Advertisement
If you have a new account but are having problems posting or verifying your account, please email us on hello@boards.ie for help. Thanks :)
Hello all! Please ensure that you are posting a new thread or question in the appropriate forum. The Feedback forum is overwhelmed with questions that are having to be moved elsewhere. If you need help to verify your account contact hello@boards.ie
Hi there,
There is an issue with role permissions that is being worked on at the moment.
If you are having trouble with access or permissions on regional forums please post here to get access: https://www.boards.ie/discussion/2058365403/you-do-not-have-permission-for-that#latest

ANOVA, post hoc tests, F statistic and p values.

  • 25-09-2009 3:29am
    #1
    Closed Accounts Posts: 2,736 ✭✭✭


    Hi.
    Say i do a one-way ANOVA looking at the difference between 4 treatment groups (4 levels of a factor).
    I get a p value of <0.05. OK that's significant.

    But let's say only one pair of the 4 treatment groups are different from each other on the post-hoc Newman-Keuls test.

    Just wanna check what that means:
    Does that mean that the significant difference between treatments really only applies to that pair and the rest are not different.

    So for the significantly different pair, i can report F=whatever, P<0.05.

    But if i wanted to report the result of one of the other pairs i couldn't say p<0.05. I'm guessing then that i can't report the F statistic either for these pairs as that is the "tethered" to the significant p value isn't it- (which in turn of course is only relevant to the significant pair).

    So is there anything (non-significant F or p values etc) that i could report about the non-significant pairs.

    Also, i'm trying to do a 2 way ANOVA on a couple of stats packages (minitab, GraphPadPrism) but because of one or two missing data points/slightly unequal cell sizes, they can't do it (useless pieces of crap :P ).
    Instead of doing a 2 way ANOVA, can you just do a separate one-way ANOVA for each of the 2 factors. I know you wouldn't get the interaction statistic but otherwise would it be OK?
    I'm guessing not. :)

    Any help would be great thanks. :)


Comments

  • Registered Users, Registered Users 2 Posts: 78,644 ✭✭✭✭Victor


    tech77 wrote: »
    So is there anything (non-significant F or p values etc) that i could report about the non-significant pairs.
    You can report that they are non-significant (whether marginal or not), stating the usual stats.


  • Closed Accounts Posts: 2,736 ✭✭✭tech77


    Victor wrote: »
    You can report that they are non-significant (whether marginal or not), stating the usual stats.

    Yeah, but what are the usual stats for them.

    They don't seem to have a p value (the p value of <0.05 refers to the significant pair only) and i don't think the F value refers to them either, does it?

    What are the stats that refer to a non-significant pair like this?

    is it the q value- that's something i've seen mentioned opposite the non-sig stuff.

    Thanks.


Advertisement