Confidence Interval for Common Mean in Interlaboratory Comparisons with Systematic Laboratory Biases

Abstract. We consider the problem of evaluation of the measurement results from the interlaboratory comparisons in metrology. It is assumed that the laboratories have either normally, uniformly, or triangularly distributed systematic errors (biases). We propose an approximate interval estimator for the common mean, i.e. the true value of the measurand. The empirical coverage probabilities of the suggested interval estimator were estimated and compared by large Monte Carlo simulations for different experimental designs. The suggested approach is based on a metrological methodology and is fully consistent with the Supplement 1 to the Guide to the Expression of Uncertainty in Measurement Propagation of Distributions Using a Monte Carlo Method, [3].