A
Adjusted R² R² penalized for the number of predictors. Used for model comparison.
AIC Akaike Information Criterion. Model selection criterion that penalizes additional parameters.
Autocorrelation Correlation of errors across observations. Common in time series data.
Arellano-Bond GMM Dynamic panel data estimator using lagged variables as instruments.
B
BIC Bayesian Information Criterion. Model selection criterion with stronger penalty than AIC.
BLUE Best Linear Unbiased Estimator. OLS under the Gauss-Markov assumptions.
Bootstrap Resampling method for estimating standard errors and confidence intervals.
Breusch-Pagan Test Test for heteroskedasticity in regression residuals.
C
CLRM Classical Linear Regression Model. The set of assumptions underlying OLS.
Consistent Estimator converges to the true parameter value as sample size increases.
Cook's Distance Measure of the influence of an individual observation on regression estimates.
Cox PH Model Semi-parametric model for survival analysis using proportional hazards.
Cross-Validation Technique for evaluating model performance by splitting data into training and testing sets.
D
Demeaning Subtracting the entity-specific mean to remove fixed effects.
Difference-in-Differences Method for causal inference comparing treated and control groups over time.
Durbin-Watson Test for first-order serial correlation in regression residuals.
Dynamic Panel Panel model that includes lagged dependent variables.
E
Efficient Having the minimum variance among unbiased estimators.
Endogeneity Correlation between regressor and error term. Causes OLS bias.
Exogeneity Uncorrelated with the error term. Required for valid inference.
Elastic Net Regularization method combining Ridge and Lasso penalties.
F
F-statistic Test for joint significance of regression coefficients.
Fixed Effects (FE) Panel data model with entity-specific intercepts.
First-Stage F Test for instrument relevance in IV regression. F > 10 indicates strong instruments.
G
Gauss-Markov Theorem OLS is BLUE under the CLRM assumptions.
GLS Generalized Least Squares. Estimation method for models with non-spherical errors.
GMM Generalized Method of Moments. Flexible estimation framework.
H
HAC Heteroskedasticity and Autocorrelation Consistent standard errors (Newey-West).
Hausman Test Test for endogeneity or for choosing between FE and RE.
Heteroskedasticity Non-constant error variance across observations.
Homoskedasticity Constant error variance across observations.
I
Instrumental Variables (IV) Method for addressing endogeneity using exogenous instruments.
Inverse Mills Ratio Ratio of the PDF to CDF, used in selection models.
J
Jarque-Bera Test Test for normality of residuals using skewness and kurtosis.
K
Kaplan-Meier Non-parametric estimator of the survival function.
L
Lasso L1 regularized regression that performs variable selection.
LATE Local Average Treatment Effect identified by IV.
Leverage Measure of how unusual an observation's X values are.
Logit Binary choice model using the logistic distribution.
LSDV Least Squares Dummy Variable. Equivalent to Fixed Effects.
M
Marginal Effects Change in probability for a one-unit change in X.
MLE Maximum Likelihood Estimation. Method for estimating model parameters.
Moran's I Test for spatial autocorrelation.
N
Negative Binomial Count data model that allows for overdispersion.
Newey-West HAC standard errors that are robust to heteroskedasticity and autocorrelation.
Nickell Bias Bias in FE estimation of dynamic panel models with short T.
O
OLS Ordinary Least Squares. The workhorse estimator of econometrics.
Overdispersion Variance greater than the mean in count data.
Overidentification Having more instruments than endogenous variables.
P
p-value Probability of observing the test statistic if the null hypothesis is true.
Panel Data Data with both cross-sectional and time-series dimensions.
Poisson Count data model assuming mean equals variance.
Probit Binary choice model using the normal distribution.
R
R² Coefficient of determination. Fraction of variance explained by the model.
Random Effects (RE) Panel data model with random entity effects.
RESET Test Test for functional form misspecification.
Ridge L2 regularized regression that shrinks coefficients.
S
Sargan-Hansen J Test for instrument validity in overidentified models.
SE Standard error. Estimate of the standard deviation of a coefficient.
Spatial SAR Spatial Autoregressive Model with spatial lag of Y.
Spatial SEM Spatial Error Model with spatial lag of the error term.
SSE Sum of Squared Errors. Unexplained variation.
SSR Sum of Squared Regression. Explained variation.
SST Sum of Squares Total. Total variation in Y.
Studentized Residual Residual divided by its estimated standard error.
T
t-statistic Test statistic for individual coefficient significance.
Tobit Model for censored dependent variables.
2SLS Two-Stage Least Squares. The standard IV estimator.
U
Unbiased Expected value equals the true parameter value.
V
VIF Variance Inflation Factor. Measure of multicollinearity.
W
Weak Instruments Instruments with low first-stage F-statistic (F < 10).
Within Transformation Demeaning variables to remove fixed effects.
Z
ZIP Zero-Inflated Poisson model for count data with excess zeros.
ZINB Zero-Inflated Negative Binomial model.