The study that's getting a lot of attention in the blogosphere, presenting data showing that differences in preschool are correlated with differences in adult political orientation, can be read here. The criticisms seem to be largely based on speculation, but I have to admit, it's hard to evaluate the study from the write-up, because it doesn't give any representation of the distribution of political orientation, and in order to get the data on things like SES, you have to read 25-year old write-ups of previous studies using the same data. It also seems natural to worry about the sample, since it comes from a town notorious for its progressive residents. As some of this post's commenters noted, it could just be that whiny kids grow up to be "hostile to their environment," and since the Berkeley environment is decidedly liberal, these particular kids grew up to be conservative. However, I will note that the study uses widely accepted methodologies and produces results that are, in many cases, highly statistically significant.
The one thing the study appears to have going for it is that its results, at least for the correlations between the adult personality measures and political orientation, are consistent with the work of John Jost and his colleagues (you can read their papers here and here, the second paper being a response to this paper). Recall that they found political conservativism to be positively correlated with death anxiety; system instability (i.e., things like social, economic, and political threats increase conservativism); dogmatism-intolerance of ambiguity; need for order, structure, and closure; and fear of threat and loss; and negatively correlated with openness to experience; uncertainty tolerance; integrative complexity (which just means cognitive complexity); and self esteem (though this negative correlation was fairly small). Since the Block and Block study found conservativism to be negatively correlated with things like "enjoys aesthetic impressions," "complicates simple situations," "has a wide range of interests," "tends to be rebellious, non-conforming," and positively correlated with things like "uncomfortable with uncertainty," and "behaves in a sex-typed manner," the two studies seem to line up fairly well. Of course, the Jost study has been widely criticized by conservatives, but the methods and data seem pretty sound to me. You can check out the papers and judge for yourself.