SE logistic with high SE

[Tommaso Salvitti]

Hi to everybody

firthlogit gruppoASDHR ANSIADEP2 i.quo3,or

Initial: Penalized log likelihood = -26.901475
Rescale: Penalized log likelihood = -26.901475
Iteration 0: Penalized log likelihood = -26.901475
Iteration 1: Penalized log likelihood = -16.835876
Iteration 2: Penalized log likelihood = -16.492021
Iteration 3: Penalized log likelihood = -16.477754
Iteration 4: Penalized log likelihood = -16.477724
Iteration 5: Penalized log likelihood = -16.477724

Number of obs = 41
Wald chi2(3) = 9.36
Penalized log likelihood = -16.477724 Prob > chi2 = 0.0249

------------------------------------------------------------------------------
gruppoASDHR | Odds ratio Std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
ANSIADEP2 | 5.000001 8.227536 0.98 0.328 .1987523 125.7848
|
quo3 |
2 | .3043478 .2510386 -1.44 0.149 .0604316 1.532768
3 | .0112721 .017632 -2.87 0.004 .0005255 .2418033
|
_cons | 3.285714 2.005822 1.95 0.051 .9931133 10.87078
------------------------------------------------------------------------------
Note: _cons estimates baseline odds.

Is it correct a model with a SE so high??

[Clyde Schechter]

With only 41 observations and 3 right hand side variables, there is nothing particularly surprising about a standard error of 8.2 around an odds ratio of 5.0.