Page 290 - Communication Theory and Research

McQuail(EJC)-3281-20.qxd 8/16/2005 12:02 PM Page 275

Lifestyle Segmentation 275

Adding media preferences, product
categories and demographics

Now, since our aim is a lifestyle typology relevant for predicting individual
differences across a wide variety of behaviour, we included four product cate-
gories in our questionnaire: cars (a classic product), tourism (a service), political
parties (a non-commercial product) and media. For both cars, tourism and polit-
ical parties, a set of potential attributes or benefits was developed:

• Cars: 14 attributes, including safety, design, powerful engine, reliability, etc.;
• Tourism: 14 attributes of the ideal holiday, including warm and sunny climate,
cultural infrastructure, luxurious, romantic, etc.;
• Political parties: 15 potential elements of party programmes, including: job
opportunities for everyone, lowering taxes on labour, raising pensions, fight-
ing unemployment, aiding the Third World, etc.

The media section focused on television, films and magazines:

• Television: 16 programme categories;
• Films: 12 ‘movie ingredients’, including romance, adventure, hard action,
humour, etc.;
• Magazines: 14 categories of magazines, including male, female, television,
general information, fashion, sports, cars, etc.

The appealing power of each of these product attributes or benefits and each of
these media categories was scored on a seven-point scale.
Finally, we added a section on demographics, asking the respondents for their
sex, age, social class and stage of life.

Segmenting the market

The questionnaire was administered to a quota sample of the Flemish population
(N = 995). In order to group these 995 respondents into more or less homogeneous
lifestyle segments, we conducted a cluster analysis (which is, in marketing
research, the dominant method for market segmentation). We selected a two-
stage approach combining both hierarchical and non-hierarchical clustering
methods (Punj and Stewart, 1983; Fournier et al., 1992: 331). Initial solutions,
using Ward’s hierarchical method, provided a preliminary indication of the total
number of clusters. The final cluster solution was then identified using the Quick
Cluster K-Means procedure. Here we identified a range of solutions (from five to
ten clusters) and chose the solution where (1) there were as many segments as
possible, but no small segments of less than 5 percent, and (2) the number of clus-
ters was justified by the results obtained through Ward’s method.
Remarkably, for values (V), life visions (L), aesthetic styles (A) and a combina-
tion of those three categories of variables (V-L-A), this yielded a seven-cluster