Contents
On this page
- The precise technical definition
- The most common misinterpretation
- Information correctly conveyed by a confidence interval
- The direct relationship between a P value and a confidence interval
- Survey evidence on the frequency of misinterpretation
- Differences between a Bayesian credible interval and a confidence interval
- Recommended phrasing for four common situations
Academy · Research Data, Analytics & AI · Correct Interpretation of a 95% CI
Interpreting a 95% Confidence Interval: Definition, Common Misinterpretation and Correct Phrasing
A 95% confidence interval is a statement about the long-run performance of the method used to construct it, not a 95% probability that the true value lies within the particular interval calculated. Survey evidence indicates that most researchers misinterpret this distinction.
A 95% confidence interval is the range produced by a procedure that, if repeated across an indefinitely large number of samples drawn the same way, would contain the true population parameter in 95% of those repetitions. That frequentist definition is deliberately a statement about the long-run reliability of the method, not a statement of probability attached to the one specific interval a researcher is looking at. Once a particular interval has been calculated from a particular dataset, the true value either lies inside it or does not, and no 95% probability remains attached to that interval; the contrary reading is the single most common misinterpretation surveyed directly in the methodological literature1.
Learning objectives
- The precise technical definition of a 95% confidence interval
- Why the statement ‘there is a 95% chance the true value is in this interval’ is not strictly correct under the frequentist definition
- How a confidence interval conveys information about precision, magnitude and direction together
Common misconceptions
- “There is a 95% chance the true effect lies within this specific calculated interval.” The widely cited guide to misinterpretations by Greenland and colleagues identifies this exact phrasing as one of the most common and most persistent errors in the description of confidence intervals in the published literature. The 95% figure describes the long-run coverage rate of the method across repeated samples, not a probability statement about a single interval already calculated.
The precise technical definition
Under the standard frequentist framework used in most clinical and epidemiological research, a confidence interval is constructed from a procedure with a known long-run coverage property: if the study were repeated an indefinitely large number of times using the same sampling method, 95% of the resulting intervals would contain the true population value1. The confidence level, 95%, is a property of the procedure across hypothetical repetitions, not a probability statement about any single interval already calculated from one dataset. This is a subtle but consequential distinction: once data are collected and a specific interval is computed, that interval either does or does not contain the true value, and no further probability attaches to it under this framework1.
The most common misinterpretation
Greenland and colleagues, writing a widely cited guide to statistical misinterpretations in the European Journal of Epidemiology, list "there is a 95% probability that the true value lies within the interval" as one of the most common incorrect readings of a confidence interval appearing throughout the published biomedical literature1. They note that this error is intuitive because it corresponds to what researchers would like the interval to mean, and because a related, technically distinct concept, the Bayesian credible interval, does carry a genuine probability interpretation; the two are easy to conflate but are built from different underlying frameworks and are not generally numerically identical even when they share the same stated level1. In practice, the safer phrasing describes the interval as the range of values compatible with the observed data at the stated confidence level, or, following du Prel and colleagues' guidance for clinical researchers, as a measure of the data's compatibility with a range of possible true effects rather than a probability statement about the truth itself2.
Information correctly conveyed by a confidence interval
Despite the subtlety in its formal definition, a confidence interval remains one of the most informative single pieces of output a study can report, because it conveys three types of information that a P value alone cannot: the magnitude of the estimated effect, the precision of that estimate, and, through the position of the interval relative to the null value, the direction and strength of evidence against no effect2. Gardner and Altman's original 1986 argument for favoring confidence intervals over P values in clinical reporting rested on this point: an interval allows a reader to see immediately whether a result is both statistically compatible with an effect and large enough, or too small, to matter clinically, a judgment a single P value cannot support on its own3.
The direct relationship between a P value and a confidence interval
One concrete demonstration that a confidence interval is a calculated range rather than a probability statement is that it can be derived directly from a reported P value and point estimate, without access to the original raw data. Altman and Bland set out this method in the BMJ, showing how to back-calculate a 95% confidence interval from a published P value and effect estimate using the relevant test statistic4. That this derivation is mechanical, following directly from the same underlying test statistic that produced the P value, reinforces that the interval and the P value are two representations of the same piece of evidence rather than two independently arrived-at probabilities; neither one is "about" the truth in the sense that a Bayesian posterior probability would be.
Survey evidence on the frequency of misinterpretation
The probability misinterpretation of confidence intervals is neither rare nor merely academic; it recurs across medical and psychological research literature and in how results are explained to clinicians and the public1. Greenland and colleagues attribute the persistence of the error partly to how confidence intervals are taught, often introduced alongside, and so easily confused with, Bayesian credible intervals, and partly to the intuitive appeal of probability language when communicating uncertainty to a non-technical audience1. The practical remedy that their guide recommends is consistent: authors should describe what the interval would capture under repeated sampling, or describe it as “the range of values compatible with these data” at the stated confidence level, and reserve explicit probability language for contexts, such as a stated Bayesian analysis, where it is justified by the method used1.
Differences between a Bayesian credible interval and a confidence interval
A Bayesian credible interval starts from a prior distribution representing existing belief about a parameter, updates that prior using the observed data, and produces a posterior distribution from which the credible interval is directly read; because a probability distribution over the parameter itself is built at every step, a statement such as "there is a 95% probability the true value lies in this interval" is a legitimate, direct description of a credible interval in a way it is not for a frequentist confidence interval1. The two intervals can look numerically similar under certain conditions, particularly with a weak or uninformative prior and a large sample size, which partly explains why the distinction is readily blurred in practice; but they answer different formal questions, and a reader cannot tell which type of interval is being described from the number itself, only from how it was built and what the author explicitly says it represents. Cochrane reviews and most clinical trial reporting use the frequentist confidence interval by default, so probability language should be treated as incorrect unless a Bayesian method is explicitly named in the methods section7.
Recommended phrasing for four common situations
For a standard frequentist interval around a single estimate, the safest phrasing is “the data are compatible with true effects in this range at the 95% confidence level.” For a random-effects meta-analysis interval, it should be added that the interval describes uncertainty about the mean effect specifically, not the spread of individual study effects, and a prediction interval should be considered alongside it8. For an interval that includes the null value, the result should be stated plainly as inconclusive rather than negative, with a note on whether the study appears adequately powered to have detected a meaningful effect if one existed. For any interval compared across two different confidence levels or two different studies, both levels should be stated explicitly, and one should not be described simply as “more confident” without noting the corresponding change in width.
A worked contrast illustrates the four phrasing situations. Consider a single trial reporting a mean difference of 5 units with a 95% confidence interval of 1 to 9; the correct phrasing is that the data are compatible with true effects between 1 and 9 units at the 95% confidence level, not that there is a 95% probability the true effect lies between 1 and 91. If the same trial's result were presented instead as a Bayesian posterior with a weakly informative prior, yielding a numerically very similar 95% credible interval of, for example, 1.2 to 8.8, the probability phrasing would become legitimate, because the credible interval was built from a probability distribution over the parameter, updated from the prior using the observed data1. The two intervals can look almost identical on the page, which is why a reader cannot infer the type of interval from the numbers alone and must instead consult the methods section for an explicit statement of which framework, frequentist or Bayesian, produced it7. A manuscript that reports a confidence interval but discusses it using credible-interval-style probability language, or vice versa, has introduced the type of mismatch between its stated methods and its stated results that a careful peer reviewer should query before publication, since this mismatch is one of the more common, and most avoidable, sources of statistical miscommunication identified across the methodological literature on this topic1.
Key points
Key points
A 95% confidence interval describes the long-run coverage of the method used to construct it, not a 95% probability attached to the particular interval calculated. It should be described as the range of values compatible with the data rather than as a probability statement about the true value, and genuine probability language should be reserved for an explicitly Bayesian credible interval.
Practical application
Practical application
- Probability language should be avoided for a single calculated interval. Authors should write ‘the data are compatible with true effects in this range at the 95% confidence level’ rather than ‘there is a 95% chance the true value is here’.
- Confidence intervals should be distinguished from Bayesian credible intervals in scientific writing. Where Bayesian methods were used, this should be stated explicitly, because credible intervals carry a direct probability interpretation that confidence intervals do not.
- All three types of information provided by an interval should be reported. The magnitude, the precision implied by the width and the position of the interval relative to the null value should all be stated, rather than the interval being reduced to a single pass/fail judgement.
Frequently asked questions
Under the standard frequentist definition, no; the 95% figure describes the long-run performance of the method across repeated sampling, not a probability statement about one specific, already-calculated interval. A Bayesian credible interval is the tool that supports genuine probability language, and it is built differently.
Because the interpretation, not just the numeric bounds, shapes how confidently a result is communicated to clinicians, policymakers, or the public; overstating certainty by using probability language for a frequentist interval is one of the most common sources of statistical miscommunication identified in the methodological literature.
References
- Greenland S, Senn SJ, Rothman KJ, Carlin JB, Poole C, Goodman SN, Altman DG. Statistical tests, P values, confidence intervals, and power: a guide to misinterpretations. European Journal of Epidemiology. 2016;31(4):337-350. read 10 October 2026. View on European Journal of Epidemiology
- du Prel JB, Hommel G, Rohrig B, Blettner M. Confidence interval or p-value? Part 4 of a series on evaluation of scientific publications. Deutsches Arzteblatt International. 2009;106(19):335-339. read 10 October 2026. View on Deutsches Arzteblatt International
- Gardner MJ, Altman DG. Confidence intervals rather than P values: estimation rather than hypothesis testing. BMJ. 1986;292(6522):746-750. read 10 October 2026. View on The BMJ
- Altman DG, Bland JM. How to obtain the confidence interval from a P value. BMJ. 2011;343:d2090. read 10 October 2026. View on The BMJ
- Altman DG, Bland JM. Statistics notes: Absence of evidence is not evidence of absence. BMJ. 1995;311(7003):485. read 10 October 2026. View on The BMJ
- Wasserstein RL, Lazar NA. The ASA Statement on p-Values: Context, Process, and Purpose. The American Statistician. 2016;70(2):129-133. read 10 October 2026. View on Taylor & Francis Online
- Deeks JJ, Higgins JPT, Altman DG (editors). Chapter 10: Analysing data and undertaking meta-analyses. In: Cochrane Handbook for Systematic Reviews of Interventions. Cochrane, 2024. read 10 October 2026. View on Cochrane Training
- Riley RD, Higgins JPT, Deeks JJ. Interpretation of random effects meta-analyses. BMJ. 2011;342:d549. read 10 October 2026. View on The BMJ
- Higgins JPT, Thompson SG. Quantifying heterogeneity in a meta-analysis. Statistics in Medicine. 2002;21(11):1539-1558. read 10 October 2026. View on PubMed
- DerSimonian R, Laird N. Meta-analysis in clinical trials. Controlled Clinical Trials. 1986;7(3):177-188. read 10 October 2026. View on ScienceDirect
Every source above is a peer-reviewed article, a publisher’s record of one, or a standard issued by the body that owns it. Each carries the date it was last read. Where a standard moves, this article is revised rather than left up with a note.
How to cite this article
Pick your referencing style and copy it. Your university or target journal may vary the punctuation — check its guide if it has one.
Pubrica Academy. (2026, October 10). Interpreting a 95% Confidence Interval: Definition, Common Misinterpretation and Correct Phrasing. Pubrica. https://pubrica.com/academy/interpretation-of-confidence-intervals/
Pubrica Academy. “Interpreting a 95% Confidence Interval: Definition, Common Misinterpretation and Correct Phrasing.” Pubrica, 10 Oct. 2026, pubrica.com/academy/interpretation-of-confidence-intervals/.
Pubrica Academy (2026) Interpreting a 95% Confidence Interval: Definition, Common Misinterpretation and Correct Phrasing. Pubrica. Available at: https://pubrica.com/academy/interpretation-of-confidence-intervals/ (Accessed: 15 October 2026).
Pubrica Academy. “Interpreting a 95% Confidence Interval: Definition, Common Misinterpretation and Correct Phrasing.” Pubrica. October 10, 2026. https://pubrica.com/academy/interpretation-of-confidence-intervals/
Pubrica Academy, ‘Interpreting a 95% Confidence Interval: Definition, Common Misinterpretation and Correct Phrasing’ (Pubrica, 10 October 2026) <https://pubrica.com/academy/interpretation-of-confidence-intervals/> accessed 15 October 2026.
Pubrica Academy. Interpreting a 95% Confidence Interval: Definition, Common Misinterpretation and Correct Phrasing. [Internet]. Pubrica; 2026 Oct 10 [cited 2026 Oct 15]. Available from: https://pubrica.com/academy/interpretation-of-confidence-intervals/
Pubrica Academy, “Interpreting a 95% Confidence Interval: Definition, Common Misinterpretation and Correct Phrasing,” Pubrica, Oct. 10, 2026. [Online]. Available: https://pubrica.com/academy/interpretation-of-confidence-intervals/
@misc{/la_interpretation_95_confidence_interval_2026,
author = {{Pubrica Academy}},
title = {Interpreting a 95% Confidence Interval: Definition, Common Misinterpretation and Correct Phrasing},
howpublished = {Pubrica},
year = {2026},
month = {oct},
url = {https://pubrica.com/academy/interpretation-of-confidence-intervals/},
note = {Accessed: <span data-cite-acc="iso">2026-10-15</span>}
}This article is written to be learned from. Cite it if you are referring to what it says; if you are using the evidence behind it, cite the sources listed above instead.
Confidence Interval Wording Checklist
This checklist may be used to verify how a confidence interval is described before it is included in a manuscript.
[ ] Removed any phrasing stating a probability about this one specific interval containing the truth [ ] Used 'compatible with the data' or equivalent phrasing instead [ ] Confirmed whether the interval is frequentist (confidence) or Bayesian (credible), and labeled it correctly [ ] Reported magnitude, precision, and direction together rather than a single pass/fail read [ ] Checked that collaborators and co-authors are using consistent language across the manuscript
Pubrica AcademyWritten and maintained by Pubrica’s subject matter experts. Articles are revised when the underlying standards change, and each article states the date on which it was last reviewed.
How Pubrica can help
Pubrica’s statistical writing review team checks the language used to describe confidence intervals line by line, identifying probability-style misstatements before they reach peer review.