Ecological Fallacy

Ecological fallacy can be defined as the erroneous inference about individuals from data or information representing the group or a geographic region. This is an important methodological problem among several disciplines in social sciences, including economics, geography, political science, and sociology, because these disciplines rely on many aggregate-level data sets such as census data. In geographic research, individual-level data may be aggregated to geographic units of different sizes or scales, such as census tract and block group, to infer individual behavior. Then the ecological fallacy is related to the scale effect, which refers to the inconsistency of analytical results when data aggregated to different geographic scales are used.

The problem of ecological fallacy is composed of two parts: how the data are aggregated and how the data are used. All data are gathered from individuals. But due to many reasons, such as privacy and security issues, individual-level data usually are not released; rather, they are aggregated through various ways in which to represent the overall situation of a group of individuals. The group of individuals can be defined by socioeconomic demographic criteria (e.g., below the poverty line, whites) such that individuals within the group should share similar characteristics. The group may also be defined geographically (e.g., within a county or a census tract) such that individuals are in the vicinity of a given location.

There are many ways in which to aggregate individual-level data. One of the most common methods is to report the summary statistics of central tendency such as the mean or median of the individual-level data. Another way is to report the number of observations possessing a specific characteristic such as the population within a given range of ages. Surely, these statistics can represent the general characteristics of the observations given that individuals have similar characteristics, but they definitely fail to describe precisely the situations of all individual observations. Not all individuals are identical, and the statistics might be good to describe only some.

Although aggregated data represent the overall situation of a group of individuals, there is nothing wrong with using the data for analysis so long as one recognizes the limitations of the data. Ecological fallacy emerges when one using the aggregated data does not recognize the limitation of using the data to infer individual situations and ignores the variability among individuals within the group.

This is a well-recognized but stubborn problem in social sciences. Many researchers attempt to solve this problem. Gary King claimed that his error-bound approach can handle the problem, but geographers are skeptical that his method can deal with the scale effect. Ideally, using individual-level data will not commit ecological fallacy. In general, less aggregated data, or data for smaller groups, are more desirable. Geographically, data representing smaller areas will be less likely to generate serious problems. Standard deviation or variance, which indicates the variability within the group, can potentially reflect the likelihood of committing ecological fallacy.