Statistics advice

#1
Hello,

I have 6 independent groups and one dependent variable. I did a univariate ANOVA analysis and the overall group significance was not significant. Is it okay to compare individual groups with the control group and do a series of unpaired t-tests? Do I need to do a Bonferroni correction? If so, how do I go about doing that in Spss?

Thank you in advance for any advice you can provide!
 

TheEcologist

Global Moderator
#2
Is it okay to compare individual groups with the control group and do a series of unpaired t-tests?
Yes, but you could also do a multiple comparison with something like a tukey HSD.

Do I need to do a Bonferroni correction?
If you do the t.tests, yes.

If so, how do I go about doing that in Spss?
Simple just use a new level of significance. You now call a p-value significant if it has a value lower than 0.05/# groups (6 in your case so your new alpha is 0.05/6 = 0.00833).
 

noetsi

No cake for spunky
#3
I believe Tukey HSD already corrects for FW error so if you use that you should not need to do bonferonni. Although my memory may betray me here...