This reminds me of the chapter from TGSOE about the domestication of the silver fox in the Soviet Union, and how certain personality traits turned out to be tied to genes controlling appearance (floppy ears, spotted coats etc). With that in mind, the same correlation in humans seem plausible.

The particular claim in the headline is, as usual, too general though. There is presumably more than one way for genes to 'cook' a broad-faced-phenotype (and obviously there are environmental factors like extra fat which can also simulate the effect). So I don't think there is really a lesson we can (or at should) take away from such a study.

The same goes for the link between testosterone levels and trustworthiness. No-one is a slave to the chemicals of which they are composed. To say that testosterone cause untrustworthiness is an especially dangerous claim to make when you consider racial differences (black males 20% higher testosterone levels than white males).