In each city, two observers—one woman and one man—collected data at all sites (in total, four women and four men, all expert ornithologists, were involved in the study).
Only four people of each gender is not a statistically relevant sample size. Whatever the birds reacted to, might have been because of random (not gender-related) differences between those two groups of four people each. And they weren’t all present at all sites, some might have done multiple sites and some only one, narrowing the effective sample size even further.
They were instructed to look directly at the birds. Maybe one or two of the four women just happened to look the birds directly in their eyes in a more intense way and skewed the data slightly. Or maybe one or two of the four men interrupted eye contact slightly more often or in more cases. Or any other thing.
Not because of their gender, just because they’re human and all humans are different. Four people are not enough to average out these differences and get at gender traits, if those do exist.
Also, the difference in Flight-Initiation-Distance (FID) was not all that big (~1 meter), barely visible in this picture and note the wide spread of the data:
As you can see in the data, the FID was mostly rounded to the nearest meter. The rounding error is about as big as the supposedly observed effect. Maybe one group tended to round down if in doubt and one group tended to round up in edge cases.
Edit: There’s more: only in the female group are there clearly visible secondary peaks at five, fifteen and twenty meters (and even at 25 m, 30 m and 35 m). This indicates, that participants tended to round not just to the nearest meter, but to the nearest five meters, casting doubt on all the measurements.
Here’s the actual study. And here’s a critical point:
Only four people of each gender is not a statistically relevant sample size. Whatever the birds reacted to, might have been because of random (not gender-related) differences between those two groups of four people each. And they weren’t all present at all sites, some might have done multiple sites and some only one, narrowing the effective sample size even further.
They were instructed to look directly at the birds. Maybe one or two of the four women just happened to look the birds directly in their eyes in a more intense way and skewed the data slightly. Or maybe one or two of the four men interrupted eye contact slightly more often or in more cases. Or any other thing. Not because of their gender, just because they’re human and all humans are different. Four people are not enough to average out these differences and get at gender traits, if those do exist.
Also, the difference in Flight-Initiation-Distance (FID) was not all that big (~1 meter), barely visible in this picture and note the wide spread of the data:
As you can see in the data, the FID was mostly rounded to the nearest meter. The rounding error is about as big as the supposedly observed effect. Maybe one group tended to round down if in doubt and one group tended to round up in edge cases.
Edit: There’s more: only in the female group are there clearly visible secondary peaks at five, fifteen and twenty meters (and even at 25 m, 30 m and 35 m). This indicates, that participants tended to round not just to the nearest meter, but to the nearest five meters, casting doubt on all the measurements.
This would be a perfect paper (including the news sites misrepresenting it too) to teach students how to critically examine science.
deleted by creator