Simple Linear Estimation of the Parameters of the Logistic Distribution from a Complete or Censored Sample

Abstract

Linear unbiased estimators which maintain high efficiency relative to the best linear unbiased estimators are proposed for the parameters of the logistic distribution. In the complete sample case, the estimators of location and scale are, respectively, a trimmed mean and the mean deviation about the sample median with an adjustment for bias. The same computational forms are applied to a Windsorized sample in the cases of symmetric and upper one-sided censoring.