Subscribe Bookmark

Community Manager


Feb 27, 2013

Probability and Multiple Choice Profiler in the Choice Platform

This article appears in JMPer Cable, Issue 30, Summer 2015.

by Melinda Thielbar melinda.thielbar, Senior Research Statistician Developer, SAS

The JMP Profiler is a powerful tool for visualizing your model. With one click, you can see what the model predicts when you change a product’s features or adjust one of your assumptions. It’s also a powerful communication tool since your audience doesn’t need a statistics background to understand the model’s message.

In JMP 11 and earlier versions, the Profiler in the Choice platform was a Utility Profiler — it showed how your product’s utility changed with changes in features. To an economist or a marketer, utility is a pretty straightforward concept. Higher utility means a more desirable product that people are more likely to buy.

JMP 12 adds the Probability Profiler and the Multiple Choice Profiler, two new tools to help visualize comparisons between competing products and predict market share for proposed new products.

Consider the two versions of the pizza data from the JMP sample data library. Let’s start with the multiple table example contained in Pizza, Pizza, and Pizza These three data sets mimic how choice data are often collected. Factors, responses, and demographic information on participants are in separate data tables.

Running the script attached to the Pizza Profiles menu produces a report with Choice model effects and Likelihood ratio tests on the resulting estimates (Figure 1). (Note: JMP performs the likelihood ratio tests by default if they can be calculated quickly. If the tests are not produced by default, you can always request Likelihood Ratio statistics from the red triangle menu.)

Figure 1 Choice model effects and Likelihood ratio tests

The effect tests show which factors are statistically significant, and the signs on the parameter estimates show whether that factor makes subjects more or less likely to buy the product. From this report, we can see that Jack cheese is less popular than Mozzarella, and there is a statistically significant interaction between Gender and preferences for Thick Crust and Pepperoni vs No Toppings.

The Utility Profiler (just called “The Profiler” in JMP 11) can help explain how these interactions affect consumer choices. Select Utility Profiler from the Choice Model red triangle menu to show the Utility Profiler. In the Gender profiler, drag the vertical line from F to M (Figure 2).


Figure 2 Utility Profiler with Gender set to M

The first profiler shows that males prefer Thick Crust pizza with pepperoni.

Changing the Gender from M to F allows us to clearly see the effect of the interactions. The slope of the line between Pepperoni and “No” toppings is negative when we set the value of Gender to “M”. When we change the value of gender to “F”, the slope becomes positive. The same is true for the slope of the line between Thick and Thin crust.


Figure 3 Utility Profiler with Gender set to F

This shows that women are more likely to choose Thin crust pizza with no toppings, and men are more likely to choose Thick crust pizza with pepperoni.

“How much more likely?” and “More likely compared to what?” your marketing manager might ask. Until JMP 12, your answer might have been, “Let me get back to you.” Now all you need is the Probability Profiler.

In JMP 12, the Probability Profiler is right below the Utility Profiler in the Choice platform’s red triangle menu. When you select Probability Profiler, you are comparing your proposed new product to a “Baseline” product. Maybe I’m planning to offer a Thick Crust pizza with Mozzarella cheese and Pepperoni to challenge my competitor’s pizza, which is a Thick Crust with Jack cheese and Pepperoni. From the Gender list in the Baseline settings, select F (Figure 4).


Figure 4 ProbabilityProfiler with Cheese set to Mozzarella

The Probability Profiler makes it easy to see that my proposed product is a good idea. When choosing between my pizza and my competitor’s, females have a 92% chance of choosing my pizza (Figure 4). I can change Gender in the Baseline settings to M to see males have a 96% chance of choosing my pizza (Figure 5). Settings related to subjects (that is, people) always appear in the Baseline settings, so you’re always comparing apples to apples when you compare choice probabilities.


Figure 5 Probability Profiler with Gender set to M

There aren’t many markets where consumers have only two choices. You might know about multiple competitor products and want to design the pizza with maximum choice probability against all of them. That’s a job for the Multiple Choice Profiler, which is in the red triangle menu just below the Probability Profiler.

Selecting Multiple Choice Profiler from the red triangle menu brings up a dialog asking how many choices you want to profile (Figure 6). Three is the default, but you can have more.


Figure 6 Setting the number of choices to profile

Now, instead of having one Baseline model and one alternative, the Multiple Choice Profiler produces a set of linked profilers — one for each alternative (Figure 7). The drop-down lists at the top allow you to choose the values of the subject variables that comprise your market segments. The profile sliders set the properties of the different products. There’s also a chart just below the header to visually show which product would have the highest predicted market share.


Figure 7 MultipleChoice Profiler for all alternatives

Now it’s easy to see that a Thin Crust pizza with Mozzarella cheese and No toppings is the most popular choice.

All of the above examples show an analysis with multiple tables. If you usually perform a one-table analysis, JMP will place your subject effects in the Multiple Choice Profiler (Figure 8). (There is no way for JMP to know whether an effect was intended as a subject effect or a profile effect, so JMP makes the choice that has the most flexibility for you the user.)


Figure 8 Subject effects in the Multiple Choice Profiler

The probability comparisons in the multiple choice model are restricted so that they must sum to one; the probability comparisons from a Choice model only make sense if you are comparing within a market segment. If you change the values of the subject variables, be sure to make them the same for all the choice probabilities. Otherwise, your predicted probabilities (and therefore, market shares) will not be correct.

The Utility Profiler has always been a useful tool for visualizing your model. JMP 12 adds the Probability Profiler and the Multiple Choice Profiler, two new tools to help visualize comparisons between different products and predict market share for proposed new products. 

To explore this topic, see JMP documentation on Choice Platform Options.

Article Tags