Determining Sample Size for Rigorous Biological Experiments: A Practical Guide

By Dr. Zubair Khalid, DVM, MS, PhD · 2026-08-27

Quick Answer

Sample size determination requires specifying the minimum effect you intend to detect, the expected variability in your measurements, and the acceptable error rates for false positives and false negatives.
Use a formal power analysis before data collection to calculate the number of biological replicates needed, then document the assumptions in your laboratory notebook or study protocol.
A calculated sample size is an estimate based on prior data or pilot studies, and real variability can differ, so plan for feasibility checks and potential adjustments.

The Cost of Arbitrary Sample Sizes

Choosing a sample size without a structured method creates two distinct failure modes in biological research. An undersized experiment lacks the statistical sensitivity to detect a real biological effect, which leads to false negative conclusions and wasted effort. An oversized experiment consumes animals, reagents, and investigator time beyond what the scientific question requires, which raises ethical and financial concerns. Both outcomes undermine the reproducibility of life science research.

The problem is common because sample size determination sits at the intersection of statistics, biology, and logistics. A researcher must know the expected magnitude of the biological effect, the natural variation in the system, and the acceptable risk of drawing a wrong conclusion. These inputs are rarely known with precision before the experiment begins. The practical solution is to use a structured approach that makes reasonable assumptions explicit and testable.

The National Library of Medicine hosts authoritative texts on biomedical research methods that describe the statistical foundations of experimental design. These references explain why sample size is not a fixed number but a calculation that depends on the specific scientific context. The same statistical framework applies across cell culture, model organisms, and field studies.

Core Principles of Sample Size Determination

Statistical Power and Error Rates

Statistical power is the probability that a study will detect an effect when a true effect exists. Power depends on four interacting factors: the magnitude of the effect, the variability in the measurements, the sample size, and the significance threshold. Increasing sample size raises power, but the relationship is not linear. Doubling the sample size does not double the power.

The significance threshold, conventionally set at 0.05, is the probability of declaring an effect when none exists. This is the Type I error rate. The Type II error rate is the probability of missing a true effect, and power is calculated as one minus the Type II error rate. A common target is 80 percent power, which means accepting a 20 percent chance of missing a real effect.

The choice of power and significance level should be justified in the study protocol. Some experiments require higher power because the consequences of a false negative are severe, such as in safety assessments or clinical translation studies. The EQUATOR Network provides reporting guidelines that help researchers document these choices transparently.

Effect Size

Effect size is the magnitude of the biological difference or association that the experiment aims to detect. A smaller effect size requires a larger sample size to achieve the same power. The effect size is not a statistical abstraction. It is a biological judgment about what difference matters for the research question.

For example, an experiment testing whether a drug reduces tumor volume must specify the minimum reduction that would be biologically meaningful. A 10 percent reduction might be irrelevant if the goal is to identify a treatment that shrinks tumors by half. The effect size should be grounded in the scientific literature, preliminary data, or clinical relevance.

The National Institutes of Health grant application process requires a rigorous experimental design, and reviewers often evaluate whether the proposed sample size is justified by the stated effect size. A grant application that does not connect the effect size to the sample size calculation is likely to receive critical review.

Variability

Variability in the experimental system directly affects the required sample size. Higher variability means more noise in the measurements, which makes it harder to detect a true effect. Variability comes from multiple sources: biological variation among individuals, technical variation in the measurement method, and environmental variation in the experimental conditions.

The standard deviation of the measurement is the key input for sample size calculations. This value can be estimated from pilot experiments, published data, or the investigator's experience with the assay. When no estimate exists, a pilot study with a small number of samples is the most reliable way to obtain it.

The National Library of Medicine research methods texts describe how to estimate variability from pilot data and how to adjust sample size calculations when the variability estimate is uncertain. A conservative approach uses a larger variability estimate to avoid an underpowered study.

At a Glance

Factor	What It Means	How It Affects Sample Size	Practical Input
Effect size	The biological difference you want to detect	Smaller effects require more samples	Prior data, literature, clinical judgment
Variability	The spread of measurements in the population	Higher variability requires more samples	Pilot study, published values, assay validation
Significance level	The accepted false positive rate	Lower values require more samples	Conventionally 0.05, adjust for multiple comparisons
Power	The probability of detecting a true effect	Higher power requires more samples	Conventionally 0.80, adjust for study importance
Experimental design	The structure of the experiment	Paired or repeated measures designs can reduce sample size	Blocking, randomization, controls

Step by Step Approach to Sample Size Calculation

Step 1: Define the Primary Question

The sample size calculation begins with a precise statement of the primary experimental question. The question must specify the outcome variable, the comparison groups, and the direction of the expected effect. A vague question produces a vague sample size calculation.

For example, a question such as "Does the drug affect cell growth?" is not specific enough. A better question is "Does treatment with 10 micromolar drug X reduce the proliferation rate of cultured fibroblasts by at least 30 percent compared to vehicle control after 48 hours?" This statement identifies the outcome, the comparison, and the effect size.

The EQUATOR Network reporting guidelines emphasize the importance of a prespecified primary outcome. The primary outcome is the single measure that the sample size calculation is based on. Secondary outcomes may require separate considerations, but the sample size is driven by the primary outcome.

Step 2: Estimate the Effect Size

The effect size estimate should come from the best available evidence. This includes published studies with similar experimental systems, preliminary data from the laboratory, or a systematic review of the literature. The estimate should be realistic and defensible.

When no prior data exist, a pilot study is the most reliable way to estimate the effect size. The pilot study should use the same experimental conditions and measurement methods as the planned experiment. The pilot data provide an estimate of both the effect size and the variability.

The National Institutes of Health grant review process expects the effect size to be justified. Reviewers will question an effect size that is unrealistically large because it leads to a small sample size that cannot detect a smaller but still meaningful effect.

Step 3: Estimate Variability

The variability estimate is the standard deviation of the primary outcome measurement. This value should be obtained from the same experimental system and measurement method as the planned experiment. Variability from a different cell line, animal strain, or assay protocol may not apply.

The variability estimate should be based on the same unit of analysis as the experiment. If the experiment compares the mean of each treatment group, the variability is the standard deviation of individual measurements. If the experiment uses paired or repeated measures, the variability is the standard deviation of the differences.

The Research Methods Resources describe how to obtain variability estimates from pilot data and how to adjust the estimate when the pilot sample size is small. A small pilot study produces an uncertain variability estimate, and the sample size calculation should account for this uncertainty.

Step 4: Choose the Significance Level and Power

The significance level and power are the statistical error rates that the experiment is designed to control. The conventional values are 0.05 for significance and 0.80 for power. These values are not mandatory, and the choice should be justified in the study protocol.

A lower significance level, such as 0.01, is appropriate when the cost of a false positive is high. A higher power, such as 0.90 or 0.95, is appropriate when the cost of a false negative is high. The choice of these values affects the sample size directly.

The Committee on Publication Ethics core practices emphasize the importance of transparent reporting of the statistical methods used in a study. The significance level and power should be stated in the methods section of the manuscript.

Step 5: Perform the Calculation

The sample size calculation can be performed using statistical software, online calculators, or published formulas. The calculation requires the effect size, the variability, the significance level, and the power. The output is the number of biological replicates needed per group.

The calculation should be performed for the primary outcome and the primary comparison. If the experiment has multiple groups, the calculation should account for the number of comparisons. Multiple comparisons require a larger sample size to maintain the overall error rate.

The Research Methods Resources provide guidance on the statistical methods for sample size calculation, including the formulas for different experimental designs. The choice of the formula depends on the design, such as a two-group comparison, a paired design, or an analysis of variance.

Step 6: Adjust for Expected Losses

Biological experiments often lose samples during the course of the study. Animals may die, cultures may become contaminated, or samples may be lost during processing. The sample size should be increased to account for the expected loss rate.

The loss rate should be estimated from prior experience with the same experimental system. If 10 percent of the samples are expected to be lost, the sample size should be increased by 10 percent. The final sample size should be rounded up to the nearest integer.

The National Institutes of Health data management and sharing policy requires that the data from the study be shared in a way that allows others to verify the results. The sample size calculation and the expected loss rate should be documented in the study protocol.

Experimental Design Considerations

Controls and Randomization

The sample size calculation assumes that the experiment is designed to minimize bias and confounding. The control group is the reference against which the treatment effect is measured. The control group should be treated identically to the treatment group except for the experimental manipulation.

Randomization is the assignment of experimental units to treatment groups by chance. Randomization reduces the risk of systematic differences between the groups. The randomization scheme should be described in the study protocol.

The EQUATOR Network reporting guidelines describe the essential elements of a well-designed experiment, including the use of controls and randomization. The guidelines help researchers report their methods in a way that allows others to evaluate the validity of the results.

Blocking and Replication

Blocking is the arrangement of experimental units into groups that are similar to each other. Blocking reduces the effect of known sources of variability, such as the day of the experiment or the batch of the reagent. The sample size calculation should account for the blocking design.

Replication is the number of independent experimental units in each group. The replication is the sample size. The replication should be distinguished from technical replicates, which are repeated measurements of the same biological unit. Technical replicates do not increase the sample size for the biological question.

The Research Methods Resources describe the difference between biological and technical replication and the implications for the sample size calculation. The sample size is based on the number of independent biological units, not the number of technical measurements.

Multiple Comparisons

When the experiment involves multiple comparisons, the sample size calculation must account for the increased risk of false positives. The significance level should be adjusted for the number of comparisons. The adjustment can be made using a method such as the Bonferroni correction or a false discovery rate approach.

The sample size calculation should be based on the adjusted significance level. The adjusted level is lower than the unadjusted level, which increases the sample size required to achieve the same power.

The Committee on Publication Ethics core practices describe the importance of transparent reporting of the statistical methods used in the study. The method used to adjust for multiple comparisons should be stated in the methods section.

Practical Workflow for Sample Size Determination

Step 1: Define the Experimental Question

Write the primary question in a single sentence. The question should specify the outcome variable, the comparison groups, and the expected effect. This sentence is the basis for the sample size calculation.

Step 2: Gather Prior Information

Collect the effect size and variability estimates from the literature, prior data, or a pilot study. Record the source of each estimate and the conditions under which it was obtained. The estimates should be relevant to the planned experiment.

Step 3: Choose the Statistical Parameters

Select the significance level and the power for the study. The significance level is the accepted false positive rate, and the power is the probability of detecting a true effect. The values should be justified in the study protocol.

Step 4: Calculate the Sample Size

Use the statistical software to calculate the sample size. The calculation should be performed for the primary outcome and the primary design. The result is the number of biological replicates needed per group.

Step 5: Adjust for Expected Losses

Increase the sample size to account for the expected loss of data during the study. The adjustment should be based on prior experience with the same experimental system. The final sample size is the number of biological replicates to be included in the study.

Step 6: Document the Calculation

Record the sample size calculation in the study protocol or the laboratory notebook. The documentation should include the effect size, the variability estimate, the significance level, the power, and the expected loss rate. The documentation allows the calculation to be verified and reproduced.

Records and Measurements

Laboratory Notebook Documentation

The sample size calculation should be recorded in the laboratory notebook with the date, the study title, and the version of the calculation. The notebook should include the source of the effect size and variability estimates, the statistical software used, and the output of the calculation.

The ORCID for Researchers resource describes how researchers can maintain a persistent identifier that links their work across publications and datasets. The identifier helps to ensure that the study design and the sample size calculation are attributed to the correct researcher.

Study Protocol

The study protocol should include the sample size calculation and the justification for the assumptions. The protocol should be written before the experiment begins and should be updated if the assumptions change. The protocol is the reference for the data analysis and the manuscript.

The National Institutes of Health grant application process requires a study protocol that describes the experimental design and the sample size calculation. The protocol is reviewed by the grant reviewers and is part of the funding decision.

Data Management

The data management plan should describe how the data will be collected, stored, and shared. The plan should include the sample size calculation and the expected loss rate. The plan should be consistent with the NIH Data Management and Sharing Policy, which requires that the data be described in a way that allows others to verify the results.

The data management plan should be written before the experiment begins and should be updated if the data collection changes. The plan should be included in the study protocol and the grant application.

Common Failure Patterns

Underpowered Studies

The most common failure pattern is an underpowered study that cannot detect the effect of interest. The study is designed with a sample size that is too small to achieve the desired power. The result is a false negative conclusion that the treatment has no effect when the effect is real.

The cause of the underpowered study is often an unrealistic effect size estimate. The effect size is larger than the true effect, which leads to a smaller sample size than needed. The effect size should be based on the literature or the pilot data, not on the desired outcome.

Overpowered Studies

The opposite failure pattern is an overpowered study that uses more samples than needed. The study detects a statistically significant effect that is too small to be biologically meaningful. The result is a waste of resources and an ethical concern about the use of animals or other biological materials.

The cause of the overpowered study is often a variability estimate that is too large or an effect size that is too small. The sample size calculation should be based on the effect size that is biologically meaningful, not the smallest effect that can be detected.

Ignoring the Design

A third failure pattern is the sample size calculation that does not match the experimental design. The calculation is based on a simple two-group comparison, but the experiment uses a more complex design with blocking or repeated measures. The sample size is incorrect because the design is not accounted for.

The sample size calculation should be performed for the specific design that will be used in the experiment. The statistical software should be used to calculate the sample size for the design, not a simplified version of the design.

Using the Wrong Variability

A fourth failure pattern is the use of a variability estimate that does not match the experimental system. The variability is taken from a different species, a different cell line, or a different assay. The variability estimate is not relevant to the planned experiment, and the sample size is incorrect.

The variability estimate should be obtained from the same experimental system and conditions as the planned experiment. The estimate should be based on the pilot data or the published data from the same system.

Limitations and Interpretation

The Sample Size Is an Estimate

The sample size calculation is an estimate based on the assumptions about the effect size, the variability, and the error rates. The assumptions are uncertain, and the actual experiment may have a different effect size or variability. The sample size calculation should be reviewed and updated if the assumptions change.

The Research Methods Resources describe the limitations of the sample size calculation and the importance of the assumptions. The calculation is a planning tool, not a guarantee of the result.

The Effect Size Is a Judgment

The effect size is a biological judgment about the difference that matters for the research question. The judgment is based on the literature, the prior data, or the clinical reasoning. The effect size is not a statistical value that can be calculated from the data.

The Variability Is an Estimate

The variability estimate is based on the pilot data or the published data. The estimate is uncertain, especially when the pilot sample is small. The uncertainty in the variability estimate should be considered when the sample size is calculated.

Safety and Regulatory Context

Animal Welfare

The sample size calculation is an ethical issue in animal research. The number of animals used in the study should be the minimum number needed to answer the scientific question. The sample size calculation is the basis for the justification of the animal number.

The National Institutes of Health grant review guidelines require that the animal use be justified in the grant application. The justification includes the sample size calculation and the statistical power of the study.

Biosafety

The sample size calculation is not directly related to biosafety, but the experimental design should be consistent with the biosafety requirements. The number of samples and the handling procedures should be described in the biosafety protocol.

The Research Methods Resources provide guidance on the safe handling of biological materials in the research laboratory. The guidance is relevant to the experimental design and the sample size calculation.

Professional Escalation Criteria

When to Consult a Statistician

A statistician should be consulted when the experimental design is complex, the variability estimate is uncertain, or the sample size calculation is not straightforward. The statistician can help with the design of the experiment and the sample size calculation.

The statistician should be consulted before the experiment begins. The statistician can help with the design of the experiment and the sample size calculation. The statistician can also help with the analysis of the data.

When to Revise the Sample Size

The sample size should be revised when the assumptions change. The effect size estimate may change based on the pilot data, the variability estimate may change based on the data, or the error rates may change based on the study context. The sample size should be recalculated when the assumptions change.

The sample size should be revised before the experiment begins. The sample size should not be revised after the data are collected, because the revision can introduce bias into the analysis.

Decision Framework for Sample Size Selection Under Resource Constraints

The Feasibility Boundary Problem

A calculated sample size often exceeds what the laboratory can realistically support. This situation is not a failure of the statistical method. It is a decision point that requires a structured framework to balance scientific rigor against practical limits. The researcher must determine whether the experiment can be conducted at all, whether the question must be narrowed, or whether the design must be changed to reduce the required sample size.

The feasibility boundary is the point where the statistically ideal sample size meets the maximum number of biological replicates that can be produced, housed, measured, and analyzed within the available time and budget. Crossing this boundary without a structured decision process leads to one of two outcomes. The researcher runs an underpowered study and accepts a high risk of a false negative. Or the researcher abandons the question entirely and loses the investment in planning.

A structured decision framework treats the feasibility boundary as a design problem instead of a failure. The framework uses the same inputs as the original sample size calculation but applies them in a different order. Instead of asking what sample size is needed, the framework asks what experimental question can be answered with the available resources.

The Three Tier Decision Framework

The framework organizes the response to a feasibility conflict into three tiers. Each tier preserves the statistical integrity of the experiment while changing the scientific question, the experimental design, or the measurement strategy.

Tier One: Narrow the Effect Size Target

The first tier examines whether the minimum biologically meaningful effect size can be revised. The original effect size is a judgment about what difference matters. That judgment may be broader than necessary for the specific decision the experiment is meant to inform.

A researcher testing a new growth medium might have specified a 30 percent increase in cell yield as the target. If the practical decision is whether to adopt the medium for routine production, a 20 percent increase may be sufficient to justify the change. The smaller effect size requires a larger sample size, so this tier does not reduce the sample size. The tier is used when the sample size calculation produces a number that is too small to be credible or when the effect size was set unrealistically high.

The more common use of this tier is the reverse. The researcher specifies the maximum sample size that can be produced and then asks what effect size can be detected with that sample size at the chosen power and significance level. This reverse calculation defines the minimum detectable effect for the feasible sample size. If the minimum detectable effect is smaller than the biologically meaningful effect, the experiment can proceed. If the minimum detectable effect is larger, the experiment cannot answer the question and the researcher must move to the second tier.

The National Library of Medicine research methods texts describe the relationship between sample size and the minimum detectable effect. The calculation is the same power analysis used for sample size determination, but the unknown is the effect size instead of the sample size.

Tier 2: Change the Experimental Design

The second tier examines whether the experimental design can be modified to reduce the required sample size without changing the scientific question. The most effective design changes reduce the variability in the measurements or increase the information obtained from each biological replicate.

A paired design is one option. If the experiment can measure the outcome before and after treatment in the same biological unit, the analysis uses the differences within each unit. The variability of the differences is often much smaller than the variability of the raw measurements. A smaller variability estimate produces a smaller sample size for the same effect size and power.

Blocking is another design option. If the experiment can be arranged into blocks that share a known source of variability, such as the same batch of reagent or the same day of measurement, the analysis can remove that variability from the error term. The sample size calculation for a blocked design uses the reduced error variability.

A repeated measures design is a third option. If the same biological unit can be measured at multiple time points, the analysis can use the within unit correlation to increase the effective information per unit. The sample size calculation for a repeated measures design is different from a simple two group comparison and requires the correlation between repeated measurements as an additional input.

The EQUATOR Network reporting guidelines describe the design elements that must be reported for a study to be evaluated. A change from an unpaired to a paired design must be documented in the study protocol with the justification for the change.

Tier 3: Reduce the Number of Groups

The third tier reduces the number of experimental groups in the study. Each additional group increases the total number of samples required. If the experiment compares four doses of a compound against a control, the sample size is multiplied by the number of groups. Reducing the number of doses from four to two reduces the total sample size by half.

This tier requires a change in the scientific question. The researcher must decide which comparisons are essential and which can be deferred to a follow up experiment. The primary comparison should be the one that the sample size calculation is based on. The secondary comparisons may be removed or moved to a separate study.

The National Institutes of Health grant review process evaluates whether the proposed number of groups is justified by the scientific question. A study with more groups than necessary is criticized for wasting resources. A study with fewer groups than necessary is criticized for failing to answer the question.

Applying the Framework in Practice

The framework is applied in a specific sequence. The researcher first calculates the sample size using the standard approach. The researcher then compares the calculated sample size to the maximum feasible sample size. If the calculated sample size is feasible, the experiment proceeds as planned. If the calculated sample size exceeds the feasible limit, the researcher moves through the tiers in order.

The sequence matters because the tiers are not interchangeable. Narrowing the effect size target changes the scientific question. Changing the design changes the statistical analysis. Reducing the number of groups changes the scope of the study. The researcher should exhaust each tier before moving to the next.

The output of the framework is a decision record that documents the original sample size calculation, the feasibility constraint, the tier that was applied, and the final sample size. The decision record is part of the study protocol and is reported in the methods section of the manuscript.

The Decision Record

The decision record is a structured document that captures the feasibility analysis. The record has four sections. The first section states the original sample size calculation with the effect size, variability, significance level, and power. The second section states the feasibility constraint, which is the maximum number of biological replicates that can be produced and measured. The third section states the tier that was applied and the revised inputs. The fourth section states the final sample size and the minimum detectable effect at that sample size.

The decision record serves three purposes. It documents the reasoning for the sample size in the study protocol. It provides the information needed to report the study methods transparently. It creates a record that can be reviewed if the study results are questioned.

The Committee on Publication Ethics core practices require that the methods of a study be reported accurately and completely. The decision record is the evidence that the sample size was determined by a structured process instead of by convenience.

The Minimum Detectable Effect Calculation

The minimum detectable effect is the smallest effect that the feasible sample size can detect at the chosen power and significance level. The calculation uses the same inputs as the sample size calculation, but the effect size is the unknown. The output is the effect size that the experiment is powered to detect.

The minimum detectable effect is a critical number for the feasibility decision. If the minimum detectable effect is smaller than the biologically meaningful effect, the experiment can answer the question. If the minimum detectable effect is larger, the experiment cannot answer the question at the feasible sample size.

The minimum detectable effect should be recorded in the decision record. The value is used to interpret the results of the experiment. If the experiment produces a null result, the minimum detectable effect defines the largest effect that the experiment could have missed.

Common Failure Patterns in the Feasibility Decision

The first failure pattern is proceeding with an underpowered study without a decision record. The researcher knows the sample size is too small but runs the experiment anyway. The result is a false negative that is reported as a true null result. The decision record would have shown that the experiment could not detect the biologically meaningful effect.

The second failure pattern is changing the effect size after the data are collected. The researcher runs the experiment at the feasible sample size and then adjusts the effect size in the analysis to make the result significant. This practice is a form of p hacking and is not a valid use of the feasibility framework. The effect size must be specified before the data are collected.

The third failure pattern is using the feasibility framework to justify an arbitrarily small sample size. The framework is not a way to avoid the sample size calculation. The framework is a way to document the decision when the calculated sample size is not feasible. The decision record must show that the researcher attempted to meet the calculated sample size and documented the constraint.

The fourth failure pattern is ignoring the minimum detectable effect in the interpretation. The researcher runs the experiment at the feasible sample size and reports a null result without stating the minimum detectable effect. The reader cannot determine whether the null result means no effect or an effect that was too small to detect.

Professional Escalation Criteria

The feasibility framework has specific escalation criteria. The researcher should consult a statistician when the tier 2 design changes are being considered. The statistician can determine whether the paired, blocked, or repeated measures design is appropriate for the experimental system and can calculate the sample size for the revised design.

The researcher should consult the institutional animal care and use committee or the ethics board when the feasibility constraint is driven by the number of animals or human subjects. The committee can advise on the minimum number of subjects that is acceptable for the scientific question.

The researcher should consult the funding agency when the feasibility constraint is driven by the budget. The National Institutes of Health grant application process allows for a justification of the sample size in the research strategy section. The justification should include the feasibility analysis and the decision record.

The researcher should consult the data management team when the feasibility constraint is driven by the data collection or storage capacity. The NIH Data Management and Sharing Policy requires that the data management plan describe the data that will be collected and shared. The plan should be consistent with the feasible sample size.

The Role of the Pilot Study in the Feasibility Decision

The pilot study has a specific role in the feasibility decision. The pilot study provides the variability estimate that is used in the sample size calculation. The pilot study also provides the data that can be used to evaluate the feasibility of the design changes in tier 2.

A pilot study with a paired design can estimate the variability of the differences. This estimate is the input for the sample size calculation for the paired design. A pilot study with a blocked design can estimate the variability within the blocks. The pilot study should be designed to provide the inputs for the design that is being considered.

The pilot study should be documented in the decision record. The pilot study data should be reported in the study protocol and the manuscript. The ORCID for Researchers resource describes how the researcher can maintain a persistent identifier that links the pilot study data to the researcher's record.

The Feasibility Decision in the Grant Application

The feasibility analysis is a component of the grant application. The National Institutes of Health grant review process evaluates the experimental design and the sample size justification. The decision record provides the evidence that the sample size is justified by the feasibility constraints.

The grant application should include the decision record as an appendix or as part of the research strategy. The application should state the original sample size calculation, the feasibility constraint, and the tier that was applied. The application should also state the minimum detectable effect for the final sample size.

The grant reviewers will evaluate whether the feasibility analysis is reasonable. The reviewers will question a decision record that shows the effect size was changed without a biological justification. The reviewers will also question a decision record that shows the design was changed without a statistical justification.

The Feasibility Decision in the Manuscript

The feasibility decision should be reported in the methods section of the manuscript. The methods should state the sample size calculation, the feasibility constraint, and the final sample size. The methods should also state the minimum detectable effect.

The EQUATOR Network reporting guidelines provide the structure for the methods section. The guidelines require the sample size calculation to be reported with the inputs and the output. The guidelines also require the statistical methods to be reported with the software and the version.

The feasibility decision should be reported in the results section when the null result is presented. The results should state the minimum detectable effect to allow the reader to interpret the null result. The results should not state that the treatment has no effect when the experiment could only detect a large effect.

The Feasibility Decision in the Data Management Plan

The feasibility decision affects the data management plan. The data management plan should describe the data that will be collected at the final sample size. The plan should also describe the data that will be shared and the format of the shared data.

The Data Management and Sharing Policy requires the data management plan to describe the data that will be collected and shared. The plan should be consistent with the feasible sample size. The plan should also describe the metadata that will be used to describe the data.

The data management plan should be written before the experiment begins. The plan should be updated if the sample size changes during the feasibility analysis. The plan should be included in the study protocol and the grant application.

Frequently Asked Questions

What is the minimum sample size for a biological experiment?

The minimum sample size depends on the effect size, the variability, and the desired power. There is no universal minimum. A sample size of three per group may be sufficient for a large effect with low variability, but a sample size of 30 per group may be needed for a small effect with high variability. The sample size should be calculated for the specific experiment.

How do I estimate the effect size for a sample size calculation?

The effect size should be estimated from the literature, the prior data, or the pilot study. The estimate should be realistic and based on the same experimental system. The effect size is the difference that is biologically meaningful for the research question.

What is the difference between biological and technical replicates?

Biological replicates are independent biological units, such as individual animals or cell cultures. Technical replicates are repeated measurements of the same biological unit. The sample size is based on the number of biological replicates, not the number of technical replicates.

How do I account for multiple comparisons in the sample size calculation?

The significance level should be adjusted for the number of comparisons. The adjusted significance level is lower than the unadjusted level, which increases the sample size required to achieve the same power. The adjustment method should be stated in the study protocol.

What should I do if the sample size calculation is too large?

The sample size calculation may be too large for the available resources. The options are to reduce the effect size, increase the variability, or reduce the power. The sample size calculation should be revised to be feasible, but the revision should be justified.

How do I document the sample size calculation?

The sample size calculation should be documented in the study protocol and the laboratory notebook. The documentation should include the effect size estimate, the variability estimate, the significance level, the power, and the expected loss rate. The documentation should be written before the experiment begins.

What is the role of the power analysis in the grant application?

The power analysis is the basis for the justification of the sample size in the grant application. The grant reviewers will evaluate the power analysis to determine whether the proposed sample size is sufficient to answer the scientific question. The power analysis should be described in the grant application.

What should I do if the variability estimate is uncertain?

The variability estimate should be based on the pilot data or the published data. If the estimate is uncertain, the sample size should be increased to account for the uncertainty. The sample size calculation should be conservative when the variability estimate is uncertain.

Using the Evidence

Source	Best use in this topic	Important limitation
Research Methods Resources	official guidance	Check the linked page for current local requirements
EQUATOR Network	official guidance	Check the linked page for current local requirements
Core Practices	official guidance	Check the linked page for current local requirements

Related Diagnostic Guides

References and Further Reading

Research Methods Resources. National Library of Medicine.
EQUATOR Network. EQUATOR Network.
Core Practices. Committee on Publication Ethics.
NIH Grants and Funding. National Institutes of Health.
ORCID for Researchers. ORCID.
Data Management and Sharing Policy. National Institutes of Health.
Laboratory Biosafety Manual. World Health Organization.
NCBI Literature Resources. National Center for Biotechnology Information.
Evidence for 28 genetic disorders discovered by combining healthcare and research data.. Nature, 2020.
The global prevalence of myocardial infarction: a systematic review and meta-analysis.. BMC cardiovascular disorders, 2023.
Series: Practical guidance to qualitative research. Part 3: Sampling, data collection and analysis.. The European journal of general practice, 2018.

This article is educational and does not replace validated laboratory procedures, institutional biosafety review, manufacturer instructions, or professional interpretation.