Distribution of radon concentrations on the ground floor of schools and kindergartens

Integrated nine-month radon measurements conducted using passive track-etched detectors in 2771 rooms on the ground floor of 461 public school and kindergarten buildings in Montenegro, showed relatively high radon activity concentrations, with some buildings and rooms exhibiting extremely high levels. The Shapiro-Wilk statistical test confirms that the distribution of 461 arithmetic mean values of radon activity concentrations on the ground floors of these public buildings follow a log-normal distribution. However, this test rejects the hypothesis that radon activity concentrations in the 2771 sampled rooms follow the log-normal distribution. Forty-four school and kindergarten buildings contain between 15 and 30 ground-floor rooms in which radon was measured, making them suitable for analyzing the distribution type of radon activity concentrations individually. The Shapiro-Wilk (S-W) test (α = 0.05) rejects the assumed normal distribution of radon concentrations on the ground floor in 23 of these 44 buildings, as well as the log-normal distribution in 9 buildings. The Anderson-Darling (A-D) test (α = 0.05) does not reject the hypothesis that the distribution of radon activity concentrations on the ground floor in each of 44 buildings can be represented by a log-logistic distribution. A direct comparison of the p-values from the S-W test (for the log-normal) and the A-D test (for the log-logistic distribution) is not possible. Therefore, a generalized likelihood ratio (GLR) test was applied to compare the log-normal and log-logistic distribution, showing that the log-logistic distribution provided a better fit than the log-normal in five more cases. The success rate of fitting radon activity concentrations data from the ground floors of 44 larger school and kindergarten buildings was 88.6% for the log-logistic distribution, 79.5% for log-normal and 47.7% for the normal distribution.