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.

Saturday, March 12, 2011

Mixed Repeated Measures ANCOVA??

So the hypothesis that I'm working with would involve running basically a mixed repeated measures ANCOVA, can we do that?  I imagine it is probably possible physically (I have to go to the library to check on SPSS later), but is that getting into a bunch of potential problems that we haven't discussed in the course because I'd be combining two already semi-complicated analyses?  Should I forget about the covariate for the purposes of this project?

Thanks.

posted by Tiffany @ 3:18 PM  

9 Comments:

  • At 12:27 PM, Blogger David said…

    From the first slide in your repeated measures notes, there are two types: studying the same outcome over time and studying the same outcome from multiple perspectives. Looking over the list of variables in the final project dataset, I'm not sure that I see something that falls in either of those categories. You might want to double check to see what you wanted your repeated measures to be.

    On a more general note, you don't have to try to make this project more complicated than it has to be. Obviously, you need to do something more complex than a oneway or a t test. But at the same time, these results aren't going to be published in a journal or be your dissertation.

    Show effort and learning in terms of scope of analysis, but I wouldn't get carried away with overly ambitious designs. It's kind of like multicollinearity in ANCOVA... don't waste power for little gain. =)

     

  • At 12:48 PM, Blogger David said…

    Just to clarify, the note above was just a suggestion for those who were feeling stressed. I didn't mean to discourage those who want to challenge themselves.

     

  • At 1:10 PM, Blogger Tiffany said…

    This comment has been removed by the author.

     

  • At 1:14 PM, Blogger Tiffany said…

    This comment has been removed by the author.

     

  • At 1:43 PM, Blogger Tiffany said…

    Ok, after much confusion in my brain, I think I understand what you mean. So I was thinking that any items that looked at mother attribute and father attribute wouldn't be independent so you'd use repeated measures. But when you use repeated measures is when two non-independent people are rating one outcome variable (eg husband and wife rating marital satisfaction). But if our one participant is rating both parents then it is not considered dependent and in need of repeated measures. Is that correct?

     

  • At 4:21 PM, Blogger David said…

    Since I'm venturing into territory where I'm not totally certain, you might have to ask Dr. Clements for the final word.

    But my understanding is that a repeated measures ANOVA is a way of dealing with one kind of dependent data... ones where "cases are measured more than once" (textbook, p. 243).

    You're right in your initial thinking that mother and father scores provided by the same participant are dependent. But I don't think RM is appropriate for this kind of dependency.

    There are other ways to deal with nested, dependent data such as HLM that may be of interest to you in the future.

    Hope that made some sense.

     

  • At 4:21 PM, Blogger Mari Clements said…

    Actually, one very common use of RM is when the same underlying construct is examined in multiple ways (e.g., self-reports of anxiety, therapist-reports of anxiety, and physiological measures...or family warmth).

    However, please refer to the slide that addresses the idea of combining ANCOVA and Repeated Measures.

     

  • At 6:02 PM, Blogger David said…

    I'm sorry for confusing you with wrong information.

     

  • At 8:08 PM, Blogger Tiffany said…

    Thank you both for replying. Since I wasn't quite sure on it, I decided to go with a different analysis (and it did seem like perhaps I was making it more complicated than it had to be). So no worries I don't have to run back to the computer lab in the middle of the night.

    Thank you for clarifying though, Dr. Clements, that is really helpful to fully understand while I start thinking more about research and analyses outside of class.

     

Post a Comment

<< Home