Abstract
Role of ‘Big Five’ personality traits as predictors of smoking and moderators of the intention-smoking relationship was tested. Five hundred and fifty-three adolescents (aged 11–12) completed measures of self-reported past smoking, gender, intentions to smoke, perceived behavioural control, family smoking, friends smoking at times 1 and 2 (6 months apart). At time 3, 2 years later, the same adolescents completed measures of the Big Five and self-reported smoking (a subset of 300 also provided an objective smoking measure). At time 4, two years after time 3, a sub-sample of 122 adolescents provided a self-report measure of recent smoking. Simple correlations indicated significant direct effects of conscientiousness (self-reported smoking, times 3 and 4), extraversion (time 4 smoking) and neuroticism (all smoking measures) on smoking. Logistic regression showed intention, and the interaction between conscientiousness and intention to significantly predict both self-reported and objectively assessed smoking (both at time 3) after controlling for other variables. Multiple regression showed intentions, family smoking and the interaction between conscientiousness and intention to significantly predict self-reported smoking at time 4 after controlling for other variables. The findings indicate that the impact of personality variables on smoking is through mediated (through cognitions) and moderated (conscientiousness by intention interaction) pathways.
Acknowledgements
This research was funded by Economic and Social Research Council (ESRC) grants to the first author: ‘Testing a social psychological tool to prevent smoking initiation in adolescents’ (reference number R-000-22-3219), ‘Understanding adolescent smoking initiation: A 4-year longitudinal study’ (reference number RES-000-22-0077); and the first, second and fourth authors: ‘Understanding smoking in young adults: A 6-year longitudinal study’ (reference number RES-000-22-1288). We are grateful for the work of Ms Dawn Greene in collecting the data for time 3 and to Professor Marco Perugini for the programme to produce the figures.
Notes
Notes
1. These data are taken from a larger dataset. Respondents were originally allocated to one of two control conditions or one of two intervention conditions (targeting implementation intentions or self-efficacy about smoking). See Higgins (Citation2002) for further details. Findings reported here are based on all respondents, however all the main findings were replicated when only those in the control conditions were included (in these analyses for self-reported smoking N = 351; for objective smoking N = 201).
2. The sample at time 4 was considerably smaller as we were reliant on participants providing names and addresses at the previous time point in order to be re-contacted.
3. This split was used to ensure that recent smoking was being measured and a good degree of correspondence with the objective measure of smoking, which is only sensitive to recent smoking. The main findings were replicated when comparing ‘never smokers’ against the other groups.
4. Splitting the objective measure at other points (e.g. 0 vs. >0) or using it as a continuous measure produced substantively identical results.
5. These results were replicated when also controlling for time 3 objectively-assessed or self-reported smoking behaviour. A similar pattern was apparent when using a logistic regression with a dichotomised measure of the number of cigarettes smoked.