On top of that, the sample sizes are too small. This lack of statistical power, coupled with a lot of freedom in how results are analyzed, means almost any result can be found.
In the past, if you wanted to get published, make a crazy claim, like that liberals have prettier children or if someone's office is messy they are racist, and throw in a p value and statistical terms, and that was fine. The results mattered, not the methodology.
That could change in 2014. Medicine expects researchers to pre-register their predictions and planned analyses of experiments before looking at the data and psychology is leading the way in doing that for social sciences.
On to: #2 Open Science And Open Access