Simple linear regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Read the full entry →Research Wiki · Methods Lab
Regression techniques
Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis.
Methods family
Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis.
Use methods only after defining the estimand, data structure, assumptions, validation plan and decision consequence.
Multiple linear regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Read the full entry →Hierarchical regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Read the full entry →Stepwise regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Read the full entry →Forward-selection regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Read the full entry →Backward-elimination regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Read the full entry →Polynomial regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Read the full entry →Piecewise regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Read the full entry →Segmented regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Spline regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Robust regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Quantile regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Median regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Weighted least squares is a statistical or analytical concept within regression techniques. It should be selected for the data-generating process and decision question rather than because software makes it available.
Generalised least squares is a statistical or analytical concept within regression techniques. It should be selected for the data-generating process and decision question rather than because software makes it available.
Orthogonal regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Deming regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Binary logistic regression models the log-odds of a two-category outcome as a function of predictors. Coefficients are commonly interpreted through odds ratios or predicted probabilities.
Multinomial logistic regression models an unordered outcome with more than two categories relative to a reference category. It estimates separate predictor effects for each comparison.
Ordinal logistic regression models an ordered categorical outcome while using the rank order of categories. The proportional-odds assumption should be assessed where applicable.
Probit regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Complementary log-log regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Conditional logistic regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Nested logit is a statistical or analytical concept within regression techniques. It should be selected for the data-generating process and decision question rather than because software makes it available.
Mixed logit is a statistical or analytical concept within regression techniques. It should be selected for the data-generating process and decision question rather than because software makes it available.
Latent-class logit is a statistical or analytical concept within regression techniques. It should be selected for the data-generating process and decision question rather than because software makes it available.
Poisson regression models count outcomes with a log link and assumes a relationship between the conditional mean and variance. Overdispersion often motivates negative-binomial or robust alternatives.
Negative-binomial regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Zero-inflated Poisson is a statistical or analytical concept within regression techniques. It should be selected for the data-generating process and decision question rather than because software makes it available.
Zero-inflated negative binomial is a statistical or analytical concept within regression techniques. It should be selected for the data-generating process and decision question rather than because software makes it available.
Hurdle models is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Tobit regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Censored regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Truncated regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Ridge regression adds an L2 penalty that shrinks coefficients to reduce variance and handle multicollinearity. It generally retains all predictors rather than setting coefficients exactly to zero.
LASSO regression adds an L1 penalty that can shrink some coefficients to zero, combining regularisation with variable selection. Results depend on penalty tuning and correlated predictors.
Elastic Net is a statistical or analytical concept within regression techniques. It should be selected for the data-generating process and decision question rather than because software makes it available.
Adaptive LASSO is a statistical or analytical concept within regression techniques. It should be selected for the data-generating process and decision question rather than because software makes it available.
Group LASSO is a statistical or analytical concept within regression techniques. It should be selected for the data-generating process and decision question rather than because software makes it available.
Generalised linear models is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Generalised additive models is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Multilevel regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Hierarchical linear modelling is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Mixed-effects models is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Random-effects models is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Fixed-effects models is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Panel regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Bayesian regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Read the full entry →Gaussian-process regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Partial least-squares regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Principal-component regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Seemingly unrelated regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Endogenous-treatment regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Instrumental-variable regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Two-stage least squares is a statistical or analytical concept within regression techniques. It should be selected for the data-generating process and decision question rather than because software makes it available.
Regression discontinuity is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Difference-in-differences estimates an intervention effect by comparing changes over time between treated and comparison groups. Its credibility relies heavily on the parallel-trends assumption and the absence of differential shocks.
Browse linked method entries
Simple linear regression
Simple linear regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodMultiple linear regression
Multiple linear regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodHierarchical regression
Hierarchical regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodStepwise regression
Stepwise regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodForward-selection regression
Forward-selection regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodBackward-elimination regression
Backward-elimination regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodPolynomial regression
Polynomial regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodPiecewise regression
Piecewise regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodSegmented regression
Segmented regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodSpline regression
Spline regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodRobust regression
Robust regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodQuantile regression
Quantile regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodMedian regression
Median regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodWeighted least squares
Weighted least squares is a statistical or analytical concept within regression techniques. It should be selected for the data-generating process and decision question rather than because software makes it available.
Open methodGeneralised least squares
Generalised least squares is a statistical or analytical concept within regression techniques. It should be selected for the data-generating process and decision question rather than because software makes it available.
Open methodOrthogonal regression
Orthogonal regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodDeming regression
Deming regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodBinary logistic regression
Binary logistic regression models the log-odds of a two-category outcome as a function of predictors. Coefficients are commonly interpreted through odds ratios or predicted probabilities.
Open methodMultinomial logistic regression
Multinomial logistic regression models an unordered outcome with more than two categories relative to a reference category. It estimates separate predictor effects for each comparison.
Open methodOrdinal logistic regression
Ordinal logistic regression models an ordered categorical outcome while using the rank order of categories. The proportional-odds assumption should be assessed where applicable.
Open methodProbit regression
Probit regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodComplementary log-log regression
Complementary log-log regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodConditional logistic regression
Conditional logistic regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodNested logit
Nested logit is a statistical or analytical concept within regression techniques. It should be selected for the data-generating process and decision question rather than because software makes it available.
Open methodMixed logit
Mixed logit is a statistical or analytical concept within regression techniques. It should be selected for the data-generating process and decision question rather than because software makes it available.
Open methodLatent-class logit
Latent-class logit is a statistical or analytical concept within regression techniques. It should be selected for the data-generating process and decision question rather than because software makes it available.
Open methodPoisson regression
Poisson regression models count outcomes with a log link and assumes a relationship between the conditional mean and variance. Overdispersion often motivates negative-binomial or robust alternatives.
Open methodNegative-binomial regression
Negative-binomial regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodZero-inflated Poisson
Zero-inflated Poisson is a statistical or analytical concept within regression techniques. It should be selected for the data-generating process and decision question rather than because software makes it available.
Open methodZero-inflated negative binomial
Zero-inflated negative binomial is a statistical or analytical concept within regression techniques. It should be selected for the data-generating process and decision question rather than because software makes it available.
Open methodHurdle models
Hurdle models is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodTobit regression
Tobit regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodCensored regression
Censored regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodTruncated regression
Truncated regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodRidge regression
Ridge regression adds an L2 penalty that shrinks coefficients to reduce variance and handle multicollinearity. It generally retains all predictors rather than setting coefficients exactly to zero.
Open methodLASSO regression
LASSO regression adds an L1 penalty that can shrink some coefficients to zero, combining regularisation with variable selection. Results depend on penalty tuning and correlated predictors.
Open methodElastic Net
Elastic Net is a statistical or analytical concept within regression techniques. It should be selected for the data-generating process and decision question rather than because software makes it available.
Open methodAdaptive LASSO
Adaptive LASSO is a statistical or analytical concept within regression techniques. It should be selected for the data-generating process and decision question rather than because software makes it available.
Open methodGroup LASSO
Group LASSO is a statistical or analytical concept within regression techniques. It should be selected for the data-generating process and decision question rather than because software makes it available.
Open methodGeneralised linear models
Generalised linear models is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodGeneralised additive models
Generalised additive models is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodMultilevel regression
Multilevel regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodHierarchical linear modelling
Hierarchical linear modelling is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodMixed-effects models
Mixed-effects models is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodRandom-effects models
Random-effects models is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodFixed-effects models
Fixed-effects models is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodPanel regression
Panel regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodBayesian regression
Bayesian regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodGaussian-process regression
Gaussian-process regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodPartial least-squares regression
Partial least-squares regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodPrincipal-component regression
Principal-component regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodSeemingly unrelated regression
Seemingly unrelated regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodEndogenous-treatment regression
Endogenous-treatment regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodInstrumental-variable regression
Instrumental-variable regression is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodTwo-stage least squares
Two-stage least squares is a statistical or analytical concept within regression techniques. It should be selected for the data-generating process and decision question rather than because software makes it available.
Open methodRegression discontinuity
Regression discontinuity is a method within regression techniques. Models that estimate relationships between an outcome and one or more predictors for explanation, control, prediction or causal analysis. Its usefulness depends on data structure, assumptions, validation and whether the output answers the intended decision.
Open methodDifference-in-differences regression
Difference-in-differences estimates an intervention effect by comparing changes over time between treated and comparison groups. Its credibility relies heavily on the parallel-trends assumption and the absence of differential shocks.
Open methodNo entries match this search.