0
$\begingroup$

I'm running four one-way ANOVAs using a different DV each time. Each ANOVA has one DV and four IV condition levels. I currently have each ANOVA set up in a different data file. I would have liked to do to MANOVAs but my data doesn't meet all the assumptions so I can't. I understand that I should use a correction (I'll likely choose Bonferroni) on the posthoc ANOVA test results. However,

  1. Does SPSS do the adjustement for me on the post-hoc tests so I still look for a p < 0.05 on the SPSS output? Or do I need to divide the p value by the number of comparisions myself?

  2. Do I need to apply a correction for the four one-way ANOVA's ontop of the multiple comparisions occuring inside each ANOVA? So assuming that the SPSS output is already adjusted for the multiple comparisions in the post-hoc tests, do I need to look at this and further divide p = 0.05 by 4 to find a significant result? If so, then is there a way to have SPSS do this even though my ANOVA's are in different data files? The only information I can find about SPSS automatically doing these adjustements is in the post-hoc tests.

Thanks very much!

$\endgroup$

1 Answer 1

0
$\begingroup$
  1. If you specify Bonferroni or another method that adjusts for multiple comparisons, then the significance values in the output would be adjusted, so you would not have to adjust them yourself.

  2. The procedure you're using to do the ANOVAs on separate data has no way to know that you're wanting to combine information across different analyses, so any multiple-comparison adjustments are confined to a particular analysis. So yes, if you want to maintain an overall error rate across multiple analyses, you'd have to manually adjust significance levels by multiplying the stated significance values on the output by the appropriate factor (4 for a Bonferroni adjustment).

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.