Biosecurity, Terrorism, and Food Consumption Behavior: Using Experimental Psychology to Analyze Choices Involving Fear
Just, David R., Wansink, Brian, Turvey, Calum G., Journal of Agricultural and Resource Economics
How would a possible food safety scare influence food consumption? Using techniques from experimental psychology, a study of 103 lunchtime participants suggests that a food scare-avian influenza-would decrease consumption of the affected food by 17% if the subjects believed it was naturally occurring, and by 26% if they believed it was the result of terrorism. While individual consumption decreased, very few eliminated all consumption of the affected food. We argue that experimental psychology is essential when attempting to study behavior in food safety where hypothetical scenarios and surveys would not capture the emotional nature of the response.
Key words: avian influenza, experimental psychology, food safety, terrorism
As food safety has become a headline issue, economists have struggled to find practical ways to assess consumer response to potential future threats to our food system. Food safety incidents, such as the outbreak of salmonella due to contaminated jalapeno peppers or peanut butter, are singular events. Consequently, while studying one outbreak using historic data may reveal some important behavior, it is difficult to generalize any results from such an approach to determine which characteristics of the outbreak drove consumer responses.
Two primary alternative methods are generally used to study food safety threats. First, survey methods allow investigators to ask consumers about hypothetical threats. Investigators can then manipulate the characteristics of the potential threat (e.g., the location of discovery) to more directly assess their influence across different segments of consumers (Turvey, Onyango, and Hallman, 2007). For example, when respondents to a national survey were asked about a hypothetical outbreak of avian influenza (AI) in the United States, their reported perceived risk was acute enough to imply that poultry markets would be devastated for at least several months (Condry et al., 2007). Yet, past research has also shown that consumers can be substantially biased when responding to hypothetical questions (Bradburn, Sudman, and Wansink, 2004). Thus, even with such dramatic survey responses, we are left to wonder "to what extent are these responses real?"
A second approach for studying food safety crises has been economic experiments that have sought to determine a consumer's willingness to pay for food safety information (e.g., Hayes et al., 1995). While this method eliminates hypothetical bias, it cannot be used to assess consumer behavior after a food safety threat has already occurred. Because of the direct risk to participants of consuming contaminated food, an economic experimenter cannot ask participants if they will eat food that has an increased risk of contamination in an incentive-compatible way.
We introduce an alternate method of assessing behavior in response to a food safety risk. Our method, derived from experimental psychology, can be used to directly assess how individuals will alter consumption choices when responding to a food safety incident. A key feature of this method is the use of deception, which is typically forbidden in economic experiments. This decision was not taken lightly. As we argue in the following sections, in situations where personal safety is at risk, deception enables us to study a new class of economic problems and may present substantial benefits to society. We propose some key characteristics of the problem that indicate when deceptive methods may be justified as an alternative to economic experiments. We describe a food choice experiment conducted in the fall of 2007 at Cornell University (Ithaca, New York). The experiment was designed to measure the actual response of participants to a perceived biosecurity threat related to the discovery of bird flu at a local processor. Information about the bird flu outbreak was delivered through the use of a confederate posing as a participant. The weight of food consumed by each participant was recorded and provides our primary measure of risk response. …