How is Response Bias factored into Population Statistics?


When a population statistic can be objectively measured, it is a great item to utilize in a discussion.


However, I often see statistics utilized in discussions were I don’t see how they could have been objectively measured. Particularity, I am referring to any statistic that relies on self-reported data about a fact about oneself that the reporter may not be willing to acknowledge even to themself.


Couple of Examples that come to Mind:

1) The percentage of men that experience sexual assault and domestic violence

2) The percentage of individuals that are sexually attracted to the same sex


How is this issue addressed in statistics? Can we even get accurate statistics for such information?

In: 3

Any decent study will detail their methodology and *as a feature* comment on its own potential weaknesses. Subsequently, anyone else referencing these will acknowledge specifically what each statistic is representing. While some older studies did try to “adjust” based on some heuristic, most of those methods are now seen as dodgy. References to previous studies for margins of error *help*, but blindly plugging them in is seen as bad science.

Meta-analysis can help to see numbers which might be “closer to the truth”, but these are always taken with a grain of salt, since populations are not uniform.