Maybe try a proportional odds model on R ( Library MASS, polr function) or nomial logistic regression ( i think the R code is multinom)
I was wondering if some one could give me advise on the appropriate multivariate statistical test to use and if they have or know of a suitable website with a worked example using R to carry out the test (including test diagnostics).
The outcome variable is data (count or a proportion) on lameness collected from a sample (generally 50 animals) from multiple consignments. In most cases the numbers of lameness is zero or low (1-5).
I have about 6 factors I want to investigate some are categorical and some continuous.
I had thought of two possible methods:
- Multilinear regression with the proportion of lameness log transformed
- Using a general linear model; family Poisson (however, sample size would not be taken into account ie in some cases less than 50 and in other cases more).
Would anyone be able to help on these two matters:
The appropriate statistical test and a worked example using R to carry out the analysis.