[R] model formula

[Eik Vettorazzi]

Am 11.08.2011 17:39, schrieb Uwe Ligges:
> 
> 
> On 11.08.2011 17:27, Bond, Stephen wrote:
>> Hello useRs,
>>
>> Pls help with removing a single interaction term from a formula:
>>
>> summary(
>>          glm.turn.2<-
>> glm(cbind(turn.cnt,tot.cnt-turn.cnt)~sn+poly(relAge,2,raw=T)+termfac+rate:termfac,data=fix,
>>
>>                    family="quasibinomial")
>>          )
>>
>> Gives
>>
>> Coefficients:
>>                             Estimate Std. Error t value Pr(>|t|)
>> (Intercept)               -7.028467   0.106002 -66.305<  2e-16 ***
>> snFeb                      0.156963   0.023660   6.634 3.27e-11 ***
>> snMar                      0.317540   0.022883  13.876<  2e-16 ***
>> snApr                      0.526084   0.022004  23.908<  2e-16 ***
>> snMay                      1.026710   0.020347  50.460<  2e-16 ***
>> snJun                      0.841044   0.021318  39.452<  2e-16 ***
>> snJul                      0.668790   0.022530  29.685<  2e-16 ***
>> snAug                      0.544267   0.022580  24.104<  2e-16 ***
>> snSep                      0.389667   0.023363  16.679<  2e-16 ***
>> snOct                      0.351294   0.023586  14.894<  2e-16 ***
>> snNov                      0.391464   0.024057  16.272<  2e-16 ***
>> snDec                     -0.373369   0.028755 -12.985<  2e-16 ***
>> poly(relAge, 2, raw = T)1  2.952887   0.067455  43.776<  2e-16 ***
>> poly(relAge, 2, raw = T)2 -1.783907   0.064074 -27.841<  2e-16 ***
>> termfac2                  -0.681719   0.128571  -5.302 1.14e-07 ***
>> termfac3                  -1.032416   0.146396  -7.052 1.77e-12 ***
>> termfac4                  -1.267011   0.108940 -11.630<  2e-16 ***
>> termfac5                  -1.009922   0.213129  -4.739 2.15e-06 ***
>> termfac6                   3.300301   0.203465  16.221<  2e-16 ***
>> termfac1:rate              0.012274   0.019895   0.617    0.537
>> termfac2:rate              0.121724   0.013472   9.036<  2e-16 ***
>> termfac3:rate              0.175232   0.018987   9.229<  2e-16 ***
>> termfac4:rate              0.197726   0.005787  34.164<  2e-16 ***
>> termfac5:rate              0.145622   0.027295   5.335 9.56e-08 ***
>> termfac6:rate             -0.362379   0.025261 -14.345<  2e-16 ***
>>
>> and I would like to remove the termfac1:rate interaction. Is there a
>> way to do that?
>> Thank you
> 
> And what do you want to do with the observations with termfac=1 then?
> Ignore? Also move to the Intercept?

Actually they are already in the Intercept because of the treatment
contrasts. The crucial part in the glm formula above is
 termfac+rate:termfac
which specifies intercepts deviating from termfac1 level and slopes for
every level of termfac. A dummy coding for different slopes should work,
since the slopes are independently modelled and tested against 0.


> 
> Uwe Ligges
> 
> 
> 
>>
>> Stephen Bond
>>
>>
>>     [[alternative HTML version deleted]]
>>
>> ______________________________________________
>> 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.
> 
> ______________________________________________
> 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.

-- 
Eik Vettorazzi
Institut für Medizinische Biometrie und Epidemiologie
Universitätsklinikum Hamburg-Eppendorf

Martinistr. 52
20246 Hamburg

T ++49/40/7410-58243
F ++49/40/7410-57790