HLM help desperately needed


New Member

I am currently learning HLM (Hierarchical linear modeling), and stuck! I will greatly appreciate if someone can help me.

Brief description of my data: I have N=2000 at the individual level (level 1), and N=267 at the branch/business unit level (level 2). My model consists of two sets of antecedents: one is at the individual level, say VA_level 1, VB_level 1, and VC_level 1. I then aggregated these three antecedents to the level 2, which forms the level 2 antecedents, say, VA_level 2, VB_level 2, and VC_level 3. In the model, I have branch size serves as a control variable for level 2 variables. In the model, I have a moderator say D at level 2, which suppose to moderate the relationship between these 6 antecedents and my dependent variable.

To test this model in HLM, my questions are:

1) What variable(s) should be included in the baseline model? I saw Raudenbush, et al, only include dependent variable, but some articles include all the antecedents and control, and some include antecedents and control variable?

2) When testing the moderating effect, do I need to "manually" create the interaction term in SPSS before creating my HLM MDM file or just enter the variable in the equitation? For some reason, I have the impression that HLM automatically generate the interaction term, but in some articles published, I saw people report how they create interaction terms. I am totally confused. If I need to create interaction term, in this case, there will be 6 interaction terms created, so what the equitation will look like?

I will greatly appreciate your assistance. By the way, I am really bad in statistics, so please use "simple" language. Thanks a lot!