Regression

View at publisher Summarise with VIRAT
Authors: David J. Hand
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Regression is one of the most widely used of statistical tools, used to answer several different types of question, but it has hidden depths and can be easy to misuse, leading to mistaken or misleading conclusions. The basic regression model relates two sets of variables, an outcome set and a regressor set. In multiple regression, the outcome set is a single variable and the regressor set contains multiple variables, relating the outcome variable to a weighted sum of the regressor variables. The weights are called regression coefficients, and are estimated by measuring the discrepancy between the values of the weighted sum and observed values of the outcome for a sample of objects for which both the outcome and regressor values are known. The most popular measure of discrepancy is the sum of squares, but other measures are also used, and the choice of measure must match the aims. Regression modelling conceals many subtleties including challenges of interpreting the coefficients, multicollinearity of regressors, the phenomenon of regression to the mean, Lord's paradox, suppressor variables, omitted variable bias, and others. Extensions of the basic model include multivariate regression, multilevel models, inverse regression, path analysis, quantile regression, generalised linear models, and many others.

Fields of Study & Subjects

Statistical Methods and Applications Advanced Statistical Methods and Models Statistical Methods and Bayesian Inference

Publication details

Journal / source
Chapman & Hall/CRC eBooks
Pages
207–230
Publication date
27 March 2026
Language
EN
Document type
Book chapter