Hello, I'm a student working on a project in which I'm testing the environmental variables that affect the presence and/or density of newts. I've been fishing in different ponds using landing nets. The number of dips is different between ponds, either because of inaccessibility or because the pond is really small. So I have to take that into account. I have two choices: either I transform my data into density (nb/sampled area) and so I do a Gamma type model (but I have a lot of 0's so a more suitable one), or I keep my count data and do a negative binomial model (because overdispersion) with offset (sampled area). I can't figure out what each of these involves and which is more appropriate, as my results differ between the two methods. Can you help me? Thank you for your kindness.
Keep the data untransformed and use a negative binomial model. The results are more transparent that way. Don't use different analyses since they will always produce different results. Just pick one and stick to it :)
Super thank you very much that's what I thought was most right. So the offset(sampled area) tool works well in my situation as I've never used this function before.
- Badges
- Science topic
- Similar topics
- Engineering
- Materials Engineering
- Powders