Rotations Within a Survey: Does JMP have any solutions?

Question

I am curious if JMP has any capabilities in helping both determine what sort of rotation method should be used and what those rotations would look like so that they can be easily programmed into a survey?

For a bit of background, we are executing a usage test for 10 products across 100 respondents but each respondent only sees 5 products. Any one product can be seen against any of the other nine and there are no exclusions.

melinda_thielba · Answer

I am assuming "rotation" means randomizing the questions, either by participant or by the order in which they are shown.

The rotation method does depend on the survey. If you are doing a consumer choice survey, the Choice Designer will generate multiple surveys (Number of surveys field, lower left-hand box). This actually treats each survey as a clocking factor, which lets you not only randomize the order but show a larger number of combinations to your survey pool.

If it isn't a Choice design, you could use formula columns in the data table to program a simple randomization. If it's more complex, the Custom Designer could almost certainly help you find a rotation scheme, but how you'd do it would depend on the details of your problem.

KarenC · Answer

This question get's a "it depends" answer. JMP has random number generators, design of experiment platforms, choice design capabilities, etc. but what might help you depends on the details of your project.   If all of your products are independent then you might just program a random selection of 5 for each subject.  If there are details such that the products fall into groups then maybe you define the groups and the assign respondents to particular groups of products.  The "rotation" method will need to be determined based on the details of the project and depending on those details JMP might have tools to help you.