Should Available Sample Size Influence the Methodological Approach?

Should You Change Your Research Approach Because You Cannot Recruit Enough Participants?

Suppose you want to investigate whether an AI-supported tutoring intervention improves undergraduate students' mathematics performance. You initially plan a quantitative experiment, but your institution can provide access to only 24 eligible students.

A colleague recommends switching to qualitative research because the sample is too small for quantitative analysis. Another suggests retaining the experiment because statistical tests can still be performed with 24 participants.

Neither recommendation resolves the central issue. The relevant question is whether the available participants, assignment procedures, outcome measures, and expected uncertainty can support the effect estimate you need.

Switching to interviews may produce valuable evidence about students' experiences, but it would not answer the original causal-effect question. Conversely, running a statistical test does not guarantee that the resulting estimate will be sufficiently precise or informative.

Sample size matters, but its significance depends on what the study is intended to establish.

Sample Size Should Influence Feasibility Without Automatically Determining the Approach

How Sample Size Affects Different Research Approaches

Why Is There No Universal Minimum Sample Size for Research?

Sample adequacy is determined by the research question, study design, analytical purpose, and characteristics of the evidence.

A quantitative study estimating a population proportion has different requirements from a randomized experiment estimating a small intervention effect. A qualitative phenomenological investigation has different sampling considerations from an ethnographic study involving several institutions.

Even within quantitative research, required sample sizes depend on outcome variability, expected or substantively important effect magnitude, desired precision, allocation, clustering, and analytical assumptions.

Within qualitative research, the relevance of selected participants, richness of evidence, scope of inquiry, and analytical orientation may matter more than reaching a predetermined count.

Consequently, rules such as "30 participants are enough for quantitative research" or "15 interviews are sufficient for qualitative research" are not defensible as universal methodological requirements.

A sample must be evaluated according to what it is expected to support.

How Does Sample Size Affect Quantitative Research?

In quantitative research, sample size influences the amount of information available for estimating numerical quantities and relationships.

Under appropriate sampling and modeling assumptions, larger samples generally reduce random sampling uncertainty. However, the relationship between sample size and precision depends on the design and data structure.

For example, the standard error of a sample mean under simple independent sampling is proportional to the population standard deviation divided by the square root of the sample size.

Increasing a sample from 100 to 400 independent observations can therefore approximately halve the standard error when the other relevant conditions remain unchanged.

However, this improvement does not correct biased recruitment, invalid measurement, unmeasured confounding, or other systematic errors.

Quantitative sample adequacy should therefore be assessed alongside the design's capacity to support the intended inference.

What Is the Difference Between Statistical Power and Precision?

Statistical power is the probability that a specified statistical procedure will reject a null hypothesis under a particular alternative, assuming the design and model conditions hold.

Precision concerns how narrowly a quantity can be estimated, often expressed through a confidence interval or another uncertainty measure.

These concepts are related but not identical.

For example, a study may have adequate power to detect a relatively large effect but insufficient precision to distinguish between two smaller effects that have different practical implications.

A nonsignificant result from a small study may be compatible with meaningful benefit, no effect, or harm.

Researchers should therefore consider whether the study is intended primarily to test a hypothesis, estimate an effect, or support a practical decision.

Where estimation is the primary purpose, a precision-based sample size justification may be more informative than relying exclusively on conventional statistical power.

How Do You Calculate a Sample Size for a Quantitative Study?

There is no single formula appropriate for every quantitative design.

For illustration, under simple random sampling and a large-population normal approximation, the initial sample size for estimating a population proportion with a specified margin of error may be expressed as:

n = Z²p(1 − p) / E²

Here, n is the initial sample size, Z is the relevant standard normal critical value, p is the anticipated population proportion, and E is the desired absolute margin of error.

Suppose the researcher uses a 95% confidence level, anticipates p = 0.50, and seeks a margin of error of 0.05.

The calculation is approximately:

n = 1.96² × 0.50 × 0.50 / 0.05² = 384.16

Rounding upward gives 385 completed observations under the stated approximation.

However, this calculation does not account for every practical consideration. Finite populations, unequal selection probabilities, clustering, nonresponse, and other design features may require adjustments.

It is also inappropriate to use this prevalence formula to justify sample size for a randomized intervention study, mediation model, or complex predictive analysis.

The calculation must follow the intended estimand and research design.

Does a Small Sample Automatically Make Quantitative Research Inappropriate?

No. Small quantitative studies can provide useful evidence when the question, design, and uncertainty are appropriately handled.

For example, a small randomized pilot study may investigate recruitment feasibility, intervention acceptability, adherence, or outcome variability.

A repeated-measures study may obtain multiple observations from each participant, although repeated observations do not automatically replace the information provided by additional independent participants.

Some experimental designs can estimate relatively large effects with limited samples under favorable measurement and variance conditions.

Nevertheless, a small study may be unable to estimate subtle effects precisely, examine numerous subgroups, or support complex models reliably.

The researcher should avoid treating the mere availability of a statistical test as evidence that the sample is adequate.

The relevant question is what the analysis can establish with acceptable uncertainty.

Why Does the Number of Independent Units Matter?

The number of recorded observations is not always the same as the number of independent units contributing information.

Suppose 200 students are distributed across four classrooms, and an instructional intervention is assigned at the classroom level.

Although the dataset contains 200 student outcomes, the intervention assignment involves only four clusters.

Students within the same classroom may share instructional experiences, teacher characteristics, and other influences.

An analysis that treats all 200 observations as independent may substantially understate uncertainty.

Cluster-randomized designs require attention to the number of clusters, cluster sizes, intraclass correlation, and the analytical method.

Similarly, repeated observations from the same participant cannot ordinarily be treated as independent participants.

Sample adequacy must therefore be evaluated at the level relevant to the research design and intended inference.

How Does Sample Size Affect Regression and Predictive Modeling?

Regression and predictive models require sufficient information relative to the outcome, model complexity, and analytical purpose.

A dataset containing many participants may still provide limited information when the outcome is rare.

For example, a student attrition model may contain thousands of enrolled students but relatively few attrition events.

Adding numerous predictors or interaction terms can produce unstable estimates or overfitting.

Simple rules based on a fixed number of participants or events per predictor may be inadequate because model requirements depend on several characteristics of the data and intended performance.

Researchers should consider anticipated outcome frequency, number of parameters, expected predictive performance, shrinkage, and validation requirements.

Riley and colleagues have proposed sample size methods for developing prediction models that account for these considerations rather than relying exclusively on simplistic thresholds.

A small sample may justify a simpler model, but reducing predictors should follow substantive reasoning and appropriate modeling principles.

Does Qualitative Research Require Fewer Participants?

Qualitative studies often involve fewer participants than population surveys, but that does not mean qualitative research is simply a solution for insufficient quantitative recruitment.

Qualitative sampling typically prioritizes participants or cases capable of providing relevant evidence about the phenomenon under investigation.

For example, a focused study of how experienced instructors negotiate AI-related assessment decisions may benefit from carefully selected participants who have direct experience with those decisions.

However, a study investigating several institutions, professional groups, and implementation contexts may require a broader range of participants and observations.

Qualitative adequacy depends on the research purpose and methodological approach, not merely the assumption that fewer participants are acceptable.

Researchers should not convert a quantitative effectiveness question into qualitative interviews solely because the original sample is too small.

What Is Information Power in Qualitative Sampling?

Malterud, Siersma, and Guassora introduced information power as a way to conceptualize qualitative interview sample adequacy.

The central idea is that a sample providing highly relevant information for a focused research question may require fewer participants than a sample addressing a broad and heterogeneous phenomenon.

Relevant considerations include the specificity of the study aim, the specificity of participants, use of established theory, quality of dialogue, and analytical strategy.

Information power does not produce a universal participant minimum.

For example, a narrowly focused investigation involving participants with highly specific experiences may obtain substantial relevant evidence from a relatively small sample.

A broad investigation of diverse student experiences across multiple institutions may require considerably more data.

The researcher should explain why the selected participants and collected material are adequate for the intended analytical contribution.

Does Every Qualitative Study Need to Reach Saturation?

No. Saturation is used in several qualitative traditions, but its meaning and appropriateness vary.

In some approaches, researchers assess whether additional data are producing new categories or relevant conceptual insights.

However, identifying recurring codes is not necessarily equivalent to developing a sufficiently nuanced understanding of their meanings.

Hennink and colleagues distinguish code saturation from meaning saturation, illustrating why identifying a range of issues may require less evidence than developing a detailed understanding of those issues.

Other qualitative approaches, including some forms of reflexive thematic analysis, question whether saturation is a coherent universal stopping criterion.

Researchers should therefore justify sample adequacy using principles appropriate to the selected methodology rather than automatically invoking saturation.

Can a Qualitative Study Be Too Large?

Yes. More qualitative data do not automatically produce stronger interpretations.

A very large collection of interviews may exceed the researcher's capacity for sufficiently detailed engagement with the material.

If analysis becomes superficial, additional participants may contribute little to the intended explanation.

However, large qualitative datasets can be appropriate when the research purpose, team, analytical strategy, and resources justify them.

The issue is not whether the participant count exceeds a particular threshold. It is whether the scale of evidence remains compatible with the methodological approach and analytical depth required.

How Should Sample Size Be Considered in Mixed-Methods Research?

Mixed-methods studies require sample adequacy to be assessed separately for each component and in relation to the integration strategy.

For example, a quantitative survey may require sufficient observations to estimate population characteristics, while qualitative interviews may use purposive selection to investigate experiences associated with selected patterns.

The qualitative sample need not be as large as the quantitative sample.

However, its selection must be appropriate for the intended explanatory or interpretive contribution.

Researchers should also consider how samples relate. They may involve identical, overlapping, nested, or distinct participants depending on the design.

A mixed-methods study does not become adequate simply because its total number of participants is large.

The important question is whether each component and their integration can support the intended conclusions.

This distinction is central to deciding when mixed methods is actually necessary.

Can You Combine Small Quantitative and Qualitative Samples to Make One Adequate Study?

Not automatically.

Suppose a study includes 20 quantitative survey respondents and 10 qualitative interview participants.

The total of 30 participants does not establish that either component has an adequate sample for its purpose.

The survey may lack sufficient precision or population coverage, while the interviews may or may not provide adequate interpretive evidence.

Combining the datasets does not eliminate these component-specific limitations.

Mixed-methods integration can provide complementary insights, but it does not convert inadequate numerical evidence into a reliable population estimate or insufficient qualitative evidence into a credible explanation.

What If Your Accessible Sample Is Smaller Than Your Planned Sample?

Researchers should first identify why the original sample size was required.

If the sample was calculated to achieve a specified level of precision, determine how the smaller sample changes the expected uncertainty.

If the sample was intended to support a causal comparison, assess whether the design can still estimate the relevant effect meaningfully.

If the sample is qualitative, reconsider whether the available participants and evidence adequately address the question's conceptual scope.

Possible responses include expanding recruitment, extending data collection, improving measurement efficiency, narrowing the target population, or revising the research question.

Researchers should not automatically redefine a confirmatory study as exploratory after observing disappointing results. Any changes to objectives or analysis should be disclosed transparently.

Should You Change the Research Question Instead of the Method?

Sometimes changing the question is the more defensible choice.

Suppose a researcher cannot recruit enough participants to estimate a small intervention effect with useful precision.

The study might instead investigate recruitment feasibility, intervention delivery, or participants' experiences, provided those questions are worthwhile and the available evidence can answer them.

However, this represents a change in research purpose.

Qualitative interviews about intervention experiences cannot be presented as a substitute for estimating effectiveness.

Similarly, a small descriptive quantitative study may be valuable without claiming to establish a precise population prevalence or causal effect.

The revised question should be chosen according to the evidence the feasible design can support.

How Should You Report a Study With a Limited Sample?

Report the planned and achieved sample sizes, relevant recruitment or selection procedures, and the consequences for interpretation.

In quantitative research, report estimates and uncertainty rather than relying only on statistical significance. Explain when the sample limits precision, subgroup comparisons, or model complexity.

In qualitative research, justify the adequacy of participants and material according to the methodological approach, including the relevance and depth of the evidence.

In mixed-methods research, explain the sample relationship across components and how limitations affect integration.

A limited sample does not automatically make a study unpublishable or uninformative. The problem arises when the conclusions require stronger evidence than the sample and design provide.

What Should You Do When Only 24 Students Are Available for an Intervention Study?

Suppose the researcher proceeds with a randomized pilot involving 12 students in each condition.

For illustration, assume the outcome has a standard deviation of approximately 15 points in each group and observations are independent. The standard error of an unadjusted difference between two group means would be approximately:

SE = √(15²/12 + 15²/12) ≈ 6.12 points.

Using a simple normal approximation, a 95% interval would extend roughly 12 points on either side of the observed mean difference. A small-sample analysis may produce a wider interval.

Such uncertainty could leave the findings compatible with substantially different intervention effects.

The study may still provide useful information about feasibility and preliminary outcome variability. However, the researcher should not treat the small sample as adequate for a definitive effectiveness conclusion merely because a statistical test can be performed.

Nor should the researcher switch to qualitative interviews and claim that the original effectiveness question has been answered.

Common Misconceptions About Sample Size and Research Methods

Let the Research Question Define Sample Adequacy Before Changing Methods

Begin by identifying what the study must establish and which sampling or information requirements follow from that inference.

Then compare those requirements with the participants, cases, or observations you can realistically obtain.

The most important distinction is between a sample that is small and a sample that is inadequate for the intended purpose.

A study may be useful with few participants when its scope and uncertainty are appropriately defined. A much larger study may still be inadequate when selection bias, measurement problems, or inappropriate comparisons undermine its conclusions.

Before Allowing Sample Size to Change Your Research Design, Check These Questions

Frequently Asked Questions About Sample Size and Methodological Choice

Sample Size Should Shape What You Can Establish, Not Automatically Dictate Your Method

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.