Testing the Effectiveness of Lottery Incentives in Online Experiments
Abstract
This article investigates the effectiveness of lottery incentive schemes for eliciting consumer valuations in large-scale online experiments. We implement a fully incentivized condition within a geographically dispersed sample of consumers in which bids for a criollo steak elicited by a BDM mechanism are realized with certainty and the products are priority-shipped in dry-ice coolers. The fully incentivized condition is compared to Between-Subject Random Incentivized Schemes (BRIS), in which only a fraction of subjects realize their choices. We tested two treatments with a 10% probability framed as a percentage or absolute number of subjects, one treatment with 1% probability, and a purely hypothetical reference condition. The results reveal that BRIS with 10% and 1% payment probabilities are effective in eliciting valuations that are statistically indistinguishable from the fully incentivized scheme. In addition to finding insignificant statistical difference between 10% and 1% and the fully incentivized scheme, all incentivized conditions mitigate hypothetical bias, resulting in lower product valuations than the purely hypothetical condition. We contribute a novel methodological framework for conducting large-scale experiments with geographically diverse and representative subjects, increasing the external validity and producing reliable valuations while significantly reducing financial and logistic constraints.
Andreas C. Drichoutis
Associate Professor of Consumer Behavior
His research interests are focused on decision making across a broad spectrum of topics pertinent to agricultural/food economics. His contributions are notably diversified spanning the area of contingent valuation and experimental auctions methods, choice under risk, inter-temporal decision making and applied demand analysis. Most of his research applies experimental economics methods to answer questions relevant to agricultural economists and decision scientists.