Using the integrative model of behavioral prediction and censored quantile regression to explain consumers' revealed preferences for food safety: Evidence from a field experiment in Kenya

Despite having poor hygiene and sanitation, and with rising concerns about food safety from production to handling, traditional food markets remain a significant point of purchase in the developing world. However, it remains unclear how consumers arrive at their valuation of food quality attributes. This study examined the determinants of actual behavior to upgrade to safety-and hygiene-enhanced fresh produce. More specifically, the integrative model of behavioral prediction was adapted with internal factors to identify determinants of real payments elicited with the Becker-DeGroot-Marschak mechanism. A field experiment (n = 147) was conducted with consumers of kale (Brassica oleracea) at a traditional pen-urban market in Nairobi, Kenya. Partial least squares path modeling and censored quantile regression analysis were used to detect different structural relationships between the set of the determinants and real payments along the range of auction prices. The results show that among consumers with low valuations, payments were reduced by salience, but increased by social anxiety and trust. In addition, trust helped to increase payments around the median bids. Among consumers with higher valuations, payments were increased by salience. These findings emphasize the importance of incorporating a non-linear setting when analyzing the influence of salience and social evaluation on auction bids. These asymmetrical structural relationships have important implications for the design and validity of monetary valuations and field auction experiments involving measurement of real payments for food product attributes. (C) 2015 Elsevier Ltd. All rights reserved.