An approach to derive economic weights in breeding objectives using partial profile choice experiments.

The aim of this study was to show how choice experiments can be used to derive economic weights in breeding objectives. In a choice experiment, respondents are asked to view various alternative descriptions of a good differentiated by their attributes and levels, and are asked to choose their most preferred alternative. Analysis of the data generated can be used to elicit a quantitative description of respondent preference for contrasting attributes and levels. We simulated a partial profile choice experiment with four different attributes (traits) each at three levels. In a partial profile design, the choices are simplified so that only a subset of traits is used for each comparison, making participation in the experimental process less onerous. Three different choice designs were compared. All three designs included four attributes each at three levels where respondents choose between two alternative genotypes. In the first design, respondents choose between two genotypes differing for all four traits simultaneously. In the second and third designs, respondents made choices based on three or two out of the four traits per choice set respectively. The effectiveness of different designs was evaluated based on comparisons between true and simulated preferences for varying numbers of respondents and choice sets per respondent. Choice design and the simulated respondent choice were analysed using a conditional logit model. Regression coefficients from the conditional logit model based on an average of 200 replicated choices across respondents were used to estimate the relative economic weights of traits. A need to account for discounted gene flow principles when formulating the survey questions was emphasised as a critical component of the method. When the relative importance's of four traits were considered, practical designs involving, e.g., 20 choice sets based on a subset of two traits at each choice, and over 30 respondents provided relatively accurate estimates of relative respondent preferences. The method based on a practical choice experiment design can be used to define economic weights for use in animal breeding selection indexes where traditional approaches such as profit equations and bioeconomic models are not practical. The approach may also be of interest to commercial breeding programs wishing to formulate a quantitative understanding of market preferences for attributes of the genestocks that they sell.