I have been asked by a researcher the following. Prior
to treatment 35 patients were asked 5 yes/no questions
regarding their habits. After therapy each patients
treatment outcome was classified using a 6 value
scale. The researcher wants to know if any of the 5
questions are good predictors of treatment outcome.
I used Stata's -ologit- on each question by itself and
did not find any question to be significantly
associated with outcome at the 5% or 10% level.
I then included all questionstios into a single
-ologit- model and got:
. xi:ologit global nf1-nf5
Iteration 0: log likelihood = -43.829309
Iteration 1: log likelihood = -38.057521
Iteration 2: log likelihood = -37.932282
Iteration 3: log likelihood = -37.931337
Iteration 4: log likelihood = -37.931337
Ordered logit estimates
Number of obs = 35
LR
chi2(5) = 11.80
Prob
> chi2 = 0.0377
Log likelihood = -37.931337
Pseudo R2 = 0.1346
------------------------------------------------------------------------------
global | Coef. Std. Err. z P>|z|
[95% Conf. Interval]
-------------+----------------------------------------------------------------
q1 | 1.548668 .76917 2.01 0.044
.0411227 3.056214
q2 | .8078226 .672176 1.20 0.229
-.5096181 2.125263
q3 | -.2580387 .7716049 -0.33 0.738
-1.770356 1.254279
q4 | -1.656282 .7447833 -2.22 0.026
-3.116031 -.1965337
q5 | 1.562217 .7666611 2.04 0.042
.059589 3.064845
-------------+----------------------------------------------------------------
_cut1 | .0145412 .7083502
(Ancillary parameters)
_cut2 | 1.922157 .7940725
_cut3 | 4.343486 1.077408
------------------------------------------------------------------------------
Can I make the statement that q4 is the most important
predictor, followed by q5, q1 and so on. That is,
ranked by their corresponding p values?
Thanks,
Ricardo
__________________________________________________
Do you Yahoo!?
Yahoo! Mail Plus - Powerful. Affordable. Sign up now.
http://mailplus.yahoo.com
*
* 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/
