Doesn't offering the incentive get you biased results, by including participants who didn't care that much for your survey's subject in the first place?
Possibly, but it was a chance to win a $25 card, not one for every participant. I still got very good results, much of which I attribute to Facebook ad targeting.
I wonder if it would make sense to do a split (meta) survey, one group gets no incentive, the other an incentive as above: and then compare conversion rates between the groups (where applicable). My intuition is that the first group will be smaller but will have a better conversion rate.
I wasn't worried about conversion rate, just number of responses. I planned to run it for a week but stopped after 3 days. My CPM ended up being 1/3 of what I budgeted per completed survey, again b/c Facebook ad targeting was already so specific. My survey was about trading card games, and I limited it to people interested in Magic: The Gathering, Pokemon, etc.
