Interpreting the K-Way and Higher-Order Effects table in SPSS Statistics Model Selection (HILOGLINEAR) output
I'm running Model Selection loglinear analysis in SPSS Statistics. I'm not certain how to intepret the entries in the K-Way and Higher-Order Effects table and don't find any specific information in the documentation or Help. How do the two sections of this table differ?
The K-way and Higher Order Effects section of the K-Way and Higher-Order Effects table provides tests of all effects at level K and higher, while the K-way Effects section provides tests of just the K-way effects.For example, suppose you have four variables specified. The line for K=4 is the same for each section of the table because there is only one 4-way effect. In the bottom part of the table, the line for K=3 provides a test that all of the 3-way terms are null, while in the upper half of the table, the line for K=3 combines the test for 3-way terms with the one for the 4-way term, testing that the 3-way and all higher order effects are null. For K=2, the bottom half shows the test for the 2-way terms, while the top half shows the combined test for 2 through 4-way effects. Finally, the line for K=1 in the lower section shows the test for main effects, while the line in the top section shows the test for all 4 levels of effects combined (which is everything other than an intercept). Note that the df in the top half of the table for each line is the sum of the df in the lower half for the same K and higher values of K (which are lower in each section of the table).