Self-Selection into Health Check-Ups

How to cite

Eyting, Markus; Kosfeld, Michael. 2022. "Self-Selection into Health Check-Ups." AEA RCT Registry. AEARCTR-0009059. https://doi.org/10.1257/rct.9059-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Markus Eyting* , Michael Kosfeld

* Primary investigator

Registry ID
AEARCTR-0009059
DOI
https://doi.org/10.1257/rct.9059-1.0
First registered
Last updated
Status
In development
Trial dates
2022-04-01 – 2022-04-30
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/9059

Abstract

Using a unique data set from a German health check-up provider we have compiled a concise and validated questionnaire tool to estimate individuals' risk for high cholesterol levels. We now apply this tool in an online experiment in order to study individual health information avoidance behavior. In particular, we use our tool to classify individuals into ‘healthy’ and ‘unhealthy’ individuals, and ask (1) which group participates more in health check-ups, (2) whether a potential difference in participation rates between the two groups is driven by selection, and (3) whether we can influence selection by increasing the salience of potentially bad news. Hereby, we base our hypotheses on a pilot study in which we find that healthy individuals have higher check-up rates than unhealthy individuals.

Intervention

We randomly divide our sample into a ‘control’ group and an ‘information’ group. In the ‘information’ group, participants receive information about adverse health risks from untreated, high cholesterol levels. Participants in the control treatment do not get this information.

Experimental design

1) Questionnaire: We ask several survey questions, in order to be able to estimate the individual health risk, as well as gather information on whether or not individuals have been to a check-up before, how often, and their likelihood to go to a check-up in the future.
2) Information: Depending on treatment, we provide individuals with different kinds of information about cholesterol levels (see intervention).
3) Willingness: We elicit individuals’ willingness to see the estimated health risk. They are informed that an algorithm was able to predict their individual risk based on answers they have given in the questionnaire at the beginning of the experiment. By paying marginal rates, participants can increase the likelihood to be provided with their calculated risk score: using a slider that is initially set at 0%, participants can increase the likelihood to see their calculated risk score. Each percentage point increase costs 1 cent.
4) Quality: We elicit individuals’ perceived quality of the classification algorithm.
5) Beliefs: We elicit individuals’ beliefs about their own health risk.
6) Depending on their previously elicited willingness, we show individuals their respective estimated individual health risk.
7) Demographics: We ask for socioeconomic information and individuals’ reasoning for their decision in 3).

Design details

Randomization method
The randomization is done by a computer.
Randomization unit
Individual
Clusters
1000 individuals
Observations
1000 individuals
Arms
500 individuals in the control treatment; 500 individuals in the information treatment
Primary outcomes
- Check-up rates in the past of healthy and unhealthy individuals - Willingness to see the estimated individual health risk
Secondary outcomes
- Planned check-up rates in the future of healthy and unhealthy individuals - Perceived individual health risk - Perceived quality of classification algorithm

Cited by

No citation data available for this registration yet.

Keywords