They offer a theory that explains their findings, but I am unconvinced their findings are accurate. The sample size of 60 is just too small to average out such subjective parameter as attractiveness, which among other things greatly depends on non-physiological parameters such as lighting during a photo shoot, or a facial expression.
For example, if the girl on the right were to pout just a bit, or had darker hair, or if her hair were darker, or if she was wearing a different color, the preferences might've fallen her way instead. There are other variable parameters affecting the selection that have nothing to do with the bone structure or hormones.
Whatever the flaws of this study may be, confusing correlation with causation is not one of them. The fact that foot size is correlated with facial prettiness is (if true) interesting; but nobody is suggesting that having small feet causes your face to get prettier, nor that having a pretty face causes your feet to shrink.
Correlation is not a causation.