Open vs. Closed Cohort: What Difference Does It Make?

Should a Cohort Allow New Participants to Enter After the Study Begins?

Suppose a university wants to examine doctoral student attrition over five years. One option is to identify all students enrolled at the beginning of the observation period and follow that group. Another is to maintain a cohort that admits newly enrolled doctoral students throughout the study.

Both approaches can generate useful evidence about attrition, but they do not necessarily describe the same population. They also differ in how follow-up begins, how much observation time participants contribute, and which students are represented in the findings.

The distinction between an open and a closed cohort is therefore more than a matter of recruitment logistics. It determines how membership is defined and influences the interpretation of outcomes observed over time.

The Difference Is Whether New Members Can Enter the Cohort

How Open and Closed Cohorts Differ in Membership, Follow-Up, and Analysis

What Is a Closed Cohort?

A closed cohort, sometimes called a fixed cohort, consists of individuals whose membership is established during a defined entry period. Once that period closes, no new members are added.

For example, researchers might identify every student who began a doctoral program during the 2026 academic year and follow those students for five years. Students who enroll in later academic years would not become members of that cohort.

The cohort remains defined by its original membership even if some students graduate, withdraw, transfer, experience the outcome, or become unavailable for follow-up.

This does not mean that every original participant must remain under observation throughout the entire study. Membership and continued observation are different concepts.

A closed cohort can have incomplete follow-up, unequal observation times, and participants who experience competing events. What makes it closed is the absence of new membership after the entry period.

What Is an Open Cohort?

An open cohort, also called a dynamic cohort, permits eligible individuals to enter during the observation period. Membership can change as new participants join and existing participants leave or cease contributing observation time.

Consider a university that maintains an administrative register of all doctoral students enrolled between 2026 and 2030. Students become eligible for cohort entry when they enroll, even if their enrollment occurs several years after the observation period begins.

Some students graduate or discontinue enrollment, while others remain under observation. The population represented by the cohort changes over calendar time.

This structure can be useful for studying outcomes in settings where entry and exit are ordinary features of the population, including universities, workplaces, healthcare systems, and geographic communities.

An open cohort should not be confused with an unrestricted convenience sample. Entry must still follow clearly specified eligibility rules and an appropriate observation framework.

How Do the Two Designs Compare?

Neither design is universally more rigorous. Their strengths depend on whether the research question concerns a fixed group or a population whose membership changes.

Does a Closed Cohort Require Everyone to Start Follow-Up on the Same Date?

Not necessarily. A closed cohort may recruit participants during a defined enrollment window rather than on a single calendar day.

For example, all students entering a doctoral program during one academic year could constitute a closed cohort even if their enrollment dates differ.

Researchers must distinguish the period during which cohort membership is established from each participant's individual start of follow-up.

Depending on the question, time may be measured from a common calendar date or from an individual event such as enrollment. The choice determines the meaning of the risk sets and follow-up durations.

The definition of the start of cohort follow-up must therefore be aligned with eligibility, exposure, and outcome measurement.

Does an Open Cohort Automatically Have Staggered Entry?

Open cohorts commonly involve staggered entry because participants join at different calendar times. However, staggered entry and open membership are not identical concepts.

Staggered entry describes when individuals begin contributing observation time. Open membership describes whether the cohort permits new members during the observation period.

A closed cohort assembled during a fixed recruitment window can also have participants whose follow-up begins on different dates.

In an open cohort, entry may depend on employment commencement, institutional enrollment, migration into a defined population, or another eligibility event.

Analyses must ensure that participants contribute time only when they meet the relevant eligibility and observation conditions. The implications of participants entering a cohort at different times extend beyond simply recording their dates of enrollment.

How Does Cohort Membership Affect the Denominator?

The denominator of an incidence measure must represent the appropriate population at risk.

For a closed cohort with complete observation over a specified period, cumulative incidence may be estimated as the proportion of initially at-risk participants who experience the outcome.

In an open cohort, participants enter and leave at different times, making a simple proportion of all individuals ever enrolled potentially misleading when observation durations differ substantially.

Incidence rates based on person-time are often useful in dynamic populations because they account for the amount of observed time contributed by participants while at risk.

However, open cohorts are not restricted to person-time analyses. Appropriately defined risks can also be estimated for eligible entrants over specified follow-up horizons using suitable survival methods and assumptions.

How Is an Incidence Rate Calculated?

An incidence rate expresses the number of new events relative to the total person-time accumulated while participants are at risk.

It is especially useful when observation periods differ across participants.

Person-time must be accumulated according to the outcome definition. For a first-event analysis, an individual ordinarily stops contributing time at risk after experiencing the first event.

If recurrent events are studied, different counting and modeling rules may apply. Researchers should not combine first-event and recurrent-event analyses without specifying the intended estimand.

Can Closed Cohorts Also Use Person-Time?

Yes. Person-time is not exclusive to open cohorts.

A closed cohort may contain participants who experience events at different times, withdraw, die, transfer, or reach the end of observation without experiencing the outcome.

These participants contribute different amounts of time at risk. Incidence rates and time-to-event analyses may therefore be appropriate even though no new members enter the cohort.

Similarly, a closed cohort can use Kaplan–Meier methods when their assumptions are appropriate, or competing-risk methods when relevant events prevent the outcome of interest from occurring.

The analytical choice depends on the outcome process and follow-up structure, not solely on whether membership is open or closed.

What Happens When Participants Leave an Open Cohort?

Leaving a cohort's observable population may occur for different reasons, and these reasons should not automatically be treated as equivalent.

A participant may experience the outcome, cease meeting eligibility criteria, transfer to another institution, withdraw from the study, or reach the administrative end of observation.

Some departures represent events of interest. Others may be censoring events or competing events, depending on the estimand.

For example, if the outcome is withdrawal from a doctoral program without completion, graduation may be a competing event because a student who graduates can no longer experience withdrawal from that same program in the same way.

By contrast, if the outcome is departure from the institution for any reason, graduation might be included in the outcome definition.

The classification must follow the research question rather than being decided after observing the data.

Incomplete observation also requires attention to loss to follow-up and its potential bias.

Can an Open Cohort Represent a Changing Population More Accurately?

It may, particularly when the target population continually receives new members.

Suppose researchers want to monitor academic attrition among all doctoral students enrolled at a university during a five-year period. A closed cohort of students enrolled only at the beginning would exclude subsequent entrants.

An open cohort can include those entrants and may therefore align more closely with the changing student population.

However, admitting new participants does not guarantee representativeness. The sampling frame must cover eligible entrants, and the observation system must record relevant entries and exits accurately.

Furthermore, an open cohort assembled from institutional records may represent only students within that institution, not doctoral students generally.

How Can Calendar Time Complicate Open-Cohort Analysis?

Participants entering at different times may experience different institutional or social conditions.

For example, students entering before the introduction of an AI-supported academic advising system may encounter different services from students entering afterward.

Differences in attrition may therefore reflect calendar-period changes rather than individual characteristics alone.

Researchers may need to account for entry cohort, calendar time, duration since enrollment, or other relevant temporal variables.

These time scales answer different questions. Duration since enrollment concerns how risk changes as students progress through their programs. Calendar time concerns changes in the institutional environment.

When both are relevant, the analytical design should represent them appropriately rather than treating all observation time as interchangeable.

Can a Cohort Change From Closed to Open?

A study may be redesigned to allow additional entrants, but doing so changes its membership rules and potentially its target population.

For example, researchers may initially plan to follow one entering class of doctoral students but later decide to include subsequent entering classes.

The expanded study can still be methodologically defensible if the change is documented, eligibility rules are consistent with the revised question, and analyses account for the resulting follow-up structure.

However, researchers should not describe a changing-membership study as though it were an unchanged fixed cohort.

Changes made after examining outcomes may also introduce selection and analytical concerns. The protocol and reporting should distinguish planned design features from later modifications.

Studying Doctoral Student Attrition With Open and Closed Cohorts

Which Design Answers the University's Question?

If the university wants to know what happened to students who began their doctorates in 2026, a closed cohort is a natural choice.

If it wants to monitor attrition across a changing population of doctoral students, an open cohort may be more appropriate.

Neither design should be selected merely because it produces more observations. The membership structure must correspond to the population and outcome the study intends to describe.

Common Misunderstandings About Open and Closed Cohorts

How to Choose Between Open and Closed Cohorts

Begin by deciding whether the target population is an initially defined group or a population whose membership changes during observation.

A closed cohort is often appropriate for studying outcomes among a specific entering class, employment intake, or group exposed during a defined historical period.

An open cohort may be more suitable when the question concerns an ongoing institutional or community population.

Before data collection or record extraction, document who can enter, when entry is permitted, when participants begin contributing time at risk, and when their observation ends.

Where exposure status changes, consider whether time-varying exposure measurement is needed. Membership rules alone do not determine how exposures should be represented.

The central objective is to ensure that the cohort's membership and observation structure support the intended population-level or individual-level inference.

Before Defining an Open or Closed Cohort

Frequently Asked Questions About Open and Closed Cohorts

Membership Rules Should Reflect the Population You Want to Study

How to Cite This Guide

This guide is intended to be read, shared, and used in research, teaching, and academic work. If you draw on its ideas, explanations, or other content, please acknowledge the source by citing the guide. Doing so gives appropriate credit and helps your readers locate the original resource.