# Sampling effort in a logistic regression (glm in R)

I would like to take into account the sampling effort in a logistic regression (in R) to give more or less weight to observations, but I don’t know how to do it. I have read many topics on the “weigths” parameter in glm but this seems not suitable in my case. I have a dataset with presence or absence at different location and the sampling effort vary between locations. I have a raster map that estimate the sampling effort at each location (depending on the number of potential observers in each area) but I don´t have precise information for each site such as the exact number of visit and the result of each visit (presence or absence). I would like to build a species distribution model based on different environmental predictors but taking into account the differences in sampling effort. Is someone have an idea on how to specify this in a glm?

## 2 Answers

Since you want to consider the sampling effort, the svyglm function should be a good option. svyglm computes the robust standard errors which consider the loss of precision introduced by sampling weights.

Before using svyglm, you might specify your survey design with the svydesign function.

• Thanks! I will read the documentation on this function. I had a brief look and it could be useful! But I still need to understand in detail how to specify the survey design. – Lisa42 Oct 2 '17 at 14:38

Can you use the sampling effort in each location as an offset? Just add an offset = term in to your glm formula.

• That doesn't sound like a good idea. – The Laconic Sep 29 '17 at 23:38
• Maybe but I am not sure to understand well in which case we should or not use an offset. I will try to find more information on this, to see if this can be suitable in my case.But thanks for the suggestion! – Lisa42 Oct 2 '17 at 14:51