|
|
Covariate types in regression |
|
Message/Author |
|
Michael posted on Friday, February 09, 2018 - 6:38 am
|
|
|
Hello, I'm looking to conceptualize a basic multi-level regression model and am asking for guidance on which standardized output option to use, 'stdyx' or 'stdy,' given my situation. Here is the basic layout: I want to examine adversity (# of adverse events) as a predictor of memory (a continuous score) controlling for basic covariates in a sample of children/siblings clustered within families. The children (and their families) also come from 2 different distinct cohorts. The variables: memory = dependent variable, continuous score adversity = # of adverse events, independent variable, continuous age = covariate, continuous sex = covariate, dichotomous cohort = covariate, dichotomous FAMID = cluster variable Age and sex are limited to the 'within' level, and cohort is limited to the 'between' level. I allow adversity to be modeled at both levels Example syntax: usevar = memory adversity age sex cohort; within = age sex; between = cohort; cluster = FAMID; ANALYSIS: type = twolevel; MODEL: %within% Memory ON adversity age sex; %between% Memory ON adversity cohort; OUTPUT: sampstat stdyx; |
|
|
For a continuous predictor, use STDYX. For a binary predictor, use STDY. |
|
Michael posted on Monday, February 12, 2018 - 9:46 am
|
|
|
Thanks Bengt, quick follow-up to see if I have it right: So for my example model I shared with both continuous and dichotomous predictors at both within and between levels, should I run the model both ways? Once with STDYX to report the results for the continuous predictors, and then again but with STDY for the results for the binary ones? In that case, which one (STDYX or STDY) would I use to report all the effects in the 'between' model? |
|
|
You don't have to run it twice - if you ask for standardized you get all the versions in the output. So you just pick from different parts of the output. |
|
Back to top |
|
|