Generalized linear models. Choose the model

[angaritadeisy16]

Hi everyone, I'm a biology student from Colombia. One year ago I take some phenology data from a tree specie. I want to now what parameters (hight, circumference at breast height and crown width) are related with the number of reproductive structures that three produce. So I tried to do multiple linear regression for each response variable (Botones, Inflorescencias and Infrutescencias) but data didn't have normal distribution. I tried to do Generalized linear models and I chose the Gaussean distribution because the residuals of linear regression looks more or less similar to a normal distribution but I think is wrong because is the same distribution that normal. I don't know a lot about these king of analysis so I require your help. I don't have a lot of gadget to pay consultancies because is a personal project but if you help me I could tech you a little of spanish.

Thanks a lot, I attach the data that I used.

 

[katxt]

but data didn't have normal distribution

I think you are on the right track. It's not the data that needs to be normal but the residuals. Do your multiple regressions and see what is significant. With 40 points the residual don't need to be perfectly normal for the regressions to be valid.

 

[hlsmith]

I would also look to the literature and see what others in this context have used.