Have an idea?

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

Would my collected data be deficient or my analysis be biased if i dont include BYO to my ACBC

First of all, I would like to thank to everyone who has helped me before. Truly great support from all practitioners and also the software team! Thank you!

I have two questions and I would be more than glad if you could help me once more.

In my survey design I am using BYO and my ACBC survey independent from each other. In other words, I am in using BYO just as fictional search engine to increase reality and interaction, in which the respondent selects few product features then  directed to my survey that is about product reviews with totally different attributes. Thus i am wondering, whether not including the BYO(together with no screening task, unacceptables) to my ACBC survey design would cause any problem for collection or interpretation of the data that i will collect for my analysis.

Additionally, I have included several conditional displays, since 3 of my attributes are interrelated. I included images to replace these conditional displays instead of one of the attributes of these three (via Position: Replace Attribute). Is there anything that needs to be done before finalizing the survey to avoid future problems for estimation, as I am planning to use Latent and SPSS for my analysis.

asked Dec 3, 2014 by Miran

1 Answer

+1 vote
ACBC indeed has an option to skip the BYO section altogether and just use Screener and/or Tournament tasks.  However, the experimental designer prior to v8.3 was not quite as good as the one we have with 8.3+, so if skipping BYO in ACBC I would recommend using v8.3+.

If you opt to skip the BYO section in ACBC, then the software will not have the ability to create near-neighbor concepts to the respondent's ideal product for evaluation in the Screener and Tournament tasks.  However, the non-adaptive design will be more statistically efficient, since it will have near-perfect level balance within each attribute.  So, you give up respondent relevancy and realism for better statistical efficiency.  It's a tradeoff and I don't know how much better or worse results would be by skipping BYO in ACBC.  I do know that respondents will not find the survey as engaging and relevant if you skip the BYO section.

Regarding analysis, it is possible to export the raw choice data to a .CHO file.  But, I doubt that SPSS or what you call "Latent" (unless you are referring to our standalone Latent Class module) will be able to handle the special characteristics of ACBC's .CHO file.  (But, please be warned that the "None" parameter for ACBC can be a strong driver of segmentation if using Latent Class MNL to estimate the utilities and find the segments simultaneously.  And, that "None" parameter is potentially strongly affected by individual respondent tendencies to be more of a "yea-sayer" or a "nay-sayer" to the Screener questions.)

However, if you are referring to using a latent class procedure (other than Sawtooth Software's) to run segmentation on the individual-level utilities produced by our ACBC/HB routine, then that's reasonable and can be done (we'd recommend you work from normalized, zero-centered diffs utilities if doing this cluster-like procedure).

Also, obviously, if you have used our ACBC/HB procedure to estimate individual-level utilities, you can read those utilities into SPSS for further analysis.
answered Dec 3, 2014 by Bryan Orme Platinum Sawtooth Software, Inc. (171,240 points)