Do attractive people tend to have more daughters?

I got a commentary in the Times today (due to the Times paywall, this is a local link to the unedited article, rather than to the published version) about a study that estimated that people rated as 'unattractive' when they were 7 years old only had 44% chance of their first child being a girl. This effect seems utterly implausible.

There are a number of subtle biases about the publicity surrounding such research: as well as the fact that only positive results tend to be published and covered in the media, 'statistically significant' findings from small studies tend to be gross over-estimates of the effects that are plausible. Andrew Gelman makes these points well in his article on a previous study by the same author. If we add in the possibility of biases in the study design, then we are left with the conclusions of one of the most downloaded and cited papers in modern medical literature - John Ioannidis claim that most published research findings are false (or at least very exaggerated).