This study was exploratory in nature and intended to provide an in-depth look at factors associated with changes in drinks per week during three time points (pre-departure, abroad, and post-return). Therefore, analyses consisted of separate repeated measures analysis of variance (ANOVAs). Sphericity was not assumed in any of the models; thus, a Greenhouse-Geisser correction was employed, ε = .75. Binary moderators were specified as fixed factors in the repeated measures ANOVAs and continuous moderators were specified as covariates. High and low values of continuous moderators were specified as one standard deviation above and below the mean in figures of interactions (Aiken & West, 1991
). A correlation matrix of drinks per week and moderators is found in . For descriptive purposes, we also evaluated main effects for peak drinks and HED.
Correlation matrix between drinking and moderators
Changes in drinking over time
contains means and standard deviations for drinking over time. A repeated measures ANOVA revealed a significant change in drinks per week over time, F (2, 350) = 45.88, p < .001; an effect best explained by a quadratic relationship where drinking more than doubled during trips abroad and subsequently reduced upon return to the U.S., F (1, 175) = 57.25, p < .001. The linear relationship trended toward significance, F (1, 175) = 2.42, p = .12 and was thus examined in more depth to determine for which participants drinking at post-return was significantly higher than pre-departure drinking. Using a repeated measures ANOVA with two time-points (pre-departure and post-return) and abroad drinking entered as a covariate, a significant time × abroad drinking interaction was found, F (1, 174) = 6.32, p < .05, with heavier drinkers abroad returning home drinking at higher levels (see ). A significant main effect, F (2, 352) = 21.95, p < .05, ε correction = .81, and quadratic trend, F (1, 176) = 29.17, p < .001 was observed for peak drinks. Peak drinks increased from a mean of 3.81 (SD = 3.34) to 5.26 (SD =3.97) while abroad and reduced to 3.62 (SD = 3.05) at post-return. Finally, a Wilcoxon signed rank test indicated the percentage of participants who engaged in HED increased from 22% at pre-departure to 34% while abroad, z = −2.89, p < .001.
Means and standard deviations of drinks per week for all participants and by region of study
Changes in drinks per week from pre-departure to post-return as a function of drinking while abroad.
Region of study
Participants traveled to multiple different countries and the sample size did not allow for comparisons of drinking by specific countries. Therefore, countries were categorized into five different world regions: Europe (e.g., Italy, France; n = 109), Asia (e.g., China, Japan; n = 28), Oceania (Australia and New Zealand; n = 7), Latin America (e.g., Ecuador, Brazil; n = 22), and non-traditional study abroad locations encompassing countries in the Middle East, Africa, South Asia, and West Asia (e.g., South Africa, Israel; n = 11). Examination of mean drinking by region revealed that participants studying abroad in Oceania and Europe drank the most while abroad, followed by Latin America, Asia, and non-traditional regions (see ). Planned comparisons with orthogonal contrast coefficients were specified prior to analysis: (1) Europe vs. Oceania, (2) Europe and Oceania vs. Asia, Latin America, and non-traditional regions, (3) Latin America vs. Asia and non-traditional regions, and (4) non-traditional regions vs. Europe, Oceania, Asia, and Latin America. There was an overall time × region of study effect, F (8, 342) = 5.24, p < .001, with significant differences in drinking over time for Contrast 2, t (171) = 2.25, p < .05. While experiencing similar increases in drinking while abroad, participants studying abroad in Europe and Oceania experienced greater changes in drinking over time compared to participants in other regions. Both the linear, F (4,171) = 2.81, p < .05, and the quadratic relationships were significant for the interaction, F (4,171) = 5.91, p < .001. Further follow-up tests revealed participants studying abroad in Latin America drank at significantly higher levels at post-return than from pre-departure, t (21) = 2.35, p < .05.
U.S. legal drinking age status
To examine if underage drinking status functioned as a moderator of changes in drinking over time, participants under 21 at pre-departure (51%) were compared with those of U.S. legal drinking age. A repeated measures ANOVA with legal drinking age status (0 = under 21, 1 = 21 and older) entered as a fixed factor revealed a significant time × legal drinking age status effect, F (2, 348) = 6.87, p < .01. Despite drinking less than participants over the legal drinking age at pre-departure, underage participants experienced greater increases in drinking over time; increasing drinking while abroad by about 170% (see ). While the quadratic effect was significant for this interaction, F (1, 174) = 7.34, p < .01, there was also a significant linear relationship, F (1, 174) = 5.09, p < .05. Follow-up paired samples t-tests revealed that participants under age 21 at pre-departure increased their drinking from pre-departure to post-return, t (89) = 2.44, p < .05. This observed effect was not a function of underage participants’ change in legal drinking status at post-return.
Changes in drinks per week over time as a function of legal drinking age status in the U.S.
Intentions to drink
Pre-departure intentions to drink while abroad associated with changes in drinking over time. A repeated measures ANOVA revealed a time × intentions interaction effect, F (2, 340) = 10.39, p < .001. This effect was best explained by a quadratic relationship, F (1, 170) = 13.28, p < .001. Those with higher pre-departure intentions to drink alcohol drank more alcohol prior to trips and increased their drinking to a greater extent than those with low drinking intentions at pre-departure (see ).
Changes in drinks per week over time as a function of pre-departure intended drinking.
Perceptions of peer behavior
Participants’ perceived drinking of other study abroad students living in their host country was related to changes in drinking over time. A repeated measures ANOVA revealed a time × perceived norms interaction effect, F (2, 340) = 5.75, p < .01. This interaction effect was best explained by a quadratic relationship, F (1, 170) = 7.04, p < .01. Participants with higher perceived norms at pre-departure drank more and increased their drinking to a greater extent than those with lower perceived norms (see ).
Changes in drinks per week over time as a function of pre-departure perceptions of student abroad student peer drinking.