On 2004-11-02, at 11.35, Ron�n Conroy wrote:
Take a step back here. Have you *graphed* your outcome against your
predictor variable?
Thanks for your advice. Yes I have graphed it. And there is a squared
component that kicks in at about 7 on the scale were probabilities
starts to rise dramatically. The graphed probabilities looks fine and
are according to theory.
The problem is the standard errors in the predicted RRR using -nlcom-.
There seem to be a paradoxical relation here: the more extreme the RRR
the LESS significant they are.
The paradox described above can be found in auto.dta also. Consider a
logit model where the probability of a car being foreign is modelled as
a function of length. Length is negatively associated with foreign
(-.0797353). Using -nlcom- a significant (p<.001) ratio of 1.3 between
the predicted probabilities are fond for length=1 vs length=10. When
length=1 is compared to length=100 the ratio increase to 764 but is no
longer significant (p=.606). Code is listed below:
sysuse auto
logit foreign length
// RRR for length=1 vs length=10
nlcom (exp(1 * _b[length] + _cons) / (1+ exp(1 * _b[length] +
_cons))) / ///
(exp(10 * _b[length] + _cons) / (1+ exp(10 * _b[length] +
_cons))) //
// RRR for length=1 vs length=100
nlcom (exp(1 * _b[length] + _cons) / (1+ exp(1 * _b[length] +
_cons))) / ///
(exp(100 * _b[length] + _cons) / (1+ exp(100 * _b[length] +
_cons))) //
I might be doing something I shouldn't and I'm happy for any advice on
how to calculate RRRs with CI from the logit model above using
auto.dta.
Michael
On 2004-11-02, at 11.35, Ron�n Conroy wrote:
Michael Ingre wrote:
A follow up on statistical power.
I have calculated a few RRRs and an interesting pattern is emerging.
Extreme comparisons give insignificant p-values but others don't.
RRR for 9.5 vs 1, p=.669
RRR for 9.5 vs 9, p=.030
RRR for 2 vs 1, p=.049
Predicted absolute probabilities are: 9.5=.33 , 9=.14, 2=.000020 &
1=.000015
What is going on here? Am I doing something wrong? I appreciate any
suggestion because this makes no sense to me.
Take a step back here. Have you *graphed* your outcome against your
predictor variable? Use a smoother to have a look at the shape of the
relationship. I sometimes use -autosmoo-, but usually do this sort of
thing in JMP, where you can vary the smoothness of a spline
interactively. It is handy to know if there is a threshold effect
(above a critical value, risk begins to rise) or even a 'normal
region' phenomenon, whereby risk is lowest in some normal region, and
rises at the high and low extremes (weight and health is a classic
example).
You may also be the victim of small numbers in some of the categories.
But relative risk ratios are a way of measuring a phenomenon. The
first thing to do is to inspect the phenomenon personally, using the
Mk I intra-ocular traumatic test.
Ronan M Conroy ([email protected]) Senior Lecturer in Biostatistics
Royal College of Surgeons Dublin 2, Ireland +353 1 402 2431 (fax 2764)
-------------------- Just say no to drug reps
http://www.nofreelunch.org/
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
------------------------------------------------
Michael Ingre , PhD student & Research Associate
Department of Psychology, Stockholm University &
National Institute for Psychosocial Medicine IPM
Box 230, 171 77 Stockholm, Sweden
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
