On the convergence rate of fixed design regression estimators for negatively associated random variables

Sharp convergence rates for fixed design regression estimators for negatively associated random variables are established. The rates are obtained by means of exponential inequalities, which hold under general conditions.