An ecological error is an error in reasoning, moving from the general to the specific. The word ecological is used to refer to a group or system, something that is larger than an individual. Ecological fallacies arise when we try to draw conclusions about individuals based on data collected at the group level. Stereotypical thinking like this assumes that groups are homogeneous, when in reality individuals do not necessarily share the characteristics of the group to which they belong.
Cause
The root cause of ecological errors is the misinterpretation of statistical information. Researchers collect statistical data with the aim of generalizing from sample to population, that is, from individual to population, not vice versa. When we collect data at the group level, it is a process similar to writing a summary : some information details will be lost or hidden.
For example, in a study on prostate cancer, researchers found that there is a correlation between high sugar intake, meat consumption, and prostate cancer mortality. Does this mean we conclude that excessive sugar and steak consumption leads to death from prostate cancer? Can we use the results as dietary recommendations? The answer is no. Although the study provides insight into prostate cancer risk factors, it does not establish causality. The data in this study are aggregated: what may apply to a population does not necessarily apply to an individual basis.
Overall, a correlation tends to be greater when an association is assessed at the group level than when it is assessed at the individual level. As a result, when studies like these are analyzed at the individual level, the relationship often disappears.
In other words, while it is true that countries with higher levels of sugar consumption have higher rates of prostate cancer deaths, this does not mean that people who eat high diets in sugar are more likely to die from prostate cancer.
How to avoid ecological mistakes?
We can avoid ecological mistakes in our own conceptions of research or in interpreting other people's research results by following the steps below:
Clearly define the unit of analysis. Before collecting your data, think about who or what it will be used for. Are they individuals, groups, photos or social interactions? For example, if you compare students in two classes on test scores, the individual student constitutes the unit. On the other hand, if you want to compare the average performance of a class, the unit of analysis is the group, because the data you are analyzing refers to the group, not the individual.
Be aware of logical leaps. When drawing conclusions from your own research, reading research articles, or viewing news articles based on data, take a minute to think: Is the claim on the same level as the data? Or does the statement concern individuals, while the data refer to a population? If so, this is a case of ecological fallacy.
Keep in mind that results from group-level data cannot be applied to individuals. If you want to investigate individuals or subpopulations within a larger population, be sure to obtain data on those individuals or subpopulations. This means that you must follow a method appropriate sampling method, such as stratified sampling, if you are interested in, for example, specific subpopulations.
EN 
FR