I have data which has one categorical variable and the response is a count,
the setup of the experiment is as such that i have 4 different categorical responses, which are in this case different types of field margins.
I would like to quantify the differences in numbers of bees (count data) that visited these field margins.
I have three locations where in each location the 4 field margins (explanatory categorical variable) are next to each other. Unfortunatly in one of those three locations two field margins did not bloom and so i could not count them.
I figured out i should run a generalized linear model with poisson distribution, i tried already doing this by adding my data up for the same sort of field margin and then provide the glm,poisson with an offset for the different margins. But as such, i come out with a model that completely fits the data, which is off course so because the glm has been fed only one count then per margin.
So i thought maybe i should give in my independant counts to the glm...
but how do i do this? can i just feed it this:
then run a glm? or how should i approach such? when i do such, R gives me an output by which i am satisfied, but is it the correct manner to approach my problem? The first count accords to the same location as the first count of the following type...
glm(formula = count ~ mengsel, family = poisson, data = countsol11)
Min 1Q Median 3Q Max
-2.1261 -1.5590 0.0000 0.4257 3.4086
Estimate Std. Error z value Pr(>|z|)
(Intercept) 3.3081 0.1104 29.957 < 2e-16 ***
mengselB -1.1109 0.2219 -5.007 5.54e-07 ***
mengselC -1.6987 0.3350 -5.071 3.95e-07 ***
mengselD -1.4363 0.2985 -4.811 1.50e-06 ***
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
(Dispersion parameter for poisson family taken to be 1)
Null deviance: 87.207 on 9 degrees of freedom
Residual deviance: 26.349 on 6 degrees of freedom
Number of Fisher Scoring iterations: 4
Advertise on Talk Stats