Professional Documents
Culture Documents
CONTACT Ahmad Ali Eslami eslamiaa@gmail.com School of Health, Isfahan University of Medical Sciences, Isfahan, Iran.
© 2017 Taylor & Francis Group, LLC
JOURNAL OF SUBSTANCE USE 269
examined to assess how well the observed measures reflect the Table 1. Description of the sample by age and gender (n = 671).
latent constructions; Secondly, the hypothesized model was tested Age Female n (%) Male n (%) Total
to examine the relationships among constructs. 14–14.9 86 (58.9) 60 (41.1) 146 (100)
15–15.9 70 (49.0) 73 (51.0) 143 (100)
Following data collection, it was determined that nearly 16–16.9 56 (29.3) 135 (70.7) 191 (100)
12% of the questionnaire data were missing. Missing data 17–17.9 113 (59.2) 78 (40.8) 191 (100)
were imputed using Full Information Maximum Likelihood Total 325 (48.4) 346 (51.6) 671 (100)
(FIML) estimation, an optimal strategy for handling missing
data (Graham, 2009).
Table 2. Correlation coefficient among self-efficacy, negative outcome expec-
Before testing the recommended model using Structural tancies, positive outcome expectancies, social influence, and substance use.
Equation Modeling (SEM), the most important assumptions of 1 2 3 4 5
the method, multivariate normality and multicollinearity, were Positive expectancies Mean:2.97
tested using Mardia’s Multivariate Normality test and the SE:.03
Variance Inflation Factor (VIF) approach, respectively. The Negative −.059 Mean:2.27
expectancies SE:.03
inclusion of scales and subscales was determined through assess- Refusal self-efficacy .269** .014 Mean:3.1
ment of existing validity studies. In order to support the validity SE:.02
of the scales in the recommended structural model and reach the Social influence −.081* .021 −.257** Mean:1.12
SE:.01
most appropriate measurement models, Exploratory (Minimal Substance use −.112** .058 −.317** .459** Mean:.7
Residual Method [MINRES]) and Confirmatory Factor Analysis behavior SE:.09
(Maximum Likelihood Estimation [MLE]) were performed, Note:*. Correlation is significant at the 0.05 level (2-tailed).
respectively (Enders & Bandalos, 2001). SEM, using AMOSE, **. Correlation is significant at the 0.01 level (2-tailed)
version 23, was run to examine whether the relationship between
social influence (SI) and substance use behavior (SU) would be positive outcome expectancies contained six observable vari-
mediated by self-efficacy (SE) and outcome expectancy (OE). ables and had loadings ranging from 0.71 to 0.86. The negative
SEM was computed using the robust method, correcting for the outcome expectance contained five observable variables and
expected non-normal distributions. had loadings ranging from 0.75 to 0.88, perceived norm drug
In order to examine the mediational role of attachment, we use contained four observable variables and had loading ran-
computed direct effects (path coefficient from SI to SU) as well ging from 0.73 to 0.88. According to Comrey and Lee (1992) we
as indirect effects (i.e., the product of the path coefficients from observed acceptable factor loadings in our model (Comrey &
SI to SE and from SE to SU and also from SI to OE to SE and Lee, 1992). The latent variables from the first-order confirma-
finally to SU). The significance of the indirect effects was eval- tory model had acceptable indices, based on goodness-of-fit
uated using 95% bootstrapped confidence intervals (MacKinnon indices (Table 3)
& Fairchild, 2009). This bias-corrected method is based on a In the second phase, the effect of exogenous variable
distribution of the product of coefficients, and generates con- include social influence with substance use behavior in
fidence limits for the true value of coefficient for indirect effects. adolescents that were mediated by negative outcome expec-
When zero is not in the confidence interval, the indirect effect is tancies, positive outcome expectancies, and refusal self-effi-
considered significant. Finally, we evaluated the proportions of cacy were tested in the primary model. The initially
the total effect that were mediated through attachment. estimated model showed that some paths were not statisti-
Model fit of all models was evaluated by χ2 values, the Root cally significant. For example, negative outcome expectan-
Mean Square Error of Approximation [RMSEA; (Browne & cies did not have any significant effects on substance use
Cudeck, 1993)], Tucker Lewis index (TLI), and the behavior in direct and indirect paths and this path was
Comparative Fit Index [CFI; (Bentler, 1990)]. Models with removed from the model. Subsequently, the final model
non-significant χ2, RMSEA less than 0.07 and CFI and TLI was re-estimated with only the significant paths (see
greater than 0.90 were considered a good fit to the observed Figure 2). The fit indexes for the estimated final model
data (Hu & Bentler, 1999). were satisfactory (Table 4).
Results
Table 3. Indices of first-order confirmatory models.
Participant: the original dataset contained responses from 720
participants. Twenty participants (2.7%) did not answer com- CMIN/
CMIN df df GFI AGFI NFI TLI CFI RMSEA
pletely and 29(4.02%) questionnaires were also eliminated due
Positive outcome 12.9 4 3.2 0.98 0.94 0.98 0.97 0.99 0.08
to doubts in validity and due to their inconformity with other expectancies
data. 48.4% (325) of the samples were female and Participants’ Negative outcome 28 9 3.1 0.97 0.94 0.97 0.97 0.98 0.07
expectancies
ages ranged from 14 to 17 years (M = 15.6, SD = 1.1) (Table 1). Refusal self- 0.005 1 0.005 0.99 0.99 0.99 0.99 0.99 0.03
Correlations between all factors were relatively meaningful efficacy
(Table 2) except for the negative outcome expectancy that did Social influence 3.50 1 3.50 0.99 0.95 0.99 0.94 0.99 0.08
Substance use 8.56 3 2.85 0.99 0.95 0.99 0.99 0.99 0.07
not have a meaningful correlation with the others. behavior
Five latent variables were specified in the first-order mea- Note: Chi-square values are significant (p < 0.05). S–Bχ2 = Satorra–Bentler Chi-
surement model. The refusal self-efficacy contained four obser- square; NFI = Normed Fit Index; CFI = Comparative Fit Index; TLI = Tucker Lewis
vable variables and had loadings ranging from 0.84 to 0.90. The index; RMSEA = root mean square error of approximation.
JOURNAL OF SUBSTANCE USE 271
Figure 2. Final model: Direct and indirect paths standard coefficients for the effect of social influence on substance use mediated by outcome expectancy and self-
efficacy. SI: social influence: (1: Father, 2: Mother, 3: Sibling, 4: Best friend), OE: Positive outcome expectancy, SE: Refusal self-efficacy, SU: Substance use behavior.
Spear (Eds.), Adolescent brain development: Vulnerabilities and oppor- psychology: A promising new cross-disciplinary field. Cambridge: MIT
tunities (pp. 1–22). New York: Academy of Sciences. Press.
Dal Cin, S., Worth, K. A., Dalton, M. A., & Sargent, J. D. (2008). Youth Mohammadpoorasl, A., Fakhari, A., Rostami, F., & Vahidi, R. (2007).
exposure to alcohol use and brand appearances in popular contem- Predicting the initiation of substance abuse in Iranian adolescents.
porary movies. Addiction, 103(12), 1925–1932. doi:10.1111/j.1360- Addictive Behaviors, 32(12), 3153–3159. doi:10.1016/j.addbeh.2007.07.014
0443.2008.02304.x Nazarzadeh, M., Bidel, Z., Ayubi, E., Bahrami, A., Jafari, F.,
Dishion, T. J., Nelson, S. E., & Bullock, B. M. (2004). Premature adolescent Mohammadpoorasl, M., . . . Taremian, F. (2013). Smoking status in
autonomy: Parent disengagement and deviant peer process in the Iranian male adolescents: A cross-sectional study and a meta-analysis.
amplification of problem behavior. Journal of Adolescence, 27, 515–530. Addictive Behaviors, 38, 2214–2218.
emcdda. (2012). European monitoring centre for drugs and drug addic- Oei, T., & Jardim, C. (2007). Alcohol expectancies, drinking refusal self-
tion. Retrieved from: http://eib.emcdda.europa.eu. efficacy and drinking behavior in Asian and Australian students. Drug
Enders, C., & Bandalos, D. (2001). The relative performance of full and Alcohol Dependence, 87, 281–287. doi:10.1016/j.drugalcdep.
information maximum likelihood estimation for missing data in 2006.08.019
structural equation models. Structural Equation Modeling: A Parolin, M., & Simonelli, A. (2016). Attachment theory and maternal
Multidisciplinary Journal, 8, 430–457. drug addiction: The contribution to parenting interventions. Front
Fathian, Z., Eslami, A., Ghofranipour, F., & Mostafavi, F. (2015a). Psychiatry, 7, 152. doi:10.3389/fpsyt.2016.00152
Psychometric properties of coping and self –efficacy scales related to Parvizy, S., & Ahmadi, F. (2009). A qualitative study on adolescence,
substance use in a sample of Iranian adolescents. Drugs: Education, health and family. Mental Health in Family Medicine, 6(3), 163–172.
Prevention & Policy Psychometric, Early Online, 1–6. doi:10.3109/ Sarafino, E. P. (2006). Health psychology: Biopsychosocial interactions. (5
09687637.2015.1028897 ed.). London, United Kingdom: John Wiley and Sons.
Fathian, Z., Eslami, A., Ghofranipour, F., Mostafavi, F., & Ebrahimi, A. Schuckit, M. A., Tapert, S., Matthews, S. C., Paulus, M. P., Tolentino, N.
(2015b). The relationship between self-efficacy, coping skill and sub- J., Smith, T. L., . . . Simmons, A. (2012). fMRI differences between
stance use in adolescent: Based on structural equation modeling. subjects with low and high responses to alcohol during a stop signal
Journal of Substance Use, Early Online, 1–7. doi:10.3109/ task. Alcoholism, Clinical and Experimental Research, 36(1), 130–140.
14659891.2015.1018973 doi:10.1111/j.1530-0277.2011.01590.x
Fu, A. T., Ko, H. C., Wu, J. Y., Cherng, B. L., & Cheng, C. P. (2007). Spear, L. P., & Varlinskaya, E. I. (2005). Adolescence. Alcohol sensitivity,
Impulsivity and expectancy in risk for alcohol use: Comparing male tolerance, andintake. Recent Developments in Alcoholism, 17, 143–159.
and female college students in Taiwan. Addictive Behaviors, 32(9), Stamates, A. L., Lau-Barraco, C., & Linden-Carmichael, A. N. (2016).
1887–1896. doi:10.1016/j.addbeh.2007.01.003 Alcohol expectancies mediate the relationship between age of first
Gorman, D. M. (1992). Using theory and basic research to target primary intoxication and drinking outcomes in college binge drinkers.
prevention programs: Recent developments and future prospects. Substance Use & Misuse, 51(5), 598–607. doi:10.3109/
Alcohol and Alcoholism, 27(6), 583–594. 10826084.2015.1126745
Graham, J. W. (2009). Missing data analysis: Making it work in the real Steinberg, L. (2005). Cognitive and affective development in adolescence.
world. Annual Review of Psychology, 60, 549–576. Trends in Cognitive Sciences, 9(2), 69–74. doi:10.1016/j.tics.2004.12.005
Hine, D. W., McKenzie-Richer, A., Lewko, J., Tilleczek, K., & Perreault, Thibodeau, E. L., August, G. J., Cicchetti, D., & Symons, F. J. (2016).
L. (2002). A comparison of the mediational properties of four adoles- Application of environmental sensitivity theories in personalized pre-
cent smoking expectancy measures. Psychology of Addictive Behaviors, vention for youth substance abuse: A transdisciplinary translational
16(3), 187–195. perspective. Translational Behavioral Medicine, 6(1), 81–89.
Hu, L., & Bentler, P. (1999). Cutoff criteria for fit indexes in covariance doi:10.1007/s13142-015-0374-4
structure analysis: Conventional criteria versus new alternatives. Urbán, R., Kökönyei, G., & Demetrovics, Z. (2008). Alcohol outcome
Structural Equation Modeling, 6(1), 1–55. doi:10.1080/10705519909540118 expectancies and drinking motives mediate the association between
Iwamoto, D. K., Corbin, W., Lejuez, C., & MacPherson, L. (2014). sensation seeking and alcohol use among adolescents. Addictive
College men and alcohol use: Positive alcohol expectancies as a Behaviors, 33, 1344–1352.
mediator between distinct masculine norms and alcohol use. Witt, E. D. (2010). Research on alcohol and adolescent brain develop-
Psychology of Men & Masculinity, 15(1), 29–39. doi:10.1037/a0031594 ment: Opportunities and future directions. Alcohol, 44, 119–124.
Jones, B. T., Corbin, W., & Fromme, K. (2001). A review of expectancy Wolf, E. J., Harrington, K. M., Clark, S. L., & Miller, M. W. (2013).
theory and alcohol consumption. Addiction, 96(1), 57–72. Sample size requirements for structural equation models an evaluation
doi:10.1080/09652140020016969 of power, bias, and solution propriety. Educational and Psychological
MacKinnon, D., & Fairchild, A. (2009). Current directions in mediation Measurement, 76(6), 913–934. doi:10.1177/0013164413495237
analysis? Current Directions in Psychological Science, 18(1), 16–20. Young, R., Connor, J., Ricciardelli, L., & Saunders, J. (2006). The
doi:10.1111/j.1467-8721.2009.01598.x role of alco- hol expectancy and drinking refusal self-efficacy
Meier, S. (2007). A survey of economic theories and field evidence on beliefs in university student drinking. Alcohol and Alcoholism,
pro-social behavior. In B. S. Frey & A. Stutzer (eds.), Economics and 41, 70–75.
Copyright of Journal of Substance Use is the property of Taylor & Francis Ltd and its content
may not be copied or emailed to multiple sites or posted to a listserv without the copyright
holder's express written permission. However, users may print, download, or email articles for
individual use.