A behavioral economic demand model modification to improve versatility

Abstract

Aims: Behavioral economic demand curves quantify the relationship between the price of a commodity and the consumption of that commodity and are often used to measure how an individual or group values a commodity, including drugs of abuse. The exponential model of demand proposed by Hursh and Silberberg (2008) quantifies this relationship and provides indices of demand elasticity (sensitivity to price) and demand intensity (amount of consumption unrestricted by price). One limitation of this model, however, is that it relies on consumption measured in logarithmic units, which prevents consumption values of zero units from being incorporated into analyses. Replacing zeroswithnonzerovalues also presents problems, with the specific value chosen often affecting the statistical conclusions. Methods: Here, we present a modification of this formula that allows the valid inclusion of unaltered consumption values of zero units. We compared the two functions on a large dataset of hypothetical drug purchase task data (n=1215) collected online. Results: Our model modification provides measures of demand intensity and elasticity that are highly correlated with and directly comparable to themeasures produced by the Hursh and Silberberg (2008) equation, but was able to fit 100% of the individual subject data series instead of just 69%. Conclusions: Our model provides a more flexible approach to analyzing demand data that can accommodate more datasets and provides estimates of demand intensity andelasticity that arebackward compatible with the popular Hursh and Silberberg (2008) model. Details of the model modification, curve fit comparisons, and discussion of advantages and disadvantages will be presented. Financial support: NIH/NCI grant U19 CA157345 and institutional funds to WKB.