I have been asked to analyse some questionnaire data- which is not data I'm
that used to dealing with. I'm hoping that I can make use of the nabble
expertise (again).
The questionnaire has a section which contains a particular issue and then
questions which are related to this issue (and potentially to each other):
1) importance of the issue (7 ordinal categories from -3 to +3)
2) impact of the impact (7 ordinal categroies from -3 to +3)
3) percentage affected by the issue (11 ordinal categories from 0, 0-10,
20-30, 30-40.....90-100)
I also have three participant predictive factors:
Gender (M/F)
Age (continuous scale)
Sector (6 nominal categories)
So that my data looks like this:
gen age sector impac importa percen
1 1 59 4 0 -3 2
2 2 64 3 2 -3 2
3 1 83 6 3 -3 2
4 1 66 5 2 -2 2
5 1 79 5 0 -1 2
6 2 63 4 0 -1 2
I have 2 questions I want my analysis to answer
1) does gender/ age/ sector affect importance, impact, reponse
2) are importance/impact/response correlated in some way
I'm thinking I need some ordered logistic or probit model (possibly using
polr() command). The problem is the multivariate aspect- I need importance,
impact and perecentage in the same model to look at the covariance between
them and affects of gender, age and sector on these covariances.
It would be good to include a latent variable- at least for the perecentage
factor.
Any help would be very much appreciated.
--
View this message in context:
http://www.nabble.com/Ordinal-response-model-tp25856728p25856728.html
Sent from the R help mailing list archive at Nabble.com.
drlucyasher wrote:> > > The questionnaire has a section which contains a particular issue and then > questions which are related to this issue (and potentially to each other): > 1) importance of the issue (7 ordinal categories from -3 to +3) > 2) impact of the impact (7 ordinal categroies from -3 to +3) > 3) percentage affected by the issue (11 ordinal categories from 0, 0-10, > 20-30, 30-40.....90-100) > > I also have three participant predictive factors: > Gender (M/F) > Age (continuous scale) > Sector (6 nominal categories) > >Gender and Sector are clear; convert these to factors, preferably giving them meaningful names (m/f, east, west), and everything will be treated correctly by most r function. Age is also clear, leave as is. There will be considerably discussion how to code the scores. If these are not heavily skewed (all -3), in some fields it is accepted to treat these as continuous. Frank Harrell would argue against it. I have revised too many manuscripts in both directions, so my opinion depends on the paper where you publish it. Anyway, Frank Harrel's lrm in Design might give you a starter. There is also a well-known book by him on the subject. Dieter -- View this message in context: http://www.nabble.com/Ordinal-response-model-tp25856728p25860439.html Sent from the R help mailing list archive at Nabble.com.