PG851: GLM--ANOVA

This blog is for the use of students enrolled in either section of PG851. It is intended to be a forum for asking, discussing, clarifying, and helping.

Wednesday, February 23, 2011

For the ANCOVA hw, I was wondering if we would use the raw or adjusted means in our write-up? And why?

I thought we always use raw means, but I don't know why.

posted by Diana Ro @ 3:27 PM  

10 Comments:

  • At 6:23 PM, Blogger David said…

    You should always report the values for the model that you actually tested.

     

  • At 12:44 PM, Blogger Diana Ro said…

    So for the ANCOVA, we used adjusted means correct? And if we do, do we need to report the standard error?

     

  • At 2:56 PM, Blogger Grant Goodman said…

    I think I understand which means to report and why, but I am trying to figure out how SD should be reported. It seems that scores that have been adjusted for an ANCOVA analysis would not only have adjusted means but also adjusted standard deviations too. However all we have is a standard error value next to our adjusted means and I am unsure what to do with that.
    Thanks for the help

     

  • At 4:55 PM, Anonymous Nikki said…

    If you report the adjusted means, then yes, you report the standard error.

     

  • At 5:34 PM, Blogger bfullmer said…

    In class, Dr. Clements said we could report either the original or the adjusted. If we report adjusted, we report SE instead of SD. My question... if we report adjusted and SE, do we report n as well or confidence intervals?

    Thanks!
    Brooke

     

  • At 6:35 PM, Blogger Jesse Malott said…

    OK - Here comes the ignorance. I am confused about what is adjusted and what is raw. Is it adjusted after I run the ancova or are we talking about after the regression analysis? I can't find the answer in my notes.

    Thanks

     

  • At 6:47 PM, Blogger Jesse Malott said…

    Or estimated means?

     

  • At 6:55 PM, Blogger Jesse Malott said…

    Also, what does "To what value of the covariate were scores adjusted?" mean? Are we to list the values that we put the adjusted values that we put in our write-up or are we putting the r-value?

     

  • At 8:05 PM, Blogger David said…

    I'm sorry for posting a comment earlier that wasn't quite right and caused more confusion. Thanks Brooke for the clarification!

    Yes, still report n even if you report adjusted means and SE. You can do CI but it's not necessary.

    Jesse, you would use the raw values when testing for the homogeneity of regression. Think of that as a preliminary step to see if an assumption has been violated.

    In ANCOVA, "the means [of the DV] are adjusted to the values that are predicted if all cases scored about the same on the CV" (p. 424). Then an ANOVA is done on the new, adjusted DV values.

    As for you last question, it's asking what score is everyone assumed to have on the CV (remember how the adjustment works). For a hint, SPSS gives it to you in the output.

     

  • At 8:36 AM, Blogger Jesse Malott said…

    Perfect! Thanks David.

     

Post a Comment

<< Home