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.

Thursday, October 27, 2011

In the HW since sphericity was assumed since the variable only has 2 level and only 1 pair can be created, in the write up do we say that sphericity was assumed and then indicate the Tests of within subjects effects between the factor and the greenhouse geiser, however that wouldn't make sense especially because the test of within subjects reveals significance. How do we report this. What kind of stats do we put in our write up since our mauchy's test of sphericity shows no significance.

posted by Unknown @ 1:27 PM  

5 Comments:

  • At 11:18 AM, Blogger T.J. said…

    1.) DO report that sphericity is assumed because you only have 2 levels.

    2.) You DO NOT need to worry about the results of Mauchly's if you have only 2 levels, so greenhouse geiser is NOT important.

    3.) Sadly, I have no idea what you are getting at with the "because the test of within subjects reveals significance". If you still have a question, can you rephrase?

    4.) You do not need to put stats on Mauchly's when it is not significant.

    Please break up your question into chunks, it would make it much easier to answer.

     

  • At 11:30 AM, Blogger Grace said…

    Hm...I'm not sure I understand your questions, so please feel free to post again if my answers aren't helpful.

    In the write up you should explain conceptually why you do or do not have concerns about sphericity. If you do have concerns, then look at the p values in Mauchly's Test of Sphericity and decide upon the appropriate adjustments when interpreting the results of the within-subjects table.

    In other words, the p values of the within-subjects table do NOT indicate sphericity.

     

  • At 12:01 PM, Blogger Mari Clements said…

    Remember that whether or not an assumption was violated (whether that assumption is sphericity, normality, or homogeneity of variance) is a completely separate question from whether or not an analysis is significant.

    That is, you can
    1. violate sphericity and have a significant effect of the IV
    2. NOT violate sphericity and have a significant effect of the IV
    3. violate sphericity and NOT have a significant effect of the IV
    4. NOT violate sphericity and NOT have a significant effect of the IV

    ANY of these are possible outcomes of your analysis.

     

  • At 12:03 PM, Blogger Mari Clements said…

    Remember also that, in a writeup, assumptions are assumed unless you say otherwise.

    Do answer specific questions outside of the writeup, but in the writeup, assumptions are assumed unless you say otherwise.

     

  • At 12:05 PM, Blogger Mari Clements said…

    Finally, the stats you report in your writeup are the stats from the analysis of the model, and they are the same ones as always. Mauchly stats never make the writeup.

     

Post a Comment

<< Home