0

I am new to using R and its bit hard for me to understand the output. I have here the glm ouptput and I am basically looking for confidence interval. Is my output (confint(fitBLA17)) showing a 97.5% confidence interval?

fitBLA16 <- glm(BLA.DOC~BLA.DO + BLA.Temp+BLA.Chlo,data=swan, family=gaussian(link=log))
> summary(fitBLA16)

Call:
glm(formula = BLA.DOC ~ BLA.DO + BLA.Temp + BLA.Chlo, family = gaussian(link = log), 
    data = swan)

Deviance Residuals: 
     Min        1Q    Median        3Q       Max  
-1.26276  -0.42484  -0.07066   0.30826   2.01178  

Coefficients:
            Estimate Std. Error t value Pr(>|t|)    
(Intercept) -3.51962    0.68938  -5.105 4.10e-06 ***
BLA.DO       0.45738    0.05926   7.719 2.25e-10 ***
BLA.Temp     0.05454    0.01398   3.903 0.000258 ***
BLA.Chlo    81.67750   12.55730   6.504 2.27e-08 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

(Dispersion parameter for gaussian family taken to be 0.4415451)

    Null deviance: 150.516  on 59  degrees of freedom
Residual deviance:  24.726  on 56  degrees of freedom
AIC: 127.08

Number of Fisher Scoring iterations: 4

> confint(fitBLA16)
Waiting for profiling to be done...
                  2.5 %       97.5 %
(Intercept) -4.86872307  -2.17627071
BLA.DO       0.34060856   0.57280469
BLA.Temp     0.02758951   0.08172916
BLA.Chlo    56.90288019 106.33717878
Glen_b
  • 257,508
  • 32
  • 553
  • 939
Sonam
  • 21
  • 1
  • 1

1 Answers1

4

No, your output is showing several 95% intervals (two-tailed, from the 2.5% point to the 97.5% point of the relevant distribution, which form the upper and lower limits of the intervals):

enter image description here

Also note that your output is not for confint(fitBLA17) but for confint(fitBLA16).

You could get one tailed 97.5% intervals (either upper or lower) by replacing the bound at the other end by $-\infty$ or $\infty$ as appropriate.

Glen_b
  • 257,508
  • 32
  • 553
  • 939
  • I found a great answer by @gung [Why do my p-values differ between logistic regression output, chi-squared test, and the confidence interval for the OR?](http://stats.stackexchange.com/a/144608/17230) though it's perhaps more than the OP currently wants to know. (+1) – Scortchi - Reinstate Monica Oct 15 '15 at 13:58