Accompanies LBS Chapters 3 (Numerical Data Description) and 9 (Categorical Data)
How do marketers segment the market? That is, how do they identify sets of consumers with propensities to buy particular types of products, and how do they reach them?
Cluster analysis identifies and classifies objects individuals or variables on the basis of the similarity of the characteristics they possess. Thus, in particular, it can be used to assist in market segmentation, the identification of life-style/buying-habit groups within the population. This is the topic of Michael J. Weiss' book The Clustering of America.
The Nova program, "We Know Where You Live," available from the UIC A&A Library and which we view in sometimes in this course, relates to material in this book and in Larson (1991).
The book discusses the PRIZM cluster system developed by the Claritas Corporation of Alexandria, Virginia. Claritas provided demographic statistics, maps, tables and bar charts. In addition, the book contains PRIZM-coded data drawn from Simmons Market Research Bureau (SMRB), Mediamark Research Inc. (MRD), R. L. Polk & Co. and Targeting Systems, Inc., under an agreement with Claritas Corp.
The book concerns social conditions in the U.S. since 1980, social life and customs since 1971, and social aspects associated with the different postal zones. It may be regarded as a particular type of social survey of the U.S.
The PRIZM zoom lens looks at America in smaller and smaller units of counties, zips and block groups. In the U.S. there are 254,000 census blocks, each containing about 340 households. Other commonly used marketing units are census tracts (about 1,270 households each) and zip codes (about 2,320 households each).
According to this population classification system, linking census data to marketing and opinion surveys, America is made up of forty (now up to 62) very different types of neighborhoods.
These forty clusters were created by Jonathan Robbin, a social scientist/computer expert turned entrepreneur who in the 1970s pioneered the computer-powered marketing technique called "geodemographics." As a programmer during the 1950s, one of Robbin's projects was to help a geologist classify Bahamian sea-bed fossils according to salt content, temperature and depth. On assignment for the Office of Economic Opportunity, he used a wealth of federal studies to draw a computer-based profile of every county in the country, resulting in a massive 187,000-page report characterizing every jurisdiction's housing, educational level, health problems, and poverty index. In 1971, Robbin started Claritas Corp. to build a cluster system of U.S. neighborhoods for marketing applications.
There were hundreds of variables, in five groupings: social rank, mobility, ethnicity, family life cycle and housing style. Robbin identified 34 key factors that account for 87% of the variation among U.S. neighborhoods. Finally, he rated each zip code on the 34 factors simultaneously in order to assign it to one of the forty clusters.
Claritas analysts tested more than three dozen experimental models, some involving one hundred neighborhood types. The forty-cluster system is a compromise between manageability and discriminating power. The forty groups range in size from 0.5 to 6 percent of all American households.
In 1974, Claritas came out with a ranking of its forty "life style segments" along a scale of affluence called a "Zip Quality" (ZQ) scale--from Blue Blood Estates (with a ZQ rating of 1) to Public Assistance (ZQ40). Some of the other clusters are named Money & Brains, Furs & Station Wagons, Young Influentials, Young Suburbia, Bohemian Mix, Shotguns & Pickups, Blue Collar Nursery, Coalburg & Corntown, Grain Belt, and Heavy Industry.
Examples. (i) "Bohemian Mix" (ZQ 11) includes Greenwich Village, New York; Dupont Circle, Washington, DC; Cambridge, Boston, MA; Lincoln Park, Chicago, IL; Shadyside, Pittsburgh, PA; Haight-Ashbury, San Francisco, CA. (ii) "Shotguns & Pickups" (ZQ 24) includes Molalla, OR; Zanesville, OH; Ringgold, PA; Monroe, IN; Jewett, WV; Moravia, NY. (iii) "Money & Brains" (ZQ 2) includes Georgetown, Washington, DC; Grosse Point, MI; Palo Alto, CA; Princeton, NJ; Park Cities, Dallas, TX; Coral Gables, FL.
In 1978, Claritas developed PRIZM-- for "Potential Rating Index by Zip Markets." This linked the clusters to dozens of surveys concerning products, media and opinion. Among the surveys used were the list of new-car buyers from R.L. Polk, the TV viewing diaries of A.C. Nielsen and the consumer buying polls of Mediamark Research Inc. and Simmons Market Research Bureau. Now, knowing only a community's cluster classification, a marketer could predict the lifestyle of residents--everything from the kinds of bread on the dining table to the magazines on the coffee table.
Such knowledge is power for target-market merchandisers who customize their products and tailor their sales pitches to fit specialized markets.
Update on PRIZM: The cluster system has been expanded from 40 to 62 life-style clusters. Some of the information is available through AOL. See also Claritas' Web site.
Larson, Erik. The Naked Consumer. Henry Holt & Co., 1992. Paper: Penguin Books, New York, 1994.
Weiss, Michael J. The Clustering of America. New York: Harper & Row (a Tilden Press book), 1988. [A profile of the U.S. based on 40 Life-Style clusters.]
Afifi, A. and Clark, V.(1990). Computer-Aided Multivariate Analysis. Van Nostrand-Reinhold, New York, 1990. [Ch. 16 is on cluster analysis.]
Aldenderfer, Mark S., & Blashfield, Roger K. Cluster Analysis. Sage Publications, Newbury Park, Cal., 1984.
Hair, Anderson, Tatham & Black. Multivariate Analysis. 5th ed.Weiss, Michael J. Latitudes and Attitudes: An Atlas of American Tastes, Trends, Politics and Passions. New York: Little, Brown and Co., 1994. (A profile of the U.S. based on Arbitron's 211 Areas of Dominant Influence (ADIs), geographically continguous areas defined by media, cultural and economic forces.)