Have an idea?

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

When does the "None" option, become dangerous?

We are hoping to include the "None" option in a CBC survey to help us capture switching costs.  However, we are concerned about a potentially large percentage of people selecting this "None" option.

Is there a point where the "None" option takes away from the strength of the attributes and levels results?  I am sure sample size plays a role in this answer, but is there a general rule of thumb that says don't include "None" if you think it will be higher than XX% of your sample.
asked Nov 18, 2011 by bsharkey (125 points)
Actually, if the "None" category is presented in the standard way (as one of the available options to choose), there is ample evidence that respondents tend to UNDER-use the None choice.  There is considerable "helping" behavior on the part of respondents.  The None usage probably understates by quite a bit the number of real buyers that would not buy the items in the set.  The "Dual-Response" None has been mentioned as a possible solution to that, and many researchers find that the None rate under DR-None is seen to be double to triple the size of the None rate under standard None placement.  Perhaps that puts the DR-None rate more in line with actual non-purchase rates, but that has not been proven yet, and is conjecture.

2 Answers

+2 votes
Indeed, each answer of a standard "none" option gives you less information by which to estimate the utilities of the attribute levels you are interested in.

The Advanced Test functionality within CBC software (later SSI Web versions) lets you simulate ahead of time how many respondents you plan to collect, and the % None that you expect.  You can examine the standard errors (precision) of the attribute level estimates under all sorts of scenarios (sample sizes and expected none usage), so you can assess the impact.

Also, consider the "Dual-Response None" option within later SSI Web versions of CBC, as it allows you to collect a None response, but doesn't lose information in the process (as it forces respondents to first select a non-none concept).
answered Nov 18, 2011 by Bryan Orme Platinum Sawtooth Software, Inc. (176,815 points)
0 votes
From my point of view the "None" cathegogy is one of the most important for the market where the category "switching" is typical . For example
"I will not buy any bacery at this prices - i will buy something else"
In order to control this category - make shure that the prices is "below market" or "in the market".
answered Nov 21, 2011 by anonymous