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.

Sunday, February 01, 2009

Reporting Factorial ANOVA Descriptives

If a factorial ANOVA has a non-significant interaction, is it necessary to report any descriptives besides group differences on levels of one IV that are averaged over other IVs (e.g., only report marginal means, associated SDs and cell sizes, rather than factorial cell means, SDs, and cell sizes)? It seems to me the only time it makes sense to report indivudal group means (factorial cells) is when there is a significant interaction...

Thanks

Chris

posted by Chris @ 11:53 AM  

4 Comments:

  • At 9:34 PM, Blogger Kris said…

    Chris
    I will start out by saying that I don't know the answer to this. However, I find that a table is a good way to list all of the information and then it might not be necessary to report individual cell descriptives in your write up for non-significant interactions.
    Kris

     

  • At 9:30 PM, Blogger Susan Sheu said…

    instead of doing a table, would it be possible to write all the means for the interaction in the result section?

     

  • At 10:24 PM, Blogger Sun said…

    Adding to Chris' Question, I would like to clarify what descriptives need to be reported in what particular cases. For example, what if there is no significant main effect and only significant interaction effect? What if there is significant main effect of one IV and no significant interaction effect? What descriptives information should I report in these various cases?

     

  • At 12:54 PM, Blogger Mari said…

    As mentioned in class, you MUST report the means, standard deviations, and ns for the model you actually analyzed. That is, even if all effects were not significant in your analysis, all effects were examined, and you must report the descriptive data in such a way that this is clear.

    This is because the inclusion of a factor (and its associated interaction terms) changes the way variance is apportioned--even when that effect is not significant.

    Also as previously mentioned, if you are comparing 3 or fewer means, you should report the means in the text. With 4 to 20, you should use a table. Twenty or more calls for a figure. This information is also in the Publication Manual.

     

Post a Comment

<< Home