Information icon.svg At All things in moderation, we are currently voting on restrictions on the use of LLMs (chatbots) in RationalWiki. Reminder: you need to have at least 75 edits and to have been registered for at least three months to be eligible to vote.

Cohort study

From RationalWiki
Jump to navigation Jump to search
Eyes wearing
inverted lenses

Philosophy of science
Icon philosophy of science.svg

A cohort study is a type of study that gathers data by observation. A typical study design would involve enrolling a large group of people and watching them for the appearance of a disease or other event of interest. This is less powerful than a randomized controlled trial ("RCT"), but allows examination of evidence that RCTs cannot evaluate. One of the more commonly cited cohort studies is the Nurse's Health Study which follows roughly 120,000 nurses, starting in 1973, gathering various information about their health, including but not limited to cancer, heart disease, and obesity.

Cohort studies are often contrasted to case-control studies, where correlations are drawn after the fact. For example, if you are interested in the association between condom use and syphilis transmission, you could use a case-control design or a cohort design. In the case-control design, you would look at people who have syphilis, and those who do not, and look at the rates of condom use in each group. By contrast, in a cohort study, you would observe a group of people, recording who uses condoms and who does not, and looking at who catches syphilis and who does not, using statistics to calculate the strength of the relationship. Cohort studies are generally more effective at demonstrating causality since subjects are observed with different risk factors before they get the disease and then followed to assess relationships between risk factors and subsequent new disease. Usually, cohort studies create more scientific jobs are more costly than case-control studies since they require larger sample sizes followed for a longer time.

Theoretically, an RCT could be designed to look at the same question, but this would often be unethical (e.g. if one is measuring the contracting of diseases). The corresponding RCT would randomly assign a group of people to use condoms, and a similar group of people to not use condoms, and then let them do what they will, and calculate syphilis rates at the end.

A RationalWiki cohort study has shown that a large percentage of regular users will, in fact, develop obsessions for goats.[citation NOT needed]