The first and last female samples are identical (except for size, they appear to be exactly the same image). Hard to believe that one was identified correctly almost 80% of the time, and the other only 30%.
Hi - I'm a Survata co-founder.
Thanks for catching that - it was an error we made in producing the chart, not a problem with the underlying study. The image has been updated. Sorry for the confusion!
In my mind, this suggests a curious error in the study (since the sample appears large enough that this shouldn't happen randomly): probably an error in reporting the results, maybe the wrong image in the results?