I've been watching these boards for some time, but I had to register so I could respond to this. In the US, research data, even at a public institution, is generally considered private to the researcher until the publication of the research. In many cases, the data does become public after a set amount of time whether there has been a publication or not. An exampe of this would be data from the Hubble Space Telescope (even though I am a computer programmer, I have a degree in astrophysics and have done analysis on satellite data). With the HST, data is kept private for 1 year, after which it is released to the public.
In the case of psychological testing, the subjects are usually given an identifier instead of their real name, such as a number or letter. The identities of the subjects normally cannot be released without concent from the subjects. Of course this is not really legally encoded per se, but normally a test subject has to sign a number of documents stating that they are a willing test subject (look up the Nuremberg Trials for information on why this is), and in those documents is usually a form that says that the personal information will not be given out by the researcher. That is a contract, and because that has been put in there the researcher cannot just give out the information. A researcher will give out this information (with consent, of course) to another researcher who wishes to do a follow up study or wishes to look at the results from a different angle, but it is unlikely that the researcher will give out this information to John Q. Public.
If a researcher disagrees with the conclusions that another researcher has come up with -- such as a corelation between variable A and variable B -- he will usually just conduct another study trying to corelate variable A with variable C, but including variable B in the data, rather than use the other researcher's data.