Are there other researchers on this topic? Seems like this is a critical set of things we should test but given the citations here, it unfortunately appears only you and your collaborators are exploring this topic. Is there related research for us to consider to make an even more compelling set of arguments?

are publications with and without publicly data different in other ways? is there a way to explore this?

why did you stratify and not do citations per year, ie rate that each item differentially accrued citations? Alternatively, this seems like a nice paired approach amenable to an effect size estimate, ie the difference in each time step between dataset and publication. Did you consider trying this? Could we?

I think there are even more powerful incentives to sharing data then citations. Has anyone does the research associated with incentives on this topic?

Do you see the F1000research data publications as a solution to the getting scooped issue you articulate here?

what is 'real' data? Is that the same as raw data? does derived data have the same benefits as those proposed here?