Recent studies have found order effects in social network data collection, where later names on a roster receive fewer nominations. Some thus argue for randomizing name orders or sampling peer names for survey rosters. We model order effects as biases in nomination choices and demonstrate observational and experimental methods for assessing these biases and illuminating their mechanisms. Employing these lenses, we find little evidence of order effects on eight sociometric questions in four middle school cohorts over six waves. To inform future work, we investigate aspects of the survey situation that may amplify or attenuate order effects. Analyzing these moderating forces offers guidance for detecting, understanding, and mitigating order effects in future research.
Keywords: Adolescent Friendships; Data Collection; Measurement Error; Name Rosters; Peer Nomination.