Small Sample Inference for Fixed Effects from Restricted Maximum Likelihood
The DOI matches its Crossref record; 23 of the 23 references it cites were found in a registry or an index.
Abstract
Restricted maximum likelihood (REML) is now well established as a method for estimating the parameters of the general Gaussian linear model with a structured covariance matrix, in particular for mixed linear models. Conventionally, estimates of precision and inference for fixed effects are based on their asymptotic distribution, which is known to be inadequate for some small-sample problems. In this paper, we present a scaled Wald statistic, together with an F approximation to its sampling distribution, that is shown to perform well in a range of small sample settings. The statistic uses an adjusted estimator of the covariance matrix that has reduced small sample bias. This approach has the advantage that it reproduces both the statistics and F distributions in those settings where the latter is exact, namely for Hotelling T2 type statistics and for analysis of variance F-ratios. The performance of the modified statistics is assessed through simulation studies of four different REML analyses and the methods are illustrated using three examples.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“Accurate model parameter estimations were ensured by use of the residual maximum likelihood approach with the Kenward and Roger, 1997 approximation of degrees of freedom.”
From Longitudinal association between social media use and psychological distress among adolescents (2020) · cites this paper as Kenward and Roger, 1997
In the citing paper
“Not only does the data in the motivating example have different flowering days for each genotype (see for example Fig. 2), but the proposed methodology also accounts for the different strata in the Wald test by approximating the denominator degrees of freedom (Kenward and Roger, 1997).”
From Incorporating environmental covariates to explore genotype × environment × management (G × E × M) interactions: A one-stage predictive model (2023) · cites this paper as Kenward and Roger, 1997
Citations
Cites
Maximum Likelihood from Incomplete Data Via the EM Algorithm
Dempster, A. P., Laird, N. M., & Rubin, D. B. (1977) · Journal of the Royal Statistical Society Series B: Statistical Methodology · doi:10.1111/j.2517-6161.1977.tb01600.x · Cited by 50,444
Random-Effects Models for Longitudinal Data
Laird, N. M., & Ware, J. H. (1982) · Biometrics · doi:10.2307/2529876 · Cited by 8,923
Recovery of inter-block information when block sizes are unequal
PATTERSON, H. D., & THOMPSON, R. (1971) · Biometrika · doi:10.1093/biomet/58.3.545 · Cited by 4,058
Maximum Likelihood Approaches to Variance Component Estimation and to Related Problems
Harville, D. A. (1977) · Journal of the American Statistical Association · doi:10.1080/01621459.1977.10480998 · Cited by 2,667 · cited 1×
“Restricted (or residual) maximum likelihood (REML) has become well established for the estimation of the parameters of Z (Patterson and Thompson, 1971; Harville, 1977; Thompson, 1980; Robinson, 1991) and the corresponding estimator of 3, 3 say, has a generalized least squares form that uses the REML estimate of Z.”
BackgroundThat BLUP is a Good Thing: The Estimation of Random Effects
Robinson, G. K. (1991) · Statistical Science · doi:10.1214/ss/1177011926 · Cited by 1,686
Gilmour, A. R., Thompson, R., & Cullis, B. R. (1995) · Biometrics · doi:10.2307/2533274 · Cited by 915
15 Design and Analysis of Cross-Over Trials
Kenward, M. G., & Jones, B. (2007) · Handbook of Statistics · doi:10.1016/s0169-7161(07)27015-4 · Cited by 871
Self-citationPrinciples of Multivariate Analysis
Ling, R. F. (1990) · Technometrics · doi:10.1080/00401706.1990.10484645 · Cited by 812
Satterthwaite, F. E. (1941) · Psychometrika · doi:10.1007/bf02288586 · Cited by 809
Approximations for Standard Errors of Estimators of Fixed and Random Effects in Mixed Linear Models
Kackar, R. N., & Harville, D. A. (1984) · Journal of the American Statistical Association · doi:10.1080/01621459.1984.10477102 · Cited by 445
A Method for Comparing Profiles of Repeated Measurements
Kenward, M. G. (1987) · Applied Statistics · doi:10.2307/2347788 · Cited by 302
Self-citationGiesbrecht, F. G., & Burns, J. C. (1985) · Biometrics · doi:10.2307/2530872 · Cited by 199
Mean Squared Error of Estimation or Prediction under a General Linear Model
Harville, D. A., & Jeske, D. R. (1992) · Journal of the American Statistical Association · doi:10.1080/01621459.1992.10475274 · Cited by 184
Ante-dependence Analysis of an Ordered Set of Variables
Gabriel, K. R. (1962) · The Annals of Mathematical Statistics · doi:10.1214/aoms/1177704724 · Cited by 143
Maximum likelihood estimation of variance components
Thompson, R. (1980) · Series Statistics · doi:10.1080/02331888008801560 · Cited by 90
Decomposition of Prediction Error
Harville, D. A. (1985) · Journal of the American Statistical Association · doi:10.1080/01621459.1985.10477152 · Cited by 90
The application of REML in clinical trials
Brown, H. K., & Kempton, R. A. (1994) · Statistics in Medicine · doi:10.1002/sim.4780131602 · Cited by 60
Jeske, D. R., & Harvallie, D. A. (1988) · Communication in Statistics- Theory and Methods · doi:10.1080/03610928808829672 · Cited by 49
AN ALGORITHM FOR CONSTRUCTING OPTIMAL RESOLVABLE ROW‐COLUMN DESIGNS
Nguyen, N., & Williams, E. R. (1993) · Australian Journal of Statistics · doi:10.1111/j.1467-842x.1993.tb01344.x · Cited by 46
John, J. A. (1987) · Springer eBooks · doi:10.1007/978-1-4899-3326-3_5 · Cited by 44
Hulting, F. L., & Harville, D. A. (1991) · Journal of the American Statistical Association · doi:10.1080/01621459.1991.10475080 · Cited by 43
Classical and Bayesian Prediction as Applied to an Unbalanced Mixed Linear Model
Harville, D. A., & Carriquiry, A. L. (1992) · Biometrics · doi:10.2307/2532693 · Cited by 40
Recovery of inter‐block information in cross‐over trials
Chi, E. M. (1991) · Statistics in Medicine · doi:10.1002/sim.4780100711 · Cited by 13
Cited by
Fitting Linear Mixed-Effects Models Using lme4
Bates, D., Mächler, M., Bolker, B., & Walker, S. (2015) · Journal of Statistical Software · doi:10.18637/jss.v067.i01 · Cited by 86,835
“Thus approximate methods for computing the approximate degrees of freedom for t distributions, or the denominator degrees of freedom for F statistics (Satterthwaite 1946; Kenward and Roger 1997), are at best ad hoc solutions.”
MethodElexacaftor–Tezacaftor–Ivacaftor for Cystic Fibrosis with a Single Phe508del Allele
Middleton, P. G., Mall, M. A., Dřevínek, P., Lands, L. C., McKone, E. F., Polineni, D., Ramsey, B. W., Taylor-Cousar, J. L., Tullis, E., Vermeulen, F., et al. (2019) · New England Journal of Medicine · doi:10.1056/nejmoa1908639 · Cited by 2,198
Khuri, A. I., & Mukhopadhyay, S. (2010) · WIREs Computational Statistics · doi:10.1002/wics.73 · Cited by 2,113
Evaluating significance in linear mixed-effects models in R
Luke, S. G. (2017) · Behavior Research Methods · doi:10.3758/s13428-016-0809-y · Cited by 1,704
“3 Both the Kenward-Roger (Kenward & Roger, 1997) and Satterthwaite (1941) approaches are used to estimate denominator degrees of freedom for F statistics or degrees of freedom for t statistics.”
MethodFatigue and cognitive impairment in Post-COVID-19 Syndrome: a systematic review and meta-analysis
Ceban, F., Ling, S., Lui, L. M. W., Lee, Y., Gill, H., Teopiz, K. M., Rodrigues, N. B., Subramaniapillai, M., Di Vincenzo, J. D., Cao, B., et al. (2022) · Brain, Behavior, and Immunity · doi:10.1016/j.bbi.2021.12.020 · Cited by 1,554
“In anticipation of marked heterogeneity, the meta::metaprop function (Balduzzi et al., 2019) was used to pool proportions, indicated as the number of cases exhibiting fatigue or cognitive impairment (events) divided by the size of the sample (observations), via the random-effects restricted maximum-likelihood model (R…”
Metadata
- Authors
- Kenward, M. G., & Roger, J. H.
- Affiliations
- University of Kent (GB)
- Venue
- Biometrics
- Publisher
- JSTOR
- Published
- 1997-09
- Volume
- 53
- Issue
- 3
- Pages
- 983
- ISSN
- 0006-341X (ISSN-L)
- OpenAlex
- W1981747359
- Cited by
- 4,552
- References
- 23
- Citations by year
- 2026: 217 · 2025: 196 · 2024: 216 · 2023: 213 · 2022: 261 · 2021: 299 · 2020: 287 · 2019: 270 · 2018: 247 · 2017: 240 · 2016: 286 · 2015: 256 · 2014: 217 · 2013: 189 · 2012: 218
- FWCI
- 7.02
- Type
- journal article
- Language
- en
- Open access
- closed
- Locations
- Biometrics (publishedVersion)
- PubMed (publishedVersion)
- Rothamsted Repository (Rothamsted Repository) (submittedVersion)
- Kent Academic Repository (University of Kent) (submittedVersion)
- Topics
- Physical Sciences › Mathematics › Statistics and Probability › Statistical Methods and Bayesian Inference
- Physical Sciences › Mathematics › Statistics and Probability › Advanced Statistical Methods and Models
- Physical Sciences › Mathematics › Applied Mathematics › Statistical and numerical algorithms
- Keywords
- restricted maximum likelihood, small sample inference, fixed effects, Wald statistic, covariance matrix estimation, small sample bias, Hotelling's T2 statistic, linear mixed models, linear Gaussian models
- MeSH terms
- Analysis of Variance, Biometry, Clinical Trials as Topic, Humans, Intermittent Claudication, Nitrogen Fixation, Plants, Probability, Regression Analysis, Research Design, Rhizobium, Soil Microbiology, Reproducibility of Results, Models, Statistical, Bias, Randomized Controlled Trials as Topic, Cross-Over Studies
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.