Dear list members,
I have a large set of variables, which are a mixture of quantitative and
categorical variables and would like to be able to combine them in some
way. The variables can be grouped under certain headings and I am
thinking of creating separate indices for separate categories.
The methods I have considered so far are standardising each of the
variables and summing them all together to form a composite score. I have
also carried out a principal components and factor analysis to reduce the
number of correlated variables.
For those variables where there is small 'unreliable' data with large
standard errors I have used the shrinkage technique to move the
'unreliable' scores towards the mean of all observations.
I have looked at some indices that already exist and have seen that some
of these apply different weights or transformations to the various
indicators making up the composite score. Are these weights applied based
purely on expert advice (i.e. prior knowledge of how important variables
are and how much they should contribute to the score) or should weights
take into account the quality of the data (i.e. missing data)?
Could anybody advise me or provide me with some good references please?
Thank you in advance,
Sarah
Sarah Jones
Statistician
Commission for Health Improvement
10th Floor Finsbury Tower
103-105 Bunhill Row
London EC1Y 8TG
Telephone: 020 7448 0815
*******************************************************************************
The information contained in this message is intended for the addressee
only. If you have received this message in error you must not read use or
disclose the e-mail contents; you should notify us immediately (by
clicking "Reply") and delete this e-mail. The unauthorised use,
disclosure, copying or alteration of this message is strictly forbidden.
Any views expressed in this message are those of the individual sender,
except where the sender specifically states them to be the views of the
Commission for Health Improvement.
Please check for viruses before accessing attachments. Although we
endeavour to keep files clean we can take no responsibility for any damage
caused by contagion.
******************************************************************************
|