*2.4. Statistical Analysis of the Data*

In order to perform a statistical analysis of the concentrations experienced by the children in the different microenvironments (in terms of PN and LDSA) a preliminary normality test (Shapiro–Wilk test) was performed to check for the statistical distribution of the data. Since the data did not meet the assumptions of Gaussian distribution, non-parametric tests and further post-hoc tests (Kruskal–Wallis test [88]) were considered in the analysis. The statistically significant result was referred to a significance level of 99% (a *p*-value < 0.01). In particular, the Kruskal–Wallis tests were performed (a) amongst the six different microenvironments for each group of children separately (S1, S2, and S3; thus 3 non-parametric tests and further post-hoc tests) and (b) amongst the three groups of children for each microenvironment separately (six microenvironments plus the whole day data, thus seven non-parametric tests and further post-hoc tests).

The PN and LDSA concentration data considered in the statistical analysis, and then shown in the result section, included all the data provided by the instrument (roughly 48 h of total sampling per each child with a sampling frequency of 10 s), thus, a huge number of values were available for each microenvironment of each children group.

On the contrary, the dose values reported and discussed in the results represented the median values (and corresponding ranges) obtained from the 20-dose data (i.e., 20 children) per each microenvironment per each children group. Thus, due to the limited number of dose data, the statistical analysis on such values was not performed as it could led to misleading results.
