Global Test for Coincidence

Sometimes, one will want to test components in one large model in one evaluation, as we have done. If the group is coincident, then one small model can be used to describe all. If not, one must test the inguinal, subclavian, forearm, or abdomen, with the IPA and IPA + CHG in individual components. First, extract the sub-models from the full model, y = bo + biXi + 2 1 + biZ2 -f 423 + bsZ4 + b(jZ.  

Regression Equation, with Time X Product Interaction, Example 9.1 

The regression equation is y = 2.21 + 0.00208x1 - 0.025zi + 0.900x2 - 0.825x3 - 0.633x4  

The test for coincidence will be for all four test sites for both products. The only way the equation can be coincidental at all sites for both products is if the equation is to explain all that is the simplest that is, y = bo + b x. So, if there is coincidence, then b2 = b i=b4 = bs = bp = 0. That is, all intercepts and slopes are identical. 

V is the number of variables in the full model minus the number of variables in the partial model. 

---