Strategies for construction hazard recognition
Figure 3: Conceptual framework of SAVES
Download 2.75 Mb. View original pdf
|
| Figure 3: Conceptual framework of SAVES Phase II – Empirical field testing using the multiple baseline design Although SAVES was designed and built based on proven techniques from modern theories of learning, we desired to empirically test its impacts on worker hazard recognition skills in the field. To accomplish this, we desired a rigorous research method capable of measuring a causal relationship (if any) between SAVES and improved performance. Therefore, pure correlation- based cross-sectional designs were dismissed (Ployhart and Vandenberg 2010). To draw causal conclusions, Davis (2002) suggests using longitudinal research designs where the response variable in measured in multiple occasions across time. In doing so, it is possible to directly capture the change in response overtime, and also eliminate issues that can arise due to inherent Instructional content Saves training cycle Training outcomes 54 differences among cohort groups because the response is compared before and after the intervention within the same independent cohort. We evaluated several longitudinal research design approaches based on their validity and reliability potential. The before-after testing method was dismissed because of the method’s inherent weakness in differentiating intervention effects and the effects of other confounding unrelated variables (Dimitrov and Rumrill 2003; Richards 1999). Similarly, reversal designs were dismissed because the withdrawal of an effective safety strategy is highly unethical or impractical in cases when the intervention involves skill acquisition (Baer et al. 1968; Barlow et al. 2009; Watson and Workman 1981). Instead, we decided to adopt the multiple baseline testing approach, which preserves internal validity by engaging multiple groups and staggering the intervention across the study period, allowing us to make comparisons within and between groups (Hawkins et al. 2007). Multiple baseline testing is structured as a series of longitudinal AB studies that are conducted concurrently and replicated within a single study, where A refers to the pre-intervention phase and B to the post-intervention phase (Barlow et al. 2009; Biglan et al. 2000; McGuigan 1997). The intervention is provided to individual groups or subjects in a time-lagged or staggered basis to verify if a change in the response variable occurs only when the intervention is induced. If a change is observed following the intervention, and the groups that are yet to receive the intervention reveals no significant change, then this is strong evidence to attribute the change to the intervention. This inference is further reinforced if the effect is replicated across multiple concurrent studies. |