http://www.sciencedirect.com/science...47272716302201

First, WP deserves a fail for running a headline implying that the effect is true of libs or cons in general, as opposed to the actual subset of the population tested, political candidates.

Second, the study itself reports the "% beauty advantage" across different regions (Table 1), but calculate it in terms of the difference/standard deviation (see description below table) rather than difference/mean (which is how most people would think of a % advantage - i.e., a US conservative candidate is xx% more attractive than a liberal candidate, not xx% of a standard deviation). Adjusting for this, the actual "beauty advantage" of US candidates was (5.66-5.47)/5.47 = 0.034, or about 3%. The same calc for MEPs makes the cons' mean advantage around 6%.

So, your average republican candidate for senate or congress in the US is 3% more attractive than your average dem candidate, and the average con MEP candidate is about 6% more attractive than the average lib MEP candidate.

Another way to conceptualize the effect is to consult a z-table and evaluate the % of lib candidates the average con is more attractive than. You can do this by cross-indexing the effect in standard deviations (0.14 for US candidates, 0.25 for MEPs) with the value in the second table, expressed as a proportion out of 1.

http://www.dummies.com/education/mat...e-the-z-table/

Finding the intersect of z = 0.14 tells us that the average rep candidate is more attractive than .5557/1 or about 56% of all the dem candidates. (for EU candidates, the corresponding value of z = 0.25 says that the average con candidate is more attractive than about 60% of all the lib candidates).

Next question is whether these fairly small differences are important enough to care about? That's for people to decide for themselves. Makes a good headline though.