Site, IPA CHG Product

Full model y = bo + b Xi + (72X2 + 63X3 + 64X4 + 65X5. 

Note that there may be a big problem. The regression fits the same slope for both products, just different at the y-intercepts. This may not be adequate for what we are trying to do. The actual data for both at time 0 are nearly equivalent in this study, differing only at the 24 h period. Hence, the predicted values do not make sense. 

There may be an interaction effect. Therefore, we need to look at the data in Table 9.3, providing the actual data, fitted data, and residual data for the model. We see a +/ pattern, depending on the x, variable, so we check out the possible interaction, particularly because for the regression is only 73.4% (Table 9.2). 

The positive interactions that can occur are x X2, XiX, X1X4, XjXs, x k, X2X4, X2X5, X1X2X3, X1X2X4, and X1X2X5. Note that we have limited the analysis to three-way interactions. 

Actual Data, Fitted Data, and Residual Data, Example 9.1 

---