Specification Tests: Ramsey RESET and Hausman Tests

College Depth 83 in the knowledge graph I know this Set as goal
Unlocks 1 downstream topic
specification hypothesis-testing model-diagnostics

Core Idea

Specification tests formally check whether key assumptions hold. The RESET test detects omitted nonlinearities by adding powers of fitted values; Hausman tests compare two estimators to detect endogeneity or misspecification.

Explainer

Your work on hypothesis testing in regression gave you the tools to test whether individual coefficients are zero. Specification tests take that logic up a level: instead of testing a coefficient, you test whether the model itself is correctly formulated. The two most important tools — the Ramsey RESET test and the Hausman test — each target a different type of misspecification.

The RESET test (Regression Equation Specification Error Test) addresses functional form misspecification. Your prerequisites covered omitted variable bias: if a variable belongs in the model but isn't included, OLS estimates are biased. Ramsey's insight was that you don't need to know what the omitted variable is — if the functional form is wrong, the fitted values ŷ will contain information about the missing structure. The procedure: run your original regression, save the fitted values, then add ŷ², ŷ³ (and optionally ŷ⁴) to the model and test their joint significance with an F-test. If those powers are significant, the original model is misspecified — something nonlinear belongs in the regression. The RESET test is a general-purpose alarm: it tells you something is wrong, but not what to add. It's useful as a quick check before reporting results.

The Hausman test operates on a different principle: comparing two estimators that both converge to the same value under the null hypothesis but differ under the alternative. The most common application is testing for endogeneity. OLS is efficient under exogeneity; instrumental variables (IV) is consistent even under endogeneity but less efficient. Under the null that OLS is consistent, the OLS and IV estimates should be close. If they differ systematically — which the Hausman statistic formalizes — that's evidence that OLS is inconsistent due to endogeneity, and IV should be preferred. The test statistic is (β̂_IV - β̂_OLS)'[Var(β̂_IV) - Var(β̂_OLS)]⁻¹(β̂_IV - β̂_OLS), which is chi-squared distributed under the null.

The broader lesson is that regression results should be reported alongside a suite of diagnostics, not just coefficients and standard errors. A model that passes the RESET test provides more credibility that the functional form is correct. A model where OLS and a valid IV give similar results provides evidence against endogeneity. Neither test is foolproof — the RESET test can miss certain misspecifications, and the Hausman test requires a valid instrument — but together they constitute a minimum standard for responsible empirical work.

Practice Questions 5 questions

Prerequisite Chain

Counting to 10Counting to 20Understanding ZeroThe Number ZeroCounting to FiveOne-to-One CorrespondenceCombining Small Groups Within 5Addition Within 10Addition Within 20Two-Digit Addition Without RegroupingTwo-Digit Addition with RegroupingAddition Within 100Repeated Addition as MultiplicationMultiplication Facts Within 100Division as Equal SharingDivision as Grouping (Measurement Division)Division: Grouping (Repeated Subtraction) ModelDivision: Fair Sharing ModelDivision as Equal SharingDivision as GroupingBasic Division FactsDivision Facts Within 100Two-Digit by One-Digit DivisionDivision with RemaindersRemainders and Quotients in DivisionDivision Word ProblemsIntroduction to Long DivisionFactors and MultiplesPrime and Composite NumbersEquivalent FractionsRelating Fractions and DecimalsDecimal Place ValueReading and Writing DecimalsComparing and Ordering DecimalsAdding and Subtracting DecimalsMultiplying DecimalsDividing DecimalsDividing FractionsMixed Number ArithmeticOrder of OperationsInteger Order of OperationsVariable ExpressionsCombining Like TermsOne-Step EquationsTwo-Step EquationsSolving Multi-Step EquationsEquations with Variables on Both SidesAngle Pairs: Complementary, Supplementary, and VerticalParallel Lines and TransversalsCorresponding AnglesAlternate Interior AnglesTriangle Angle Sum TheoremExterior Angle TheoremTriangle Inequality TheoremSimilar Triangles: AA SimilaritySimilar Triangles: SSS and SAS SimilarityProportions in Similar TrianglesRight Triangle Trigonometry IntroductionTrigonometric Ratios ReviewRadian MeasureConverting Between Degrees and RadiansThe Unit CircleGraphing Sine and CosineGraphing Tangent and Reciprocal Trigonometric FunctionsDerivatives of Trigonometric FunctionsAntiderivativesIndefinite IntegralsBasic Integration RulesRiemann SumsDefinite Integral DefinitionProbability Density Functions and Continuous DistributionsCumulative Distribution FunctionsContinuous Random VariablesNormal DistributionCentral Limit TheoremConfidence Intervals for MeansZ-Tests and T-Tests for MeansOne-Sample Z-Test for MeansOne-Sample and Two-Sample T-TestsOne-Way ANOVAF-Test and Joint SignificanceR-Squared and Model FitOmitted Variable BiasSpecification Tests: Ramsey RESET and Hausman Tests

Longest path: 84 steps · 422 total prerequisite topics

Prerequisites (2)

Leads To (1)