BIOSTATISTICS • COMPARING MULTIPLE GROUPS

Multiple Comparisons & Family-Wise Error — Multiple comparisons and family-wise error concepts (intro)

Why testing many hypotheses simultaneously inflates your false-positive rate—and what to do about it.

Historical Context & Motivation

The problem of drawing conclusions from many simultaneous statistical tests has deep roots in the development of experimental science. As the twentieth century progressed, researchers in agriculture, medicine, and psychology increasingly designed experiments that compared not just two groups, but three, four, or even dozens. Each additional comparison introduced a subtle but critical danger: the accumulation of false positives. A single t-test with α = 0.05 carries a 5% chance of a Type I error, a risk most scientists find acceptable. But when the same researcher performs ten, twenty, or a hundred such tests, the probability that at least one test falsely rejects a true null hypothesis climbs steeply—often to near certainty.

This realization did not arrive all at once. It emerged through decades of work by statisticians who grappled with the practical challenge of analyzing multi-group experiments while maintaining scientific rigor. The timeline below traces the key intellectual milestones that gave rise to modern multiple comparison procedures and the formal concept of family-wise error rate (FWER).

1925
Fisher's ANOVA Framework
Ronald A. Fisher published Statistical Methods for Research Workers, introducing analysis of variance (ANOVA) as a global test for differences among multiple group means. ANOVA told researchers whether any groups differed, but not which ones—creating the need for follow-up comparisons.
1953
Tukey's Honest Significant Difference
John Tukey introduced his HSD test, designed to control the overall error rate when comparing all possible pairs of group means after ANOVA. This was among the first procedures to explicitly protect against inflated Type I error across a family of tests.
1959
Dunn and the Bonferroni Approach
Olive Jean Dunn formalized the application of the Bonferroni inequality to multiple comparisons, showing that dividing α by the number of tests provides a simple, conservative bound on the family-wise error rate. This "Bonferroni correction" became one of the most widely used adjustment methods in biostatistics.
1979
Holm's Step-Down Procedure
Sture Holm proposed a sequential refinement of the Bonferroni method that rejects hypotheses in order of their p-values, uniformly more powerful than the classical Bonferroni correction while still controlling FWER strongly.
1995
Benjamini–Hochberg FDR
Yoav Benjamini and Yosef Hochberg introduced the false discovery rate (FDR) framework, shifting the focus from FWER to the expected proportion of false positives among rejected hypotheses—a paradigm especially influential in genomics and high-dimensional biostatistics.

The central question these developments address is deceptively simple: How do we maintain a credible false-positive rate when our experiment involves many statistical tests? This lesson introduces the conceptual foundations—what the multiple comparisons problem is, why it matters for biostatistical inference, and how the family-wise error rate provides a principled framework for thinking about it.

Core Principles & Definitions

Before exploring correction techniques, we need a firm grasp of the foundational ideas that underlie the multiple comparisons problem. The concepts below form the vocabulary and logic you will encounter throughout biostatistics whenever more than one hypothesis is tested within a single study.

1

Type I Error (α)

The probability of rejecting a true null hypothesis in a single test. Conventionally set at 0.05, meaning a 5% chance of a false positive for each individual comparison.
2

Family of Tests

A collection of hypothesis tests that are analyzed together within the same study or experiment. The definition of a 'family' depends on the research question; it may be all pairwise comparisons among treatment groups, or all endpoints in a clinical trial.
3

Family-Wise Error Rate (FWER)

The probability of making at least one Type I error anywhere in the family of tests. Controlling FWER means ensuring this probability stays at or below a pre-specified level (typically 0.05).
4

Comparison-Wise Error Rate

The Type I error rate for each individual test, considered in isolation. When no correction is applied, the comparison-wise rate equals α, but the family-wise rate exceeds α whenever more than one comparison is made.
5

Multiplicity Adjustment

Any statistical procedure—such as Bonferroni, Tukey, or Holm—that modifies the significance threshold or p-values to account for multiple tests, thereby controlling FWER or another error metric.
KEY TAKEAWAY
Think of each hypothesis test as rolling a 20-sided die: you "false-alarm" when you roll a 1 (a 5% chance). One roll is fine—a 1 is unlikely. But if you roll the die 20 times, the probability of seeing at least one '1' jumps to about 64%. The multiple comparisons problem is exactly this: more rolls (tests) mean more chances for a false alarm, even when every null hypothesis is true. Family-wise error rate quantifies that cumulative risk, and multiplicity adjustments are the rules we impose to keep it under control.

Visualizing Error Accumulation

The most powerful way to internalize the multiple comparisons problem is to see how the family-wise error rate grows as the number of independent tests increases. The diagram below plots FWER against the number of tests, assuming each test uses a comparison-wise α of 0.05 and all null hypotheses are true.

The curve follows the formula FWER = 1 − (1 − α)m. Notice how quickly it departs from the red dashed line at α = 0.05. By m = 10 tests the probability of at least one false positive is already about 0.40, and by m = 20 it exceeds 0.64.

The diagram makes a key insight visually immediate: a researcher who tests twenty hypotheses at α = 0.05 without any correction faces a roughly 64% chance of reporting at least one spurious finding, even if every null hypothesis is true. In genomics, where thousands of genes may be tested simultaneously, uncorrected FWER approaches 1.0—meaning a false positive is virtually guaranteed. This is why multiplicity corrections are not merely an academic nicety but a practical necessity in modern biostatistical research.

Mathematical Framework

The mathematical treatment of family-wise error begins with a straightforward probability argument. Consider a family of m hypothesis tests, each conducted at significance level α. If all m null hypotheses are true and the tests are independent, the probability of correctly retaining any single null is (1 − α). The probability of correctly retaining all of them is the product of these probabilities.

FAMILY-WISE ERROR RATE (INDEPENDENT TESTS)
FWER = 1 − (1 − α)ᵐ
where α is the per-comparison significance level, m is the number of independent tests, and FWER is the probability of committing at least one Type I error across the family.

This formula assumes independence among tests, which is often approximately (but not exactly) true for pairwise comparisons derived from the same ANOVA. When tests are positively correlated—as they often are in biomedical settings—the true FWER may be somewhat lower than this formula suggests, but the qualitative lesson remains: FWER grows rapidly with m.

The Bonferroni Correction

The simplest and most widely known multiplicity adjustment is the Bonferroni correction. It derives from Boole's inequality (also called the union bound), which states that the probability of the union of events is at most the sum of their individual probabilities. Applied to hypothesis testing, it yields a corrected significance level α* that guarantees FWER ≤ α regardless of the dependence structure among tests.

BONFERRONI-CORRECTED THRESHOLD
α* = α / m
Each individual test is conducted at significance level α* instead of α. Equivalently, each raw p-value can be multiplied by m and compared to α.

For example, with m = 10 tests and α = 0.05, the Bonferroni-corrected threshold is α* = 0.05 / 10 = 0.005. Only p-values below 0.005 would be declared significant. This stringent threshold keeps FWER ≤ 0.05 but comes at a cost: reduced statistical power. Real effects may go undetected because the bar for significance has been raised.

Šidák Correction

ŠIDÁK-CORRECTED THRESHOLD
α* = 1 − (1 − α)^(1/m)
This is the exact solution when tests are independent, derived by inverting the FWER formula. For m = 10 and α = 0.05, this gives α* ≈ 0.00512, slightly more liberal than the Bonferroni value of 0.005.
📌 Bonferroni vs. Šidák
The Bonferroni correction is always valid (regardless of dependence), while the Šidák correction is exact only under independence. In practice the difference is small, but Bonferroni's universality makes it the more commonly cited method in biostatistics journals.

When & Where Multiple Comparisons Arise

Multiple comparisons are not confined to a single type of study design. They arise naturally in a variety of biostatistical scenarios, from the classic one-way ANOVA to modern high-throughput genomics. Recognizing when you are performing a family of tests is the first step toward deciding whether—and how—to adjust. The diagram below classifies common situations in biostatistics that generate multiple comparisons.

Four common scenarios in biostatistics that generate families of tests. The bottom box shows how the number of pairwise comparisons grows quadratically with k groups, following the formula C(k, 2) = k(k − 1) / 2.

The combinatorial formula C(k, 2) = k(k − 1)/2 explains why even a modest number of groups can generate a surprisingly large family of tests. With only five treatment groups in a clinical trial, a researcher who wishes to compare every pair faces 10 pairwise comparisons, yielding an uncorrected FWER of approximately 1 − (0.95)¹⁰ ≈ 0.40. In genome-wide association studies (GWAS), the situation is orders of magnitude more extreme: a million SNPs tested demands a Bonferroni threshold of roughly 5 × 10⁻⁸, the now-standard genome-wide significance level.

FWER and Bonferroni-corrected significance thresholds for k groups at α = 0.05
Number of Groups (k)Pairwise Tests C(k,2)Uncorrected FWERBonferroni α*
330.1430.0167
460.2650.0083
5100.4010.0050
6150.5370.0033
10450.9010.0011

Worked Example: Drug Trial with Four Treatment Arms

A clinical researcher compares four treatments (A, B, C, D) for reducing systolic blood pressure. After performing a one-way ANOVA and finding a significant omnibus F-test, the researcher plans all pairwise comparisons. The six resulting p-values from independent-samples t-tests are: pAB = 0.003, pAC = 0.012, pAD = 0.041, pBC = 0.087, pBD = 0.210, pCD = 0.560. Determine which comparisons remain significant after Bonferroni correction at FWER = 0.05.

Bonferroni Correction for Pairwise Comparisons
1
Step 1 — Count the Number of ComparisonsWith k = 4 groups, the number of pairwise comparisons is C(4, 2) = 4 × 3 / 2 = 6. So m = 6.
m = 6 comparisons
2
Step 2 — Compute the Bonferroni-Corrected ThresholdDivide the overall significance level by the number of comparisons: α* = α / m = 0.05 / 6 ≈ 0.00833.
α* ≈ 0.00833
3
Step 3 — Compare Each p-Value to α*Evaluate each p-value against the corrected threshold of 0.00833: pAB = 0.003 < 0.00833 → Significant pAC = 0.012 > 0.00833 → Not significant pAD = 0.041 > 0.00833 → Not significant pBC = 0.087 > 0.00833 → Not significant pBD = 0.210 > 0.00833 → Not significant pCD = 0.560 > 0.00833 → Not significant
4
Step 4 — Equivalently, Compute Adjusted p-ValuesMultiply each raw p-value by m = 6 and compare to α = 0.05: pAB(adj) = 0.003 × 6 = 0.018 < 0.05 → Significant pAC(adj) = 0.012 × 6 = 0.072 > 0.05 → Not significant The remaining adjusted p-values exceed 0.05 and confirm the same conclusion.
Only the A vs. B comparison survives the Bonferroni correction.
5
Step 5 — InterpretWithout correction, three comparisons (A–B, A–C, A–D) appeared significant at α = 0.05. After Bonferroni correction, only A vs. B retains significance. The other two findings cannot be distinguished from chance at the family-wise level. This illustrates how multiplicity adjustment protects against overinterpreting noisy data, at the cost of reduced sensitivity.
Conclusion: Treatment A differs significantly from Treatment B in reducing systolic blood pressure (adjusted p = 0.018), but no other pairwise differences survive correction.

Comparison of Common Correction Methods

The Bonferroni correction is just one of several approaches to controlling family-wise error. Different methods balance conservatism (how strongly they protect against false positives) against power (how likely they are to detect true effects). The table below compares four widely used methods at an introductory level.

Comparison of common multiplicity correction methods
MethodApproachControlsPowerBest Used When
BonferroniDivide α by mFWER (strong)Low–moderateSmall m; any dependence structure
Holm (step-down)Sequentially adjusted thresholdsFWER (strong)ModerateGeneral; uniformly more powerful than Bonferroni
Tukey HSDStudentized range distributionFWER (strong, for all pairs)Moderate–highAll pairwise means after balanced ANOVA
Benjamini–HochbergOrdered p-values × m/rankFDR (not FWER)HighLarge m (genomics); tolerates some false positives
KEY TAKEAWAY
Choosing a correction method is like choosing a security screening at an airport. Bonferroni is the strictest checkpoint: very few contraband items (false positives) get through, but many legitimate travelers (true effects) are delayed. Benjamini–Hochberg is a risk-based screening: it accepts that a small fraction of flagged items may be harmless, in exchange for processing everyone faster. The right choice depends on how costly a false positive is in your specific research context.
⚠️ A Note on FWER vs. FDR
The Benjamini–Hochberg procedure controls the false discovery rate (the expected proportion of false positives among rejected hypotheses), not the family-wise error rate. FDR is less conservative than FWER, which is why it dominates in high-dimensional settings like GWAS and RNA-seq. A full treatment of FDR is beyond this introductory lesson but will be covered in subsequent material.

Connecting to Advanced Theory

The introductory concepts presented in this lesson—FWER, the Bonferroni inequality, and the basic landscape of correction methods—form the foundation for a richer set of ideas that you will encounter in advanced biostatistics coursework. Understanding where these introductory tools sit relative to their more sophisticated counterparts helps frame what you have learned and motivates what comes next.

From introductory concepts to advanced extensions
Introductory ConceptAdvanced Extension
FWER = 1 − (1 − α)ᵐ (independence)Simes inequality, closure principle, and resampling-based FWER estimation for dependent tests
Bonferroni correction (single-step)Holm, Hochberg, and Hommel step-wise procedures; gatekeeping strategies in confirmatory trials
Tukey HSD for all pairwise meansDunnett's test (comparisons to control), Scheffé's method (arbitrary contrasts), Games–Howell for unequal variances
Family-wise error rateFalse discovery rate (FDR), positive FDR (pFDR), local FDR, Bayesian multiple testing approaches
Fixed α thresholdAdaptive α-spending functions in sequential trials (O'Brien–Fleming, Lan–DeMets)

As you advance, you will find that the core tension introduced here—between controlling false positives and maintaining power—recurs at every level of sophistication. The Bonferroni correction is often criticized as overly conservative, especially when tests are correlated, and advanced methods address this limitation through permutation testing, bootstrap resampling, or closed testing procedures that exploit the logical structure among hypotheses. The false discovery rate framework represents a fundamental philosophical shift: rather than asking 'did I make any false positive at all?' (FWER), FDR asks 'among the findings I declared significant, what fraction are false?' This question is often more aligned with the practical goals of exploratory research.

Practice Problems

PROBLEM 1CONCEPTUAL
A researcher conducts a single t-test at α = 0.05 and finds p = 0.03. She concludes significance. Now imagine she had planned 20 t-tests from the start, all at α = 0.05, with all null hypotheses true. Explain, in your own words, why the probability of observing at least one p-value below 0.05 is much higher than 5%, and state what this probability approximately equals.
PROBLEM 2BASIC CALCULATION
A one-way ANOVA with k = 5 groups yields a significant F-test. The researcher plans all pairwise post-hoc comparisons. (a) How many pairwise comparisons are there? (b) What is the Bonferroni-corrected significance threshold if FWER is to be controlled at 0.05? (c) What is the uncorrected FWER if each test uses α = 0.05?
PROBLEM 3INTERMEDIATE
A study compares three diets (Low-Fat, Mediterranean, Low-Carb) on weight loss. After ANOVA, the researcher obtains these p-values from pairwise t-tests: p₁ = 0.008 (LF vs. Med), p₂ = 0.025 (LF vs. LC), p₃ = 0.040 (Med vs. LC). Apply both the Bonferroni correction and the Holm step-down procedure at α = 0.05. Do the two methods yield different conclusions?
PROBLEM 4APPLIED
A pharmaceutical company runs a Phase III trial comparing a new drug to placebo on three co-primary endpoints: reduction in LDL cholesterol, reduction in systolic blood pressure, and change in HbA1c. The FDA requires that the drug demonstrate efficacy on all three endpoints with FWER ≤ 0.05. The observed p-values are: LDL: 0.018, BP: 0.032, HbA1c: 0.009. Using the Bonferroni correction, can the company claim the drug is effective across all three endpoints? Discuss the implications.
PROBLEM 5CRITICAL THINKING
A biostatistician argues: 'The Bonferroni correction is always too conservative because it assumes the worst-case scenario of independence among tests. In reality, comparisons from the same ANOVA are positively correlated, so the true FWER is lower than what Bonferroni predicts.' Evaluate this argument. Is the Bonferroni correction truly 'too conservative' because of independence assumptions? Under what conditions might Bonferroni actually be appropriate or even slightly anti-conservative?

Lesson Summary

When a study involves more than one hypothesis test, the multiple comparisons problem arises: the probability of making at least one Type I error (false positive) increases with each additional test. The family-wise error rate (FWER) quantifies this cumulative risk as the probability of one or more false rejections across the entire family of tests. For m independent tests at significance level α, the exact FWER is given by FWER = 1 − (1 − α)ᵐ, which climbs rapidly toward 1.0 as m grows.

To maintain a credible error rate, researchers apply multiplicity adjustments. The Bonferroni correction (α* = α / m) is the simplest and most universally valid method, though it trades statistical power for conservatism. Step-down methods like Holm's procedure improve upon Bonferroni without sacrificing FWER control. For high-dimensional problems (e.g., genomics), the false discovery rate (FDR) offers a less conservative alternative by controlling the expected proportion of false positives among discoveries rather than the probability of any false positive at all. Selecting the appropriate correction depends on the number of tests, the dependence among them, and the scientific consequences of false-positive versus false-negative errors.

Varsity Tutors • Biostatistics • Multiple Comparisons & Family-Wise Error