Hi Jeremy,
I am also interested in your position on 1 and 2 tailed tests. Someone
asked you: " Are you saying that you should always make a two-tailed
prediction and perform a two-tailed test? Or are you saying that even
if you have made a one-tailed prediction, you should still perform a
two-tailed test?"
And you replied..."The latter. You can make a prediction that it will
go in a particular direction, but you still should use a 2 tailed test.
Note that lots of tests don't exist in one tailed forms - ANOVA,
chi-square, for example."
I follow your latter point that not all tests have 1 tailed forms, but
I have never heard of running a 2 tailed test with a 1 tailed prediction
before and I can't see members in my dept. using such an approach
either. Would you mind elaborating on your reasons for this? Also, in a
related vein, I was wondering what your thoughts are on making
directional predictions when the theory states that the results should
go a certain way, but there is not any research to suggest that they
will e.g., we would expect 2 tests of the same construct to converge
with high correlations (if they are measuring the same underlying
construct), but there is no research correlating these particular tests.
Is this grounds for a directional hypothesis? I have heard some
researchers say that we can make directional predictions based on theory
alone, whilst others prefer theory and pervasive trends in the
literature.
Thanks
Kathryn
>>> Jeremy Miles <[log in to unmask]> 22/03/2007 18:59:12 >>>
On 22/03/07, leah quinlivan <[log in to unmask]> wrote:
> Ok,
> I meant partial Eta squared, is that the wrong symbol?
>
No, that was right.
> What I actually did is run 4 custom model factorial MANOVAs in SPSS.
I
> attempted to hold some predictors constant (this was before Kathryn
gave me
> great advise about running hierarchical multiple regression).
>
MANOVA's not a great word, because it means so many things. It stands
for multivariate analysis of variance, which is anova, with multiple
outcomes. But there is also a manova approach to repeated measures
analysis of variance. And people often use MANOVA to mean the
procedure from SPSS, called MANOVA, which can do lots of things (like
multivariate regression, or even principal components analysis) so I'd
like to know what you actually did.
> I had to run 4 factorial MANOVAs, as many of the predictors (6
continious)
> had a number of levels (3 continuous DVs). The custom models crossed
over
> the factors to test the main effect of the group, and the interaction
of the
> discrete predictors. I applied the Bonferroni procedure, alpha at
.125.
>
So there were 3 DVs, and 6 continuous predictors, and a group factor
(with 2 groups?)
> The results were nonsignificant (fine, also interesting), but some of
the
> effect sizes were large.
>
Ask for options, parameter estimates. That helps to make sense of it
all, and allows you to express things in their original units.
Personally, I'd avoid MANOVA, cos it's weird and hard and complicated,
and run three anovas (or strictly, ancovas).
> I have read over the multiple regression material, and your book
"applying
> regression and correlation", but at this stage (project due next
week), Ive
> run out of time. For me to understand stats, i have to deeply
understand the
> theory, data, and so forth. For me to get to grips with multiple
regression
> enough to FULLY understand and grasp it, is not feasible in this time
frame.
> I think i have a hold of MANOVA but that could easily be a
crocodile!!
>
Drop MANOVA, and/or read Andy Field's book on SPSS.
> However, my research life has just began, and I'm very excited about
the
> sheer number of multivariate approaches. Nonetheless, stats are easy
to do
> bad, and I want to do them well. For my own interest, Im going to
conduct a
> hierarchical multiple regression, putting the predictors in
sequentia
l
> order, according to the effect size found. However, me n' multiple
> regression have got to spend some quality time together!
>
I wouldn't do it according to the effect size found. The advantage of
hierarchical is that it lets you put things in in the order that your
theory says it should, not the order that the data says it should.
It sounds like you're struggling with a complex design, and not a
large enough sample (which is a pretty common problem - I have it most
days. I'm actually having it today). Are you interested in the
groups, or in the continuous predictors? Are the groups randomised?
if they are, then adding the predictors should only shrink the
standard errors.
Jeremy
> Thanks once again,
>
> Leah
>
>
>
>
>
>
> On 22/03/07, Jeremy Miles <[log in to unmask]> wrote:
> > Hi Leah
> >
> > On 22/03/07, leah quinlivan <[log in to unmask] > wrote:
> > > Thanks Jeremy,
> > >
> > > Its brilliant to get advice. Will have to take a bit of time to
digest.
> > > Sorted the typo, and report confidence intervals.
> > >
> > > This listserve is great, as there is a lack of stats courses in
the
> republic
> > > of Ireland.
> > >
> > > I have another question for anyone whos interested: in a MANOVA
is it ok
> to
> > > say for e.g " the partial Þ2 was quite strong accounting for 37%
of the
> > > variability to this interaction (Þ2 =.37)".
> > >
> >
> > Possibly, but MANOVA is a rather general term - can you explain
what
> > you did. (Eta squared in RM anova is weird, for e.g.).
> >
> > > Also, I have found strong nonsignificant interaction effect
sizes. I
> have
> > > interpreted this as due to small sample size (N=51, but with
missing
> data,
> > > and between groups, small n) ?? Im going to use the strong effect
sizes
> on
> > > the predictors as a guide for inputting predictors in a
hierarchical
> > > multiple regression.
> > >
> >
> > That sounds sensible.
> >
> > > I want to give the service users who participated in this
research as
> much
> > > respect as possible by conducting rigorous statistical analysis.
Ive
> kinda
> > > taken off a bit more than I can chew, but I love a challenge!
> > >
> >
> > :)
> >
> > > Thanks for advice, as sometimes with stats, Im swimming a stormy
ocean
> in
> > > the dark! Its great to get a point the right direction.
> > >
> > >
> >
> > Or a log to hang onto. Unless it turns out to be an alligator.
> > (Maybe I took that analogy too far.)
> >
> > Jeremy
> >
> > --
> > Jeremy Miles
> > Learning statistics blog:
> www.jeremymiles.co.uk/learningstats
> >
>
>
>
> --
> Leah Quinlivan
--
Jeremy Miles
Learning statistics blog: www.jeremymiles.co.uk/learningstats
|
