Asymptotic results for truncated-censored and associated data (Q2188759)

Let \(\{ Y_i ; i=1,\dots,N\} \) be a sequence of strictly stationary associated random variables (rvs) with continuous cumulative distribution function with bounded density. Let \(\{ T_i ; i=1,\dots,N\} \) and \(\{ X_i ; i=1,\dots,N\} \) be sequences of independently and identically distributed absolute continuous rvs. Assume that the sequences \(\{ T_i\}\) and \(\{ X_i\}\) are independent of the sequence \(\{ Y_i \}\). In the random left truncated and right censored (LTRC) model one can observe \((Z_i,T_i,\delta_i))\) if \(Z_i \geq T_i\), where \(Z_i = \min (Y_i,X_i)\) and \(\delta_i = I_{\{ Y_i \leq X_i\} }\) is the indicator of censoring status, and when \(Z_i < T_i\) nothing is observed. The authors derive strong uniform consistency rates the product-limit and cumulative hazard function estimates in the case of the LTRC model with associated observations. As an application, they give a strong uniform consistency rate for the kernel hazard rate function estimator.