I've recently been given a dataset with the following variables:

-BMI -Type of Therapy (indicated by 1, 2, or 3) -ADH Levels -HIV positive or negative (0,1 dummy variables) -Systolic blood pressure -Diastolic blood pressure -CD4 Cell Count -HCV Status (discrete 0 to 4) -HIV RNA viral load -Education Level (discrete: 0, 1, 2, 3) -Age -Protease Inhibitor Use

I'm being asked to do the following:

  1. Describe the distribution of BMI (continuous and using the CDC categories: http://www.cdc.gov/healthyweight/assessing/bmi/adult_bmi/ ) by 5 year age categories, race/ethnicity, education, HIV status, HCV status, viral load, Hypertensive >=140/90 vs. non-hypertensive, current IDU, antiretroviral drug categories

  2. Are each of the predictors associated with BMI, our outcome variable?

  3. Which predictor(s) were statistically significantly associated with BMI in a full multivariate mode?

Can all this be done in Excel or do I need to use some other statistical software? I think what would help is if you can specify the method I should be using for each and if there's anything I need to watch out for. Thanks.

  • 1
    $\begingroup$ You could do almost anything in Excel given enough time and effort but I probably wouldn't do it that way. $\endgroup$
    – Glen_b
    Dec 6, 2015 at 4:41
  • $\begingroup$ What kind of methods should I use and what kind of software would be the best approach then? $\endgroup$
    – cambelot
    Dec 6, 2015 at 5:18
  • $\begingroup$ The questions are rather oddly phrased; it seems that the second item is probably asking for pairwise correlations (to what end I cannot fathom, since we know there are other important variables -- what could the pairwise correlation tell us that's very useful?), and the third presumably seeks multiple regression. You can do both in Excel, but I'd use a proper statistics package myself, since they will provide (or makes easy to obtain) particular diagnostics (Excel's limited diagnostics are mostly either misleading or useless). ...ctd $\endgroup$
    – Glen_b
    Dec 6, 2015 at 11:32
  • $\begingroup$ ctd There are several free ones, such as the ones mentioned here. The most widely used of the free ones is probably R. $\endgroup$
    – Glen_b
    Dec 6, 2015 at 11:32


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.