Include covariate in analysis - Statistica Visual Basic - Statistica - Dell Community

Include covariate in analysis

Include covariate in analysis

This question is not answered

Hello,

Thank you to those of you who responded to my earlier posts, but unfortunately, I am stuck. I have 10 subjects and multiple conditions, so I have a repeated measures ANOVA with one dependent variable. A reviewer asked me to consider another measured value (he termed it a "covariate") in my analsyis. I believe that I want to do an analysis of covariance so I can basically in account for the effects of my covariate on my dependent variable and still see if there is significance of my dependent variable for the various conditions.

Q. What are the steps to performing this analysis?

    a) How do I arrange my data? For my repeated measures ANOVA I had a column for every factor, a row for each subject. When I add the covariate, do I add these as extra rows or extra columns? When I try extra columns, I have too many factors for the experimental units.

   b) What analysis am I looking for?  Part of the reason I am stuck here is because I am not sure how to answer a) above.

Please help! I am really stuck here and am unaware of where else to turn. If you have any suggestions of other resources (or perhaps other forums?) where I may be able to find the answer, I would appreciate that as well. Thank you!

 

All Replies
  • Hello Cheryl,

    I believe one of the issues you are having is that you are taking repeated measurements of your covariate, if I remember correctly.  In that case it is a repeated measure, not a covariate.  In an ANCOVA your are partialing out the variance that was previously attributed to error but was actually attributed to the covariate.  You will only have one variable per covariate.  So in your example, you would have one case per participant, one variable for each of your repeated measurements, one variable for each categorical predictor and then one variable for each covariate.  You have to watch for degrees of freedom here especially with your small sample size.

      RM1 RM2 RM3 Cat Pred Cat Pred CV1
    Particpant 1            
    Participant 2            
    Participant 3            

    If you set up your data this way you can do a Repeated Measures ANCOVA with your within effects assigned to the three repeated measures (RM1-3), two categorical predictors, and then on covariate.

    Hopefully that helps.

  • Thank you datasleuth for the comment. I think we are getting closer... Your comment that my "covariate" should be used as a repeated measure makes a lot of sense to me. Unfortunately, I don't believe that I hvae any categorical predictors though. I only have 1 subject population, so I have no categorical predictors when my data are setup in this manner. Statistica will not allow me to run an ANCOVA analysis without any categorical predictors, so I am stuck again. Please advise! Thank you.

  • The categorical predictor is the grouping variable since with ANOVA (or ANCOVA) you are wanting to see how groups (signified by the grouping variable or categorical predictor) differ on the continuous dependent variable.  If you do not have a categorical predictor or grouping variable you would not be able to run an ANOVA.  Another analysis type would be most appropriate.  You can find a good guide of statistical tests here:  http://www.csun.edu/~amarenco/Fcs%20682/When%20to%20use%20what%20test.pdf.

  • StatSoft also has two tools to help with "What analysis am I looking for?".

    We have a Statistical Advisor and Commonly Used Statistical Tests.

     

    Angela Waner
    Forum Admin