Experimental Methods in Consumer Preference Studies

Controlled experimental auctions can be used to elicit preferences for food products. We describe results from two series of experiments in which subjects revealed their willingness-to-pay for safer food. In one series, the risk reduction technology was not specified; in the other, it was identified as food irradiation. The results provide some evidence on the acceptability of food irradiation as a risk reduction technology. Disciplines Agribusiness | Agricultural and Resource Economics | Agricultural Economics | Economics Comments This article is from Journal of Food Distribution Research, 27 ( July 1996): 1–7. This article is available at Iowa State University Digital Repository: http://lib.dr.iastate.edu/econ_las_pubs/8 Experimental Methods in Consumer Preference Studies John A. Fox, Dermot J. Hayes, Jason F. Shogren, and James B. Kliebenstein Controlled experimental auctions can be used to elicit preferences for food products. We describe results from two series of experiments in which subjects revealed their willingness-to-pay for safer food. In one series, the risk reduction technology was not specified; in the other, it was identified as food irradiation. The results provide some evidence on the acceptability of food irradiation as a risk reduction technology. Recently, considerable attention has been requiring acceptance or rejection of the reduced given to the issue of food safety (Caswell, 1995). risk state at a given cost. Open ended questions The publicity given to outbreaks of foodborne produce more data and the data are easier to indisease causing death and serious illness has foterpret. Discrete choice questions correspond cused the attention of policymakers on means of more closely to real world situations since, for the improving food safety, particularly for meats. majority of food purchases, the decision is either Naturally, such improvements will come at a cost, to buy or not buy at the posted price. a cost that will ultimately be borne by the conBut regardless of how well a survey is desumer. An interesting question, both from a pubsigned and executed, people still know they are lic policy standpoint and for the food industry, is valuing a hypothetical scenario. The absence of the extent to which consumers perceive that market discipline, applied in the real world by benefits from safety improvements outweigh their budget constraints and the availability of substicost. tutes, creates an environment conducive to quesThe methods for estimating consumer tionable responses. Values from contingent benefits from reductions in health risk can be valuation surveys have exhibited inconsistencies broadly categorized as direct or indirect. Indirect such as a lack of responsiveness to the scale and methods use market data on costs of illness inscope of proposed benefits [see recent papers by cluding medical costs and lost wages to derive Diamond and Hausmann (1994), and Hanemann values for reduced incidence of adverse health (1994) for a discussion of the pros and cons of outcomes (Roberts, 1989; Buzby et al. 1995). Dicontingent valuation]. rect methods elicit respondents' subjective values Recently, economists have developed exfor reductions of a specified risk and thus allow perimental methods which can serve as a useful for additional factors such as the value of forecomplement to surveys (Hayes et al., 1995; Shin gone leisure time, avoidance of pain and discomet al., 1994; Shogren et al., 1994). Experimental fort, and peace of mind. auction markets use real money and real goods to Direct elicitation procedures usually employ create a scenario in which the participants give contingent valuation type surveys in which a sceundivided attention to the valuation task. While nario involving choice between two risk states is the laboratory situation is admittedly artificial, it presented to respondents. The actual value eliciis certainly no more so than the typical scenario tation question can be either open-ended (how presented in a survey. The experimental market much would you pay?) or a discrete choice format should, in fact, have advantages over surveys because it involves real monetary payments and a situation wherein the respondent chooses beFox is an Assistant Professor, Dept. of Agricultural Ecotween, and subsequently consumes, one or annomics, Kansas State University, Manhattan, KS 66506; other food product (typical or new/improved). Hayes is a Professor, Dept. of Economics, Iowa State UniThe choice of an appropriate auction mechanism versity, Ames, IA 50010; Shogren is Stroock Distinguished Professor of Natural Resource Conservation and Manage(where the high bidder pays the 2nd highest bid ment, Dept. of Economics and Finance, University of Wyoprice) can create additional incentives for particiming, Laramie, WY 82071; Kliebenstein is a Professor, Dept. of Economics, Iowa State University, Ames, IA 50010. 2 July 1996 Journal of Food Distribution Research pants to reveal actual willingness to pay for the native methods of improving food safety. In one option to consume the improved product. series, participants can upgrade from a "typical" An additional advantage of the laboratory food product (i.e. typical risk of illness) to a experiment is the option of having several rounds product described as having been screened for of bidding for the same product. This process pathogens and whose consumption offers a concreates an opportunity to inject additional inforsiderably lower risk of producing the associated mation about the product being valued and to illness. The screening technology is not explicitly measure the effect of that information on the identified and thus could represent a quality convaluation by participants. When each round of trol process such as Hazard Analysis and Critical bidding has an equal probability of being the Control Points (HACCP). In the second series, the binding round, incentives to reveal true values are safety enhancing technology is identified as food preserved and wealth effects (an effect associated irradiation and a detailed description of the food with being the winner in a previous round) are irradiation process is provided. eliminated. The samples are limited and no claim is Experimental methods, however, do have made that results can be extended to the general considerable limitations. Variable cost per parpopulation. However, a comparison of bidding ticipant runs between $30 and $60, approximately behavior in the two series of experiments gives double that of a survey. A significant time some indication as to the relative acceptability of commitment approximately two hours, is refood irradiation as a means of enhancing safety. quired of subjects, necessitating some level of The second section of this paper describes financial compensation to reduce sample selecthe laboratory experiment in more detail. The tion effects related to opportunity cost of time. third section presents the results of the bidding The nature of the experiment also imposes geoprocess in both series of experiments. graphic restrictions on sample selection, a restriction not generally faced by mail or telephone surExperimental Procedures veys. Compared to surveys, however, the effects of non-response bias can be minimized by providAt the beginning of each experiment, subing a vague description of the experiment at the jects were given an I.D. number, assigned to a time of recruiting. Higher costs and the restricseat and asked not to communicate with other tions on samples suggest that experimental marparticipants. Participation fees ($15-$20 for stukets can best be used as a complement to other dents; $25-$30 for adults) were paid in cash at the survey methods. Used in this manner, experibeginning of the experiment. Participants were ments can allow for calibration of survey values then asked to sign a consent form and to complete with a somewhat more reliable baseline, a short questionnaire dealing with knowledge of This paper describes two series of experifood safety issues, demographic and socioments investigating consumer values for altereconomic characteristics (see Table 1). Table 1. Demographic Characteristics of Participants. -----Food Safety Experiments--------Irradiation Experiments----Bidding for Bidding for E.coli Salmonella Irradiated Non-Irradiated