T-Statistic for Individual Coefficients

Graduate Depth 91 in the knowledge graph I know this Set as goal
Unlocks 2 downstream topics
hypothesis-testing inference coefficients

Core Idea

The t-statistic tⱼ = (β̂ⱼ - βⱼ₀) / se(β̂ⱼ) tests individual coefficients under normality, following a t-distribution with n - k degrees of freedom under H₀: βⱼ = βⱼ₀. It is used to construct confidence intervals and conduct significance tests on individual parameters.

Explainer

You already know from the normal linear regression model that OLS estimates β̂ are random variables — different samples would give different estimates. The question inference asks is: given the estimate we got, what can we conclude about the true population parameter β? The t-statistic is the tool that answers this question for individual coefficients, one at a time.

The formula tⱼ = (β̂ⱼ − βⱼ₀) / se(β̂ⱼ) has a clear structure. The numerator is the distance between your estimate and the null hypothesis value (usually βⱼ₀ = 0, meaning "does this variable have any effect?"). The denominator — the standard error of the estimate — measures how much sampling variability we'd expect in β̂ⱼ. Dividing by the standard error rescales the distance into units of "how many standard errors away is the estimate from the null?" If the true parameter equals the null value, this ratio follows a t-distribution with n − k degrees of freedom (n observations minus k parameters estimated), which you can look up in tables or evaluate with software. Larger absolute t-values are less likely to arise by chance when the null is true, so they constitute stronger evidence against it.

The mechanics of hypothesis testing with the t-statistic follow directly from your work on hypothesis testing in regression. Choose a significance level α (typically 5%), find the critical value t* such that P(|t| > t*) = α under the null, and reject H₀ if |tⱼ| > t*. For large samples, the t-distribution approaches the standard normal, so t* ≈ 1.96 for a two-sided test at 5%. Many regression outputs report the p-value directly — the probability of observing a t-statistic at least as extreme as the one computed, if H₀ is true. A p-value below 0.05 means the result is "statistically significant at the 5% level," which is shorthand for "we'd see a t-statistic this large less than 5% of the time if the true coefficient were zero."

The t-statistic also underlies confidence intervals: β̂ⱼ ± t* · se(β̂ⱼ) gives an interval that, in repeated samples, would contain the true βⱼ 95% of the time (for t* chosen to give 95% coverage). This is more informative than a yes/no reject/fail-to-reject decision because it shows you both the plausible range of the effect and its precision. A key caution: the t-test on individual coefficients does not tell you whether a *group* of coefficients is jointly significant — for that, you need an F-test. Testing many individual t-statistics inflates the chance of false positives, a problem that builds toward the topic of multiple testing corrections.

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 BiasCausal Inference and the Identification ProblemPotential Outcomes and the Rubin Causal ModelSelection BiasInstrumental VariablesTwo-Stage Least Squares (2SLS)Reduced Form and First-Stage EquationsTest of Overidentification: Hansen J-TestF-Statistic for Overall Model SignificanceT-Statistic for Individual Coefficients

Longest path: 92 steps · 517 total prerequisite topics

Prerequisites (4)

Leads To (1)