Factors associated with missing data in an experience sampling investigation of substance use determinants

The Experience Sampling Method (ESM) collects data repeatedly over time, and is therefore prone to missing observations. Little is known about the characteristics of the subjects and of the ESM procedure associated with unanswered records. Through an ESM investigation of substance use determinants, these characteristics were able to be analyzed.

Participants (n=224) were undergraduate university students enrolled for a study of substance use factors, providing data through the use of classic questionnaires and through the Experience Sampling Method (ESM) using palmtop computers. For the ESM, they were signaled five times per day for 7 days (7840 records). Characteristics of the ESM procedure and of the participants were analyzed jointly. The probability of an unanswered ESM record was analyzed using a random-intercept logistic regression, fitting a multivariate mixed-effect model for repeated measurements.

Factors significantly associated with an unanswered record were: male gender, being a Sport Science student, having higher scores of novelty seeking and of persistence, and being a poly-substance user. Unanswered records were also more frequent in the middle of the week and at the beginning of the day.

Findings are discussed in term of the possible impact of missing observations. In particular, the lower compliance of poly-substance users with the ESM protocol may curtail the validity of the method, since ESM records are less representative of all moments in these persons daily life. Thus, results from ESM studies of substance use should be regarded cautiously and complemented with other data gathering procedures.