Requests for Additional Help
If you think you currently need additional help with the class, please submit your request for that help to me, not to the TAs. Send me an email at clements@fuller.edu.
This is the only way that I can make sure that all students are given equal and fair access to additional help. Should you need help beyond the time available to you from the TAs, further assistance from the departmental teaching fellow is available, but again, you will need to request that directly from me.
This is the only way that I can make sure that all students are given equal and fair access to additional help. Should you need help beyond the time available to you from the TAs, further assistance from the departmental teaching fellow is available, but again, you will need to request that directly from me.
posted by Mari Clements @ 2:39 PM
2 Comments:
At 6:21 AM,
Anonymous said…
A 3 (cohabitation type) X 2 (gender) X 2 (time) ANOVA
on relationship quality indicated a significant main effect of
type of premarital cohabitation, F(2, 112) 11.57, p .01.
Planned comparisons revealed that the before-engagement
group (M 120.92, SD 14.92) reported lower relationship quality than both the after-engagement group (M
129.18, SD 10.84), t(92) 2.85, p .01, ES .64,
and the at-marriage group (M 132.11, SD 11.76),
t(71) 7.53, p .01, ES .84. ----
-I read this data and wonder: does there use of ANOVA compromise their P level since they are doing multiple comparisons? Also, what is the difference between alpha level and p level? Because earlier in this study the researchers reported alpha levels of .74 and .69, but report p levels of less than .01 - i thought they were the same thing??
At 10:08 AM,
Mari Clements said…
Please do not bring up new topics in existing threads, but rather create a new post.
As we discussed in class, ANOVA was designed for this very situation, and in fact, prevents the inflation of alpha by looking at all groups simultaneously. Once an overall significant finding is obtained, the comparisons completed afterward are (a) justified and (b) typically adjusted using one of the post-hoc comparison methods available in SPSS under post-hoc.
More on that topic next week, but both the book and your handouts show you how to conduct these multiple comparisons without compromising actual alpha.
The alpha levels of .74 and .69 that the authors refer to are almost certainly Cronbach's alphas (which are measures of internal consistency of a questionnaire or scale, and have nothing to do with the statistical test you describe above).
Alpha for ANOVA is set at .05, and is the critical value for p (i.e., the value p must be lower than for you to reject the null).
The obtained p value is the probability that results as large or larger as those obtained in the current analysis were obtained just by chance.
Post a Comment
<< Home