Assessing self-selection biases in online surveys: Evidence from the COVID-19 Health Behavior Survey

In the majority of the countries, the respondents recruited through the athlete images (images 1 and 2) and the nose-blowing images (images 3 and 4) reported higher adoption of face masks, except in France for the “male athlete” image (image 1) and in the Netherlands for the “woman blowing nose” image (image 3). [...] From the figure it stands out that the model estimates for the threat perception to oneself are consistently higher for the low threat perception in comparison to the estimates for the threat perception to the family. [...] On the contrary, we found no effect of the advertising image on the respondents’ perceived threat of COVID-19 to themselves in the final model in France, Spain, the Netherlands, and the United Kingdom. [...] In Spain and the Netherlands, the effect of the various images on the survey outcome of wearing a face mask was fully explained by adding the month of survey participation to the model. [...] Second, the results from the regression models revealed that the impact of the ad image on the various survey outcomes was explained by adding either the month of survey participation or the respondents’ age to the model.