# Interpretating poisson regression output correctly to wirte in results from R

#### Francesca.123

##### New Member
Hi. I've done a poisson regression on count data in r. I need to write up the results for it but have very little idea where to start because the p value shown are linked to the other variables.
I've searched online and thought I was getting somewhere but I've seen so many contradictory answers I'm just getting confused. I found that many of the sites explained how to manipulate the data this way and that but wouldn't explain what was the useful bit you want when writing up and explaining the results.

My code is
summary(glm(Number.fo.fish~Julian.date*size.class, na.action=na.omit, poisson)

Deviance Residuals:
Min 1Q Median 3Q Max
-3.7713 -1.1880 -0.0631 -0.0292 10.9927
Coefficients:
Estimate Std. Error z value Pr(>|z|)
(Intercept) -0.0315824 0.4773666 -0.066 0.947
Julian.Date 0.0003701 0.0022630 0.164 0.870
Size.Classmedium 0.1224086 0.6034808 0.203 0.839
Size.Classsmall 3.6317021 0.4924481 7.375 1.65e-13 ***
Julian.Date:Size.Classmedium -0.0005007 0.0029317 -0.171 0.864
Julian.Date:Size.Classsmall -0.0156465 0.0024656 -6.346 2.21e-10 ***
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
(Dispersion parameter for poisson family taken to be 1)
Null deviance: 2196.0 on 355 degrees of freedom
Residual deviance: 1368.1 on 350 degrees of freedom
(298 observations deleted due to missingness)
AIC: 2211.1
Number of Fisher Scoring iterations: 6

I have several of these models which are similar to each other but change the variable ie (Julian date to temperature)

Then I have a model with all the variables (except for the categorical variables bar one). These two different models are meant to interact somehow and tell me which variable is significant. But as I've said already I'm completely clueless on this and how I would even interpret that in my results.

> summary(glm(Number.of.Fish~Temperature+Water.Level+Water.velocity+Julian.Date*Size.Class,na.action=na.omit,poisson))

Call:
glm(formula = Number.of.Fish ~ Temperature + Water.Level + Water.velocity +
Julian.Date * Size.Class, family = poisson, na.action = na.omit)

Deviance Residuals:
Min 1Q Median 3Q Max
-5.3244 -1.6002 -0.4139 0.1217 10.6913

Coefficients:
Estimate Std. Error z value Pr(>|z|)
(Intercept) -2.227250 1.707518 -1.304 0.1921
Temperature -0.006308 0.035440 -0.178 0.8587
Water.Level 0.654394 0.055940 11.698 < 2e-16 ***
Water.velocity 0.450130 0.059297 7.591 3.17e-14 ***
Julian.Date -0.000537 0.010647 -0.050 0.9598
Size.Classmedium -0.382517 1.826293 -0.209 0.8341
Size.Classsmall 3.198100 1.702639 1.878 0.0603 .
Julian.Date:Size.Classmedium 0.003230 0.011322 0.285 0.7754
Julian.Date:Size.Classsmall -0.013923 0.010566 -1.318 0.1876
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

(Dispersion parameter for poisson family taken to be 1)

Null deviance: 1827.6 on 207 degrees of freedom
Residual deviance: 1124.3 on 199 degrees of freedom
(446 observations deleted due to missingness)
AIC: 1669.3

Number of Fisher Scoring iterations: 6

I would really appreciate if anyone could help me with this.