Dear ACBC user,

A bug has been discovered in ACBC (Adaptive CBC) that could potentially affect proper utility calculations. The bug involves the "price adjustments" feature (when you click Price Adjustments... from the Pricing tab). Note that this is not referring to standard level-based prices for summed pricing, where you specify price levels associated with different levels of your attributes; this involves more advanced price adjustments based on the joint occurrence of specific levels between multiple attributes.

If the following combination of settings occurred in your study, your part-worth utility calculations were not done quite properly:

1. Price adjustments in place for summed pricing (an advanced option beyond just the standard level-based prices).

2. For the attribute reflecting the price adjustment, blank fields rather than explicit zeros provided in the price fields on the pricing tab.

3. Piecewise function or log-linear function for price. If linear price function used, there is no difficulty for proper utility calculation due to this bug.

If the first two elements above occurred for your study, you should also be aware that the price shown in the first calibration concept (a repeat of the BYO concept) would have been computed incorrectly. But, the incorrect price shown to respondent for that first calibration concept would also be recorded in the data file, so the calibration data would be internally consistent and should be good for the purposes of computing the new None threshold.

We're sorry for this error, and hope it didn't occur for any of your studies. We expect it to be a rather rare case, as few studies actually involve "price adjustments" and the default price function estimation (linear) is unaffected. If you have questions regarding your data sets, or how to remedy the problem, please call us.

We are working on a fix to the problem, and will notify you soon when that is available. A workaround is to use zeros instead of blanks in the pricing tab. The presentation of the questionnaire will be the same to the respondent whether using zeros or blanks, but the presence of zeros instead of blanks in the study file avoids the bug.


Sawtooth Software