Universities often draw from their student body when conducting human subject studies. Unfortunately, as with any longitudinal human studies project, data quality problems arise from student's waning compliance to the study. While incentive mechanisms may be employed to boost student compliance, such systems may not encourage all participants in the same manner. This paper coupled student's compliance rates with other personal data collected via Fitbits, smartphones, and surveys. Machine learning algorithms were then employed to explore factors that influence compliance. With such insight, universities may target groups in their studies who are more likely to become non-compliant and implement preventative strategies such as tailoring their incentive mechanisms to accommodate a diverse population. In doing so, data quality problems stemming from failing compliance can be minimized.
