[last update
11/26/03]
Ecological
studies are merely epidemiologic studies in which the unit of measurement is an
aggregate rather than an individual. To explain the distinction between
aggregate-level and person level measurements, consider a study on smoking, age,
and respiratory disease. For example, a person-level study on smoking and
health might collect information on whether each person smokes (SMOKE), the
person’s age (AGE),
and whether the person has respiratory disease (RESP):
PERSON SMOKE AGE
RESP
1 Y 46 Y
2 N 52 N
3 Y 49 N
etc.
In
contrast, aggregate-level studies collect data on group characteristics only.
For example, an aggregate-level study might collect information on each region's prevalence of smoking per 100
(SMOKE_P),
average age (AVG_AGE),
and prevalence of respiratory disease per 100 (RESP_P):
DATA TABLE 2: AGGREGATE-LEVEL DATA
STATE SMOKERS_P AVE_AGE RESP_P
FLORIDA 8 49 24
NEW
YORK 21 39 15
CALIF 18 29 14
etc.
Ecological studies often use readily available data, such as from surveillance and vital statistics sources (a primary benefit of this method). However, ecological studies suffer from multiple weaknesses, such as aggregate data are often inaccurate, contributors to disease are often unavailable, and they may suffer from a form of bias known as the ecological fallacy (or aggregation bias). The ecological fallacy occurs when you try to extrapolate findings from ecological data to the individual, but it does not apply (see pp. 195 – 196 for illustrative examples).