Stata commands to conduct hierarchical multiple logistic regression

[Matt Mathew]

Hello.
I am new in Stata. I would like to conduct multiple logistic regression to analyze factors associated with anemia among children using Rwanda DHS data. could anyone help me with Stata commands to use. Thank you

[Clyde Schechter]

I think this question is too broad and vague for anybody to give a useful answer. I suggest you post back with more information.

What is your specific research question? ("Analyze factors associated with anemia among children" is not a research question. It's a generic term for several things you might do in the course of answering a research question. What specific factors do you hypothesize might be associated with anemia?) What other factors are either known or likely to be associated with anemia that you might have to adjust your analysis for? What variables in the data set you have are measures of those factors, or can be used to proxy those factors? What variables in your data set enable you to identify cases of anemia? Do you really want to look at the dichotomous outcome of anemia? Or might you be better off examining the effects of your risk factors on the hemoglobin level? What ages define childhood for the purpose of your study? Is it plausible that the same factors apply equally across that age range, or might it be necessary to analyze smaller age ranges separately? Are all children in the data set suitable for inclusion in your study? If not, what are the inclusion and exclusion criteria?

These are just some of the questions you need to think about and arrive at answers to before you are ready to think about specific Stata commands. You need a study design first. Then you need a plan that maps the concepts and variables in your study to variables in your data set. It is likely that some data management will be required to get the data set ready for analysis. Then you need to think about how the effects of the risk factors you are studying are likely to be quantitatively related to anemia (or hemoglobin level, or whatever actual outcome variable you settle on) so that you can write down a statistical model. Only then are you at the point where you can begin the actual analysis, i.e fitting the model and interpreting the results.