3
$\begingroup$

I ran an orthogonal design for my conjoint plan and ended up with 16 cards to test. However, not all combinations make sense in the real world. I'm afraid that if I exclude the cards that are unrealistic from the survey and data analysis, then the conjoint utility scores won't be robust. On the other hand, if I do include these unrealistic combinations in the utility calculation, then the results might not be realistic. Unfortunately, I do not have access to software like Sawtooth that accounts for Vetoes in data, and I am limited to SPSS. So how do you deal with Vetoes (unrealistic combinations) in conjoint when working with SPSS?

$\endgroup$
  • $\begingroup$ You might run the generating of cards again and again and select the collection which has the least number of unrealistic cards. Also, nobody forces you to present all cards to respondents: you might leave the response for the unrealistic card missing and then do some form of imputation. Of course, this may have sense if there are few unrealistic cards. $\endgroup$ – ttnphns Mar 6 '15 at 7:43
1
$\begingroup$

Dropping the unrealistic combinations may appear to spoil a nice, neat computational system. But that has to take a back seat. You do not want to waste your survey respondents' time or risk squandering their good will, and thus their willingness to continue participating (or participating genuinely), by presenting them with scenarios that may seem bizarre, off-putting, or overly confusing.

Moreover, conjoint results such as utility scores are not sacrosanct; often they need to be converted into more user-friendly results anyway, if your audiences are like most. For example, instead of utility scores you might ultimately present rankings.

If necessary, you may even want to forego conjoint analysis and design the survey items in such a way that you can analyze them using, say, ANOVA (to disentangle the role played by different factors). It may even be sufficient for you simply to report descriptive results for each combination or each factor level.

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ Classic conjoint in SPSS is main-effects ANOVA. $\endgroup$ – ttnphns Mar 6 '15 at 7:31

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.