Compared with the population, men and women employed in the restaurant business scored significantly higher on
extraversion (P -C 0.000). Women scored significantly lower on neuroticism (P < 0.000). while no difference was found on
neuroticism for the men.
Bivariately a high extraversion score, belonging to the young age-group, living alone, or working as a waiter increased the
probability of heavy drinking. The proportion of heavy drinkers among those scoring low, intermediate, and high on
extraversion was 4.5, 5.4, and 8.0%, respectively (P = 0.001).
The logistic regression analysis of the main effects of personality showed that extraversion significantly predicted heavy
drinking (Table 1, Model 1). The 50% increase in risk associated with a high score on neuroticism was not significant. When
control for sociodemographic variables was introduced, the odds for heavy drinking associated with extraversion were
somewhat reduced (Table 1, Model II). No interaction-effect between extraversion and neuroticism or the two personality
scores and the background variables seemed important in predicting heavy drinking. Additional analyses with extraversion
and neuroticism used as continuous variables (not shown) gave adjusted ORs of 1.08 (95% CI I .Ol-1.16, P = 0.037) for
extraversion and 1.07 (95% CI 1.01-l .14, P = 0.033) for neuroticism.
Based on the models in Table I, the odds ratios for heavy drinking according to different combinations of scores on
extraversion and neuroticism were calculated (Table 2). Model I shows the odds ratios adjusted only for the other variable.
while model II presents the odds ratios adjusted by gender, age-group, household type and occupation in addition (odds
ratios for background variables are not given). The probability of heavy drinking was increased by almost two and a half