Multiple baseline testing protocol and analysis

93 prior to the introduction of the intervention. The intervention was introduced to each crew in a time lagged manner in both project sites. During the training sessions the researchers and safety managers introduced the SMQM intervention and mnemonic cues independently to each work crew. During the training the instructors illustrated the use of the tool using various simulated work scenarios with a number of visual photographs. The time-series regression models described by Huitema and Mckean (2000;2007) were compared to identify the appropriate statistical model for analysis. These models are shown and described in Table 2. When determining the most appropriate model we used the flowchart shown in Figure 1. The process began with estimating the parameters of model I and II by regressing Y (dependent variable) on their respective predictor variables. Once the parameters were determined, a model comparison testing the null hypothesis that a trend/slope does not exist in either phase (β
1 =
β
3
= 0) was tested using the test statistic shown in Equation 2. This statistic is used to test if either a slope change was induced as a result of introducing the intervention or a slope existed in the baseline phase. In the test, if the null hypothesis is accepted, then the parameters β
1 and would assume the value zero, thus reducing the equation to represent only the level change effect as in the Model II equation (see Table 2) In this case, Model II will better represent the observed data with higher power for statistical inference. But if the null hypothesis is rejected, then Model II may provide a biased estimate of the level change as the slope is not represented.

94

Download 2.75 Mb.

Share with your friends: