Strategies for construction hazard recognition
Download 2.75 Mb. View original pdf
|
| Figure 5: Results of Case 1- Multiple baseline study on oil and gas facility Table 3: Results of Case 1- Multiple baseline study on oil and gas facility Predictor Coefficient Std. Error t value p value Model test (Fcritical = 2.807) r 2 D-W test p H-M test (α = 0.05) p Levene's (α = 0.05) p A-D (α = 0.05) Crew Civil Constant 42.575 2.205 19.309 0.000 Model II F obt = 0.209 0.880 2.073 0.696 0.057 0.570 D 29.337 2.790 10.516 0.000 Crew 2: Constant 56.220 1.933 29.087 0.000 Model II Fobt = 1.694 0.792 2.550 0.288 0.128 0.379 D 19.942 2.733 7.296 0.000 Crew 3: Mechanical Constant 35.414 4.070 8.701 0.000 Model II Fobt = 4.118 0.931 2.742 0.219 0.556 0.659 Time -0.639 0.656 -0.974 0.349 D 43.704 5.930 7.370 0.000 Time 0.157 1.568 0.100 0.922 0 10 20 30 40 50 60 70 80 90 100 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Ha za rd re co gn iti on le ve l ( %) Work Period (Time -->) Crew Crew Crew Intervention 64 The Levene’s test for the homogeneity of error variances and the Anderson-Darling test for the normality of errors yielded a p-value above 0.05 for each case. Thus, equality of variance and the normality of error variance can be reasonably assumed. Also, for all cases, the Durbin-Watson test and the Huitema-McKean test indicated that the errors were independent. As such, additional autocorrelation parameters were not required and the selected models as presented in Table 3 were appropriate for statistical inference. Comparing HR performance between the baseline and the treatment phase, crew 1 indicated a level-change improvement of 44% (p<0.05). The level-change indicates the difference between the expected performance of the crew in the absence of the intervention and the observed performance immediately after the intervention. For crew 1, the projected performance in the seventh work-period based on the pre-intervention data is 43% ( β 0 + β 2 D), where D assumes the value zero for the baseline phase and the fitted value in the treatment phase is 72% (D=1). Thus, the difference between the two values yields the level change coefficient of 29% (72%-43%). The associated significance level (p<0.05) indicates a statistically significant level-change improvement. Similarly crew 2 revealed a statistically significant level-change improvement of 20% For crew 3, the projected performance based on the baseline data for the eleventh work period (T=11) is 35% ( β 0 + β 1 T + β 2 D + β 3 SC) where D and SC assumes the value zero for the baseline phase and the fitted value in the treatment phase is 79% (D=1, SC=0). Thus, the level- change for this case is 44% (79%-35%). Combining the level-change coefficients of all three 65 crews using the reciprocal of error variance as indicated in equation 3, the crews revealed an overall level-change improvement of 26% (p<0.05). The slope change coefficient for crew 1 and 2 is zero because the null hypothesis β 1 = β 3 = 0 was accepted and model II was selected as being appropriate. However, crew 3 revealed a slope change coefficient of 0.157. Therefore, the slope in the intervention phase is equal to -0.482, which is the difference between the two slopes in the baseline and intervention phase. This represents a decline in the HR index (0.48% per work period, however, the associated p-value (p=0.922) indicates the absence of any statistical significant difference in the slope change coefficient. Therefore, this decline in performance can be attributed to random error rather than true performance. The corroborative test involving the pretest and a post-test using representative construction work scenarios also revealed similar findings. After testing for normality using the Anderson- Darling test, the two-sample t-test for independent measures was used to test the null hypothesis that the HR index was equal before and after the intervention. Crew 1, 2 and 3 exhibited an increase in hazard recognition of 32%, 31% and 36%, respectively. The p-values for each case was less than 0.05, indicating highly significant improvements. Download 2.75 Mb. Share with your friends:
|