I don't know much about these CESD or Internet addiction scales, but they appear to be non-negative. Non-negative variables are never normally-distributed, but in some sufficient cases (i.e. asymptotic normality) you can get really good approximations. I recommend adding constraints like non-negativity to the model when it is feasible.
With that table, you are in N-dimensional space, so you can’t really plot it. What I like to do is stick with the main effects and the interaction. That way I can plot the response surface in 3 dimensions. Most people do simple slopes (and I do too) but I just love proc gplot and throwing a wireframe graph into my paper. I am getting pretty tired of SAS university edition though and am trying to learn to do that in R instead. I can’t wait for the semester to end. I’m totally signing up for your R class. Flexplot saved my bacon on my latest paper because I never did get the hang of proc mixed and all your mixed model vids with R were exactly what I needed.
There are different notions of plotting that can illustrate multiple dimensions. One of my favorites are parallel axis plots, although even they get a bit incomprehensible for a sufficiently-large number of variables.
9:45 Hot political takes! "Especially when you look at this far-right plot, you see that ... the slope of non-religious people is far less steep than religious people".
Looking forward to going through this sort of analysis hands on in a few weeks' time.
Me too!
Please talk about Propensity score and the inverse probability weighting it’s super complicated and it’s widely used now
I don't know much about these CESD or Internet addiction scales, but they appear to be non-negative. Non-negative variables are never normally-distributed, but in some sufficient cases (i.e. asymptotic normality) you can get really good approximations. I recommend adding constraints like non-negativity to the model when it is feasible.
With that table, you are in N-dimensional space, so you can’t really plot it. What I like to do is stick with the main effects and the interaction. That way I can plot the response surface in 3 dimensions. Most people do simple slopes (and I do too) but I just love proc gplot and throwing a wireframe graph into my paper. I am getting pretty tired of SAS university edition though and am trying to learn to do that in R instead. I can’t wait for the semester to end. I’m totally signing up for your R class. Flexplot saved my bacon on my latest paper because I never did get the hang of proc mixed and all your mixed model vids with R were exactly what I needed.
There are different notions of plotting that can illustrate multiple dimensions. One of my favorites are parallel axis plots, although even they get a bit incomprehensible for a sufficiently-large number of variables.
Gracias bro
9:45 Hot political takes! "Especially when you look at this far-right plot, you see that ... the slope of non-religious people is far less steep than religious people".
You know me--Mr Controversy!