No Validity Without Reliability: Measurement in Relationship Science, 1994–2024

Abstract

ABSTRACT Measurement reliability is necessary for validity and foundational to theory, inference, and cumulative knowledge in relationship science, yet concerns about questionable measurement practices persist. We examined long‐term trends in measurement reliability and reporting practices across 482 empirical articles published in three leading relationship‐focused journals ( JPSP: IRGP , JSPR , and PR ) in 1994, 2004, 2014, and 2024. Using multilevel models to account for measures nested within studies nested within articles, we assessed change over time and journal differences in internal consistency reliability, scale length, citation age of measures, sample size, sample items, and use of ad‐hoc and single‐item measures. Results revealed steady increases in internal consistency reliability (driven by mean inter‐item correlation increases), alongside accelerating growth in sample sizes and a marked decline in ad‐hoc measures after the mid‐2000s. At the same time, authors increasingly relied on older measures, and trends in scale length, single‐item usage, and sample‐item inclusion varied by journal. Despite gains in reliability, validity reporting remained rare. These findings document meaningful progress in some aspects of measurement practice while underscoring the continued neglect of validity reporting. We conclude with recommendations for researchers, reviewers, and editors aimed at strengthening measurement reporting and cumulative progress in relationship science.