Hi,
Is anyone out there aware of data on the impact of high Click Counts per page on data quality for online studies?
I just analyzed the results of a study hosted on Qualtrics (with MTurk workers recruited through CloudResearch) and it turns out that the key expected finding, which is not quite significant when including everyone, is significant when excluding participants with exceptionally high Click Counts (on the study's main Directions page and the main Stimulus page). This makes me wonder if a very high click count may at times index unfocused or even anxious participants.
Interestingly, if "speeders" (who completed the whole study exceptionally fast) are excluded instead of high-clickers, the finding is also significant. And it's even more strongly significant (and with a higher N) if both speeders and high-clickers are excluded.
In short, there seems to be quite a bit of convergent evidence suggesting that, at least in this case, high-clickers and speeders were similarly unfocused and had a similar dampening impact on the study's finding.
When I searched the literature on Click Counts, however, I only found discussion of how very low Click Counts can be dubious (ie, if lower than the number of items on a page, it's indicative of software instead of a human).
In any case, I'm curious if anyone else has had a similar experience or is aware of studies on Click Counts that may shed light on this.
Thanks!