Bias and measurement errors in radioactivity data from four European radiation research laboratories

Abstract

Four radiation research laboratories participated in an interlaboratory comparison exercise to assess precision in the measurement of radionuclide concentrations. Each laboratory was required to report results for ¹³⁷Cs, ¹³⁴Cs and ⁴⁰K in accordance with their usual procedures. A normal probability plot of the pooled data showed ten of the 161 readings to be obvious outliers and seven of these were attributed to one of the laboratories. One laboratory consistently overestimated relative to the global mean, another laboratory consistently underestimated while the other two showed positive and negative bias dependent on isotope. Both bias and measurement errors were found to increase in the order ¹³⁷Cs, ¹³⁴Cs, ⁴⁰K. A need for greater standardization of analytical techniques was identified.