Jul 25, 2017 · 1 min read
The word “significant” is unfortunate. As you say, all it means is that you’ve ruled out the role of chance in the association you’ve found. If you use multivariate methods (and who doesn’t love and use mulitvariate methods?) you can also rule out the role of things like age and gender. Not really any different from controlling for chance, but you still see people writing papers where they list and compare p-values and never give you a glimpse of the absolute effect. One might almost suspect them of a little data manipulation to pad their publication records……
