So this can be applied to a correlational analysis.

It still doesn't get us there if it's purely correlational of course.

The example I used here today was just cross sectional correlational data.

Uh, but again, at least in this example, we're just dealing with correlational data.

There are all sorts of possibilities for confounds and in fact, this is truly just a correlational study.

So just remember that the validity of any argument based on correlational analyses depends on all these assumptions.

And that kind of correlational, analysis is what was done after inaudible introduced the test to really drive home the point that it, it truly was culturally biased.

This is merely correlational data, not causal, and it's about happiness in the first Hollywood sense I'm going to talk about happiness of ebullience and giggling and good cheer.

So we might have evidence for full mediation in a statistical sense But if we're just dealing with cross sectional, correlational data it doesn't mean we can say anything, make any strong arguments about causality.

I'm, but still, even if we tested it using structural equation modelling, we are dealing with cross sectional correlational data, it's hard to make strong causal themes. that said, we can use mediation with true independent variables.

And one of the reasons we're not quite ready to jump into inferential statistics is you saw that the, the validity of correlation based arguments depends on the reliability and validity of the measures that go into the correlational analysis.

More importantly, and this is an, this is critical for just consumers of statistical information, the validity of any argument that's, that comes from a correlational analysis depends on these assumptions, so these assumptions need to be met for that argument to have any weight.