In this study, the normality and variability of heavy-weight floor impact sound insulation are examined statistically using field data from 106 dwelling units across eight complexes and 19 floor plans, all with identical floor-assembly designs. Measurements in one-third-octave bands followed KS F ISO 16283-2, and single-number quantities were derived per KS F ISO 717-2. The overall standard deviation was 1.75 dB, with normality tests indicating a Gaussian distribution. When grouped by site and floor plan, the variability decreased and normality improved. Nested analysis of variance identified significant site and floor-plan effects, accounting for 45 % and 13 % of the total variance, respectively. After removing these systematic effects, the residual standard deviation was 1.14 dB, attributed to random construction deviations and measurement uncertainty. Since random deviations occur even when site and floor-plan factors are controlled, securing an adequate design margin that accommodates statistical deviations is crucial for in-situ performance verification. These findings highlight the need for sample surveys based on statistical confidence intervals during performance inspections.
Seong et al. (Wed,) studied this question.