Estimating equation

An estimating equation sets an empirical average $n^{-1}{\sum }_i\psi (X_i;\theta )$ equal to zero to define an estimator of $\theta$.

For independent clusters with mean vector $m_i(\theta )$ and positive definite working covariance matrix $V_i$, a generalized estimating equation solvesCorrect mean specification and regularity can give a consistent estimator of identifiable mean coefficients despite a misspecified working covariance. A sandwich covariance matrix accounts for actual cluster variation. Mean equations alone cannot identify two parameters that always enter through the same combination.

A working correlation matrix $R_i(\alpha )$ is a positive-definite proposed correlation structure used to weight a generalized estimating equation, with $A_i$ the diagonal matrix of marginal variances. It need not equal the actual correlation matrix. Correct marginal means and independent sampling clusters keep the estimating equations unbiased; a cluster-level sandwich covariance matrix estimates their actual large-sample covariance. For three exchangeable observations, positive definiteness requires $-1/2<\rho <1$.

For independent subjects with repeated Bernoulli observations, specify marginal means $m_{ij}={\mathrm{logit}}^{-1}(w_{ij}^T\gamma )$. A generalized estimating equation uses their derivative matrix $D_i$ and working covariance matrix $V_i$ to solve ${\sum }_i{D}_i^T{V}_i^{-1}(Y_i-m_i)=0$. Subject-level sandwich covariance matrices give robust uncertainty when the mean is correctly specified but the working correlation is not. The coefficients describe population odds ratios; this mean specification does not prescribe a complete joint distribution within a subject.

For monotone missing data, let $R_{ij}$ indicate observation and let ${\rho }_{ij}$ be the product of the conditional retention probabilities through visit $j$. Under sequential missing at random, conditioning on the complete response history makes these retention probabilities depend only on the history already observed at each decision. Therefore $\mathbb{E}(R_{ij}/{\rho }_{ij}\mid Y_i,X_i)=1$. Weighting a working-independence generalized estimating equation by $R_{ij}/{\rho }_{ij}$ restores its mean-zero property. This needs correct retention probabilities, positivity and regularity; uncertainty must account for estimating the weights.

A population-averaged log-rate model with exposure $P_{it}$ gives an annual rate ratio $e^{{\beta }_1}$. A generalized estimating equation can account for repeated observations within clusters through a working correlation and sandwich covariance matrix. Independent Poisson standard errors do not account for clustering. An exposure-weighted population rate and an equally weighted average of cluster rates are different estimands.

A Z-estimator is a root of an estimating equation. Under differentiability, identification, and a central limit theorem, its first-order expansion has sandwich covariance.

Suppose random continuous real functions have unique zeros. If pointwise convergence in probability gives opposite strict limiting signs at the two ends of every sufficiently small interval about the target, the intermediate value theorem traps their zeros in those intervals with probability tending to one. The limiting function need not be continuous and the random functions need not be monotone.