Linked from

The 33 pages that link to Multiple comparisons problem, each with the reason it gives.

• Statistical inferenceRelated: Testing many claims changes the reliability of unadjusted inferential results.

• P-valueRelated: Repeated testing can produce small p-values even when every null hypothesis is true.

• Statistical powerRelated: Adjusting for many tests often raises the sample size needed to preserve power.

• Analysis of varianceRelated: A significant omnibus ANOVA does not itself identify differing groups, and unadjusted follow-up tests inflate error.

• Statistical significanceRelated: Repeated testing can produce significant results by chance more often than a single-test threshold suggests.

• False positiveRelated: More tests create more opportunities for at least one false positive.

• Functional magnetic resonance imagingRelated: Testing thousands of brain locations requires correction to control spurious fMRI findings.

• Exploratory data analysisRelated: Repeatedly searching data for striking patterns can make chance findings appear convincing.

• A/B testingRelated: Trying many metrics or variants can make a seemingly winning result appear by chance.

• Null hypothesisRelated: Testing many null hypotheses can make nominal significance thresholds misleading without adjustment.

• Standard errorRelated: Accurate standard errors do not by themselves correct the error rate across many tests.

• StatisticsRelated: Large-scale testing requires adjustment or validation to keep apparent discoveries credible.

• Permutation testRelated: Permutation procedures can also estimate joint null behavior for adjusted comparisons.

• Statistical hypothesis testingRelated: Repeated testing can make nominal significance levels understate the overall false-positive risk.

• Frequentist statisticsRelated: Repeated testing can inflate frequentist error rates unless procedures adjust for multiplicity.

• Model selectionRelated: Searching many candidate models can make the apparently best result overly optimistic.

• Type I and type II errorsRelated: Repeated testing can raise the overall type I error rate beyond each test’s stated level.

• Egon PearsonRelated: Controlling a single test's error rate does not automatically control error across many tests.

• Margin of errorRelated: Separate margins do not automatically provide joint coverage across many reported estimates.

• BiostatisticsRelated: Large biological datasets often require many comparisons, making error control essential.

• Student's t-testRelated: Running many t-tests can inflate the chance of finding at least one misleading result.

• Null hypothesis significance testingRelated: Repeated thresholded tests can produce apparently significant results by chance.

• Frequentist inferenceRelated: Repeated testing can invalidate the error-rate guarantees of individual frequentist tests.

• Nonparametric statisticsRelated: Large collections of rank-based or exploratory tests still require error-rate control.

• 27 ClubRelated: Searching across ages and celebrity groups can make some age appear unusually dangerous by accident.

• Extrasensory perceptionRelated: Searching many outcomes can produce apparently significant ESP results by chance.

• Statistical hypothesis testRelated: Repeated tests can invalidate error guarantees unless the analysis accounts for multiplicity.

• Bonferroni inequalitiesCompared with: Its Bonferroni procedure shares the name but addresses error control, not inclusion–exclusion bounds.

• Microarray analysisRelated: Thousands of probe-wise tests require adjusted significance criteria.

• Boole's inequalityRelated: Boole's inequality bounds the chance of at least one false rejection across tests.

• Eyeball theoremRelated: Scanning many plotted patterns can make chance features look convincing.

• Law of Truly Large NumbersRelated: Testing many hypotheses makes at least one spurious result more likely.

• Sequence analysis (bioinformatics)Related: Large database searches require corrected significance estimates because they compare enormous numbers of sequences.