The AEA RCT Registry will be down for maintenance on Tuesday, April 20th, from 9pm EDT to 11pm EDT to perform necessary upgrades. The site will be in maintenance mode during that time. We apologize for any inconvenience this may cause.
Simplicity is the ultimate sophistication? Giving information on time inconsistency to increase sophistication and commitment demand
Initial registration date
March 05, 2021
March 08, 2021 10:25 AM EST
This section is unavailable to the public. Use the button below
to request access to this information.
Erasmus University Rotterdam
Other Primary Investigator(s)
Additional Trial Information
Physical Activity (PA) is a popular target for behavioural interventions, but very few sustainably solve the problem of time inconsistency in PA. One exception is commitment devices, which have been shown to be effective, but take-up is typically low. I run a randomized experiment to test whether a simple information intervention can increase sophistication about PA time inconsistency and in turn boost the demand for commitment. The experiment is run through a three-questionnaire longitudinal online general population survey. Survey respondents are randomly allocated to one of three arms - a control arm or one of two information treatment arms in which they are given information on either their past time inconsistent preferences or time inconsistent behaviour.
O Ceallaigh, Diarmaid. 2021. "Simplicity is the ultimate sophistication? Giving information on time inconsistency to increase sophistication and commitment demand." AEA RCT Registry. March 08.
I test two different interventions which aim to increase sophistication about time inconsistency in physical activity (PA) and in turn increase demand for a PA commitment device. The expectation of an increase in sophistication having a positive effect on Demand for Commitment is based on the prediction of several of the most popular economic models of self-control that increasing sophistication increases demand for commitment. The first intervention gives subjects information about their past time inconsistent preferences in PA, while the second gives information on their past time inconsistent behaviour. Further details are available in the hidden Intervention details section which will be visible after the completion of the study.
Intervention Start Date
Intervention End Date
Primary Outcomes (end points)
- Sophistication about PA TI
- Demand for a PA commitment device
Primary Outcomes (explanation)
Primary outcome 1 - Sophistication about PA TI: Measured as P/A, where P = Predicted PA for a given two-week period and A = Actual PA for the same two-week period. I will have this measure of sophistication at two time points - pre-intervention and post-intervention. The two variables used to construct the sophistication measure are measured as follows:
Predicted PA: "Own Prediction"
For approx. 66% of subjects (randomly chosen) Predicted PA is measured using a self-report survey question asking the subject to state their predicted hours of their own physical activity for the next two weeks. See actual formulation of question in attached pdf "Supplemental material". "Prediction for a similar other" For approx. 33% of subjects (randomly chosen), Predicted PA is measured using a survey question asking the subject to state their prediction of how many hours of physical activity another subject with whom they've been matched (their "partner subject") will do in the next two weeks. A subject is matched with another subject who has given the same or similar responses to the Ideal PA (see Secondary Outcomes section) and Actual PA (see below) questions in a given questionnaire. The prediction for the partner subject is then used as a proxy for the the subject's prediction of their own behaviour. See actual formulation of question in attached pdf "Supplemental material".
Which of these two questions a participant receives is determined by the randomization of the participant to an Incentive type (See Experimental Design section below). Participants are randomised to different incentive types to facilitate another separate experiment on incentives being run in parallel with the experiment described here (see more detail in the Experimental design section below).
The Predicted PA measure is measured in each of questionnaires 1 and 2.
Measured using a self-report survey question asking the subject to state their actual hours of physical activity for the previous two weeks. See formulation of question in attached pdf "Supplemental material". Measured in each of the three questionnaires.
Primary outcome 2 - Demand for a PA commitment device: Summary index of four measures obtained from four survey questions which elicit a subject's demand for a hypothetical monetary PA commitment device. The four questions are as follows:
1. "House Money" (HM) commitment demand - Binary: Subjects are asked whether they would prefer to receive €25 per week for the next 4 weeks unconditionally, or to receive it each week conditional on doing at least 2.5 hours of PA in that week.
2. HM commitment demand - Continuous: Subjects are presented with the above two options except that €25 in the unconditional option is replaced by €X. They are asked what is the minimum value of €X for which they would choose the unconditional option.
3. Self-funded commitment demand - Binary: Subjects are asked if they want to take up a commitment device where they pay a deposit of €100 now and are repaid €25 at the end of the week for each of the next 4 weeks if they do at least 2.5 hours of PA in that week. If they fail to meet that target in a given week, they lose the €25 for that week.
4. Self-funded commitment demand - Continuous: Subjects are presented with the commitment device in (3) above except that the deposit amount of €100 is replaced by €X. They are asked what is the maximum value of €X at which they would be willing to use the commitment device.
These four measures will be used to construct a Demand for Commitment summary index as described by Anderson (2008). This is calculated by standardizing the four measures and then getting a weighted average, where the measures are weighted by their correlation with other measures and the number of missing values (a measure that is less correlated with other measures in the index will be given greater weight as it provides “new info”). Additionally, a measure with less missing values is given more weight. This summary index will then be used as the primary outcome measure. This summary index is calculated for each of questionnaire 1, 2 and 3, so I will have measures of commitment demand at three different time points. Two consistency check questions are also asked in questionnaire 1. One question presents the subject with the same options as in question 1 above except that the conditional option states that they must do "no more than" 2.5 hours of physical activity to be paid the monetary reward of €25. The second question presents the commitment device in question 3 above altered in a similar manner.
See the formulation of these questions in the attached pdf "Supplemental Material".
Anderson, M. L. (2008). Multiple inference and gender differences in the effects of early intervention: A re-evaluation of the Abecedarian, Perry Preschool, and Early Training Projects. Journal of the American Statistical Association, 103(484), 1481-1495.
Secondary Outcomes (end points)
- Ideal PA
- Predicted PA
- Actual PA
Secondary Outcomes (explanation)
- Ideal PA: Measured using a self-report survey question asking the subject to state their ideal hours of physical activity for the next two weeks. See actual formulation of question in attached pdf "Supplemental material". Measured in each of the three questionnaires.
- Predicted PA: See explanation in Primary outcomes section above
- Actual PA: See explanation in Primary outcomes section above
I randomly assign respondents to a control arm or to one of two different treatment arms - the TIP treatment or the TIB treatment. Subjects in the TIP treatment arm receive the TIP intervention described in the "Intervention" section above, and subjects in the TIB treatment arm receive the TIB intervention described above. Further details are available in the hidden Experimental Design details section which will be visible after the completion of the study.
Experimental Design Details
Participants are first randomized to one of three incentive types with equal probability. This is done for the purposes of another separate experiment run in parallel with the experiment described here. Then, stratified randomisation of subjects to one of the three arms in this study occurs (Control, TIP, TIB). One stratification factor is used - incentive type received by subject. Qualtrics randomizer elements are used to carry out all randomisation. See further details in the Experimental Design section above.
Was the treatment clustered?
Sample size: planned number of clusters
Sample size: planned number of observations
Invitations to take part in this survey will be sent out to approximately 40,000 members of the Lifelines cohort study, which should give an estimated sample size of 4,800 based on a 12% response rate observed in a pilot study with this cohort.
Sample size (or number of clusters) by treatment arms
TIP arm: 2,133
TIB arm: 1,067
Control arm: 1,600
Reason for unequal distribution between arms is explained in the experimental design section above.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
INSTITUTIONAL REVIEW BOARDS (IRBs)
Medisch Ethische Toetsingscommissie METC UMC Groningen
IRB Approval Date
IRB Approval Number