Point and Interval Estimation

Point and interval estimation are the two basic ways of summarising what a sample tells us about an unknown population quantity. A point estimate is a single best guess - for example, the sample mean as an estimate of the true mean - while an interval estimate surrounds that guess with a range of values that plausibly contains the true quantity. Reporting both a point estimate and an interval communicates not only the magnitude of an effect but also how precisely it has been measured.

Definition

A point estimate is a single value computed from sample data to approximate an unknown population parameter; an interval estimate is a range of values, derived from the same data and a stated method, intended to contain the parameter with a specified level of confidence.

Scope

This topic covers what makes a point estimator good (such as being unbiased and efficient), how the standard error quantifies the precision of an estimate, and how point estimates are extended into interval estimates. It treats estimation as a reference methodology for designing and appraising studies, not as a clinical rule.

Core questions

• What is the single best estimate of the population quantity of interest?
• How precise is that estimate - how much would it vary across repeated samples?
• What range of values is plausibly consistent with the data?
• What properties make one estimator preferable to another?

Key concepts

• Estimator and estimate
• Population parameter
• Unbiasedness
• Efficiency and precision
• Standard error
• Sampling distribution
• Margin of error
• Maximum likelihood estimation

Mechanisms

A point estimator is a rule that maps sample data to a number approximating a parameter; the sample mean, sample proportion, and regression coefficients are common examples. Because a different sample would give a different value, every point estimate has a sampling distribution whose spread is summarised by the standard error - smaller standard errors mean more precise estimates. An interval estimate is then built by combining the point estimate with a multiple of its standard error (or, for bounded quantities such as a proportion, with exact methods like the Clopper-Pearson construction). Good estimators are typically judged on bias, efficiency, and consistency, so that as sample size grows the estimate concentrates on the true value.

Clinical relevance

Effect sizes reported in health research - mean differences, relative risks, prevalence figures - are point estimates, and their accompanying intervals tell the reader how much to trust them. Recognising that a point estimate without a measure of precision is incomplete is a core appraisal skill. This entry explains how such estimates are formed and is not a basis for individual clinical decisions.

Evidence & guidelines

Methodological guidance in the health sciences has long urged authors to present effect estimates with their precision rather than rely on significance verdicts. Gardner and Altman's influential argument for interval reporting, and the later misinterpretation guide by Greenland and colleagues, frame the conventions now expected in medical journals.

History

Point estimation was placed on a rigorous footing by Fisher's work on maximum likelihood in the 1920s, while interval estimation grew from the same period, including exact interval constructions such as the Clopper-Pearson limits for a binomial proportion in 1934. The emphasis on routinely reporting estimates with intervals in medicine was consolidated later in the twentieth century.

Key figures

• Ronald A. Fisher
• Jerzy Neyman
• Egon Pearson
• Douglas G. Altman

Seminal works

• gardner-altman-1986
• clopper-pearson-1934

Frequently asked questions

What is the difference between a point estimate and an interval estimate?

A point estimate is a single number, such as the sample mean, used as the best guess for an unknown quantity; an interval estimate is a range around it that conveys how precisely the quantity has been measured.

What does the standard error measure?

It measures the variability of an estimate across hypothetical repeated samples - in effect, the precision of the estimate. A smaller standard error means the point estimate is more tightly determined by the data.