Exponent Settings

Top  Previous  Next


The Exponent is used as a multiplier for the utilities during market simulations.  An Exponent of 1.0 means multiplying the raw utilities by 1.0 prior to conducting the market simulation, meaning no change to the scaling of the utilities. For the Share of Preference and Randomized First Choice simulation methods, entering a value >1 causes the resulting shares of preference to become more extreme (steeper), whereas entering a value between 0 and 1 causes the shares of preference to become less extreme (flatter).  See more details about the Exponent.


Tuning the Exponent is useful when the response error for respondents to the conjoint questionnaire doesn't match the error in judgments for the out-of-sample choice events to be predicted (such as actual market choices).  Often, researchers find that the response error in conjoint questionnaires is too low (resulting in shares of preference that are too extreme) relative to actual market choices, so they find themselves tuning the Exponent typically somewhere in the range of about 0.3 to 0.8. Exponent adjustments below about 0.2 (for conjoint part-worths estimated via logit-based methods) would seem extreme and point to possible problems in the data (either the part-worth utilities or the holdout judgments being used to tune the exponent).


Many researchers tune the Exponent manually, until the desired degree of steepness or flatness of the simulated shares of preference is accomplished.  This dialog also includes a search tool wherein you can specify multiple scenarios (product specifications and target shares of preference) and the simulator will iterate across different Exponent settings until the Exponent that produces shares of preference that most closely match the target shares (according to a sum of squared errors criterion) is found.

Page link: http://www.sawtoothsoftware.com/help/lighthouse-studio/manual/index.html?exponentsettingswindow.html