[R] A simple question about summary.glm

michal33 michal_sa at hotmail.com
Mon Jul 30 05:05:27 CEST 2007


I am new to R and have tried to search similar questions but could not find
exactly what I am looking for, but I apologize if the question was already

I have 10 different treatments and want to know whether they affect the sex
ratios of insect emergence. After running the glms I got this table:

      Df Deviance Resid. Df Resid. Dev      F   Pr(>F)   
NULL                    133     9250.3                   
sex    1    481.5       132     8768.9 7.7212 0.006314 **
trt    9   1099.1       123     7669.7 1.9585 0.049780 * 

But now I would like to know WHICH of the treatments was significant. I
tried to use Tukey test but for some reason it does not work. 
My question is:
I used the following function: 
>summary(file.name, corr=F)
and got the following table:
Deviance Residuals: 
    Min       1Q   Median       3Q      Max  
-14.118   -4.808   -1.466    2.033   33.882  
              Estimate Std. Error   t value Pr(>|t|)    
(Intercept)  8.696e+00  1.893e+00     4.594 1.06e-05 ***
sexm        -3.791e+00  1.364e+00    -2.779  0.00631 ** 
trtccc      -1.050e+00  4.325e+00    -0.243  0.80859    
trtcga3      2.450e+00  4.325e+00     0.566  0.57211    
trtcga4     -2.300e+00  4.325e+00    -0.532  0.59584    
trtg         1.550e+00  2.497e+00     0.621  0.53593    
trtga4      -5.550e+00  4.325e+00    -1.283  0.20183    
trtp         5.422e+00  2.566e+00     2.113  0.03658 *  
trtpg       -1.850e+00  2.497e+00    -0.741  0.46019    
trtw        -3.634e-17  2.497e+00 -1.46e-17  1.00000    
trtwg       -3.750e+00  2.497e+00    -1.502  0.13573    

What do the stars  mean? Is it the same as Tukey test that tells me which
treatment is different from which? i.e. is trtp (with *) significantly
different to the control (which, by the way do not appear in this list and I
do not know why)? 


View this message in context: http://www.nabble.com/A-simple-question-about-summary.glm-tf4167757.html#a11857514
Sent from the R help mailing list archive at Nabble.com.

More information about the R-help mailing list