Have an idea?

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

ACBC or CBC with a Conditional Pricing Lookup Table


I have a client who want's to conduct a study which has a moderate number of attributes and an equally modest number of levels for those attributes (approximately 4^5). The survey is going out to a very diverse respondent group which ranges from small domestic to large commercial. There are roughly 10 subsamples for which all the attributes and levels are relevant except one - Price.

While all the attributes and levels are relevant the price variable is likely to range from $50-$100 at the low end to many thousands of dollars at the high end.

In approaching this study what would be the optimal approach? Would it be to use an ACBC with a self assessed base price which could be drawn from earlier in the survey i.e. a question could be asked such as "what is the average yearly cost for your service" which could then be drawn into the choose your own section of the ACBC as a base price then varied by +/- 30%  OR would a more austere approach be to simply have respondents self select into one of the 10 subsamples and simply use a conditional pricing table with prices ranges that are relative for each of the groups?

I note that this study doesn't really need the full functionality of ACBC but could benefit from the base price feature.

Any help with this would be great.  

Many thanks
asked Sep 6, 2017 by Jasha Bowe Bronze (1,745 points)
retagged Sep 6, 2017 by Walter Williams

1 Answer

+1 vote
Best answer
Jasha, don't forget that you can trick CBC software into using a respondent's self-declared earlier price to customize the prices shown.  For example, in a CBC attribute, you can use 5 levels of price:  -30%, -15%, +0%, +15%, +30%.  You can ask respondents prior the CBC questions to provide a price value.  Then, for each of the price level text you can use Sawtooth script to multiply the answer to that previous question by 0.7, 0.85, 1.0, 1.15, or 1.30.  But, analysis proceeds with part-worth (effects-coded) model rather than linear price, so make sure to think through your analysis.

But, ACBC as you mention should work too.  Just think through the analysis and practice with practice data (clear through analysis).
answered Sep 6, 2017 by Bryan Orme Platinum Sawtooth Software, Inc. (162,590 points)
selected Sep 7, 2017 by Jasha Bowe
Thanks so much Bryan,

Brilliant answer as usual. However, while I know the trick for adjusting the base price value in ACBC, I don't know the trick in CBC. Could you point me in the right direction for incorporating the custom price variable in CBC, to trick it as you suggest. I haven't done this before.

Any help would be greatly appreciated!
OK, let's assume that in question Q1 respondents type a number.  This is a numeric question you set up in Lighthouse Studio.

Now, as the text for your price attribute, you can specify these three labels for your three levels of price:

$[% ROUND(Q1*.75,2) %]
$[% ROUND(Q1*1,2) %]
$[% ROUND(Q1*1.25,2) %]

This script first prints a "$" followed by a value equal to 25% less, 0% more, or 25% more than the respondent typed.  Plus, we round it to the nearest 2 decimal places.