Fisher's sons' effect in sexual selection: absent, intermittent or just low experimental power?

The Fisherian sexual selection paradigm has been called the null model of sexual selection. At its heart is the expectation of a genetic correlation (r_G ) between female preference and male trait. However, recent meta-analysis has shown estimated correlations are often extremely weak and not statistically significant. We show here that systematic failure of studies to reject the null hypothesis that r_G  = 0 is almost certainly due to the low power of most experimental designs used. We provide an easy way to assess experimental power a priori and suggest that current data make it difficult to definitively test a key component of the Fisher effect.