4 ms·
There's a major validity problem with the sampling method, and exterm points out a pretty glaring bias, so high numbers don't necessarily improve validity, only
by CognitiveLens 15y ago
There's a major validity problem with the sampling method, and exterm points out a pretty glaring bias, so high numbers don't necessarily improve validity, only reliability. Random sampling would be much better than opt-in sampling, but it's obviously pretty difficult to do that in practice.
- Steko 15y agoWould be nice if they could control for household income as well, you'd see much of the differences disappear.
- code_duck 15y agoAll browsers are provided free of charge and will run equally well on an average piece of hardware. Why do you think browser use is related to income?
- Steko 15y agoFirst we know IQ is correlated with income. We might guess Mac (hence Safari) users are correlated with income. We might guess Opera users are correlated with income for several reasons. We might guess Chrome users are correlated with income. We might guess Firefox users are correlated with income. We could guess the same of Lynx or any other niche browser more likely to be popular with well educated, higher income bracket people.
- code_duck 15y agoWe might guess those things, I suppose. I was wondering why specifically. Are you saying you think it is mainly related to education, and education is correlated to high income, this that's the link and not something else?