Some scientists horde data, on the theory that this gives them an edge that other scientists in the community lack so that they can publish on something that no one else can publish upon. Other scientists share data, on the theory that this allows more people to engage with their work and by doing so recognize it.
In academia, the first law of career advancement is publish or perish. More publications are good. But, academia recognizes superstars based not only upon how many articles one publishes, but how many times those articles are cited by others.
Empirically, if you are a scientist who manages to publish at all (and many people in academia publish very little once they receive tenure), the best way to advance your citation count and thrust yourself into academic superstar status, is to share your data, according to a short preprint examining the question released Sunday, at least in astrophysics, but probably in a much broader array of other disciplines as well.
Overall, making your underlying data available will increase you citation count for a paper by 20% on average.