The blog post has a mistake, there were only 292 not 2923 respondents (he copied a superscript 3 from the PDF). The mailing list had 1939 members[1], so only 15% responded. All the data quoted in the paper discussed in the blog is survey data... so it seems that a large proportion of academics that respond to surveys about their political ideology are liberal. This could also be explained by a large number of non-liberal academics who don't care for answering surveys about political ideology. Not sure what counts for rigorous exposition in the social sciences, but I had to laugh when the authors actually said that these ideology surveys overestimate the proportion of liberals because conservatives don't want to self-identify, and that this supports their point!

[1] http://yoelinbar.net/papers/political_diversity.pdf