An ecological study is a study design that uses group-level or aggregate-level data to figure out if there is a potential association between two variables.
For example, let’s say you want to figure out if women that eat more fat have a higher risk of breast cancer.
Perhaps you have information about the average dietary fat intake per capita as well as breast cancer rates in 10 different countries.
You could plot this information on a graph with dietary fat intake on the x-axis and yearly incidence of breast cancer - which is the number of new cases of breast cancer in a year, per 100,000 women - on the y-axis.
In Japan, the dietary fat intake per capita is around 650 calories, and the incidence of breast cancer is 50 cases per 100,000 women.
On the flip side, in the United States the dietary fat intake per capita is around 1400 calories, and the incidence of breast cancer is 235 cases per 100,000 women.
The rest of the countries - Switzerland, Denmark, France, Norway, Australia, Spain, Hong Kong, and Romania - have an average fat intake and breast cancer rates somewhere between Japan and the U.S.
Generally, we can see that the more fat a country consumes, the higher the rate of breast cancer is for that country, and at this point we might conclude that eating more fat leads to breast cancer.
The problem with this conclusion is that we only have information from the whole country, and we don’t have information about each individual within that country.
In fact, we might get very different results from a study that uses individual-level data!
For example, let’s say you conduct a cohort study using individual-level data from all 10 countries.
Cohort studies look at individuals in a group who have a certain exposure, as well as individuals in a cohort who have not had that exposure, to compare their rates of a certain outcome in the future.
So, you could follow 500,000 individuals that have a high fat diet, the exposed group, and 500,000 individuals that have a low fat diet, the non-exposed group, for ten years.