Voluntary survey completion among team members: implications of noncompliance and missing data for multilevel research.

We explored whether voluntary survey completion by team members (in aggregate) is predictable from team members' collective evaluations of team-emergent states. In doing so, we reanalyze less-than-complete survey data on 110 teams from a published field study, using so-called traditional and modern missing data techniques to probe the sensitivity of these team-level relationships to data missingness. The multivariate findings revealed that a greater within-team participation rate was indeed related to a higher team-level (mean) score on team mental efficacy (across all four missing-data techniques) and less dispersion among team member judgments about internal cohesion (when the 2 modern methods were used). In addition, results show that a commonly used approach of retaining only those teams with high participation rates produces inflated standardized effect size (i.e., R²) estimates and decreased statistical power. Suggestions include research design considerations and a comprehensive methodology to account for team member data missingness.