0

I just fitted a GLM with a Gamma distribution and somehow my output gives me a pseudo R². (Nagelkerke). As far as I know a pseudo R² is only used when the response is ordinal or nominal, but my response is numerical. Why do I get a pseudo R²? Below you can see the structure of my data and the output of the model. Thanks in advance!

glm(DGUnkraut ~ Feuchte * Entfernung, data = BFK_oa, family = "Gamma") -> model2

ibble [396 x 16] (S3: tbl_df/tbl/data.frame)
 $ Datum     : Factor w/ 6 levels "2021-03-17","2021-04-07",..: 1 1 1 1 1 1 1 1 1 1 ...
 $ Soll      : chr [1:396] "119" "119" "119" "119" ...
 $ Plot      : chr [1:396] "1" "2" "2" "3" ...
 $ Behandlung: Factor w/ 2 levels "a","b": 1 1 2 1 2 1 2 1 2 1 ...
 $ Entfernung: Factor w/ 2 levels "2","5": 1 2 2 1 1 2 2 1 1 2 ...
 $ Feuchte   : num [1:396] 33.5 32.5 34.1 34.1 33.1 ...
 $ Korn      : chr [1:396] "63.6" "41.9" "47.2" "77.6" ...
 $ DGUnkraut : num [1:396] 1 1 0.6 0.3 0.3 0.5 0.3 0.2 0.3 0.3 ...
 $ Ertrag    : num [1:396] 63.6 41.9 47.2 77.6 71.6 53.5 56.7 48.5 51.6 65.5 ...
 $ Datum.y   : Factor w/ 3 levels "2021-04-08","2021-05-17",..: 1 1 1 1 1 1 1 1 1 1 ...
 $ DGKultur  : int [1:396] 17 27 42 7 27 40 38 50 48 65 ...
 $ DGGes     : int [1:396] 17 27 42 7 27 40 38 50 48 65 ...
 $ Hoehe_FA  : int [1:396] 12 13 13 9 12 12 12 14 14 15 ...
 $ IQR       : num [1:396] 18 18 18 18 18 18 18 18 18 18 ...
 $ q25       : Named num [1:396] 51.8 51.8 51.8 51.8 51.8 51.8 51.8 51.8 51.8 51.8 ...
  ..- attr(*, "names")= chr [1:396] "25%" "25%" "25%" "25%" ...
 $ q75       : Named num [1:396] 69.8 69.8 69.8 69.8 69.8 69.8 69.8 69.8 69.8 69.8 ...
  ..- attr(*, "names")= chr [1:396] "75%" "75%" "75%" "75%" ...

enter image description here

Effigy
  • 51
  • 5
  • 1
    It's not the case that pseudo r-square measures are used only for models with nominal or ordinal responses. Potentially, they can be used with a variety of generalized linear models. A true r-square would be applicable only to a general linear model. – Sal Mangiafico Feb 17 '22 at 14:10

0 Answers0