sgdesign

I'm saying it's a fallacy is you use sampling bias on one factor to discredit another unrelated factor.

If you pick a sample to test a new painkiller, you should include an equal amount of men and women, old and young people, people of different ethnic backgrounds, etc. to have a balanced sample.

On the other hand, you do not need to have equal shares of people whose favorite color is blue, red, green, etc. because that clearly has no relation with how people would react to medicine.

Author here, I appreciate the feedback but I feel like there's a falacy here. If we asked respondents if they liked ice cream, and it turned out 95% of survey respondents hate it, would you say the rest of the data is not valid because the survey is clearly not representative of the general population?

The fact that the survey may not be representative on things like immigration or its Japanese audience has no relation to the data's validity on completely unrelated questions such as developer's outlook on AI.

A recap of the recent State of Web Dev AI survey results.

Thanks for sharing!