Have an idea?

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

Random selection of questions


I am trying to measure the similarity between pairs of pairs of products. In total I have 7 products (A, B, C, D , E, F, and G). This means that there are 7*6/2=21 pairs of products. Thus in total there are 21*20/2=210 possible pairs of pairs of products.

I want respondents to answer the following question: Which pair of products are more similar to one another? (select one)

I am going to analyze the similarity data with mutidimensional scaling, and I want data on the full design.

Since I don't want to burden respondents with 210 questions I would like them to answer a subset of the 210 questions (say 15). Is there any way to show respondents a subset of questions? And how does one randomize these? I want to show 1 question per page.

I'm using version 8.0.0. If you require additional info just ask.
Thanks in advance,
asked Aug 22, 2012 by Sander (360 points)
retagged Sep 13, 2012 by Walter Williams

1 Answer

0 votes
If you wanted to randomly pull 15 of 210, you could use the randomized blocks functionality.  In this case each question would just be it's own block, and then you would set up a constructed list that adds 210 items from a parent list, randomizes, and then use ListMax(15) to cut the list down to 15 item.  That would control which blocks each respondent sees.

Another potential way of doing this would be with MaxDiff.  Generate a design from 210 items, but change the settings to only show 15 questions.  By generating hundreds of versions you effectively get a random grouping of questions for each person.  It might have the added benefit of generating aggregate item scores if you had a really big sample size and Latent Class.
answered Aug 22, 2012 by Brian McEwan Gold Sawtooth Software, Inc. (38,015 points)
is this possible with lists within lists