Analysis of variability in laboratory airborne sound insulation determinations

In an investigation of the overall precision of airborne sound insulation determinations in laboratory conditions 10 different samples were measured, according to the general procedure described in ISO 140/III, in an horizontal transmission chamber, with lateral flanking transmission. Measurements were carried out according to a prescribed programme to investigate the repeatability, r, and reproducibility, Re, in the results, within the same laboratory. Two different operators carried out 15 replicates, five per day during three consecutive days. Each operator performed the test under repeatability conditions: that is, using the same instrumentation, without removing the samples and with the environmental conditions kept constant. Between operators, certain parts of the experimental set-up were different: i.e., the measuring samples had been removed and reinstalled and the environmental conditions were slightly different. The statistical parameters, r and Re′, were first obtained for each sample level separately in order to investigate any dependence with level. Successively, in order to get an improvement in precision, an overall factorial variance analysis was carried out for all levels simultaneously. The whole treatment was performed for each third octave frequency band between 100 and 3150 Hz separately, as well as for the weighted apparent reduction index, R′_w.