This research project has three goals: (1) to present an overview on the use of Amazon Mechanical Turk (MTurk) as a data collection tool in alcohol-related research, and discuss the potential impacts of important MTurk-specific methodological decisions; (2) to extend our understanding of peer influence on alcohol use among college students utilizing a social network analysis (SNA) approach; and (3) to establish best practices for operationalizing perceived peer alcohol use within SNA research. The variety of uses for MTurk to facilitate alcohol-related investigations will be discussed. Alcohol researchers interested in collecting data using MTurk are encouraged to pay particular attention to methodological best-practices detailed in this report. To explore the influence of peer drinking on personal behavior, the author will describe results from a comparison of two distinct assessment strategies for measuring perceptions of peer drinking: (1) perceptions of the "typical" student's drinking behaviors, versus (2) egocentric social network measures, in which respondent's report on perceptions of the drinking behaviors of personally identified peers. Overall, egocentric network measures explained markedly greater levels of variability in peer influence on personal alcohol consumption, compared to global typical student measures. Proximal peers have a greater influence on personal alcohol consumption than "typical students" at the same institution. Finally, the author will report on whether using different strategies for measuring referent/peer alcohol use (i.e., single-item or two-item approaches versus multiple item assessments) within alcohol-related SNA research impacts study results. Commonly, SNA research employs single-item or two-item assessments (i.e., quantity, frequency, or quantity/frequency) to measure individuals' perceptions of peer drinking. Findings suggest utilizing a minimum of three items (i.e., frequency, quantity, and frequency of heavy drinking) is necessary to provide adequate insights into alcohol consumption patterns of respondents. Future research would benefit from utilizing an egocentric network approach to examine the complex, interpersonal nature of alcohol use among college students. Future investigations utilizing an egocentric network approach may fail to capture valuable insights if using simple quantity/frequency assessments of nominated peers' alcohol use. Moreover, this can enhance the effectiveness of programming efforts aimed at reducing heavy drinking among college students within social contexts.
