A psychology problem :
variable to be explained: a score which is the sum of the answers to 21
questions (so continuous data based on an arbitrary scale)
independent variables : 5 other scores obtained in the same way
current situation for the variable to be explained :
lots of zero, lots of 1 (up to 8) but very few scores between 9 and 21.
aim :
see whether the dependant variable can be explained by the other variables
questions :
what information could be extract for each of the three measurements ?
score (continuous / arbitrary scale) :
1 would the proportional odds regression for ordinal data useful ?
2 what kind of conclusion does this method leads toward to ?
3 is there any other way to approach this variable ?
number of observations per category
1 would the zero inflated poisson model be of any help
2 what information could be extrated looking at the data this way ?
0 or not zero (discrete data with two levels)
1 would the proportional hazard model help in predicting the risk of
having 0 or not 0 ?
2 would it be another way to look at this ?
Could you add limits and advantages of each of the methods you know about
and details about SPSS procedures in your answers ?
Regards,
Véronique Bourcier
http://veronique.bourcier2.free.fr
|