**5.3. Clustering consumers**

Based on the previous section results, different combinations of socio-demographic indicators and psychographic variables have been implemented to determine the optimal segmentation strategy. For the purpose of this study, broad segmentation is defined as a segmentation strategy that has non-distinct segments while good segmentation is defined as a segmentation strategy with very distinct segments. The idea is to maximize intra-group homogeneity and intra-group heterogeneity. This allows for more robust profiling, as consumers will behave in the same way when they belong to the same segment and will behave differently if they belong to different segments. Note that homogeneity and heterogeneity are defined with regards to the segmenting variables. For the purpose of having a good measure of inter-group heteroge‐ neity, several ANOVAs were run to make sure that consumers in different segments have different profiles. All tests were conclusive.

The three attributes driving the purchasing behavior and the consumption patter are: organic, food mileage, and healthiness. Consumers were clustered using those 3 variables; all are measured using a 5-point bipolar scale. Results of 2-step cluster analysis show 2 groups of consumers that act in a very distinctive way (cf. Table 5). It is clear that cluster 1, referred to as hardcore local food consumers, is composed of consumers that look for higher and elaborate values than only local. They are looking for local organic food products. Further, they score very high on food mileage and healthiness showing that these attributes are crucial in their purchase decisions. Conversely the regular local food consumers are not looking for any organic dimension rather these consumers look for basic food attribute criteria to make their purchases. These results are in line with the previous results but add a clustering dimension to the results.


**Table 5.** Cluster Analysis Results
