[R] Coefficients, OR and 95% CL

Frank E Harrell Jr f.harrell at vanderbilt.edu
Tue Sep 23 14:37:02 CEST 2008


Eik Vettorazzi wrote:
> l.mod<-glm(menarche~age,binomial)
> 
> you will get odds-ratios by exponentiating the coefficients of this 
> model, so
> 
> exp(coef(l.mod))
> 
> will do this job. You may notice that this will produce an "OR" for the 
> intercept part as well - which is not interpretable.
> For the confidence intervals for this odds ratio you have to 
> exponentiate the borders of the "standard" confidence intervals for the 
> log-odds-ratios,
> 
> exp(confint(l.mod))
> 
> hth.

The approach of anti-logging a coefficient only works in the simple case 
in which the variable is binary (or a one-unit change is of interest and 
the variable is linear), doesn't interact with another variable, and the 
reference cell desired equals the reference cell coded.

Frank

> 
> Luciano La Sala schrieb:
>> Dear R-users,
>>
>> After running a logistic regression, I need to calculate OR by 
>> exponentiating the coefficient, and then I need the 95% CL for the OR 
>> as well. For the following example (taken from P. Dalaagard's book), 
>> what would be the most straightforward method of getting what I need? 
>> Could anyone enlight me please?  
>> Thank you!
>> Lucho
>>  
>>> summary(glm(menarche~age,binomial))
>>>     
>>
>> Call:
>> glm(formula = menarche ~ age, family = binomial)
>>
>> Deviance Residuals:      Min        1Q    Median        3Q       Max  
>> -4.68654  -0.13049  -0.01067   0.09608   2.35254 
>> Coefficients:
>>             Estimate Std. Error z value Pr(>|z|)    (Intercept) 
>> -17.9175     1.7074  -10.49   <2e-16 ***
>> age           1.3549     0.1296   10.45   <2e-16 ***
>> ---
>> Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
>> (Dispersion parameter for binomial family taken to be 1)
>>
>>     Null deviance: 974.31  on 703  degrees of freedom
>> Residual deviance: 223.95  on 702  degrees of freedom
>>   (635 observations deleted due to missingness)
>> AIC: 227.95
>>
>> Number of Fisher Scoring iterations: 9
>>
>>
>>
>>
>>
>>
>> ______________________________________________
>> R-help at r-project.org mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-help
>> PLEASE do read the posting guide 
>> http://www.R-project.org/posting-guide.html
>> and provide commented, minimal, self-contained, reproducible code.
>>   
> 


-- 
Frank E Harrell Jr   Professor and Chair           School of Medicine
                      Department of Biostatistics   Vanderbilt University



More information about the R-help mailing list