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.

Tuesday, January 18, 2011

3 Questions

1.) On the Review sheet, #3, it tells us to know which variable types are appropriate as IVs and which are appropriate as DVs in ANOVA. I know we went over variable types in class and have them on our slides; however, I cannot find which are appropriate as IVs and which as DVs. In the book, it simply says that IVs include both qualitative and quantitative, and DVs, continuous, ordinal, and discrete. Any further clarification would be greatly appreciated.

2.) 1st Handout, page 4, slide "So You Have Heterogeneity of Variance", Understand Why... 1. Why is it that the spread of scores is different in different groups... is this because of the IV or ?? and 2. Is it that the level of IV affects participants differently??

3.) At a review session I wrote in my notes, that "a smaller alpha is less conservative"; however, a greater alpha would increase a type I error, would it not? I would think it would think that a smaller alpha should be more conservative? Clarification would be greatly appreciated.

Thank you for your help!

Brooke Yetter

posted by bfullmer @ 10:26 PM  

3 Comments:

  • At 11:22 PM, Blogger David said…

    This comment has been removed by the author.

     

  • At 11:29 PM, Blogger David said…

    1) In all of the examples you've seen in class (and the HW if you've started on it), what kind of variables were the IVs? The DVs? Are they continuous or categorical? That should give you a hint.

    2) This is more of a theoretical question you would ask yourself if you have heterogeneity of variance. For instance, if SAT scores at two schools had different variances, perhaps you need to check if one school was a selective private school and the other was a diverse public school. Or perhaps a certain treatment works better for men than women, creating different variances.

    3) You're right about alpha being the probability of a Type I error. A smaller alpha would reduce your probability of a Type I error (which you might say is more "conservative") but the trade-off is more Type II error.

     

  • At 12:09 AM, Blogger bfullmer said…

    Thank you, David!

     

Post a Comment

<< Home