Beyond Random Effects: When Small-Study Findings Are More Heterogeneous