I am an undergraduate economics student and I am familiar with difference-in-difference estimators: For example, to evaluate the effect of a policy change, measure the outcome of interest before and after the policy change took place both in the treatment and control group, and then compare the difference in the changes in the treatment and control group.

Doing literature research, I recently came across the term 'cohort study', which seems to be more used in epidemiology. To my understanding, in a cohort study basically the same is being done as in difference-in-difference estimators: compare change in prevalance levels of some disease before and after one group was exposed to to treatment with the change in prevalence levels in an unexposed control group.

Am I right, that these are just different names in different disciplines for the same econometric technique or am I missing something?

I haven't found anything on that in the literature, so I would be grateful for any help!

Best, Lucas


The key difference between these terms is the phase of research they describe. A cohort study is a method, first and foremost, of data collection. Data collected in a cohort study can be analyzed using a wide range of analytic approaches, including difference-in-difference. In contrast, a difference-in-difference analysis is a method of analyzing data. This analytic approach can be applied to data collected in a wide range of ways, including a cohort study.

Briefly, a cohort study is an observational study in which subjects are enrolled on the basis of some characteristics and then followed over time to determine how their characteristics change, but without any specific intervention by the researchers.

There are two main types of cohort studies.

The first, which is (probably) most commonly described in epidemiology courses/ textbooks, involves recruiting patients on the basis of some specific exposure variable of interest. Patients with and without the exposure are recruited and are often matched on other characteristics. For example, a study could recruit smokers and non-smokers, matched on age, gender, and education levels. Then the frequency (rate, or probability) of the outcome in both groups is compared at some later time. Or, a difference-in-difference method can be applied to compare how the outcome of interest changes in the two groups over time.

The second, which is (probably) most commonly used in epidemiology research, is to enroll a group of people based on a particular population of interest: for example, physicians, or all people living in a specific town. These cohort studies tend to be much bigger and allow for assessment of many different types of exposure variables. Individuals are often followed for life (or as long as funding allows), and many different outcomes are assessed. A classic example of this type is the Framingham Heart Study.

| cite | improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.