Peter Koval posted on Wednesday, October 12, 2016 - 9:33 pm
Apologies if this duplicates an existing thread/post: I have a question about how to properly standardize a between variable in a two-level model when I have unequal cluster sizes. I have data from 89 participants (clusters) each of whom has between 37 and 90 rows of data (occasions). My question is: how do I standardize a "between" variable (no within-person variation) when I have unequal cluster sizes? I want to ensure that it has a mean of zero and an SD of 1 when entered into my model. However, no matter how I standardize it (either before importing data into Mplus or using Mplus STANDARDIZE option) my variable doesn't seem to have a mean of zero or a variance of 1. How can I ensure that I have properly standardized my between variable?
Can't you just do a Type=Twolevel Basic run where the between-level variable is put on the Between list? That gives you its mean and variance over the 89 subjects. And then you can use that to standardize in Define in subsequent runs.