JMIR Publications

ABSTRACT

Background:

Ecological momentary assessment (EMA) methods can provide assessment of alcohol-related beliefs and behaviors when people are in their natural environments. An increasingly common EMA approach involves the use of passive sensors (e.g., continuous sharing of Bluetooth and global positioning system [GPS] cell phone data) to collect rich data and trigger alcohol assessment in specific contexts. While collecting real-time assessments allows researchers to explore contextual factors impacting alcohol use, there is a lack of research examining compliance rates to alcohol-related EMAs that use daily questions for study periods over two weeks. EMAs over longer periods allow researchers to assess meaningful changes over time and sample low-base-rate events. Further, there is limited research exploring how EMA compliance is influenced by individual-level factors (e.g., privacy concerns, alcohol use). Identifying factors associated with compliance can help inform EMA protocol design.

Objective:

The current study had two primary goals. First, we add to the literature by examining the feasibility and acceptability of a 60-day college student alcohol-EMA that utilizes phone-based, passive sensors. Second, we examined whether compliance was associated with event-level alcohol use, privacy concerns, and types of questions asked (i.e., questions triggered by time or GPS location).

Methods:

: College students (N = 68) who reported regular alcohol use completed the 60-day EMA study. At baseline, participants completed an online survey that included an assessment of privacy concerns. During the study period, participants completed up to 15 questions a day, including evening and morning self reports of alcohol use. Feasibility was assessed based on enrollment, retention, and compliance rates. Acceptability was examined at the end of the study period using open- and closed-answered questions on perceptions of the EMA experience.

Results:

The findings supported the feasibility of the 60-day EMA with enrollment exceeding benchmarks and high compliance across the study period (average compliance = 90%). Overall, students found the protocol acceptable with 88% indicating that they would be willing to complete another EMA study. Participants’ compliance was not associated with alcohol use at the event-level (OR = 0.94, 95% CIs [0.80, 1.10]) or at the trait level (OR = 0.92, 95% CIs [0.73, 1.17]). Participants’ compliance was greater for time-trigger assessment than GPS triggered questions (p < .001). Privacy concerns were not associated with reduced responding.

Conclusions:

The current study has implications for those designing alcohol EMAs for college populations that involve the use of passive sensors, including recommendations for balancing predictable timing of questions and variability in question content.