Menu
Home
Log in / Register
 
Home arrow Environment arrow Research Methods in Anthropology: Qualitative and Quantitative Approaches
Source

MAKING UP QUESTIONS FOR A FORMAL CONSENSUS TEST

A really important part of running a formal consensus analysis is building good test items. This takes work. Start with systematic and in-depth interviews with knowledgeable informants about the domain you’re investigating and do a close analysis of those qualitative data so that you can create sensible questions—that is, questions that reflect the content of the domain. We expect professors who make up course exams to know the material cold, and you should expect no less of yourself when you make up questions for a formal consensus analysis. About half the questions should be positive (true, yes, agree) and about half should be negative (false, no, disagree).

Jeffrey Johnson and David Griffith (1996) studied what people in North Carolina know about the link between seafood safety and coastal pollution using the formal cultural consensus model. First, they asked some expert informants to free-list (1) the kinds of pollution along the Atlantic coast (examples included acid rain, chemical runoff as the result of coastal erosion, and so on); (2) the species (clams, crabs, tuna) that might be affected by the various types of pollution; and (3) problems that might arise, like diseases in fish. From these data, Johnson and Griffith identified 12 types of pollution, 11 causes of pollution, and 10 species. Next, they asked a convenience sample of commercial fishermen and local residents of various ethnic backgrounds to (1) pile sort the types of pollution; (2) link each pollutant to the problems they cause; and (3) identify the species that each pollutant affected the most.

Johnson and Griffith asked the informants to explain why they made all these linkages and transcribed the results. Then, three coders went through the transcripts, tagging statements about the how seafood, pollutants and various risks to health and the environment were related. There were 53 statements identified by at least two of the three coders. The researchers turned these into a true-false knowledge test, with half the statements positive (true) and half being negative. For example, ‘‘heavy metals cause sores on both fish and people’’ is true (positive), but ‘‘heavy metals are necessary nutrients for both fish and people’’ is false (negative). Finally, they gave the test to 142 people in the area including: (1) a representative sample of 132 people, stratified by residence (rural-urban), income (high and low), and ethnicity (white and black); (2) 10 students from their university; and (3) 10 marine scientists. They analyzed the 142(people)-by-53(statement) matrix using the formal consensus model. The first eigenvalue in the analysis of the agreement matrix was over seven times bigger than the second. Those 142 people, despite coming from such different backgrounds, drew from a single culture in answering the 53 questions on the test about seafood and pollution.

Fine-grained analysis, though, turned up some interesting differences. The consensus was that the following statement was true: ‘‘Much of the pollution dumped into coastal and ocean waters has no effect on the flavor or seafood.’’ The scientists, however, disagreed.

 
Source
Found a mistake? Please highlight the word and press Shift + Enter  
< Prev   CONTENTS   Next >
 
Subjects
Accounting
Business & Finance
Communication
Computer Science
Economics
Education
Engineering
Environment
Geography
Health
History
Language & Literature
Law
Management
Marketing
Mathematics
Political science
Philosophy
Psychology
Religion
Sociology
Travel