Have an idea?

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

Valid response rate, conjoint?

Hi Forum,

Is there an accepted minimum respondent N-count for study validity? Would it be the same as normal sample sizes? Does conjoint have an impact (up/down) on what's considered a valid sample size?

asked Nov 5, 2018 by WTW_cal Bronze (1,595 points)

1 Answer

0 votes
That a study includes a conjoint analysis does not change the rules of sample size - more will still be better and quadrupling sample size will improve your precision by a factor of two.

It is a good idea to test your design to make sure that the standard errors around your estimates are such that you can live with them.  

Also, it's helpful to think of how much precision you want around your simulation shares - in fact, since simulators are often the primary deliverable for a conjoint study, that's probably the most important way  to think about sample size.  Samples of size 100 will have shares accurate to about +/- 10 percentage points, samples of size 400 will have shares accurate to about +/-5 percentage points and so on.
answered Nov 5, 2018 by Keith Chrzan Platinum Sawtooth Software, Inc. (105,750 points)