A researcher runs a hierarchical regression predicting employee burnout from age and gender (Step 1), then adds weekly workload hours (Step 2). In SPSS, Step 2 shows R² Change = .03 with Sig. F Change = .18. However, the coefficient for workload in Model 2 is significant (p = .02). What is the best conclusion?
Step 2 did not significantly improve overall model fit; the significant workload coefficient may reflect shared variance/suppression and should be interpreted cautiously.
Because workload is significant, Step 2 must be significant; SPSS is contradictory and you should ignore the R² Change test.
Since R² Change is small (.03), workload has no relationship with burnout at all.
The correct fix is to remove age and gender because control variables are not allowed in hierarchical regression.