Strategies for construction hazard recognition

162 Therefore, this research makes valuable contributions towards improving hazard recognition in the construction industry, which is a prerequisite to making any improvements in safety performance. These proactive methods of hazard recognition methods overcome many limitations associated with traditional methods.
METHODICAL CONTRIBUTIONS
Between 1993 and 2007, according to Taylor and Jaselskis (2009), 27% of research published in the Journal of Construction Engineering and Management (JCEM) employed techniques to make causal inferences. The goal of these research studies was to test hypnotized relationships between treatment variables and outcome variables. However, as pointed out by Shadish et almost published research findings and associated effect sizes are confounded by various nuisance variables that are of no interest to the researcher. Therefore, often, the computed effect sizes may not be reflective of the true relationship between the treatment variable and the outcome variable. He argues that causal inferences can only be made when the following criteria are warranted
• Sufficient evidence that the effect or outcome variable occurs as a consequence of introducing a specific treatment variable
• Clear indication of the absence of any alternate plausible explanation for the effect observed and
• Evidence that the causal factor or treatment variable precedes the occurrence of the observed effect In light of these requirements, cross-sectional research that measures outcome variables at a single point in time inherently fails to provide adequate evidence for causal inference. In fact, it is impossible to provide evidence to assert that the causal factor preceded the occurrence of the

163 observed effect (Shadish et al. 2011). Also, such studies do not adequately control for extraneous or alternate plausible explanations for the observed effect (Diggle 2009). But shockingly, there is some evidence suggesting that a large proportion of construction research relies on cross- sectional data to make causal inferences. For example, Deng and Smyth’s (2013) review of firm performance studies from four well-embraced journals in construction ASCE Journal of
Construction Engineering and Management (JCEM), ASCE Journal of Management in
Engineering (JME), Construction Management and Economics (CME), and International
Journal of Project Management (IIJPM) revealed that 89% of studies used cross-sectional data for making inferences. Several other studies such simple pretest and post-test designs with a control group have been used in construction research (Bernold and Lee 2010). However, such comparative two-wave studies that involve measuring the response variable once at the pre-intervention and post- intervention phase are highly susceptible to measurement error that are confounded with true changes, do not measure changeover time, and cannot be used to distinguish delayed effects
(Ployhart and Vandenberg, 2010). Moreover, cross-sectional and two wave studies assume that the variables being studied are static in nature, whereas most variables such as productivity levels, safety performance, etc are dynamic in nature (i.e. vary with time. To minimize the above mentioned methodical issues with causal inferences, researchers have suggested the use of longitudinal studies with control groups such as Randomized controlled experiments (Solomon et al. 2009). Researchers have elevated Randomized controlled experiments as being the gold standard for conducting scientifically credible research because (i) the method can provide sufficient evidence to indicate that the observed effect occurs as a consequence of introducing the treatment variable (ii) controls of alternate plausible explanation for the observed effects

164 iii) can show that the treatment preceded the observed effect, (iv) efficiently measures changeover time and can capture delayed effects of the treatment (v) accounts for the dynamic nature of the dependent variable, (vi) highly resistant to measurement error due to random error or variability due to the repeated nature of data accusation (Robinson et al. 2007). While the strengths of longitudinal randomized controlled experiments are obvious, they cannot be easily adopted in applied settings when randomization is not practical and when randomized group may not reflect real-life situations (Kirk 2013). Further, the most important limitation of the method is that it maybe unethical to adopt such methods when the treatment variable is beneficial to the participants of the research study (Rossi et al 2004). For example, in safety research, depriving the control group of a positive safety intervention that can reduce injures is unethical. Therefore, when conducting intervention studies, it is necessary to choose a research method that is both (i) rigorous to make valid causal inferences, and (ii) also that which is ethical to adopt in practice. The multiple baseline testing approach meets both the above mentioned criteria to conduct rigorous research. The multiple baseline testing approach involves a series of replicated and simultaneously conducted AB (i.e. before-after) studies in which the intervention is introduced to each baseline in a staggered, or time-lagged fashion (Barlow et al., 2009; Biglan et al., 2000; McGuigan,
1997). Hence, when a given subject or group receives the intervention, the other groups serve as control. Simultaneous comparisons can be made within and between groups allowing us to reject any alternate or plausible explanations for the intervention effect. If similar patterns of changes are observed when, and only when, the intervention is introduced, then the observed effects can be confidently imputed to the intervention instead of other nuisance variables and causal inferences can be made (Bulté and Onghena, 2009).

165 This study is among the few published work in construction research that has adopted the
Mutliple baseline technique. Chapter 3, 4, and 5 clearly describes the procedures that were followed to make valid causal inferences in dynamic construction settings. It is expected that these studies will encourage researches to adopt more experimental studies and will minimize the perception among researchers that methodological rigor to conduct scientifically credible conclusions is impractical in dynamic construction environments (Levin 2005).

Download 2.75 Mb.

Share with your friends: