[R] prevalence ratio and confidence intervals

Roberta Pereira Niquini robertaniquini at ensp.fiocruz.br
Fri Dec 12 20:23:45 CET 2008

Hi everybody,

I would like to estimate prevalence ratio and confidence intervals. 

I tried to do a log-binomial regression, but there was a failure of 
Now, I would like to learn how to do a poisson regression with robust 
I am trying to estimate coefficients with poisson regression and then get 
standard errors that are adjusted for heteroskedasticity. 

glm22<- svyglm(y~x1+x2+x3+offset(log(x4)), data = banco,  family = poisson, 
design= design_tarv)

# Y has a binomial distribution (0/1)
# X1, X2, X3 e X4 are categorical variables.
#I am using the library(survey) because it is an analysis of Complex Sample 
Survey Data .


svyglm(y~x1+x2+x3+ offset(log(x4)),data = banco, family = poisson, design = 

Survey design:
svydesign(ids = ~conglomerado, strata = ~estrato, data = banco, 
    weights = ~peso)

            Estimate Std. Error t value Pr(>|t|)    
(Intercept) -5.61224    0.07223 -77.699  < 2e-16 ***
x1           0.33847    0.07428   4.557 0.000155 ***
x2           0.17745    0.07059   2.514 0.019765 *  
x3           0.33508    0.09447   3.547 0.001808 ** 
x4           0.24382    0.08808   2.768 0.011217 *  
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 

(Dispersion parameter for poisson family taken to be 0.7535822)

Number of Fisher Scoring iterations: 5

# Using family=quasipoisson, I found the same values.



         (Intercept)        x1           x2           x3        x4
(Intercept)1.060857e-12-1.306035e-13-5.139155e-13 -9.788354e-13 -3.428080e-13
x1 -1.306035e-13  7.237868e-13   -3.263182e-13  -1.620593e-13  1.704392e-13
x2 -5.139155e-13  -3.263182e-13  1.250564e-12   7.207572e-13   -9.350062e-13
x3 -9.788354e-13  -1.620593e-13  7.207572e-13   1.707176e-12   -2.244859e-13
x4 -3.428080e-13   1.704392e-13   -9.350062e-13  -2.244859e-13   2.031640e-12


 (Intercept)       x1        x2            x3             x4
1.029979e-06 8.507566e-07 1.118286e-06 1.306589e-06 1.425356e-06 

I think these standards errors are very small. 

Is this the correct form to do poisson regression with robust variance?

Thank you for the help, 

More information about the R-help mailing list