Reactive Sticks A Randomized Control Trial on the Effects of Sunk Costs on Completions Rates of a Substance-Use-Disorder Self-Help Program

Individuals in the treatment groups will be asked to complete a standard slider task, in order to have subject exert effort in a way that is as uniform as possible across individuals. (Gill & Prowse https://web.ics.purdue.edu/~gill53/GillProwseSliders.pdf)

The task consists of several sliders positioned at 0, which have to be repositioned at the mid-point using the mouse, trackpad or touch screen.

Low-effort version: 20 sliders
High-effort version: 40 sliders

2021-12-13

2022-01-31

Quit rates at intervention points, completion rate and length of engagement with the program (number of modules completed and time elapsed from the start to the last completed task), self-assessment about reaching initially stated goals, self-reported level of cravings, motivation rating at the end of module 1.

We will measure the rate of program completion for treatment and control groups as well as the average length of engagement with the program both in terms of the number of modules completed and in terms of the time elapsed from the start of the program to the last completed task. We will assume that an individual has quit the program if he/she has been inactive for 15 days.

heterogeneity analysis

We will run a heterogeneity analysis to see whether the effect (if any) is concentrated in a particular subgroup of the population of subjects. This could be done by leveraging on data from the companion data-science project to identify individuals who are likely to have “weak motivation”. Our previous study shows that those are the ones who are most likely to be benefitted by the intervention.

Jellinek offers an online self-help program targeted at individuals that self-report the wish to diminish use or quit alcohol, smoking, cannabis, cocaine or gambling. The program is anonymous and free of charge for participants and consists of several modules

Rough data: 85% fails to finish the program. Yet of the 15% who finishes does not necessarily reach their goal.

At the start, subjects communicate their use and their goal (quit, reduce use, …). At the end, they report their use (hence one can assess whether they reached their goal).

The experiment will run for 1.5 months — from 13 December 2021 to 31 January 2022 — and the data collection will take place until 30 March 2022 in order to follow the performance of all subjects, including those who enrolled towards the end of the trial period.

It is crucial that during this period no changes be made to the program.

Subjects will be randomly allocated to 1 of 4 groups:
1. Control group: Motivation / cravings rating before the start and at the end of module 1.
2. Condition A: At enrollment: motivation / cravings rating followed by low-effort task. At the end of module 1: motivation / cravings rating and no task.
3. Condition B: At enrollment: motivation / cravings rating followed by low-effort task. At the end of module 1: motivation / cravings rating followed by low-effort task.
4. Condition C: At enrollment: motivation / cravings rating followed by low-effort task. At the end of module 1: motivation / cravings rating followed by high-effort task.

The idea of the treatments is to test the effect of having to overcome a small hurdle before being able to go on with the program. Condition A tests the effectiveness of one intervention. Condition B tests whether repeating the effort task a second time further boosts resolve. The first and the second tasks have the same degree of difficulty. Condition C tests the effect of escalating the difficulty of the task from the first to the second intervention.

Computerized randomization, blocked per condition.

Individual

No

We expect a total sample size of about 1200 subjects based on enrollment in 2020 (9255 subjects) and an intervention period of 1.5 months. We expect the sample to be composed as follows: 33% alcohol, 29% smoking, 21% cannabis, 9% cocaine and 8% gambling.

1200 individuals

300