paint-brush

This story draft by @escholar has not been reviewed by an editor, YET.

Fitbit Attrition, Leaderboard De-Adoption, and Additional Robustness Checks

EScholar: Electronic Academic Papers for Scholars HackerNoon profile picture

Authors:

(1) Muhammad Zia Hydari, Katz Graduate School of Business, University of Pittsburgh and Corresponding author;

(2) Idris Adjerid, Pamplin College of Business;

(3) AAaron D. Striegel, Department of Computer Science and Engineering, University of Notre Dame.

Table of Links

Abstract and 1 Introduction

2. Background and 2.1. Leaderboards

3. Effect of Leaderboards on Healthful Physical Activity and 3.1. Competition

3.2. Social Influence

3.3. Moderating Effects of Prior Activity Levels and Leaderboard Size

4. Data and Model

4.1. Data

4.2. Model

5. Estimation and Robustness of the Main Effects of Leaderboards

5.2. Robustness Check for Leaderboard Initiation

5.3. Fitbit Compliance

5.4. Fitbit Attrition, Leaderboard De-Adoption, and Additional Robustness Checks

6. Heterogeneous Effect of Leaderboards

6.1. Heterogeneity by Prior Activity Levels

6.2. Interaction of Leaderboard Size, Rank, and Prior Activity Levels

6.3. Summary of Findings from Heterogeneous Effect Analysis

7. Conclusions and Discussion, Endnotes, and References

5.4. Fitbit Attrition, Leaderboard De-Adoption, and Additional Robustness Checks

Related to the challenge of compliance, we also consider the role of attrition from the sample due to Fitbit abandonment, which has generally been noted in the popular press for health wearables.18 If sample attrition is related to leaderboard adoption, it may introduce bias in our analysis. For example, lower performers may abandon their Fitbit device after joining leaderboards because it reveals to them that they are less active than their peers. We examine this concern extensively and identify no relationship between leaderboard adoption and sample attrition for lower performers, and no differences in physical activity and similar leaderboard effects for those who eventually leave the sample compared with those who report data throughout (see the Online Appendix, Section F). We also consider whether individuals who eventually hide themselves from the leaderboard (the main mechanism for de-adoption) impact our results. As we mentioned previously, this was rare for leaderboard adopters (approximately 5%), and excluding these individuals results in consistent estimates of leaderboard effects (see Table 2, column 7).


5.4.1. Outliers and Falsification with Negative Control Treatments. We also evaluate the potential for a particular individual (or time period) in the data to be an outlier driving our results. Specifically, we systematically “leave out one” individual (or time period) and re-estimate our model (see the Online Appendix, Section G). We find consistent treatment effects of leaderboards that are always statistically significant, suggesting minimal risk from outliers in the data. Furthermore, we constructed a negative control treatment (NCT), as the focal user’s leaderboard with no other active users. Such leaderboards exist because other users may accept a request to connect but then become inactive on the platform and thus neither provide competition nor reference points (see the Online Appendix, Figure D.II and associated discussion). Thus, the absence of any other active users of such leaderboards should result in no effect on the user’s physical activity. Indeed, we find a null effect of such leaderboards on steps (see the Online Appendix, Section D). This falsification test with an NCT strengthens the plausibility of the common trends assumption.


This paper is available on arxiv under CC BY 4.0 DEED license.


L O A D I N G
. . . comments & more!

About Author

EScholar: Electronic Academic Papers for Scholars HackerNoon profile picture
EScholar: Electronic Academic Papers for Scholars@escholar
We publish the best academic work (that's too often lost to peer reviews & the TA's desk) to the global tech community

Topics

Around The Web...