Have an idea?

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

ACBC must haves and unacceptables not working for one variable (pricing)

We have set up an ACBC where pricing is calculated within the questionnaire based on the level (basic, medium, top) and type (single, couples, family) of health cover a person has. This price is then varied by -20%, -10%, base price, +10% and +20% (so there are 5 levels for every respondent, but the actual price shown depends on level and type of cover they hold so it will be relevant to them. The price is based purely on two questions in the survey that ask about level and type, and the value of the pricing variable (all 5 levels) is calculated on this basis.

Everything seems to be working fine EXCEPT price seems to be working oddly in the must haves and unacceptables.  Two problems:
1. I am consistently choosing only the -20% price, yet only the +20% price is coming up in the unacceptables. Why aren't the -10% price, base price, and +10% price, all of which I am not choosing, coming up here?
2. In the must haves, it is asking if the -20% price is a must have TWICE - once as "At most: price" and then again as "Price". This seems weird - why would it come up twice?

I can't understand why this function isn't working properly, and it is making me nervous about launching the study. Can anyone shed any light on this?
asked Sep 23, 2019 by Sharon Morris (275 points)
edited Sep 23, 2019 by Sharon Morris

1 Answer

0 votes
When we ask unacceptables questions for attributes with Sequence in Preference order, we don't want to give respondents the ability to say that too many levels are unacceptable for that attribute too early on in the questionnaire.  So, if there are 5 price levels and the respondent only says the cheapest level is acceptable in the Screening section to this point, then the first time the unacceptable question appears, it will present the top price level as a potential unacceptable level.  Then, if the respondent says "yes, this is unacceptable" then continues rejecting everything but concepts with the lowest price in the next screener questions, the next time the Unacceptable question probe appears, it will show the next price level beneath the one the respondent previously said was unacceptable, etc.  This is by design and the behavior is set to be this way in the software.

There are two types of probes: unacceptable, and must haves.  These are two separate questions and ACBC can ask about Price attribute twice.  Once as an unacceptable and once as a must-have level.

It sounds like you are doing some customized scripts to "trick" ACBC into doing conditional pricing.  Please think this through and consider how you are going to conduct your analysis.  ACBC will, generically, estimate utilities for the 5 levels of your price attribute.  If you are conducting market simulations, then you'll just need to keep track of what level "1" actually means for price when combined with the other attributes that triggered a conditional price during the questionnaire.   Just remember that in the end ACBC will just estimate part-worth utilities for the 5 generic levels of price.
answered Sep 23, 2019 by Bryan Orme Platinum Sawtooth Software, Inc. (179,015 points)
Thanks for the clarification - very helpful. I'm still unclear though why the  same price is asked twice within the one 'must-have' question : "At most: price" and then again as "Price" - all within the same question. It looks odd.

Regarding the customised pricing, the intent is simply to test the five levels for every respondent - -20%, -10%, base price, +10%, +20%, and all analyses will refer to these constructs. Ideally this is what we would have shown respondents, but we know lots of people struggle with percentages . In addition, we know that a large proportion of private health insurance holders have no idea what they are paying (in Australia, there are two possible types of insurance - hospital (used for hospital admissions) and extras (used for health services outside a hospital such as dental, optical, physiotherapy, etc).  Most people have both these types of insurance, and so while they may know what they pay for both combined, most people wouldn't know how much they are paying for each component, so we needed to present price levels rather than base price because people don't really know what the base price is even though they are paying it.
Hmmm, if I'm understanding your comment, then you are saying that within the SAME must-have question, your price attribute is listed twice.  Once as "at most: price" and once again just as "price".  This is unexpected to me and could you please take a screenshot of this behavior and forward the screen shot together with your .ssi file to support@sawtoothsoftware.com so they can look into the issue?
Hi Bryan,

We seem to have fixed the problem by toggling within ABC Must Haves -->  Question Settings --> Attribute Level Settings --> At least/At most labels from All Attributes to Per attribute (and then specified the price attribute). Not sure why this would make a difference but the problem seems to be fixed.
I have exactly the same problem like Sharon: One level appears twice in the must-haves, one time with "at most" and one time without. The change in Question Settings didn't work. Any other suggestions?
Would you be willing to share your study setup with our technical support group so we can look into this?  It is unexpected that the same level of price appears twice in must-haves.  You can write our technical support group at support@sawtoothsoftware.com.
Hi Bryan,

for me the same problem appeared not with price but with some "normal" attributes with preference order. The "best" level often showed up twice: One time with "at most" and one time without. I've already sent my study to your technical support and they have sent me a custom JavaScript function to change this behaviour. The double appearance is gone now but I have the impression that there are much less must haves now. I've set up the same study with SSIWeb v7 and there the problem didn't occur.
Hi Guys  - I also have this issue in my study - could you please share the code?
also, my must haves dont seem to be working properly...
Hi, I got the following code from Sawtooth support. Hope it is ok to share it here.
"Adding this JavaScript to the Must Haves question header or footer should remove any "at most" or "at least" items if an equivalent item without "at most" / "at least" is an option. Lines 4 and 5 would need to be updated if you change the ACBC settings associated with those lines."
<script>
$(document).ready(function(){
    // Parameters
    var atMostText = 'At most:';
    var atLeastText = 'At least:';
    // Run
    var atMostOrLeast = [];
    var minifiedLabels = {};
    
    var rows = $('.acbc_rules > tbody > tr');
    for (var r = 0; r < rows.length; r++) {
        var label = $(rows).eq(r).find('label').text();
        var minified = label.replace(atMostText, ' ').replace(atLeastText, ' ').replace(/\s+/g, ' ');
        if (label.indexOf(atMostText) != -1 || label.indexOf(atLeastText) != -1) {
            atMostOrLeast.push({
                row: r,
                min: minified
            });
        }
        else {
            minifiedLabels[minified] = true;
        }
    }
    
    atMostOrLeast.forEach(function(at){
        if (minifiedLabels[at.min]) {
            $(rows).eq(at.row).hide();
        }
    });
})
</script>  
thank you!
...