Environmental Engineering Reference
In-Depth Information
material is often in a database or in a format that can easily transfer into a data-
base, extensive manipulation—sorting, comparing, contrasting, aggregating,
and synthesizing—is even simpler. In my higher-education studies, most
content analysis takes place online or after downloading to a database.
Management philosophy is easily detected in online meeting minutes, budgets,
arguments, and policy statements (and drafts). A brief review of a department's
budget provides vital information about its values: People put their money into
areas they care about. A comparison of content analysis data with interview and
observational data can significantly enhance the quality of findings.
STATISTICS
Ethnographers use nonparametric statistics more often than parametric statis-
tics because they typically work with small samples, the assumption of nor-
mality is not met, and outcomes are frequently categorical or ordinal variables.
Parametric statistics require large samples for statistical significance. The
use of nonparametric statistics is also more consistent with the needs and con-
cerns of most anthropologists. Ethnographers often collect data that are in the
form of frequencies (“How many times did the director address the staff as
'guys'?” and “How often does the gang change hand signals in a 6-month
period?”), ranks (“What is the order of the six administrators in power within
the organization?”), or quantified names (“Agnostics are scored '1,'
Fundamentalists are scored '2,' etc.”). Like sociologists, political scientists, and
many other social scientists, ethnographers seldom resort to well-established
metrics, such as the gram-centimeter-second world of the physical sciences or
even the IQ scale and test scores of the psychologist. As a consequence of this
measurement style, ethnographers usually employ methods of statistical analy-
sis that more resemble the sociologist's chi-square contingency table analysis or
rank correlations and distribution-free tests (Friedman Rank Test, Mann-
Whitney U test, etc.) than they resemble the analysis of variances, t tests, and
regression analyses of the psychologist or economist.
Anthropologists typically work with nominal and ordinal scales. Nominal
scales consist of discrete categories, such as sex and religion. Ordinal scales
also provide discrete categories as well as a range of variation within each
category—for example, reform, conservative, and orthodox variants within the
category of Judaism. Ordinal scales do not determine the degree of difference
among subcategories.
The Guttman (1944) scale, also known as cumulative scaling or scalogram
analysis (Trochim, 2006a), is one example of an ordinal scale that is useful in
ethnographic research. In studying folk medicators, I used a Guttman scale to
Search WWH ::




Custom Search