CROSS-SECTIONAL AND LONGITUDINAL STUDIES
Most surveys are cross-sectional. The idea is to measure some variables at a single time. Of course, people’s attitudes and reported behaviors change over time, and you never know if a single sample is truly representative of the population. Many surveys are conducted again and again to monitor changes and to ensure against picking a bad sample. Multiple cross-sectional polls use a longitudinal design. The daily—even hourly— tracking polls in U.S. presidential elections are an extreme example, but in many industrialized countries some questions have been asked of representative samples for many years.
The Gallup Poll, for example, has been asking Americans for about 70 years to list ‘‘the most important problem facing this country today.’’ The data track the concerns of Americans about unemployment, the quality of education, drugs, street crime, the federal deficit, taxes, health care costs, poverty, racism, AIDS, abortion. . . . There are not many surprises in the data (people in the United States are more worried about the economy in recessions, less worried when the economy is clicking along) but data from the Gallup Poll, and others like it, are important because they were collected with the same instrument. People were asked the same question again and again over the years. After several generations of effort, longitudinal survey data have become a treasured resource in the highly industrialized nations.
Longitudinal studies by anthropologists are rare and are also great treasures. Beginning in 1961, Robert Edgerton studied a sample of 48 mildly retarded people in California who had been released from a state mental institution (Edgerton 1967). This was full-blown participant observation: hanging out, following people around, doing in-depth interviews, taking field notes. Edgerton was able to interview 30 of the same people in 1975 (Edgerton and Bercovici 1976) and 15 members of the sample in 1982 (Edgerton et al. 1984). Edger- ton last interviewed ‘‘Richard’’ in 1988, just before Richard died at age 68 (Edgerton and Ward 1991). As a result, we know more about how the mildly retarded get through life— how they make ends meet; how they deal (or don’t deal) with personal hygiene; how they get people to do things for them, like write letters; how people take advantage of them financially—than we could learn from any cross-sectional study.
Two of the best-known longitudinal studies in anthropology are the Tzintzuntzan project in Mexico and the Gwembe Tonga project in Zambia.
The Gwembe Tonga Project began with visits in 1956 by Elizabeth Colson and Thayer Scudder. Some 57,000 Gwembe Tonga were being resettled to make way for the lake that would form in back of the Kariba dam on the Zambezi River, and Scudder and Colson were studying the effects of that resettlement. In 1962-1963, they realized the ‘‘long-term possibilities involved in a study of continuity and change among a people who, having been forcibly resettled in connection with a major dam, were soon to be incorporated with the independent nation of Zambia,’’ as the colonial period came to an end (Scudder and Colson 2002:200). Colson and Scudder continued their work and began recruiting colleagues into the project, including Lisa Cliggett, who now manages the project (Cliggett 2002; Kemper and Royce 2002:192). The last full census was done in 2000-2002 and the data from two of the four original sending villages are updated regularly (Lisa Cliggett, e-mail, August 26, 2009). Like the Tzintzuntzan project, the Gwembe project has incorporated indigenous members on the team
Neither of these important projects started out as longitudinal studies. They just went on and on and on. The field notes and other data from these projects grow in importance every year, as more information is added to the corpus. All cross-sectional studies, including Master’s and Ph.D. projects, should be designed as if they were the start of a lifetime of research. You never know (Further Reading: longitudinal research in anthropology).