Jackknifed variance estimators for simulation output analysis

We develop new point estimators for the variance parameter of a steady-state simulation process. The estimators are based on jackknifed versions of nonoverlapping batch means, overlapping batch means, and standardized time series variance estimators. The new estimators have reduced bias---and can be manipulated to reduce their variance and mean-squared error---compared with their predecessors, facts which we demonstrate analytically and empirically.