Statistical analysis
The main variables tested in this experimental study were total
respirable dust concentrations and respirable quartz dust concentrations.
The reduction in the two variables produced by the units
with controls (DustBubble) compared to that produced by those
without controls during concrete drilling processes were estimated
through the following equation:The data are generally assumed to follow the lognormal distribution,
when samples are collected over time (NIOSH, 1977).
Therefore, all the data has been tested firstly for a lognormal distribution.
Descriptive statistics were used to describe total respirable
dust and silica dust concentrations in terms of arithmetic and geometric
means (AM and GM) as well as the corresponding geometric
standard deviations (GSD) and ranges (min–max), while inferential
statistics were conducted to compare the differences of dust generation
between the two situations.
According to Hornung and Reed (1990), for the non-detectable
values, there are two simple ways to reach enough accuracy:
1/ ffiffiffi 2
p
of nondectable values when the data are not highly skewed;
1/2 of nondectable values when the data are highly skewed
(Geometric Standard Deviation approximately 3.0 or greater).
However, if more than half of one set of data are non-detectable,
only the percentage of samples below LOD and the range of
remaining samples were reported. Statistical analyses were conducted
with SPSS statistical software (version 17; SPSS Inc.).