27.10.2014 Views

Russel-Research-Method-in-Anthropology

Russel-Research-Method-in-Anthropology

Russel-Research-Method-in-Anthropology

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

326 Chapter 12<br />

12 societies and 11 traits. Carneiro, you’ll recall, coded 100 societies for 354<br />

traits and then went look<strong>in</strong>g for subsets of the data that showed the desired<br />

pattern. When he did this <strong>in</strong> the 1960s, it was heroic work. Today, Anthropac<br />

(Borgatti 1992a, 1992b) has a rout<strong>in</strong>e for look<strong>in</strong>g at big matrices of pluses and<br />

m<strong>in</strong>uses, rearrang<strong>in</strong>g the entries <strong>in</strong>to the best pattern, calculat<strong>in</strong>g the CR, and<br />

show<strong>in</strong>g you which units of analysis and traits to drop <strong>in</strong> order to f<strong>in</strong>d the<br />

optimal solution to the problem.<br />

Data Scale, Variables Don’t<br />

DeWalt (1979) used Guttman scal<strong>in</strong>g to test his <strong>in</strong>dex of material style of<br />

life <strong>in</strong> a Mexican farm<strong>in</strong>g community. He scored 54 <strong>in</strong>formants on whether<br />

they possessed eight material items (a radio, a stove, a sew<strong>in</strong>g mach<strong>in</strong>e, etc.)<br />

and achieved a remarkable CR of .95. This means that, for his data, the <strong>in</strong>dex<br />

of material style of life is highly reliable and differentiates among <strong>in</strong>formants.<br />

Remember: Only data scale, not variables. If the items <strong>in</strong> a cumulative<br />

<strong>in</strong>dex form a Guttman scale with 0.90 CR or better, we can say that, for the<br />

sample we’ve tested, the concept measured by the <strong>in</strong>dex is unidimensional.<br />

That is, the items are a composite measure of one and only one underly<strong>in</strong>g<br />

concept. DeWalt’s data show that, for the <strong>in</strong>formants he studied, the concept<br />

of ‘‘material style of life’’ is unidimensional—at least for the <strong>in</strong>dicators he<br />

used.<br />

The Guttman technique tests whether unidimensionality holds for a particular<br />

set of data. An <strong>in</strong>dex must be checked for its Guttman scalability each time<br />

it is used on a population. My hunch is that DeWalt’s material-style-of-life<br />

scale has its analog <strong>in</strong> nearly all societies. The particular list of items that<br />

DeWalt used <strong>in</strong> rural Mexico may not scale <strong>in</strong> a middle-class neighborhood of<br />

Ulan Bator, but some list of material items will scale there. You just have to<br />

f<strong>in</strong>d them.<br />

The way to do this is to code every household <strong>in</strong> your study for the presence<br />

or absence of a list of material items. The particular list could emerge from<br />

participant observation or from <strong>in</strong>formal <strong>in</strong>terviews. Then you’d use Anthropac<br />

to sort out the matrix, drop some material items, and build the material<br />

<strong>in</strong>dex that has a CR of 0.90 or better. Greg Guest (2000) did this <strong>in</strong> his study<br />

of 203 households <strong>in</strong> an Ecuadorian fish<strong>in</strong>g village. He gave each household<br />

a score from 1 to 7, depend<strong>in</strong>g how many material items they had. That score<br />

correlated significantly with the education level of the head of each household.<br />

(We’ll get to correlation and statistical significance <strong>in</strong> chapter 20.) S<strong>in</strong>ce we<br />

expect a correlation between wealth and education, this adds construct validity<br />

to Guest’s scale.<br />

Careful, though. Oliver Kortendick tried to develop a Guttman scale of

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!