Concept Architecture
Concept
Theoretically, Within Subject Variation is the variability observed in repeated measurements taken from the same individual under similar conditions. It reflects biological fluctuation, measurement error and other sources of intra-individual variability, distinct from variation between different individuals. The concept is fundamental to repeated measures analysis, longitudinal studies, measurement reliability and mixed-effects modelling, where separating within-subject and between-subject sources of variation is essential for valid statistical inference.
Mathematically, within-subject variation is represented by the within-subject variance component in hierarchical and repeated measures models. It is estimated from repeated observations after accounting for systematic effects and represents the residual variability attributable to repeated measurements on the same individual. This variance component contributes directly to estimates of the standard error, intraclass correlation coefficient and overall covariance structure.
In practice, within-subject variation is estimated using repeated measurements analysed with repeated measures analysis of variance, linear mixed-effects models or variance components analysis. In health economics, it is used when analysing longitudinal healthcare costs, quality-adjusted life-years, utility scores and patient-reported outcome measures, ensuring that the correlation between repeated observations is appropriately incorporated into statistical analyses.
Purpose
Used to quantify the variability of repeated measurements within the same individual, supporting longitudinal analysis, variance components estimation, reliability assessment and mixed-effects modelling.
Mathematical Formulae
Primary Formula
��within = Var(X?? ? ??)
where:
- X?? = observation j for subject i
- ?? = mean for subject i
Supporting Formulae
Linear mixed-effects model:
Y?? = X??? + Z??u? + �??
where:
�?? ~ N(0, ��within)
Intraclass correlation coefficient:
ICC = ��between / (��between + ��within)
Related Mathematical Methods
- Variance Components Analysis
- Linear Mixed-Effects Models
- Mixed Model Repeated Measures
- Repeated Measures Analysis of Variance
- Intraclass Correlation Coefficient
- Random Effects Models
Example
A health economist measures EQ-5D utility scores for 250 patients at baseline, 3 months and 12 months. A linear mixed-effects model estimates a between-subject variance of 0.082 and a within-subject variance of 0.018. The relatively small within-subject variance indicates that repeated measurements for the same patient are more similar to one another than measurements across different patients, supporting efficient longitudinal estimation.
Excel Implementation
| Function | Example Formula | Health Economics Application |
|---|---|---|
| VAR.S | =VAR.S(B2:B4) | Calculate the variance of repeated measurements for an individual. |
| AVERAGE | =AVERAGE(B2:B4) | Calculate the individual's mean repeated measurement. |
| STDEV.S | =STDEV.S(B2:B4) | Estimate within-subject standard deviation. |
| AVERAGE | =AVERAGE(H2:H251) | Summarise within-subject variances across all study participants. |
VBA (Optional)
Automate calculation of within-subject variance components from repeated measurements and prepare outputs for longitudinal health economic analyses.
Sources
- Fitzmaurice GM, Laird NM, Ware JH. Applied Longitudinal Analysis.
- Verbeke G, Molenberghs G. Linear Mixed Models for Longitudinal Data.
- Diggle PJ, Heagerty P, Liang KY, Zeger SL. Analysis of Longitudinal Data.
- Briggs A, Claxton K, Sculpher M. Decision Modelling for Health Economic Evaluation.
- Drummond MF, et al. Methods for the Economic Evaluation of Health Care Programmes.
Related Concepts (2)
Library
Publications
1
Statistical Analysis of Cost-Effectiveness Data — Willan & Briggs, 1st Edition ed., 2006 (John Wiley & Sons)
A synthesis of statistical methods for analysing cost-effectiveness data, including net-benefit regression, confidence intervals for the ICER, cost-effectiveness acceptability curves, and covariate adjustment. Part of the Wiley Statistics in Practice series.
BookView source →
Frequently Asked Questions (6)
What is within-subject variation?
The degree to which repeated measurements on the same individual fluctuate over time, unlike between-subject variation reflecting differences across individuals.
Source: Friedman LM, Furberg CD, DeMets DL, Reboussin DM, Granger CB. Fundamentals of Clinical Trials. 5th ed. Springer; 2015. doi:10.1007/978-3-319-18539-2.
What does within-subject variation describe about one person?
Within-subject variation is the fluctuation in repeated measurements taken on the same individual over time, reflecting how much one person's readings wobble around their own typical value. It arises from genuine short-term change, measurement error, and the conditions of each occasion, and it is distinct from the differences between one person and another. It matters because it sets how reliably a single measurement pins down a person's true level, and how many readings are needed to do so. A person's own scatter over time is what it captures. Kirkwood and Sterne (2003) describe this.
Source: Kirkwood & Sterne 2003
How does within-subject variation differ from between-subject variation?
Within-subject variation is the variability in repeated measurements on the same individual, while between-subject variation is the variability across different individuals. Within-subject variation reflects fluctuation or measurement variability within a person, whereas between-subject variation reflects genuine differences between people. So the two differ in their source, with within-subject variation coming from differences within a person across measurements and between-subject variation from differences between people, and together they make up the total variability, with study designs and analyses distinguishing them because they have different implications for how outcomes are compared and how precisely effects can be estimated.
Source: Friedman, Furberg & DeMets 2015
Why does within-subject variation matter?
Within-subject variation matters because it affects the reliability of individual measurements and the precision of within-person comparisons: high within-subject variation means a single measurement may not represent a person well, and it influences the design and analysis of studies using repeated measures. So within-subject variation matters for measurement reliability and for the power of within-person designs, since large fluctuation within individuals can obscure true changes and may require multiple measurements to characterise a person accurately, which is why understanding it informs how many measurements to take and how to analyse repeated-measures data.
Source: Friedman, Furberg & DeMets 2015
What causes within-subject variation?
Within-subject variation is caused by genuine biological or behavioural fluctuation in the measured quantity over time, such as natural variation in blood pressure through the day, and by measurement variability, including imprecision of the instrument or procedure. So within-subject variation arises from both real fluctuation within the individual and error in measurement, and distinguishing these sources can be important, since real fluctuation reflects the variability of the quantity itself while measurement error reflects the reliability of the method, and both contribute to how much an individual's repeated measurements differ, affecting the interpretation of a single measurement.
Source: Friedman, Furberg & DeMets 2015
How is within-subject variation handled in analysis?
Within-subject variation is handled in analysis by using methods for repeated measures that account for it, such as mixed models that separate within-subject from between-subject variation through their structure, and by taking multiple measurements to average out fluctuation where a stable estimate of an individual is needed. So within-subject variation is handled by appropriate designs and models that recognise the repeated measurements within individuals, which allows the variation to be quantified and accounted for, and this is why repeated-measures methods distinguish within-subject from between-subject variation, since correctly modelling the within-person variability is necessary for valid inference from repeated measurements.
Source: Friedman, Furberg & DeMets 2015
Trust Record
Verified by Dr Darrin Baines
British health economist
Professional identity: darrinbaines.org
Verification date: 26 Dec 2025
Content version: 1.0.0
Canonical Identity
- Term code
- HE-ES-SA-237
Stable URI · Machine-readable · Resolvable · CC BY 4.0