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.

Friday, January 21, 2011

Questions on OneWay ANOVA Hmwrk

Ashli asked:
For the OneWay ANOVA Homework, would you like us to answer the questions and write our APA style paragraphs within the assignment document? Or would you prefer for us to answer all the questions about all the types of tests in one results section as if it were part of a single APA article? Or would you like us to have separate paragraphs as if each were its own separate results section in APA format?

Also for the violation of normality on the homework do we need to go back and recode the 4-6 outliers or since it passed the Fmax is it okay?

posted by Nikki Frederick @ 11:50 AM  

6 Comments:

  • At 12:10 PM, Blogger Nikki Frederick said…

    For the first question, you do not need to have answers to every single question in APA format. The questions that ask for a "write-up, being careful to note APA style" are the questions when you will want to enter a paragraph noting the results according to APA style.

    So, you could type directly onto the HW page, as it is a word document,and save that as your document to turn in, as long as you make sure the paragraphs that need to be in APA are in fact in that style (font, indentation, etc). Or you could create a Word document from scratch and just put the answers to the questions according to the HW sheet, making sure that the answers to separate questions begin new paragraphs so it's clear in grading which answer goes to which question.

    For your second question, the Fmax is used to test the homogeneity of variance, not the assumption of normality. Please over slides 19-20 on the first handout about the assumption of normality, noting when ANOVA is robust to this assumptions and what to do if it is indeed violated. I would say that unless you have been taught to recode outliers, that probably isn't required for this HW.

     

  • At 12:14 PM, Blogger Nikki Frederick said…

    also, do not forget to submit your spss output separately per the instructions for HW1.

     

  • At 12:49 PM, Blogger Ashli said…

    Hi Nikki, For the homework the normality part is looking like positive skew with a few outliers. I looked over those slides. The slide says to do "square root, logarithm, inverse, or dichotomize" for positive skew but we haven't really talked about how to do any of that. I know for outliers we can recode or delete but we haven't really talked about if that is okay to do for more than one. I know I can use the standard deviation to recode though... should I do that or just document that I chose to leave them in? I know the skew part can be an issue but I definitely don't know how to fix that apart from just deleting.
    Thanks!!
    Ashli

     

  • At 2:26 PM, Blogger David said…

    If you look at your notes, there are a certain set of criteria for which ANOVA is robust to deviations in normality. Let us know if those are met or not.

    What follows that in the notes is how to tell if the deviation is severe for large and small samples. It may be helpful to know that a sample is considered "small" when N is around 20.

    If you can justify that the normality assumption is sufficiently met, you do not need to transform or recode.

     

  • At 2:59 PM, Blogger David said…

    here is a relevant comment I left under a more recent post:

    My advice would be to take a look at ch. 2 in the textbook on outliers (p. 46-53). The author discusses two variables (var_a and var_c), one of which is sufficiently normal and another which is problematic (fig. 2.5). That should give you a good idea as to whether your distribution in your HW is "normal enough."

     

  • At 8:16 AM, Blogger Mari Clements said…

    For the future, please do insert your answers into the Word homework document (it makes it much easier to find each answer).

    As Nikki noted, you will want to make sure that APA formatting is used correctly in those portions which are write ups (as opposed to just answering a question).

    Finally, on assignments in which multiple analyses are computed (like this one), please treat each one as a separate Results section. That is, do not say, "As in the previous analysis,..." or anything like that. Make sure that you act as if this section were the only section your reader has access to.

     

Post a Comment

<< Home