Dear all
Can anyone help?
I am about to conduct a conjoint analysis.
n individuals each give 8 preference ratings i.e. one rating is given to
each of 8 combinations. One of the combinations could be
"circle your choice of either 0,1,.....10 to the following:
I should give a high priority to A
I should give a high priority to B
I should give a low priority to C
I should give a low priority to D
Choice 0 1 2 3 4 5 6 7 8 9 10 "
Conjoint analysis allows 'Part worth' values to be derived for each of
the attributes. We can thus work out 'relative' importance of each of
the attributes.
Books seem to deal with the case of just one person's ratings and do
not mention the topic of standardisation. However, here we have n
individuals. I have been reading some papers which deal with n cases and
they mention standardising. However, they are not really clear as to
whether they mean:
1) 'standardise each person's original ratings (i.e. 8 values per person
where each value can be within the range 0 to 10) prior to calculating
the part worth values.
OR
2) 'standardise the part worth values for each person'
OR
3) something else!
One paper (Moore, 1980) seems to imply it is 1) above.
Does anyone have any views on this?
Many thanks,
Kim .
|