NoSeeds

"The bitter taste of HumaN health and physical well-being - the paradOxon of SEEDless table grapeS"

Interdisciplinary research project in cooperation with Prof. Dr. Simone Graeff-Hönninger, Dr. Nikolaus Merkt and Dr. Gebhard Bufler (all University of Hohenheim) financed by the TransFak fund of the University of Hohenheim (funding code BA 520071), 2013-2014.

Within this project, consumer preferences regarding visible and invisible grape characteristics in monetary units were determined. The Attribute-based Choice Modeling (ABCM) method was applied for the monetary evaluation of individual product attributes.

For this purpose, a questionnaire was created in which respondents had to choose between different fictitious grape varieties. The fictitious grape varieties were each identified by six characteristics, which in turn were divided into different attribute levels: skin thickness (thin, thick), fruit size (small, medium, large), number of seeds (0,1,2,4), resveratrol content (zero, low, medium, high), number of treatments with pesticides per year (1,3,5) and price per kilogram (between 1,99€/kg and 6,99€/kg in one-Euro steps). The 36 choice sets and 72 fictitious grape varieties resulting from the survey format were created with the software NGene in a way such that they meet the standardized requirements for an efficient survey design (minimization of standard errors in model estimation).

Stated preference methods, such as ABCM experiments and contingent valuation studies, necessarily require interviewees to make a choice in a hypothetical situation. This procedure is in danger of producing biased results, if respondents do not behave as they are supposed to behave. A possible threat to the validity of the results of choice experiments is that interviewees might not consider the trade-offs between the different options in the choice sets as thoroughly as researchers want them to, but, instead, always choose the option in the same position across different choice sets. This phenomenon is termed "non-trading" in the literature. Non-trading can be seen as an extreme form of attribute non-attendance: It is the case where a respondent chooses to ignore not only some, but all attributes of the different options offered in a choice set. Consequently, researchers may exclude the answers from non-traders from their data set because their responses provide no information on the trade-offs.

In our study, we found that non-trading behaviour is more likely to occur among respondents who show protest beliefs regarding the practical impact of surveys in general. It also showed that they take less time to answer the questionnaire than the average respondent does. These results help to identify non-traders and to distinguish them from those respondents whose true preferences lead them to tick always the options in the same position in the choice sets. We ran the same regression model (for the identification of determinants of grape consumers' purchasing behaviour) over two different versions of our survey data set. The data set for the first regression analysis included all respondents of our survey, while for the second regression analysis the non-traders were removed from the survey sample. A comparison of the likelihood ratio chi-square values of the two regression analyses showed that the significance of the regression results increased as the non-traders were eliminated from the survey sample. This underlines the usefulness of identifying non-traders in choice experiments.