Share Adjustment

After making appropriate adjustments for respondent weighting, product distribution, product awareness, and tuning the steepness or flatness of the scale (via the Exponent), researchers may find that the simulated shares of preference still do not closely match real world market shares.  Many managers take this in stride, recognizing that the choice simulator provides an indication of what their product might achieve under assumptions of long-range equilibrium and equal effectiveness of advertising/sales force effectiveness.  To find that current market share falls significantly short of predicted share of preference represents an opportunity to investigate why the product is not achieving its potential.

It is also possible that simulated shares of preference do not match actual market shares due to other more pernicious issues such as: not all attributes that strongly affect product choice were included in the study, weaknesses in questionnaire wording, etc.  

Sometimes researchers are tasked with making market forecasts, so the conjoint analysis information must be used (often along with other data) to make predictions about the future.  We believe that conjoint analysis data are extremely valuable as part of a more complete market forecasting model.  When the researcher absolutely must make the simulated shares of preference for a base case match actual market shares (and after the other procedures involving properly weighting respondents, adjustments for product distribution, awareness, and scale have been implemented), then something else must be done.  Applying additional adjustments to simply close the gap between simulated shares of preference for a base case and actual market shares for that base case is to admit that we cannot explain the remaining differences.  Closing the gap via adjustment factors critically assumes that the actual market share information you have is accurate (it sometimes is not) and that the adjustments you make to the simulated base case scenario so that the predictions match the market reality will also be appropriate when you modify the specifications of the base case scenario (they may not be).

Utility Adjustment Method

Given the warnings and considerations above, the recommended method for share adjustment is the Utility Adjustment Method.  This method simply adds a global constant utility (same across all respondents) to each product's total utility.

When you go to the My Scenario Settings tab and check Apply Share Adjustment, you can click the icon to edit the settings for this option.  

A table opens with as many rows as products in your market simulation scenario.  You may specify a global utility adjustment for each product in the Change to Utility column.  This adjustment may be any positive or negative value, including many decimal places of precision.

Because it would be a long trial-and-error process to figure out how much share adjustment must be made for each product in the simulation scenario to result in target shares of preference, we have provided a simple tool for automatically finding the utility adjustments and inserting these within this table.

Solve for utility adjustment to match shares

To let the simulator automatically search for the utility adjustments so that the simulated shares of preference match certain target shares, simply specify the target shares in the Target Share column.  The shares you specify must sum to 100.  To specify a market share of 50% means typing "50" (not 0.50).  

Click the Solve button.  The simulator does multiple iterations until the utility adjustments lead to predictions that very closely (but not perfectly) match the target market shares.

Note that under certain simulation methods (especially First Choice), it may be impossible for the simulated shares to match target shares beyond a certain degree of precision.  For example, under the first choice rule and n=50 (with equal respondent weights), shares of preference can be 50% or 52% (25 or 26 out of 50 respondents are predicted to choose the product), even though you may have specified a target share of 51%.  It is not possible with the First Choice simulation rule and n=50 to achieve an odd share of preference value; only even values are possible.

Aggregate Share Adjustment (Legacy Method)

Prior to the 2016 release of Sawtooth Software's Choice Simulator, the only method we supported formally for implementing External Effects was the Aggregate Adjustment method.  This method was shown to be inferior to other methods by Orme and Johnson (2006 Sawtooth Software Conference, "External Effect Adjustments in Conjoint Analysis").  As just one example, applying aggregate adjustments to account for differences in product distribution change the price sensitivities (price elasticities of demand) for the simulated products, when this is not logically correct.  Despite the weaknesses of the aggregate adjustment approach, for historical purposes we have chosen to include this option within this current market simulator.

The Aggregate Adjustment modifies product shares of preference by applying a multiplicative External Effects factor to each product's preference share.  A value of 1 introduces no effect, a value greater than 1 increases a product's preference share above what the model would otherwise predict, a value less than 1 decreases the predicted preference below this value, and a value of 0 eliminates the product from the model.

Setting the factor for each product so that simulated shares match target shares is a simple procedure:

The Solve procedure will perform this adjustment automatically for you if you specify the target shares within the software dialog and click Solve.