Alcohol-specific parenting, adolescent alcohol use and the mediating effect of adolescent alcohol-related cognitions

Abstract

Objectives : Previous research indicated that alcohol-specific parenting is an important precursor of adolescent alcohol use, but failed to define the underlying mechanism. Based on social cognitive theory, alcohol-related cognitions such as alcohol refusal self-efficacy and alcohol-related expectancies were hypothesised to mediate this link.

Design : A cross-sectional survey included 1349 mothers and their sixth grade (11–12 years old) adolescent offspring. Structural equation modelling was employed to test the association between alcohol-specific parenting and adolescent alcohol use, mediated by adolescent alcohol-related cognitions.

Main outcome measures : Adolescent alcohol use, drinking refusal self-efficacy and alcohol expectancies.

Results : The associations between frequency of communication, maternal alcohol use and adolescent alcohol use were mediated by negative alcohol-related expectancies. The associations between quality of communication, rules and disclosure and adolescent alcohol use were mediated by self-efficacy.

Conclusions : The present study provides a first indication that the underlying mechanism of the association between the most important alcohol-specific parenting practices and adolescent alcohol use can be contributed to the mediating effect of alcohol-refusal self-efficacy.

Keywords:

• adolescent alcohol use
• alcohol-specific parenting
• alcohol expectancies
• self-efficacy
• structural equation modelling

Acknowledgements

This study was supported by a grant from Zorg-onderzoek Nederland en Medische wetenschappen (ZonMw) [50-51,300-98-007].

My coauthors and I do not have any interests that might be interpreted as influencing the research.

Notes

1. To reduce possible error, a measurement model was computed in which each variable was represented as a latent factor indentified by several parcels consisting of two or three items. This measurement model showed satisfactory fit to the data (χ ²(296) = 906.35, p < .001; CFI = .97; RMSEA = .04), with factor loadings ranging between .60 and .93. We computed an identical SEM model as the SEM model using manifest variables, which produced nearly identical results.