After the Wizard completes its work on a cluster model file, your newly formatted workbook is revealed.
Most of the new formatting is common to all models. You can review this by clicking here. On this page we will introduce formatting explicitly for cluster models and walk through an interpretation of model results.
Notice that in column H all of the values have been turned boldface and colored blue. All R2 values above the arbitrary cutoff value of 0.2 are formatted in this way.
Although no examples exist here, any p-value greater than the cutoff 0.05 is displayed in boldface red. You can see this as an exercise if you over-type one of the entries in column G with a value larger than 0.05.
The pattern displayed on the Model 4 Parameters sheet above could not be more striking. The green cells in column B suggest that respondents in Cluster1 tend to report few of the five symptoms. Conversely, respondents in Cluster4 tend to report all of them. Cluster2 tends to report back and neck symptoms; Cluster3, joint, swelling and stiffness. A look at the Profile and ProbMeans sheets confirms this distinction. If you wish, you can enter these characterizations in selected cells B1:E1, as below.
By default, cluster models include a table of bivariate residuals, which the Format Wizard imports into its own sheet, like the one in the following image.
The Wizard adds conditional formatting to all the numerical cells in the sheet. If any bivariate residual exceeds the cutoff value in cell I1, the residual value will turn boldface red. In this example no bivariate residual exceeds 3.84, so no red cells appear. Just to illustrate the capability, we arbitrarily change the cutoff value to 0.5. The worksheet reveals two bivariate residuals with values that exceed this amount.
That concludes our tutorial on formatting cluster model output. Click on one of these links to see tutorials on DFactor models, regression models and choice models.