|
Hi
all,
I am trying to
analyse my data using an ordinal logistic regression. More specifically, I am
investigating the degree of infestation of a parasitic weed using a
scale 1 to 6 (1 minimum infestation, 6 heavy infestation) within four
geographical regions and 2-6 sub-regions within each region. Different
environmental variables (predictors) have been measured such as humidity,
soil pH, soil type, nitrogen content, temperature degree days, precipitation
etc. as well as some phenological data regarding certain characteristics of
the parasitic weed under investigation (such as height, number of
flowers, length of inflorescence etc).
The most of
these data are continues while some others are discrete (number
of flowers) or even qualitative (no numerical, such as soil
type). How can I incorporate these kind of data into the model ?
Does it makes any
difference if the number or sub-regions within each region is unbalanced (in
region 1, for example, the number of sub-regions is 4 whereas in region 3
the number is 2).
If the D or Chi
square value given by the statistical package I am using regarding the model is
greater than the critical value from the corresponding Chi-square
distribution table with the same df, does this mean that my data do not fit the
model adequately? If yes, how can I overcome this problem ? Do I have to
consider other techniques of analysis?
Thanks a
lot
Nicholas
Korres
|
Title: Message
