Author Correction: The effectiveness of public health advertisements to promote health: a randomized-controlled trial on 794, 000 participants

Elad Yom-Tov, Jinia Shembekar, Sarah Barclay, Peter Muennig

Introduction

This article serves as a formal correction to the study "The effectiveness of public health advertisements to promote health: a randomized-controlled trial on 794,000 participants," originally published in npj Digital Medicine. The correction addresses critical statistical reporting errors regarding the demographic composition of the study's massive participant pool.

The Nature of the Correction

In the original publication, the authors investigated how digital public health advertisements influence health behaviors across a cohort of nearly 800,000 individuals. A fundamental requirement for such a randomized controlled trial (RCT) is the assessment of demographic balance between treatment and control groups. If groups are not balanced—for instance, if one group is significantly younger or has a different gender distribution—the internal validity of the trial's outcomes may be compromised.

The primary correction concerns Table 1 of the original paper. The initial caption incorrectly stated that there were no statistically significant differences between the study groups. However, a re-evaluation of the data revealed that significant demographic imbalances did, in fact, exist.

Statistical Findings and Demographic Skew

The revised analysis indicates that the $p$-values for both age and gender distributions across the groups were less than 0.05 ($p < 0.05$), contradicting the original report of $p > 0.05$. This shift from "not significant" to "statistically significant" highlights several key trends in how the digital advertisements were distributed:

  • Youth Over-representation: In the "all user" treatment population, the 18–24 year old demographic was significantly over-represented.
  • Older Adult Under-representation: Conversely, the 50–64 year old demographic was under-represented in both the "tracked" and "all user" populations.
  • Gender Imbalance: Statistical differences were also confirmed in the gender distribution of the participants.

Implications for Digital Health Research

The correction underscores a vital challenge in digital health interventions: the inherent bias of advertising platforms. Even with a sample size approaching one million participants, the algorithms used by digital platforms to serve ads may not distribute information uniformly.

For researchers, this serves as a cautionary note. Large-scale digital RCTs must account for the structural biases of the platforms they utilize. The "aha moment" of this correction is the realization that massive scale does not automatically guarantee demographic parity. Researchers must rigorously monitor and statistically control for these imbalances to ensure that the observed effects of a health intervention are not merely artifacts of the platform's delivery logic.

Conclusion

While the correction does not explicitly alter the broader conclusions of the original study regarding ad effectiveness, it provides the necessary statistical transparency required for the scientific record. It ensures that the demographic nuances of the 794,000-person cohort are accurately represented, allowing for a more precise interpretation of the trial's results.

Summary updated

Abstract

In the original version of the published Article, there was an error in the caption to Table 1 which stated “None of the differences are statistically significant (χ2, two-sided, p &gt; 0.05)”. This has been changed to “The 18–24 year old are over-represented in the all user treatment population, while the 50–64 year old are underrepresented in both tracked and all user population, p-values were &lt;0.05 for age groups and gender.” This has been corrected in the HTML and PDF version of the Article.
BibTeX
@article{yomtov2018author,
  author = {Elad Yom-Tov and Jinia Shembekar and Sarah Barclay and Peter Muennig},
  title = {{Author Correction: The effectiveness of public health advertisements to promote health: a randomized-controlled trial on 794, 000 participants}},
  journal = {Npj Digit. Medicine},
  year = {2018},
  doi = {10.1038/s41746-018-0047-z},
}
Figure