3.2. Conjoint analysis
The conjoint model was estimated using ordinary least squares regression analysis. The estimated model establishes the relative importance of the attributes, as well as the part-worth of each level of the attributes. The accuracy of the estimation was tested by the Pearson correlation coefficient (0.961). The high value found indicated that the model predicts to a high degree the preferences of the consumers. Table 4 shows the aggregate results for the whole sample.
The percentage in the right-most column of each attribute is its relative importance. Also, the value of each level's part-worth is given with its corresponding sign. A positive sign indicates that, for this survey, the presence of that level of the attribute adds that amount of utility to the product (for two levels with positive signs, that of greater value is the one that provides greater utility). A negative sign, on the other hand, implies that the presence of that level of the attribute in the product lessens its utility.
The most important attribute (30.27%) in forming the preferences was the colour of the meat, followed by origin (25.68%), and the price (22.64%). The attribute that least affected the choice of the different types of meat was the production system, with 21.41% of relative importance.