college students (Clements, 1999; Hingson, Heeren, Zakocs, currently predicting drinking quantity and frequency (Leigh,
Kopstein, & Wechsler, 2002), it deserves further examination. To 1989). Further, longitudinal evidence has shown that positive
attitudes toward alcohol use are positively correlated with future
alcohol use among college drinkers (Stacy, Bentler, & Flay, 1994).
Susan E. Collins and Kate B. Carey, Center for Health and Behavior, The subjective norm is a target person’s perception of others’
Syracuse University. evaluations of the target person performing a behavior (Ajzen,
Susan E. Collins is now at the Smoking Cessation Research Group, 1991). This construct may be broken down into two components:
University Hospital Tübingen. (a) perception of others’ evaluations (also referred to as normative
This research was supported by a Creative Research Grant from the beliefs) and (b) the importance of the others’ opinions to the target
College of Arts and Sciences, Syracuse University, to Susan E. Collins and
person (representing motivation to comply with perceived norms).
by National Institute on Alcohol Abuse and Alcoholism Grants R01-
AA12518 and K02-AA15574 to Kate B. Carey.
In other words, a person is more likely to perform a behavior if
Correspondence concerning this article should be addressed to Susan E. others, whose opinions the person values, approve of the person
Collins, Arbeitskreis Raucherentwöhnung, Universitätsklinikum Tübingen, performing the behavior. A conceptualization of HED among
Herrenbergerstr. 23, D-72070 Tübingen, Germany. E-mail: college students that includes subjective norms is important be-
susan.collins@gmx.net cause adolescents and college students report being influenced
498
THEORY OF PLANNED BEHAVIOR AND COLLEGE DRINKING 499
more by their peers than by adults (Beck & Treiman, 1996; however, have been difficult to develop because it is unclear
Perkins, 2002). Injunctive norms, as commonly conceptualized in whether people can accurately judge the habituality of their own
the college drinking literature, resemble the evaluative component behavior (Ouellette & Wood, 1998). Difficulties defining and
of subjective norms in the TPB, and ample evidence has linked not directly testing past behavior as a measure of habit have thus
only descriptive norms (perceptions of how others drink) but also complicated the interpretation of past behavior as a predictor in
injunctive norms (perceived peer approval of drinking) to college TPB models.
drinking (Larimer, Turner, Mallett, & Geisner, 2004; Perkins, Second, proponents of the inclusion of past behavior have
2003). asserted that it is a consistently strong predictor of future behavior
Perceived behavioral control, or the perceived ease or difficulty that should be accounted for (e.g., Stacy et al., 1994). In the
of performing a behavior, is believed to both indirectly (through its interest of correct model specification or the inclusion of all
association with intentions) and directly influence behavior variables known to be important in predicting a dependent vari-
(Ajzen, 1988, 1991). Both self-efficacy, which has been defined as able, it can be argued that past behavior should be included in
one’s perceived control over a certain behavior in a specific models of TPB. However, it has been asserted that past behavior’s
situation (Bandura, 1977; Marlatt & Gordon, 1985), and control- predictive strength may be due to a methodological artifact: com-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
lability beliefs, or beliefs that the performance of a behavior mon method variance, or the correlation between identical indica-
depends on the individual alone, make up the concept of perceived tors repeatedly assessed using the same method (Ajzen, 2002c). To
behavioral control (Ajzen, 2002b). Perceived behavioral control test the effect of common method variance when predicting future
and drinking refusal self-efficacy have been supported in the behavior on the basis of past behavior, researchers in one study
literature as strong predictors of college drinking, such that greater used both self-report and observation methods in the alcohol-use
drinking refusal self-efficacy and perceived behavioral control assessment (Conner et al., 1999). Contrary to hypotheses, past
predict lower drinking quantity, decreased frequency, and fewer drinking explained variance above and beyond the TPB variables
problems (Johnston & White, 2003; Young, Connor, Ricciardelli, even when common method variance was eliminated. This study,
& Saunders, 2006). however, did not use a simultaneous confirmatory test of the TPB
and did not focus on HED specifically. Considering these differing
TPB and Past Drinking Behavior perspectives and findings, an empirical comparison of competing
theoretical models (inclusion vs. exclusion of past HED) would be
Including past behavior in the TPB model may improve the a helpful addition to the literature.
prediction of future behavior, but this modification has engendered
controversy. The debate regarding the role of past behavior in TPB TPB and HED
models typically centers on challenges in defining the past behav-
ior construct and the relative importance of including a significant Three studies to date have examined the influence of attitudes,
yet perhaps atheoretical predictor of future behavior. Contrasting subjective norms, and perceived behavioral control on intention to
arguments for and against the inclusion of past behavior in TPB engage in HED (also referred to as “to drink too much” and “binge
models are presented below. drinking” in Johnston & White, 2003; Norman et al., 1998; Wall
First, it has been asserted that consistent performance of a et al., 1998). Wall et al. (1998) used stepwise or statistical regres-
behavior represents habit, which may be a relevant predictor of sions (conducted separately by gender) to test the ability of con-
various types of behavior (Ouellette & Wood, 1998). The results of currently assessed TPB variables to predict intention and HED. For
a meta-analysis of 64 studies, which evaluated past behavior women, attitudes, perceived behavioral control, and sociability
prediction of future behavior, indicated that past behavior was a expectancies predicted intention, whereas attitudes, perceived be-
stronger predictor than intention of behaviors repeated on a weekly havioral control, and subjective norms predicted intentions in men.
basis in stable, conducive contexts (rs ⫽ .45 and .27, respectively; Intentions, perceived behavioral control, and assertiveness expect-
ps ⬍ .001; Ouellette & Wood, 1998). For behaviors that were not ancies predicted HED among women, whereas intentions and
performed regularly, such as voting or getting a flu shot, past expectancies of higher sexual functioning predicted HED among
behavior was a weak predictor of future behavior relative to men.
intention (rs ⫽ .12 and .82, respectively; ps ⬍ .001). Another Another study used partial correlations and regressions to test
study found that past or habitual behavior significantly predicted the concurrent associations among attitudes, perceived behavioral
future behavior even after controlling for a range of other cognitive control, subjective norms, beliefs (i.e., behavioral, normative, and
components, including attitude accessibility, self-concept, intent, control beliefs), and HED (Norman et al., 1998). In this study,
attitudes, subjective norms, and perceived behavioral control gender explained 9% and TPB variables an additional 29% of the
(Ouellette, 1996). These authors have argued that, taken together, variance in HED. Of the TPB variables, only perceived behavioral
this evidence supports past behavior as a habit construct that may control and positive control beliefs reached significance. Thus, in
be helpful in understanding longitudinal behavior patterns. spite of its high multiple correlation, this study provided only
However, it has been argued that the habitual nature of partic- partial support for the concurrent prediction of HED by TPB
ular behaviors may not be assumed simply because they are variables.
regularly performed or because other variables do not account for Finally, a more recent study used a longitudinal model to test the
the variance in future behavior (Ajzen, 2002c). Instead, it has been prediction of intentions and HED by TPB variables (Johnston &
proposed that a direct measure of habit should be used to establish White, 2003). The results of the multiple regression indicated that
its role in behavior prediction (Ajzen, 2002c; Verplanken, Aarts, attitudes, subjective norms, and self-efficacy significantly pre-
van Knippenberg, & van Knippenberg, 1994). Measures of habit, dicted 69% of the variance in intention to engage in HED. Addi-
500 COLLINS AND CAREY
structural equation modeling) would provide a more appropriate Demographic information. The Personal Information Ques-
test of the model. Finally, despite the fact that research supports a tionnaire was used to assess participants’ age, gender, year in
strong correlation between past and future drinking behavior (Con- college, ethnicity, residence, and membership in an on-campus
ner et al., 1999; McMillan & Conner, 2003; O’Callaghan et al., Greek organization.
1997), none of the studies mentioned above accounted for this Intention factor. Intentions were measured by the Behavioral
effect. Intentions Questionnaire (BIQ; Neal & Carey, 2004). The BIQ
consists of seven items asking participants to rate their intentions
of increasing and decreasing their drinking quantity or frequency,
Proposed Study peak drinking, and HED in the next 2 weeks. Responses were
scored on a scale of 1 to 6, where 1 ⫽ definitely will not do and
The TPB is a potentially useful conceptual framework for un- 6 ⫽ definitely will do. Questions 2, 4, and 6, which assessed the
derstanding drinking among college students. The studies that have probability of decreased alcohol use, were reverse scored. The BIQ
tested the prediction of HED by TPB model components have not was supplemented with an item that asked participants to use the
provided methodologically optimal tests of the theory. Our goal in 6-point scale described above to rate the probability of their
this study was therefore to improve on the methodology used in drinking to get drunk in the next 2 weeks.
previous studies by using a longitudinal design, a simultaneous and At the suggestion of the scale’s author (D.J. Neal, personal
confirmatory rather than an exploratory testing method, and error- communication, November 7, 2005), the seventh BIQ item was
free latent factors (i.e., latent factors from which measurement used separately from its original scale. An exploratory principal
error is estimated and removed). Further, in this study, we com- components analysis was conducted on the modified six-item scale
pared models including and excluding the potential contribution of and indicated a one-factor solution (eigenvalue ⫽ 3.82), which
past HED to the prediction of future HED in the context of the accounted for 64% of the variance and evinced acceptable internal
TPB. In doing so, we aimed to provide a better test of whether past consistency (␣ ⫽ .88). In the analyses, three indicators represented
HED, attitudes, self-efficacy, and subjective norms predict behav- Time 1 intention: (a) the sum of the six BIQ items, (b) participants’
ioral intention and future HED among college students.1 intention to engage in a heavy-drinking episode (as defined by
Wechsler et al., 2002) in the next 2 weeks, and (c) participants’
intention to drink to get drunk in the next 2 weeks. The internal
Hypotheses consistency of the three indicators was acceptable (␣ ⫽ .83).
Attitudes factor. The two attitude indicators in the current
The first set of hypothesized models included a past HED factor study were selected from the Global Attitude Scale (AS; Simons
(Time 1 HED). As shown in Figure 1, it was predicted that Time and Carey, 1998). The AS has previously evinced good reliability
1 HED, subjective norms, drinking refusal self-efficacy (referred (␣ ⱖ .91) and discriminant validity (Simons & Carey, 1998).
to from this point as self-efficacy), and attitudes would predict Participants rated their overall opinions about drinking to get
Time 1 intention. In turn, Time 1 HED and intention would predict drunk along an unnumbered, 9-point scale framed by two opposing
future HED (Time 2 HED). To test Ajzen’s (1991) assertion that word pairs on either end. The two global attitude word pairs used
self-efficacy both indirectly (via intention) and directly predicts in the current study were like and dislike and desirable and
behavior at Time 2, we also used nested models to test whether undesirable. In this study, higher attitude scores represented more
direct effects of self-efficacy on Time 2 HED would improve positive attitudes toward drinking. These items evinced good in-
model fit over a reduced model in which self-efficacy served as an ternal consistency (␣ ⫽ .87).
indirect explanatory variable via intention (represented by a dotted Subjective norms factor. Two items from the Subjective
line in Figure 1). The second set of models excluded Time 1 HED. Norms Questionnaire (SNQ; based on guidelines from Ajzen,
It was hypothesized that self-efficacy, attitudes, and subjective
norms would predict intention, which would, in turn, predict Time 1
In the current study, self-efficacy, but not beliefs about controllability,
2 HED (see Figure 2). It was further predicted that the model was directly assessed. Thus, perceived behavioral control was operation-
containing Time 1 HED would be a better fitting model than the alized as a unitary self-efficacy factor in this study and is referred to as
model excluding Time 1 HED. such from this point forward.
THEORY OF PLANNED BEHAVIOR AND COLLEGE DRINKING 501
T1 HED
+
+ D D
ATT
+
+
INT T2 HED
+
NORMS - -
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
SE
This document is copyrighted by the American Psychological Association or one of its allied publishers.
.38*
T1 HED
.20 .38*
ATT .23*
.26*
-.24* INT T2 HED
.47*
.19
D D
NORMS -.32*
-.52*
SE
-.46*
Figure 1. Top panel: Hypothesized model including Time 1 drinking. ⫹ ⫽ hypothesized positive association;
- ⫽ hypothesized negative association; D ⫽ disturbance (unexplained variance in endogenous latent factors); T1
HED ⫽ Time 1 heavy episodic drinking; ATT ⫽ Time 1 attitudes; NORMS ⫽ Time 1 subjective norms; SE ⫽
Time 1 self-efficacy; T2 HED ⫽ Time 2 heavy episodic drinking. Double-headed arrows imply nondirectional
covariance between the variables. The dotted line represents the nested model including a direct effect for SE
on T2 HED. For ease of interpretation, measured indicators are not shown. Bottom panel: Final model including
Time 1 drinking. *p ⫽ .05.
2002a) served as indicators of the subjective norms latent factor. factor. On a 6-point Likert-type scale, participants indicated their
Participants reported how much an average American college confidence they could resist drinking to get drunk in each of the
student and their closest friend would approve or disapprove of hypothetical situations presented. Good internal consistency (␣s ⫽
their drinking to get drunk on a 5-point Likert-type scale, where .87–.94) and concurrent and discriminant validity have been es-
1 ⫽ highly disapprove and 5 ⫽ highly approve. Next, participants tablished for this measure (Baldwin, Tian, & Young, 1993; Young
rated the importance of these individuals’ opinions to them on a et al., 1991). In this study, summary scores for the Social Pressure
scale ranging from 1 ⫽ highly unimportant to 5 ⫽ highly impor- and Emotional Relief Scales were formed, such that higher scores
tant. Subjective norm indicators were ascertained for each group indicated higher self-efficacy to resist drinking to get drunk (␣s ⫽
by multiplying the approval of the target group by the participants’ .88 and .93, respectively).
report of the importance of the target group’s opinion (Ajzen, For the third indicator of this factor, participants answered the
1991, 2002a). The average American college student and closest question “How confident are you that, if you wanted to, you could
friend indictors evinced good reliability (␣ ⫽ .91). cut down on your drinking?” Responses were made on a 7-point
Self-efficacy factor. Total scores for the Social Pressure and Likert-type scale, where 1 ⫽ not at all confident and 7 ⫽ very
Emotional Relief Scales of the Drinking Refusal Self-Efficacy confident. Pilot testing indicated that higher scores on this item
Questionnaire (DRSEQ; adapted from Young, Tian, & Crook, were significantly associated with lower 1-week drinking quantity
1991) served as two of the three indicators of the self-efficacy and fewer drinks consumed on the heaviest drinking day (rs ⫽
502 COLLINS AND CAREY
ATT
D
D
+
+
+ INT T2 HED
NORMS
-
-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
SE
This document is copyrighted by the American Psychological Association or one of its allied publishers.
.49*
ATT
.33*
.14 .46*
-.53* NORMS INT T2 HED
-.34*
D D
-.48* SE
Figure 2. Top panel: Hypothesized model excluding Time 1 drinking. ⫹ ⫽ hypothesized positive association;
- ⫽ hypothesized negative association; D ⫽ disturbance (unexplained variance in endogenous latent factors);
ATT ⫽ Time 1 attitudes; NORMS ⫽ Time 1 subjective norms; SE ⫽ Time 1 self-efficacy; T2 HED ⫽ Time
2 heavy episodic drinking. Double-headed arrows imply nondirectional covariance between the variables. The
dotted line represents the nested model including a direct effect for SE on T2 HED. For ease of interpretation,
measured indicators are not shown. Bottom panel: Final model excluding Time 1 drinking. *p ⫽ .05.
⫺.33 and ⫺.34, respectively, ps ⬍ .05). The alpha coefficient for who reported having experienced a heavy-drinking episode in the
all three indicators indicated an internal consistency of .65. past 2 weeks (n ⫽ 131) were randomized to receive no treatment
HED factors. Two open-ended quantity and frequency items or one of two minimal interventions involving single-session de-
measured by the F-Q questionnaire (adapted from Collins, Carey, cisional balance exercises (see Collins & Carey, 2005). Approxi-
& Sliwinski, 2002) served as the indicators for Time 1 and Time mately 2 weeks after the interventions, all participants attended a
2 HED: number of drinks consumed during the peak drinking posttest questionnaire session. Participants were given course
occasion in the past 2 weeks and number of heavy-drinking epi- credit for their participation.
sodes. Heavy-drinking episodes were assessed using two gender-
specific items and were defined as having consumed five or more Data Analysis Plan
drinks on one occasion for men and four or more drinks on one
occasion for women (Wechsler et al., 2002). The internal consis- The main hypotheses in this study were tested with a series of
tencies of the Time 1 and Time 2 HED indicators were good (␣s ⫽ latent factor structural equation models (SEMs) using the EQS 6.1
.78 and .82, respectively). program (Bentler, 2004). This type of analysis allows for simul-
taneous confirmatory model testing, which was deemed appropri-
ate for evaluating an established model like the TPB. All models
Procedure
except the multiple indicators and multiple causes (MIMIC) model
College students (N ⫽ 234) who signed up to participate in an described below made use of the full information maximum like-
alcohol-use study gave written, informed consent and filled out all lihood method to estimate parameters and standard errors for the
questionnaires listed in the Measures section as part of a larger entire data set. Two types of model fit assessment were used in the
questionnaire battery (see Collins & Carey, 2005). Participants current study: (a) descriptive goodness-of-fit indices (comparative
THEORY OF PLANNED BEHAVIOR AND COLLEGE DRINKING 503
fit index [CFI] ⱖ .95 and root-mean-square error of approximation preliminary independence SEMs, which tested the hypothesis that
[RMSEA] ⱕ .06 were interpreted as indications of close model fit; the variables were uncorrelated, were easily rejected (all ps ⬍
Hu & Bentler, 1999) and (b) exact model testing using the model .001).
chi-square. Traditionally, exact fit is determined when the null
hypothesis is accepted using the model chi-square test (Hayduk &
Glaser, 2000). Although exact fit is the recognized term for meet- Models Including Time 1 HED
ing this criterion, it does not imply perfect model fit and is not Evaluation of the measurement models. The baseline measure-
interpreted as such in this article. ment model included the following five latent factors (measured
The SEM analyses were conducted according to the two-step indicators in parentheses): attitudes (AS dislike–like, AS
modeling approach (Anderson & Gerbing, 1988; Pentz & Chou, desirable– undesirable), subjective norms (SNQ college student,
1994). In the first step, saturated measurement models, which are SNQ closest friend), self-efficacy (DRSEQ social, DRSEQ emo-
essentially confirmatory factor analyses, were used to test the tional, general self-efficacy), Time 1 HED (number of drinks
relationships of the measured variables to their hypothesized latent consumed on the peak drinking occasion, frequency of heavy-
constructs. After the appropriateness of the measurement model
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
was established, hypothesized relationships among the latent fac- drunk, intention to engage in a heavy-drinking episode). All latent
tors were tested in the second step. factors were initially allowed to covary. To ensure model identi-
fication, we placed equality constraints on the error variances of
Results the indicators of subjective norms, attitudes, and Time 1 HED,
respectively.
Exploratory Data Analysis
The first model tested was the baseline measurement model.
At the baseline assessment, participants reported having con- Although fit indices indicated good fit of the hypothesized model,
sumed an average of 28.24 (SD ⫽ 25.39) drinks and having CFI ⫽ .97, RMSEA ⫽ .05, the robust scaled chi-square test did not
experienced 2.62 (SD ⫽ 2.23) heavy-drinking episodes in the past indicate exact fit, T(47, N ⫽ 131) ⫽ 65.38, p ⫽ .04. To improve
2 weeks. On their peak drinking occasion, participants reported model fit, we examined the multivariate Wald test and, accord-
having consumed 8.95 drinks (SD ⫽ 5.77). ingly, dropped a nonsignificant covariance path, F1F3 (Time 1
The HED variables (i.e., number of drinks consumed at the peak drinking, subjective norms), from the model. The resulting hypoth-
drinking occasion and number of heavy-drinking episodes) exhib- esized model evinced adequate fit, T(48, N ⫽ 131) ⫽ 65.68, p ⫽
ited positively skewed distributions and were therefore trans- .05, CFI ⫽ .97, RMSEA ⫽ .05.
formed using a square-root transformation for the main analyses. The next step confirmed the appropriateness of the longitudinal
Other variables (i.e., the subjective norm variables, the Emotional measurement model. Except for the addition of the Time 2 HED
Drinking Scale of the DRSEQ, and two intentions items) did not factor, the measured indicators and latent factors were identical to
exhibit normal distributions but could not be successfully trans- those in the baseline model. When using a longitudinal model, one
formed. For this reason, only robust statistics (i.e., Santorra– must also run tests of measurement invariance to ensure that
Bentler T and other tests using robust standard errors) are reported factors represent the same constructs across time and may be
(Bentler, 2004). consistently interpreted (Grouzet, Otis, & Pelletier, 2006; Little,
As shown in Table 1, bivariate Spearman correlations between 1997). In accordance with these standards, the final measurement
the measured variables provided further support for the construct model fulfilled the criteria for configural, metric, and partial scalar
validity of the proposed factors. Although not reported here, all invariance (see Thompson & Green, 2006) and exhibited good
Table 1
Bivariate Spearman’s Rho Correlations Between Model Indicators
Measure 1 2 3 4 5 6 7 8 9 10 11 12 13 14
model fit, T(66, N ⫽ 131) ⫽ 81.03, p ⫽ .10, CFI ⫽ .98, .10. Finally, the multivariate Lagrange multiplier (LM) test, which
RMSEA ⫽ .04. is a modifier index indicating possibilities for improving model fit,
MIMIC model. Because the second data collection period in- revealed a nonsignificant improvement in fit with the addition of
volved a treated sample, a preliminary model was run to test the the direct self-efficacy effect, 2(70, N ⫽ 131) ⫽ 2.34, p ⫽ .13. On
appropriateness of collapsing across treatment conditions. It was the basis of these tests, this model was rejected.
hypothesized that no group effects would be found on Time 2
HED, as observed in a previous study (Collins & Carey, 2005). To Models Excluding Time 1 HED
test this hypothesis, we ran a MIMIC model, which is the SEM
equivalent of multiple regression with dichotomous predictors The next set of models tested the TPB excluding Time 1 HED
(Hancock, 2004). MIMIC models allow for the inclusion of one or (see Figure 2 for the hypothesized model). Tests of the saturated
more dummy-coded variables to test population differences on a measurement model indicated exact fit, T(N ⫽ 131, df ⫽ 46) ⫽
specific dependent variable within an SEM model. This procedure 47.93, p ⫽ .39, CFI ⫽ 1.00, RMSEA ⫽ .01. No post hoc modi-
was deemed appropriate because (a) the sample size in the current fications were made. Structural model tests indicated that the
study was too small to conduct a structured means analysis, (b) any hypothesized model evinced exact fit, T(N ⫽ 131, df ⫽ 49) ⫽
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
square testing can benefit from smaller sample sizes because the Bentler, P. M. (2004). EQS 6 structural equations program manual.
associated lower power is more likely to produce nonsignificant Encino, CA: Multivariate Software.
effects. Thus, despite these encouraging findings, future studies Burnham, K. P., & Anderson, D. R. (2004). Multimodal inference: Under-
using SEM to replicate these results should clearly aim for larger standing AIC and BIC in model selection. Sociological Methods &
sample sizes. Research, 33, 261–304.
Clements, R. (1999). Prevalence of alcohol-use disorders and alcohol-
Finally, the fact that this study was conducted using data from
related problems in a college student sample. Journal of American
a randomized study complicates the prediction of Time 2 HED. College Health, 48, 111–118.
Specifically, participants underwent a brief intervention for at-risk Collins, S. E., & Carey, K. B. (2005). Lack of effect for decisional balance
drinking between Times 1 and 2. The preliminary MIMIC model as a brief motivational intervention for at-risk college drinkers. Addictive
confirmed the Collins and Carey (2005) findings that there were no Behaviors, 30, 1425–1430.
significant group differences in participants’ Time 2 drinking. Collins, S. E., Carey, K. B., & Sliwinski, M. J. (2002). Mailed personalized
However, this analysis could not rule out Group ⫻ TPB Variable normative feedback as a brief intervention for at-risk college drinkers.
interactions or undetected differences in covariance structures. Journal of Studies on Alcohol, 63, 559 –567.
Despite its limitations, this study provided overall support for Conner, M., Warren, R., Close, S., & Sparks, P. (1999). Alcohol consump-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Neal, D. J., & Carey, K. B. (2004). Developing discrepancy within self- Stacy, A. W., Bentler, P. M., & Flay, B. R. (1994). Attitudes and health
regulation theory: Use of personalized normative feedback and personal behavior in diverse populations: Drunk driving, alcohol use, binge
strivings with heavy-drinking college students. Addictive Behaviors, 29, eating, marijuana use, and cigarette use. Health Psychology, 13, 73– 85.
281–297. Terry, D. J., & O’Leary, J. E. (1995). The theory of planned behavior: The
Norman, P., Bennett, P., & Lewis, H. (1998). Understanding binge drink- effects of perceived behavioural control and self-efficacy. British Jour-
ing among young people: An application of the theory of planned nal of Social Psychology, 34, 199 –220.
behavior. Health Education Research, 13, 163–169. Thompson, M. S., & Green, S. B. (2006). Evaluating between-group
O’Callaghan, F. V., Chant, D. C., Callan, V. J., & Baglioni, A. (1997). differences in latent variable means. In G. R. Hancock & R. O. Mueller
Models of alcohol use by young adults: An examination of various (Eds.), Structural equation modeling: A second course (pp. 119 –169).
attitude– behavior theories. Journal of Studies on Alcohol, 58, 502–507. Greenwich, CT: Information Age.
Ouellette, J. A. (1996). How to measure habit? Subjective experience and Verplanken, B., Aarts, H., van Knippenberg, A., & van Knippenberg, C.
past behavior (Doctoral dissertation, Texas A&M University, 1996). (1994). Attitude versus general habit: Antecedents of travel mode
Dissertation Abstracts International, 57, 4093. choice. Journal of Applied Social Psychology, 24, 285–300.
Ouellette, J. A., & Wood, W. (1998). Habit and intention in everyday life: Wagenmakers, E., & Farrell, S. (2004). AIC model selection using Akaike
The multiple processes by which past behavior predicts future behavior. weights. Psychonomic Bulletin & Review, 11, 192–196.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Psychological Bulletin, 124, 54 –74. Wall, A.-M., Hinson, R. E., & McKee, S. A. (1998). Alcohol outcome
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Pentz, M. A., & Chou, C.-P. (1994). Measurement invariance in longitu- expectancies, attitudes toward drinking and the theory of planned be-
dinal clinical research assuming change from development and interven- havior. Journal of Studies on Alcohol, 59, 409 – 419.
tion. Journal of Consulting and Clinical Psychology, 62, 450 – 462. Wechsler, H., Lee, J. E., Kuo, M., Seibring, M., Nelson, T. F., & Lee, H.
Perkins, H. W. (2002). Social norms and the prevention of alcohol misuse (2002). Trends in college binge drinking during a period of increased
in collegiate contexts. Journal of Studies on Alcohol, Suppl. 14, 164 – prevention efforts. Journal of American College Health, 50, 203–217.
172. Young, R. M., Connor, J. P., Ricciardelli, L. A., & Saunders, J. B. (2006).
Perkins, H. W. (Ed.). (2003). The social norms approach to preventing The role of alcohol expectancy and drinking refusal self-efficacy beliefs
school and college age substance abuse: A handbook for educators, in university student drinking. Alcohol and Alcoholism, 41, 70 –75.
counselors, and clinicians. San Francisco: Jossey-Bass. Young, R. M., Tian, O. P. S., & Crook, G. M. (1991). Development of a
Satorra, A., & Bentler, P. M. (2001). A scaled difference chi-square test drinking self-efficacy questionnaire. Journal of Psychopathology and
statistic for moment structure analysis. Psychometrika, 66, 507–514. Behavioral Assessment, 13, 1–15.
Schwarz, G. (1978). Estimating the dimension of a model. Annals of
Statistics, 6, 461– 464.
Simons, J., & Carey, K. B. (1998). A structural analysis of attitudes toward Received September 19, 2006
alcohol and marijuana use. Personality and Social Psychology Bulletin, Revision received February 6, 2007
24, 727–735. Accepted February 8, 2007 䡲