BIAS-CORRECTED MAXIMUM LIKELIHOOD ESTIMATION OF THE PARAMETERS OF THE TWO-PARAMETER LINDLEY DISTRIBUTION

In this paper, some bias correction methods are considered for parameter estimation of the two-parameter Lindley distribution. We have adopted a 'corrective' approach to derive modified MLEs that are bias-free to second order O(n(-2)) that index the two-parameter Lindley distribution. Efron's bootstrap resampling and Monte Carlo simulations are conducted to compare the performance between the proposed and maximum likelihood estimators (MLEs). The simulation study and applications to two real data sets have shown that the proposed estimators are extremely effective even for very small sample sizes in terms of biases and root mean squared errors.