How to interpret GLM coefficient (binomial, continue value)

Question

I have difficulty to understand how to interpret my coefficient in R with a GLM

My formula is :

glm(IND ~(TEI+TDF+SAB) , family=binomial, data=CN, weights=N_1)

and my coefficient are :

Coefficients:
          Estimate Std. Error z value Pr(>|z|)    
(Intercept) -12.957596   4.740681  -2.733 0.006271 ** 
TEI          0.087508   0.023516   3.721 0.000198 ***
TDF         -0.004049   0.009586  -0.422 0.672782    
SAB_         0.049415   0.015133   3.265 0.001093 **   
---

So, i want to know for a increase (or decrease) of 1 x unit ( TEI, TDF and SAB), what is the increase for y (IND) ?

TEI, TDF and SAB are continued value

Since the link function is not linear, you can't answer that question. It depends on the starting point. — Roland, Jan 24 '19 at 13:54
https://stackoverflow.com/questions/41384075/r-calculate-and-interpret-odds-ratio-in-logistic-regression — G. Grothendieck, Jan 24 '19 at 14:02

score 1 · Answer 1 · answered Jan 24 '19 at 13:59

You are looking at the wrong thing - your coefficients will tell you the log odds of the outcome being Y=1, so for every unit increase in one of those variables holding all others constant - you would expect on average the log odds to increase/decrease your probability of being classified as Y= 1 - please provide some data for a reproducible example

score 0 · Answer 2 · answered Jan 24 '19 at 14:21

You can't look at the increase in IND - it's a dichotomous variable (or should be).

The most intuitive way to interpret logistic regression is to look at the odds ratios, not the parameter estimates you have shown. The odds ratios show how the odds of the dependent variable change with a unit increase in the independent variable. You can get R to show you the ORs, or you can calculate them yourself: They are $e^{\beta}$ where $\beta$ is your parameter estimate.

How to interpret GLM coefficient (binomial, continue value)

2 Answers2

Linked