Testing homogeneity of several exponential populations using censored data

Consider k two-parameter independent exponential populations E(mu(i), theta(i)), where mu(i) (-infinity < mu(i) < infinity) and (theta(i)(theta(i) > 0) are the location and the scale parameters, respectively, i 1/4 1, :::, k: Based on doubly Type II censored samples, separate classes of tests are proposed for two problems, namely: (i) Testing H-0(l): mu(1) = ::: = mu(k) against H-l(1) : l(1) <= ::: <= theta(k), with at least one strict inequality, under the assumption that h(1) = ::: = hk; and (ii) Testing H-s(0) : h(1) = ::: = hk, against H-s(1) : h(1) <= ::: <= hk, with at least one strict inequality. The test procedures are inverted to obtain the associated onesided simultaneous confidence intervals. To facilitate the implementation of the members of proposed classes in each case, selected values of critical constants, computed numerically, are tabulated. Simulation study has been performed to check the correctness of numerically computed critical constants and to obtain the power of some members of the proposed classes of tests. The working of the test procedures is illustrated with the help of a real-life data.