I'm working with ACA and am testing out the survey but I'm noticing that in my paired questions the bests are always juxtaposed against the worst. Is this right?
It may be right, and it may not be right. There are three things you'll need to do to make sure you're testing properly.
- Make sure you are testing your survey and not previewing it. Only when you test the survey will it be able to pull one of the surveys created from the design algorithm. Otherwise, when you preview, it is randomly pulling in attributes and may produce the undesirable combinations you are experiencing.
- Ensure that your a priori relationships are set in the proper order. If your levels are (1) $50,000, (2) $75,000, and (3) $100,000, but you've defined them as best-to-worst (rather than worst to best), it may produce the undesirable combinations you are experiencing.
- Make sure that when you are testing the survey, you aren't just randomly picking answers. Adaptive Conjoint Analysis means that it is taking your answers into consideration when creating other questions, so if you start to randomly answer questions, it will mess up the behind-the-scenes logic and produce the undesirable combinations you are experiencing.