Between-Subject Variation

The variability in an outcome measure observed across different individuals in a study, unlike within-subject variation in a single person's repeated measurements.

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.

Between-subject variation is the spread in an outcome measure from one individual to another, reflecting how much people naturally differ in the quantity being studied. It describes the diversity across a population, as opposed to how much a single person's own measurements fluctuate over time. This variation matters because the more people differ from each other, the larger a study must be to detect a treatment effect against that background noise. Differences from person to person are what it captures. Kirkwood and Sterne (2003) describe this.

Source: Kirkwood & Sterne 2003

Between-subject variation is the variability across different individuals, while within-subject variation is the variability in repeated measurements on the same individual. Between-subject variation reflects genuine differences between people, whereas within-subject variation reflects fluctuation or measurement variability within a person over time or occasions. So the two differ in their source: between-subject variation comes from differences between people and within-subject variation from differences within a person across measurements, 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

Between-subject variation matters because it contributes to the noise against which treatment effects must be detected: when people differ greatly, larger samples are needed to distinguish a real effect from the variability between individuals. It also affects the precision of estimates and the choice of study design. So between-subject variation matters for the power and efficiency of a study, since high variability between people makes effects harder to detect and may favour designs, such as within-subject or crossover designs, that remove between-person variation, and understanding it helps in planning sample sizes and choosing analyses that account for the differences among individuals.

Source: Friedman, Furberg & DeMets 2015

Between-subject variation is handled in analysis by choosing designs and models that account for it: parallel-group comparisons treat it as part of the error against which effects are judged, sometimes reduced by adjusting for baseline characteristics; crossover and other within-subject designs remove it by using each person as their own control; and mixed models can separate between-subject from within-subject variation explicitly. So between-subject variation is handled by design choices and statistical models that either account for or remove it, since recognising and appropriately treating the variation between individuals improves the precision and validity of estimates of treatment or exposure effects.

Source: Friedman, Furberg & DeMets 2015

Between-subject variation affects sample size because greater variability between individuals means a larger sample is needed to detect a given effect with adequate power, since the effect must be distinguished from the noise of individual differences. Sample size calculations use the expected variability, so higher between-subject variation increases the required number. So between-subject variation is a key input to sample size determination, with more variable outcomes demanding larger studies, which is why estimating the variability in advance is important for planning, and why designs that reduce between-subject variation, such as within-subject comparisons, can achieve the same power with fewer participants.

Source: Friedman, Furberg & DeMets 2015