Strategies for construction hazard recognition
Testing protocol and multiple baseline testing analysis
Download 2.75 Mb. View original pdf
|
| Testing protocol and multiple baseline testing analysis The field testing protocol began with systematically measuring the HR index at the pre- intervention phase, as discussed in the previous section. These measurements were gathered simultaneously for three crews from each case site. The HR indices at the pre-intervention phase provided measures of the crew’s initial hazard recognition capability, variability, and trends in performance before the intervention was introduced. After gathering adequate baseline data, the 57 intervention was introduced in a staggered basis to all crews in both sites. During this session, the crews were independently introduced to the energy mnemonics and SAVES. As indicated above, this session focused on assessing the crew’s hazard recognition performance in the augmented virtual environment, and provided active performance feedback. As such, the program exposed the crews to various case scenarios and illustrated the use of the energy-based retrieval mnemonics to efficiently identify hazard-signals in diverse conditions. From the baseline performance data (i.e. pre-intervention performance) we were able to extrapolate or project the HR index for each crew in the second (i.e. post-intervention) phase in the hypothetical case that they were not exposed to the intervention. In other words, the baseline data was used to estimate the expected performance had the intervention not been introduced. This projected HR index value, when compared with the observed actual performance in the field, provides a reliable measure of the intervention effect. This was accomplished using interrupted time-series regression analysis. We adopted and compared the four time-series regression models suggested by Huitema and Mckean (2000; 2007) to identify the most appropriate statistical model for analysis. These models were used to estimate intervention effects and are presented and described in Table 2. As can be seen, intervention effects can be expressed as level change, slope change or both. Level change refers to immediate changes that result due to the intervention introduction, whereas slope change refers to any delayed effects that result in gradual (slope) changeover time. Like conventional regression models, model I and II assume that the errors are independent while model III and IV are relevant when the assumption of independent errors is violated. That is, 58 model III and IV apply when the errors are assumed to be dependent or autocorrelated, which is common when repeated process measures are gathered. Download 2.75 Mb. Share with your friends:
|