03 · What You Need to Know
How to Determine Whether Multiple Research Sites Are Necessary
What Counts as a Research Site?
A research site is a setting, institution, location, or organizational unit where some defined aspect of the research takes place. Depending on the study, sites might be hospitals, schools, university campuses, workplaces, laboratories, clinics, communities, or administrative jurisdictions.
The definition must reflect the actual research design. Two classrooms in one school are not necessarily two independent research sites. Conversely, two independently administered campuses within the same university system may reasonably be treated as separate sites when institutional conditions differ in ways relevant to the investigation.
A site is also not necessarily equivalent to a statistical cluster. Researchers must identify the level at which participants share conditions, receive interventions, or produce potentially correlated observations.
When the Research Question Concerns Differences Between Settings
Multiple sites become necessary when differences between settings are part of what the research intends to investigate.
Suppose a researcher wants to understand whether institutional support for online learning differs between public and private universities. Studying only one university cannot establish the proposed between-institution comparison, even if that university contributes hundreds of respondents.
The relevant requirement is not simply the number of participants. It is the presence of sufficiently varied and appropriately selected institutional units to support the intended comparison.
Similarly, a study investigating how rural and urban healthcare facilities implement the same policy must include settings representing the contextual distinction under investigation.
When site characteristics form part of the research question, the choice between single-site and multisite research becomes a substantive design decision rather than an administrative preference.
When One Site Cannot Provide the Required Participants
A second reason for involving multiple sites is recruitment feasibility.
A single hospital might treat too few patients with a particular condition to recruit the required number within the study period. Several hospitals could collectively provide access to an adequate pool of eligible participants.
This is particularly relevant to studies involving uncommon conditions, narrowly defined eligibility criteria, or populations that are difficult to reach through one institution.
However, recruitment across sites does not automatically solve sample-size problems. Participants from the same institution may share experiences, characteristics, or exposures, creating statistical dependence. If clustering is relevant, the sample-size calculation and analysis must account for it.
Researchers should also assess whether participating sites recruit comparable populations. Increasing the number of participants may introduce additional variation in case mix, referral pathways, or eligibility assessment.
When Findings Must Be Examined Across Different Contexts
A study conducted in one setting provides evidence about that setting and, depending on its design, may support carefully qualified inferences beyond it. Nevertheless, some research questions explicitly concern whether an intervention, relationship, or process operates across different institutional conditions.
For example, a digital learning intervention might perform differently in institutions with reliable connectivity, extensive instructional support, and established learning management systems than in institutions with fewer technological resources.
Including sites with relevant differences can help researchers investigate the conditions under which findings appear consistent or variable.
This is one of the potential benefits of adding research sites, but the benefit depends on how those sites are selected and analyzed.
A convenience sample of three highly similar institutions may contribute less contextual information than a deliberately selected set of institutions representing theoretically important differences.
When the Study Is Intended to Support Implementation Across Institutions
Research evaluating programs, organizational practices, or implementation strategies may require multiple sites when the intended conclusions concern operation across institutions.
Consider an intervention intended for adoption across a network of community clinics. If the study evaluates only whether it can be delivered successfully at one highly resourced clinic, that evidence may establish local feasibility but cannot adequately characterize implementation across the network.
Including clinics with different staffing arrangements, patient populations, and operational capacities may help identify implementation barriers and contextual influences.
Yet multisite implementation research also requires clarity about what is being evaluated: the intervention itself, the strategy used to implement it, or both.
When Site-Level Effects Are Part of the Hypothesis
Some research questions involve characteristics that exist primarily at the institutional or organizational level.
Examples include whether school leadership practices are associated with teacher retention, whether hospital staffing policies relate to patient outcomes, or whether institutional research support predicts faculty research participation.
If the objective is to estimate relationships involving site-level characteristics, one site generally provides insufficient between-site information.
Moreover, two sites may permit a descriptive comparison but ordinarily provide a weak basis for estimating a generalizable site-level relationship. The required number depends on the estimand, variation between sites, design, model complexity, and desired precision.
Researchers should plan how to account for site effects before recruitment begins, particularly when observations within sites may be correlated.
When a Single Site May Be Methodologically Sufficient
Multiple sites are not needed for every research objective.
A single-site design may be appropriate when:
- The research question concerns a particular institution, program, community, or organizational process.
- The study is an in-depth qualitative case study focused on understanding one bounded setting.
- The intended population can be adequately sampled within one site.
- The purpose is an initial feasibility, pilot, or methodological investigation rather than a broad effectiveness claim.
- Adding sites would introduce contextual variation unrelated to the immediate research objective.
Single-site research can produce valuable theoretical, methodological, and practical contributions. Its adequacy depends on the fit between the question, design, evidence, and claims.
Qualitative case studies, for example, may prioritize contextual depth and analytic interpretation rather than statistical representation of institutions.
Does More Than One Site Automatically Improve Generalizability?
No. Generalizability depends on more than the number of sites.
In quantitative research, population inference depends partly on sampling procedures, coverage, selection mechanisms, and assumptions about the target population. Including several conveniently accessible institutions does not transform a nonprobability sample into a representative probability sample.
For intervention studies, transportability also depends on whether important effect modifiers and implementation conditions in the study settings resemble those in the intended target settings.
In qualitative research, researchers may instead discuss transferability by providing sufficient contextual detail for readers to assess relevance to other situations.
The appropriate claim should follow the inferential logic of the design rather than the number of locations listed in the methodology.
How Many Sites Are Enough?
There is no universal minimum that makes a study legitimately multisite or ensures adequate inference.
Two sites may be sufficient for a narrowly defined comparison between two particular institutions. They would generally be inadequate for estimating how institutional characteristics influence outcomes across a broad population of organizations.
For cluster-randomized studies, the number of clusters can be especially consequential for statistical power and valid inference. Increasing participants within existing clusters cannot always compensate for having too few clusters.
For qualitative multisite research, site selection may instead be guided by case-study logic, theoretically meaningful variation, feasibility, and the depth of evidence required.
The correct question is therefore not simply how many sites can be recruited, but how many and what kinds of sites are needed to support the planned conclusions.
What Additional Design Obligations Come With Multiple Sites?
Adding sites changes the operational and analytical demands of a study.
Researchers may need site-specific approvals, coordination agreements, harmonized eligibility criteria, staff training, secure data-sharing arrangements, and procedures for monitoring protocol deviations.
They must also decide whether recruitment procedures should be standardized and whether measurement procedures need to be identical across settings.
Where differences are unavoidable or scientifically meaningful, researchers should document them and consider their implications for interpretation.
Watch Out
Do not add sites after discovering that the initial site has produced an inconvenient result merely to obtain a preferred finding. Changes to site selection, recruitment targets, or the analysis plan should be scientifically justified, documented, and appropriately reported.
