Table 3.

An Example of Using EMA to Assess Regulatory Flexibility as a MOBC in Behavioral Treatment for Substance Use Disorders

Study Design  • Randomized controlled trial comparing cognitive-behavioral therapy and motivational    interviewing in the treatment of substance use disorder  • Participants complete EMA measures via smartphones for a 2-week period immediately before    treatment and then again immediately after treatment  • Participants are prompted to answer a set of questions on their smartphone on a quasi-random    basis 5 times per day, with one randomly scheduled prompt in each of five 3-hour time periods    from 8:00AM to 10:00PM.  • Substance use will be measured at 3 months following the completion of treatment.

EMA Items for each prompt:  • Urge → “Right now, how strong is your urge to drink/use?” (0 = none/minimal, 4 = very strong)  • Stress → “Right now, how stressed are you?” (0 = not stressed at all, 4 = very stressed)  • Substance Use Availability → Right now, are you in a situation where BOTH of the following    are true: you can get access to alcohol/drugs and it would be possible to actually drink/use if you    decided to? (0 = No, 1 = Yes)  • Urge-Specific Self-Regulation → Please indicate whether you did any of the following since the    last recording to STOP YOURSELF FROM USING DRUGS OR DRINKING HEAVILY    WHEN YOU HAD AN URGE: (0 = No, 1 = Yes) Display all the urge-specific self-regulation    items in Table 1  • General Stress Self-Regulation → Please indicate whether you did any of the following since the    last recording to HANDLE OR MANAGE GENERAL STRESS: (0 = No, 1 = Yes) Display all    the stress-specific self-regulation items in Table 1

Analytic Strategy  • Use multilevel modeling where assessment prompts (Level 1) are nested within person (Level 2)  • Extract urge—self-regulation slope scores for each participant in which urge level at time t-1 is a    predictor of self-regulation at time t. (A lagged association is computed to establish temporal    ordering, i.e., the regulatory behavior comes after the urge). Self-regulation is computed as the    total number of urge-specific strategies used at time t. In extracting the slopes, only use data from    prompts in which urge is 1 or greater and substances are available (i.e., answered yes to    substance use availability question).  • Extract general stress—self-regulation slope scores for each participant in which stress level at    time t-1 is a predictor of self-regulation at time t. Self-regulation is computed as total number of    general stress strategies used at time t. In extracting slopes, only use data from prompts in which    stress is 1 or greater and urge is 0.  • Use the self-regulation slopes as mediator variables in a single-level mediation analysis with    treatment condition as the independent variable and substance use at the month 3 follow-up    assessment as the outcome.