BUS602: Marketing Management

To assess current clothing consumption behavior, we created artificial consumer segments by employing a cluster analysis. However, because sociodemographics seemed to have lost their predictive power through consumer fragmentation, we achieved our main aim of identifying consumption patterns and their potential relations to different related environmental aspects by eschewing the socioeconomic clustering variables common in segmentation strategies, as well as values or attitudes toward a product domain, general lifestyle, or actual reported behavior. Rather, we defined our segments based on purchasing behavior, building the different segments to sort the heterogeneous sample into more homogenous subgroups whose members resemble each other on the clustering variables. At the same time, to account for as many intergroup differences as possible, we identified our consumer groups based on the amount and type of clothing bought and then compared them based on both purchasing and environmentally related behaviors across the consumption phases. This process enabled us to compare, for example, high volume and budget brand buying consumers with low volume premium buying consumers with regard to discard behavior.

To achieve our aim, we combined domain specific (general fashion) and product category-specific (jeans and t-shirts) variables. As the segmentation base, we included on the domain-specific level only the following purchase characteristics: material purchased (new, conventional; new, organic; recycled; and second-hand) and acquisition mode, divided into first market acquisitions, including high street, shopping mall, online shopping, mail order, small boutiques, and supermarket; and second market acquisitions, including second-hand purchases and swapping (see also Table 1). We then assessed the type of brands purchased (budget, casual/medium and premium), the number of purchased items, and spending over the last 3 months on both the domain- and product category-specific levels. To determine the number of clusters, we employed a hierarchical cluster analysis with the squared Euclidean distance as the distance measure and then used Ward's algorithm to link consumers. In line with the Duda–Hart stopping rule, we created five clusters by running a k-means clustering analysis on a 3984 respondent sample determined by missing values in the segmentation base variables (n = 633) and case-wise deletion. The resulting segments are: (1) low consumption - budget brands; (2) low consumption - casual/medium brands; (3) medium consumption - budget brands; (4) medium consumption - casual/medium brands; and (5) high consumption - casual/medium and premium brands.

We then compared these five consumer segments on sociodemographics, use and maintenance behavior, discard behavior, and environmentally related behavior employing either ANOVA or Kruskal–Wallis equality-of-populations rank tests dependent on the measurement level and variable distribution across segments. In presenting the results, instead of p-values, we report only the differences that are statistically significant (the group comparison results are available from the authors upon request). After first describing the consumer segments' demographic characteristics and reported purchase behaviors, which serve as the segmentation base, we compare their use and maintenance, and discard behaviors. We then investigate intersegmental differences in environmentally friendly clothing consumption behavior.