MedStats Google Group

Re: {MEDSTATS} Re: Multivariate regression, cell with few counts

tfr...@gmail.com (Thomas Fröjd) — Thu, 26 Nov 2009 09:23:25 UT

Frank: Actually it is a count distribution but I use a negbin
distribution function in the model so for this exercise I belive we
can treat it as continious.

Ted: Thanks, that settles it.

As an experiment I tried to run your code but switching the N for

Z=A Z=B
X=A 590 1
X=B 490 46

RE: {MEDSTATS} Re: Multivariate regression, cell with few counts

ted.hard...@manchester.ac.uk (Ted Harding) — Wed, 25 Nov 2009 14:52:45 UT

Just to follow up on my suggestion (the rest snipped).
I have run an R simulation of a model (with non-zero interaction terms)
to check on what I suggested previously. R code with comments follows.
With the system of contrasts used in the regression, the true
effect and interaction values are:
Intercept = 0.0000

Re: {MEDSTATS} Re: Multivariate regression, cell with few counts

f.harr...@vanderbilt.edu (Frank Harrell) — Wed, 25 Nov 2009 14:48:58 UT

Ted's response is a good one. It would still be nice to know more
about Y. Is is continuous? Ordinal? If ordinal what is the
distribution of frequencies?

Frank

Re: {MEDSTATS} Re: Multivariate regression, cell with few counts

tfr...@gmail.com (Thomas Fröjd) — Wed, 25 Nov 2009 11:32:21 UT

Hi,
I should have learnt to provide context when asking question by
now....

Groups are naturally occuring subgroups. The first variable is
exposure to an event(was there and physically affected, was there,
weren't there). Second variable is lost a relative or friend in the
event (yes or no).

RE: {MEDSTATS} Re: Multivariate regression, cell with few counts

ted.hard...@manchester.ac.uk (Ted Harding) — Tue, 24 Nov 2009 15:05:35 UT

Well said, Frank!
Regarding the OP's question, I think the simplest approach (and the
most transparent) would be to simply apply a standard regression
analysis to the data as they stand, using the desired model
Y = a + X + Z + X*Z

In the result, there will be a reported estimate of the effect

Re: Multivariate regression, cell with few counts, how should it be handled?

f.harr...@vanderbilt.edu (Frank Harrell) — Tue, 24 Nov 2009 14:01:28 UT

Please tell us more about Y and about what was known apriori about the
likely interaction between X and Z.

I assume you meant to say 'multivariable' instead of 'multivariate'
because your problem is univariate with only one dependent variable.

Frank

Multivariate regression, cell with few counts, how should it be handled?

tfr...@gmail.com (Thomas Fröjd) — Tue, 24 Nov 2009 13:45:32 UT

Hi this is a pretty basic question but I haven´t encountered it before
so I would like to ask for the proper way to handle this.

X and Z are both categorical variables and I am interested on how they
together predict Y. X has three possible values(ABC), Y has 2(AB). The
model I am interested in is Y =a + X + Z + X*Z.

Re: {MEDSTATS} Re: sample size for sensitivity/specificity

sos_st...@yahoo.co.uk (Evie) — Tue, 24 Nov 2009 10:47:19 UT

Thank you all for your advice which has been really helpful

Re: {MEDSTATS} Re: sample size for sensitivity/specificity

m861h...@btinternet.com (Martin Holt) — Tue, 17 Nov 2009 17:52:02 UT

If I understand correctly, Evie will need to be able to have confidence intervals simultaneously for sensitivity and specificity. Googling, I found this (it comes with no guarantees). Beware the wrap-over.

[link]

Re: {MEDSTATS} Re: sample size for sensitivity/specificity

brown...@gmail.com (Barry Brown) — Tue, 17 Nov 2009 16:43:38 UT

I have written a program called CONFINT that calculates the sample size
necessary to
have a specified probability that the confidence interval length is not
longer than some
specified size. The program is free and includes Fprtran90 source, Win32
and Mac
executables. Download from
[link]

Estimating test retest and inter rater reliability

s.wh...@sgul.ac.uk (s.white@sgul.ac.uk) — Tue, 17 Nov 2009 12:44:07 UT

Dear All

Is it possible to obtain estimates of test retest reliability and
inter rater reliability using the following design?

Time 1 Time 2
S1 rated by R1 S1 rated by R1
S2 rated by R1 S2 rated by R2
S3 rated by R2 S3 rated by R2

Re: {MEDSTATS} Re: sample size for sensitivity/specificity

sos_st...@yahoo.co.uk (Evie) — Tue, 17 Nov 2009 11:12:22 UT

That's great, thank you all for your help.

Evie

Re: {MEDSTATS} Re: sample size for sensitivity/specificity

tfr...@gmail.com (Thomas Fröjd) — Tue, 17 Nov 2009 09:34:59 UT

Just for your information, if I remember it correctly there is a typo
in the graph suggested below in the pdf version. The correct graph can
be had at the BMJ website.

On 16 Nov, 19:27, Pedro Emmanuel Alvarenga Americano do Brasil

Re: {MEDSTATS} Re: Sample size for Repeated Measures

d.e.kornb...@herts.ac.uk (kornbrot) — Tue, 17 Nov 2009 07:54:43 UT

G*Power does this and more
[link]
diana

Professor Diana Kornbrot
email: d.e.kornb...@herts.ac.uk
web: [link]
Work
School of Psychology
University of Hertfordshire
College Lane, Hatfield, Hertfordshire AL10 9AB, UK

Re: {MEDSTATS} Re: Sample size for Repeated Measures

e.sabagh...@gmail.com (ehsan sabaghian) — Tue, 17 Nov 2009 05:50:31 UT

Thank you for your advice, but I have to analyze data via Repeated Measures
ANOVA method,
so I have to estimate sample size for this kind of test.