Multilevel Multivariate Meta-analysis with Application to Choice Overload.

Area of Science:

• Psychological research methodology
• Statistical modeling
• Meta-analysis

Background:

• Contemporary psychological research generates complex data with nested structures.
• Previous meta-analyses have limitations in fully accounting for data complexity.

Purpose of the Study:

• Introduce a novel multilevel multivariate meta-analysis methodology.
• Address the complexity of psychological research data, including variation, covariation, and nesting.
• Provide richer insights into the choice overload hypothesis.

Main Methods:

• Developed a multilevel multivariate meta-analysis technique.
• Directly modeled observations, accounting for variation and covariation across dependent measures, moderators, and nested structures (papers, studies, subjects, conditions).
• Applied the methodology to data from choice overload hypothesis studies.

Main Results:

• The new methodology fully accounts for choice overload data complexity.
• Choice overload significantly varies across six dependent measures and four moderators.
• Identified potentially important interactions among dependent measures and moderators.
• Demonstrated that up to the fifth (paper) level of nesting is necessary to capture variation and covariation.

Conclusions:

• The developed methodology offers a more comprehensive analysis of complex psychological data.
• Results provide substantial implications for future choice overload research.
• Highlights the importance of considering multilevel structures and interactions in meta-analyses.