Meta-Analyzing Apples and Oranges: How to Make Applesauce Instead of Fruit Salad

2019/12/09 by Christopher J. Carpenter · 1 voice · 3 citations
Psychology · #Attachment and Relationship Dynamics #Evolutionary Psychology and Human Behavior

paper · doi:10.1093/hcr/hqz018

openalex publication_date 2019/12/09 · openalex created_date 2025/10/10 · openalex updated_date 2026/06/13

Abstract

Abstract This essay attempts to describe the apples and oranges problem in meta-analyses. Essentially, some meta-analyses combine original studies of various variables that are not the same pairs of variables. Metaphorically, they meta-analyze the effects of fruit when they should conduct separate meta-analyses of apples and oranges. This practice is inconsistent with the assumptions behind the meta-analytic formulae concerning sampling error and makes meta-analytic estimates difficult to interpret. Meta-analysis teams are advised to justify their choices and types of evidence are described to assist researchers and reviewers in assessing and justifying when constructs can and cannot be combined together in a meta-analysis.

Cited by

• Temporal dynamics of media and communication processes: a review of theories and meta-analysis of empirical studies
• The State of Evidence in Digital Hate Research: An Umbrella Review
• Media literacy interventions: meta-analytic review of 40 years of research

Discussions

• I agree with your analysis. Your comments on mixing outcomes are consistent with some arguments I made in a paper you may find interesting doi.org/10.1093/hcr/... [bsky, 2 points, 0 comments]

Related

• orchaRd 2.0: An R package for visualising meta‐analyses with orchard plots
• How to do a meta‐analysis
• What can we learn from 1,000 meta-analyses across 10 different disciplines?
• A CALL FOR CAUTIOUS INTERPRETATION OF META-ANALYTIC REVIEWS
• An overview of meta-analytic methods for economic research