3 ms·
Do you know what a confidence interval is?
by psychphysic 3y ago
Do you know what a confidence interval is?
- Mathnerd314 3y agoI think that's the point, the paper is saying "a linear model was a good fit" when P=0.11. It's not at all a strong fit, compared to e.g. physics where p=1e-5 is desired. You'd think with 300k data points the results would at least be "statistically significant", p<0.05. The only data they present is averages over countries so I wouldn't be surprised if there was a huge effect of which site the worker was at, and their analysis deliberately ignores this effect by averaging, instead of including it as an explicit variable that explains most of the variation.
- psychphysic 3y agoR coefficient is how you assess fit to model. P is for significance you might use it to assess which model is better (typically using an interaction term). That 0.11 shows that the alternative model is not significantly better than the model presented. That might be confusing because typically everyone just thinks p<0.05 is good.
- scythe 3y agoI work in this field. I know lots of statistics. Do you?
- psychphysic 3y agoX DOUBT.