Glossary

Reliability

Reliability is the consistency of a measurement instrument — the degree to which it produces stable and reproducible results under consistent conditions. A reliable scale yields similar scores when administered to the same individuals on different occasions (test-retest reliab...

Definition

Reliability is the consistency of a measurement instrument — the degree to which it produces stable and reproducible results under consistent conditions. A reliable scale yields similar scores when administered to the same individuals on different occasions (test-retest reliability), when alternate forms are used (parallel-forms reliability), or when internal consistency is high (all items measure the same underlying construct).

Why It Matters

Unreliable measurements introduce random error that obscures true relationships and reduces statistical power. Even the most sophisticated analysis cannot rescue data that are dominated by noise. Establishing reliability is therefore a prerequisite for validity: an instrument cannot accurately measure a construct if its scores fluctuate unpredictably. Reliability coefficients are required by journals, funding bodies, and ethics committees before instruments are used in primary research.

Example

A fitness tracker claims to measure resting heart rate. A physiologist measures 50 volunteers three times in one hour under identical conditions. The intraclass correlation coefficient (ICC) between the three measurements is 0.94, indicating excellent reliability. The tracker can be trusted in research settings. A second device yields an ICC of 0.45, indicating poor reliability; its data would be too noisy for any meaningful inference about cardiovascular health.

Related Terms

Software Notes

  • SPSS: Analyze > Scale > Reliability Analysis for internal consistency. Analyze > Correlate > Bivariate with Pearson r for test-retest reliability. For ICC: Analyze > Scale > Intraclass Correlation Coefficient.
  • R: psych::alpha() for internal consistency. psych::ICC(data) computes intraclass correlation coefficients. cor.test(time1, time2) for test-retest Pearson correlation.
  • Stata: alpha item1 item2 item3 for Cronbach's alpha. icc rating target rater for intraclass correlation. pwcorr time1 time2 for test-retest correlation.