Individual level moderator in HLM? Need help, please!
I'm working on a paper about the influence of a group-level variable on a individual-level variable and want to add an individual-level moderator. I'm totally new to the whole HLM field and I found articles talking about group-level moderators, but none that dealt with individual-level moderators, so I was wondering if including those was possible at all. Any suggestions? Your help would be GREATLY appreciated!
Re: Individual level moderator in HLM? Need help, please!
yes, you can include moderators at the individual level (level 1) or group level (level 2) or, as you know, the most popular version is when level 2 moderates level 1 estimates. there are quite a few problems with estimating level 1 moderators, however. do keep in mind that you will be addding another random coefficient to be estimated within your model at level 1. since you're at the risk of running into problems with multicollinearity (the "death star", if you forgive my star wars lingo, of hierarchical linear models) so you'll have to mean-centre the predictors AT THE GROUP LEVEL (you cannot use grand-mean centering because you'd be brining influence of the other groups into each of your random moderation coefficients')... and i know a few people dont like to mean-centre at the group level because then you have to add back the effect of the removed mean as a level 2 predicior... which means you'll need a bigger sample size then (to take care of power issues) and the list goes on and on... oh, gebeezuz, in my years working with these i dont think i have ever seen a paper where moderation at level 1 wasnt addressed only as either an example or in theory... they can mess up the whole model very quickly because they need to become random effects at some point in the analysis