Have an idea?

Visit Sawtooth Software Feedback to share your ideas on how we can improve our products.

Share of preference calculation for Groups in Maxdiff

We have done a Maxdiff study (Best – Worst response) with 18 statements and after generating utilities and taking exponential of the utilities we have arrived at the share of preference for each of the statements as mentioned below.
Code    Statements    Netting    Share of Preference
1    Statement 1    Group 1    4.37
2    Statement 2    Group 1    5.15
3    Statement 3    Group 1    3.76
4    Statement 4    Group 2    8.61
5    Statement 5    Group 2    7.27
6    Statement 6    Group 2    2.20
7    Statement 7    Group 2    6.02
8    Statement 8    Group 2    5.31
9    Statement 9    Group 3    7.19
10    Statement 10    Group 4    3.37
11    Statement 11    Group 4    3.63
12    Statement 12    Group 4    5.03
13    Statement 13    Group 5    3.29
14    Statement 14    Group 5    4.20
15    Statement 15    Group 6    15.47
16    Statement 16    Group 6    7.19
17    Statement 17    Group 6    1.30
18    Statement 18    Group 6    6.64

Now the 18 statements can be clubbed into 6 groups based on their characteristics and the number of statements varies from one group to another. We need to find out the share of preference for the 6 groups and for that we are simply adding up the statements falling in each group. However, the group having more statements may have more SOP which will better its rank of order. The table below has the number of statements and share of preference for each groups.
Groups    Statements    SOP
Group 1    3    13.28
Group 2    5    29.41
Group 3    1    7.19
Group 4    3    12.03
Group 5    2    7.49
Group 6    4    30.60

Is there any other / better way of calculating the share of preference for the groups nullifying the effect of number of statements in the group? Look forward for your response.
asked Jun 20 by Suman

1 Answer

0 votes
Are the items within a group mutually exclusive?  If so, you might compute the SOP of a group by counting only the highest utility item in the group for a given respondent.
answered Jun 20 by Keith Chrzan Platinum Sawtooth Software, Inc. (111,275 points)
Thanks...But what if in Group 1 there are 3 statements (the highest and 2 lowest); in group 2 there are 3 statements (all in the middle rank). Then if we take Max SOP for groups Group 1 will be my top group even though Group 2 has better 3 statements if compared with Group 1. Is there any way we can take into consideration all the statements and the gap between them?
If items within a group are mutually exclusive then I think it may make sense to look at the maximum utility item as the best representation of the group; if they are not then you could look at the sum (which you find problematic) or maybe the mean - though I don't see a lot of sense in looking at the mean.
Thanks...
...