Monday, March 17, 2008
Update on Adaptive CBC
At the 2007 Sawtooth Software Conference, we presented our latest research into Adaptive CBC. Our current stream of research is a significant departure from earlier Adaptive CBC approaches we
have tried and described at earlier conferences. And, we’re happy to say, this approach seems to work better than the traditional CBC for complex studies involving about five or more attributes.
Importantly, respondents find the interview more engaging, realistic, and focused on levels more relevant to their choices.
Our approach involves first asking respondents to indicate which product they’d most likely purchase. We’ve formatted that phase as a BYO (configurator) task, but it probably could also be done as an indication of
“most likely” or “preferred” levels for attributes. In the second phase of the adaptive interview, we ask respondents to screen product concepts that resemble their “most likely” product.
Respondents indicate whether each concept is a possibility or not. After a few choices, if the respondent seems to be using non-compensatory rules (i.e. “must have” or “unacceptable” levels), we
identify the possible rule and ask the respondent to confirm or skip the rule. This process of observing respondent choices and following up with the opportunity to define decision rules is repeated.
And, of course, if the respondent indicates that a particular level is either required or unacceptable, only products meeting the criteria will be shown throughout the remainder of the interview.
In the final phase, we ask the respondent to compare screened-in products using standard CBC choice tasks. This is a round-robin tournament that identifies an overall winning concept.
To complete an online Adaptive CBC survey yourself, visit:
To this point, we have completed two studies. We have published the findings in the Technical Papers library on our website, in an article entitled:
“A New Approach to Adaptive CBC.”
What are we doing now? We are currently wrapping up a third study, and will be presenting the results at the 2008 A/R/T Forum Conference in June. In this latest research, we are testing
whether the Adaptive interviews can be made significantly shorter without losing much in terms of predictive accuracy. We’ve implemented an improvement to the experimental designs that may
allow us to get away with shorter interviews, so this latest work involves more than just shortening the interview.
We are also in development of a beta version for Adaptive CBC. Because this is such a new approach, it is critical that we obtain more data points and greater experience prior to launching
a commercial v1 product. We’ll announce the beta software when it is available, and will be enlisting your help to further test this promising methodology.
If you would like to tell us about your interest in becoming a beta tester for Adaptive CBC, please email Bryan Orme at bryan@sawtoothsoftware.com.
Tuesday, December 11, 2007
CVA/Web Available
We have completed the upgrade to CVA/Web (the newest component within the SSI Web platform).
CVA is our package for traditional, full-profile conjoint (the original conjoint method from the 1970s). Our CVA
system offers many improvements over the original method. We offer both single card at a time (card-sort) or pairwise
designs (two cards at a time). Plus, we have an excellent experimental designer, based on optimization searches to
maximize D-Efficiency. Add to that the option of the HB estimation add-on and you can see that this isn’t your father’s
conjoint analysis program.
In this era where CBC modeling is king, why use CVA? Small sample sizes submarine some potential CBC studies. CVA’s
results tend to be similar to CBC’s, but you can use it effectively for even the smallest of sample sizes (even to n=1).
When using CVA, we suggest including a few holdout tasks (CBC-like holdouts) so that the scale factor in the market
simulator can be tuned for the sample to approximate choice scaling.
Improvements made in this latest version include:
- Ability to generate and field up to 10 versions (blocks) of the questionnaire (previously, only 1 version of the questionnaire was supported). Fielding multiple versions helps control psychological order/context effects.
- Conditional pricing and conditional graphics.
- Advanced prohibitions: In additional to the standard prohibitions between levels of two attributes, prohibitions between levels of three or more attributes considered simultaneously are also supported. “Null-level” prohibitions have also been implemented.
To speak with someone about the improvements and changes, please call Aaron Hill or John Howell at 360/681-2300, or
contact your Sawtooth Software representative.
Friday, September 14, 2007
SSI Web v6.2 Available!
We’re pleased to announce the availability of SSI Web v6.2, as a free update to all v6 users.
With this update, our capable programmers have greatly enhanced some nuts-and-bolts aspects of
the software that affect most every SSI Web project. We have updated the following three major areas:
- Respondent Passwords
- Quota Control
- Online Administrative Module
We have paid close attention to user requests in making improvements, and we appreciate those of you who
have sent suggestions or reviewed our initial planning documents.
If you are a licensed v6 user, you may download the new update directly from our website (no new userid file is required).
Because v6.2 uses new file formats, you shouldn’t use v6.2 to modify a current v6.0 study in field.
Download SSI Web v6.2
And, of course, if you are using a version prior to v6, call today for upgrade pricing and to talk to our helpful staff.
To read more about this release, visit: http://www.sawtoothsoftware.com/version/ssiweb/ssiweb_history.html.
To speak with someone about the improvements and changes, please call Aaron Hill or John Howell at 360/681-2300, or
contact your Sawtooth Software representative.
Thursday, August 16, 2007
151 Already Registered for Conference, Oct 15-19, Santa Rosa, CA
The early-bird registration cutoff was yesterday, and as of today (Aug 16), 151 people are registered for the
2007 Sawtooth Software Conference. If the pattern of registration from last year's event holds,
there will be about 200 people attending. Registration is still open, but at the regular rate of $1,000
for the 2.5-day conference. Optional tutorials are extra.
Although the unexpected turnout seems like universally good news, one problem is that the room block we had reserved for
the hotel at the discounted rate has already been filled. We are working with the hotel to see if additional rooms
can be opened at the discounted rate for attendees. So, even if you initially discover that the room block is closed, please
write danell@sawtoothsoftware.com to determine if additional rooms will become available at the discounted
rate.
Review the program and register today!
Thursday, July 12, 2007
Conference Early Bird Deadline Aug 15
The early-bird deadline for registration for the 2007 Sawtooth Software conference is on August 15.
If payment is not received by then, the cost increases from $850 for the basic registration to $1,000.
The Sawtooth Software Conference comes around just every 18 months. We had 170 people at the last
conference. This conference brings together academics and practitioners from a variety of market
research consultancies and end-user companies. About 25 papers will be presented by folks like these,
and optional 4-hour tutorials for more in-depth learning are also available.
Review the program and register today!
Tuesday, July 17, 2007
Latent Class v4 Released!
Recent surveys of our customers reveal that the majority of our CBC users employ HB estimation to develop their final market simulation models. Indeed, HB is a powerful tool, significantly reducing IIA (red-bus/blue-bus problems) and improving the accuracy of market simulations. A related technique, Latent Class, is commonly used for finding segments of respondents with similar preferences and estimating logit-based utilities for each segment. Latent Class shares much in common with the notion of cluster analysis, though it is a more statistically sound method of finding segments from choice data than clustering on HB utilities.
The good news is that in the next few weeks, we will be releasing a new version of Latent Class (v4), which borrows the same powerful user interface as CBC/HB v4. So, current CBC/HB v4 users will feel at home using the Latent Class system. In addition to the improved interface, the new v4 system will offer two new capabilities: ability to estimate models for constant-sum and dual-response none CBC studies.
We feel that Latent Class has an important role in choice modeling, complementary (rather than replacing) HB. In general, we recommend using HB utilities in market simulations. However, much value can be had in using Latent Class to find needs-based segments, and in merging the segment membership information into SMRT for use as banner points (respondent filters), while retaining the underlying utility HB run during simulations.
Because Latent Class v4 uses the same underlying interface programming as CBC/HB v4, we are offering it at a significant discount for CBC/HB v4 users for a limited time (until October 31, 2007). The standard price is $1,000, but CBC/HB v4 users may purchase it for $600 (first user, $180 each additional user) during this promotional period.
For current Latent Class v3 users, the upgrade to Latent Class v4 is free. Those who wish to upgrade may write or call us, and we'll prepare your user id and download link.
|
Technical Papers
About Us