Regression to the mean
- Concept · Statistics and research methods
- Strong evidence
- Published October 20, 2026
- 11 sources
- 4 min read
In short
Related in this topic
On this page
Key facts
- Field
- Statistics and research methods
- Originators
- Francis Galton, in studies of inherited height (1886)
- First described
- 1886
- Also known as
- Regression toward the mean; Regression towards mediocrity; Statistical regression
- Topics
- How psychotherapy works
State of the evidence
Strong evidenceRegression to the mean is a ubiquitous property of repeated measurement, larger with greater measurement error and when follow-up is limited to people selected for extreme baseline values[1]. In depression, wait-list groups improve by about 10% to 15% over a few weeks[2], and untreated arms of three-armed trials change significantly from baseline[3], although such change also reflects natural course and not only regression.
Evidence labels summarise the research literature; they are not recommendations. How labels are assigned
Regression to the mean is a statistical phenomenon in which unusually large or small measurements tend to be followed by measurements that are closer to the average. It can make natural variation in repeated data look like real change[1]. It is not a force acting on people and does not mean that individuals become more average over time; it arises when measurements contain chance variation and attention is drawn to values that happen to be extreme[1].
The phenomenon matters to anyone weighing evidence about psychological treatment. When people are selected for treatment or for a study because their scores are high, the conditions that magnify the effect apply[1], and some of the improvement that follows would have occurred anyway.
Galton’s heights
The term comes from Francis Galton. In “Regression towards mediocrity in hereditary stature” he related the heights of children to the average height of their two parents, which he called the mid-parent height; parents and children had the same mean height, but the spread of heights differed[4][5]. The statistical term regression, from a Latin root meaning going back, was first used in that paper[5].
In data of this kind, measurements that are far from the average on one variable tend to be followed by values closer to the average on the other, and the same pattern appears whenever repeated measurements are imperfectly related[1].
Why extreme scores move toward the average
Every measurement of a psychological state combines a person’s underlying level with variation from the day, the situation and the instrument. A score that is extreme on one occasion is likely to be extreme partly because the variation happened to push it there; on the next occasion the variation is unlikely to push as far in the same direction, so the score tends to be less extreme. Barnett and colleagues identified two conditions that make the effect larger: greater measurement error, and following up only a subgroup selected because of its baseline values[1].
Both conditions apply when symptom measures carry error and participants are enrolled because their scores exceed a severity threshold. Under those conditions, average improvement at follow-up is expected even if nothing has been done. Regression to the mean can lead to the conclusion that a change was produced by treatment when it was produced by chance, an error that Morton and Torgerson describe as affecting clinical, public health and managerial decisions[6].
Starting treatment at the worst point
Studies of people who wait for treatment show how much change occurs without it. In 19 psychotherapy trials that randomized adult outpatients with major depression to a waiting list, scores on clinician-rated depression scales fell by an average of 11.9% over 2 to 20 weeks, and in a subsample with individual data about one in five wait-listed participants improved enough to count as a positive response[2]. A model drawing on wait-list trials and cohort studies estimated that in adults recruited from primary care, 23% of untreated cases of major depression would remit within three months and 53% within a year[7]. In trials of cognitive behavior therapy for tinnitus distress, wait-list groups improved slightly, by 3% to 8% over 6 to 12 weeks, with considerable variation between trials[8].
These figures combine regression to the mean with the natural course of a condition and with the effects of being in a study, and the designs cannot by themselves separate the three. A meta-analysis of trials with no-treatment, placebo and active arms found significant change from baseline in all three, with the untreated arms accounting on average for about a quarter of the change seen with active treatment[3]. Waiting itself may not be neutral: in depression trials, people on waiting lists fared worse than people given no treatment, raising the possibility that a waiting list acts as a nocebo condition, as discussed under placebo effect[9] .
Control groups and other defenses
The principal defense is a randomized control group. People allocated to a control condition are selected in the same way and measured at the same times as those who receive treatment, so they are subject to the same regression; the difference between the groups at follow-up is then not attributable to it. Barnett and colleagues note that the effect can also be reduced at the design stage and through suitable statistical methods[1].
Uncontrolled before-and-after comparisons offer no such protection. Change from baseline in a single treated group combines treatment effects with spontaneous improvement and placebo effects[3], and regression to the mean is one source of that spontaneous change[1]. A large improvement in an uncontrolled study, or in an individual who began treatment during a crisis, is therefore weak evidence that the treatment caused it.
Regression in judgment and in research
People do not reason about regression intuitively. Kahneman and Tversky traced fallacious intuitions about statistical regression to the representativeness heuristic, by which people predict the outcome that appears most representative of the evidence and so neglect the reliability of that evidence[10].
Researchers can fall into the same trap. In the imagination inflation procedure, participants rated how likely childhood events were to have happened to them, imagined some of the events, and rated them again; increases for imagined events had been taken as evidence that imagining an event can plant a false memory. Pezdek and Eddy replicated the increase for events first rated as unlikely but argued that the result was a statistical artifact of regression toward the mean[11].
Sources11
Numbered in order of first citation.
- 1.Barnett, A. G., van der Pols, J. C., & Dobson, A. J. (2005). Regression to the mean: What it is and how to deal with it. International Journal of Epidemiology, 34(1), 215–220. https://doi.org/10.1093/ije/dyh299
- 2.Posternak, M. A., & Miller, I. (2001). Untreated short-term course of major depression: A meta-analysis of outcomes from studies using wait-list control groups. Journal of Affective Disorders, 66(2-3), 139–146. https://doi.org/10.1016/S0165-0327(00)00304-9
- 3.Krogsbøll, L. T., Hróbjartsson, A., & Gøtzsche, P. C. (2009). Spontaneous improvement in randomised clinical trials: Meta-analysis of three-armed trials comparing no treatment, placebo and active intervention. BMC Medical Research Methodology, 9(1), 1. https://doi.org/10.1186/1471-2288-9-1
- 4.Galton, F. (1886). Regression towards mediocrity in hereditary stature. The Journal of the Anthropological Institute of Great Britain and Ireland, 15, 246. https://doi.org/10.2307/2841583
- 5.Bland, J. M., & Altman, D. G. (1994). Regression towards the mean. BMJ, 308(6942), 1499. https://doi.org/10.1136/bmj.308.6942.1499
- 6.Morton, V., & Torgerson, D. J. (2003). Effect of regression to the mean on decision making in health care. BMJ, 326(7398), 1083–1084. https://doi.org/10.1136/bmj.326.7398.1083
- 7.Whiteford, H. A., Harris, M. G., McKeon, G., Baxter, A., Pennell, C., Barendregt, J. J., & Wang, J. (2013). Estimating remission from untreated major depression: A systematic review and meta-analysis. Psychological Medicine, 43(8), 1569–1585. https://doi.org/10.1017/S0033291712001717
- 8.Hesser, H., Weise, C., Rief, W., & Andersson, G. (2011). The effect of waiting: A meta-analysis of wait-list control groups in trials for tinnitus distress. Journal of Psychosomatic Research, 70(4), 378–384. https://doi.org/10.1016/j.jpsychores.2010.12.006
- 9.Furukawa, T. A., Noma, H., Caldwell, D. M., Honyashiki, M., Shinohara, K., Imai, H., Chen, P., Hunot, V., & Churchill, R. (2014). Waiting list may be a nocebo condition in psychotherapy trials: A contribution from network meta‐analysis. Acta Psychiatrica Scandinavica, 130(3), 181–192. https://doi.org/10.1111/acps.12275
- 10.Kahneman, D., & Tversky, A. (1973). On the psychology of prediction. Psychological Review, 80(4), 237–251. https://doi.org/10.1037/h0034747
- 11.Pezdek, K., & Eddy, R. M. (2001). Imagination inflation: A statistical artifact of regression toward the mean. Memory & Cognition, 29(5), 707–718. https://doi.org/10.3758/BF03200473
Further reading
- Barnett, A. G., van der Pols, J. C., & Dobson, A. J. (2005). Regression to the mean: What it is and how to deal with it. International Journal of Epidemiology, 34(1), 215–220. https://doi.org/10.1093/ije/dyh299
- Morton, V., & Torgerson, D. J. (2003). Effect of regression to the mean on decision making in health care. BMJ, 326(7398), 1083–1084. https://doi.org/10.1136/bmj.326.7398.1083
Related entries 11
- How psychotherapy worksTopic
- Placebo effect (psychology)Concept
- Publication biasConcept
- Replication crisisConcept
- Base-rate fallacyConcept
- Dunning–Kruger effectConcept
- Gambler’s fallacyConcept
- Hot-hand fallacyConcept
- Regression (defense mechanism)Concept
- Representativeness heuristicConcept
- Cognitive biasesTopic
Cite this entry
APA 7
Psych Dictionary. (2026, October 20). Regression to the mean. In Psych Dictionary. https://www.psychdictionary.com/concepts/regression-to-the-mean
MLA 9
“Regression to the Mean.” Psych Dictionary, 20 Oct. 2026, www.psychdictionary.com/concepts/regression-to-the-mean.