# what statistical test should i use for my count data?

i want to test the number of varroa mites in individual apiaries is higher in the south of england vs the north I'm working with count data (mites) and catergorical data (location) i used a bionominal glm however the output produces a EXTREMELY high over dispersion (around 1600) is there any way to handle this? or should i be using a different statistical test? used this code glm(formula = Varroa ~ location, family = poisson) this is the subsequent output the two locations are south and north Coefficients:

(Intercept) locationsouth 7.73165 0.09428 Degrees of Freedom: 1999 Total (i.e. Null); 1998 Residual Null Deviance: 3387000 Residual Deviance: 3380000 AIC: 3394000 –

• You may need to give us more detail but surely Poisson regression would be more appropriate for a count outcome? Dec 29, 2017 at 13:46
• Location is rarely modeled as categorical, due to the likelihood that relationships among location matter because they reflect a huge constellation of potentially relevant, but unmeasured, variables that are spatially correlated. Unless you have extremely few locations, look into models that accommodate spatial relationships in some manner.
– whuber
Dec 29, 2017 at 14:00
• definitely would like more detail, both on the data and on the analysis you ran. As @mdewey says, you probably want a Poisson GLM (or negative binomial, glm.nb from the MASS package) Dec 29, 2017 at 14:09
• hi so i used this code glm(formula = Varroa ~ location, family = poisson) this is the subsequent output the two locations are south and north Coefficients: (Intercept) locationsouth 7.73165 0.09428 Degrees of Freedom: 1999 Total (i.e. Null); 1998 Residual Null Deviance: 3387000 Residual Deviance: 3380000 AIC: 3394000
– ella
Dec 30, 2017 at 14:19
• You might find it useful to read the vignette for the pscl package cran.r-project.org/web/packages/pscl/vignettes/countreg.pdf which has an extensive set of examples comparing different models. This would be especially helpful if you have an excess of zeroes. Dec 31, 2017 at 14:18