Offer clear definitions on rules for rejection rates on Prime Panels. Currently the site (and communications with support) say "Please note the maximum rejection rate is 30 %." What I learn when using prime panels in practice is that this means 30% of the size of the sample you requested, not a rate of people taking the survey that you reject, i.e., denominator for calculating the rate is x instead of (x + # rejected) Since this affects strictness of the screens you can use, it's important to make it clear up front so users can specify a rule they will be able to apply throughout.
hyper and micro batch combined
In case you are developing the site further, an option that allows researchers to combine the advantages of hyper and micro batches would be quite useful. Hyperbatches complete quickly and for this reason, one will get a certain demographic, especially in terms of geography as people from different places are active at different times. To get more “continuous” data collection, one tends to use micro batches.The issue is that this takes forever, and if one can not repost the HIT before the previous batch is completed it takes a very long time for Turkers to take the last HIT from a batch. What would be super cool, is to be able to micro batch, but release more than 9 HITs at a time (say, 50), and have the ability to repost, say every 30 minutes, even if only 45 people have completed the study from the previous batch. The current system would have us create MANY HITs manually to achieve this for hyperbatching.
Timed study start, close and user notification
Allow the option to close studies at certain times of day/week. Also all for a timed start of studies and send email notifications to included workers on launch