The most recent Quirks newsletter had an interesting piece about mitigating response bias from socially acceptable responses. While the item count method presented there can reduce the stigma of having to actually check a box associated with any particular risky behavior, I think part of the trouble with surveying about socially unacceptable behavior is also about identity. The respondent doesn't even want to admit to themselves that they perform the activity.
This was particularly important in my previous work on bill payments. People don't want to identify as a person who doesn't pay their bills. Few people want to admit to paying bills late or not at all, however people were much more willing to complain about having to pay late fees. Sometimes little tweaks can help navigate around socially acceptable responses.
I would be curious if the authors have conducted any research-on-research to see what the difference in estimated prevalance is between a direct response and the item count method.
This was particularly important in my previous work on bill payments. People don't want to identify as a person who doesn't pay their bills. Few people want to admit to paying bills late or not at all, however people were much more willing to complain about having to pay late fees. Sometimes little tweaks can help navigate around socially acceptable responses.
I would be curious if the authors have conducted any research-on-research to see what the difference in estimated prevalance is between a direct response and the item count method.
RSS Feed