A New Way to Look at the Data: Similarities Between Groups of People Are Large and Important

摘要

Most published research focuses on describing differences, while neglecting similarities that are arguably at least as interesting and important. In Study 1, we modified and extended prior procedures for describing similarities and demonstrate the importance of this exercise by examining similarities between groups on 22 social variables (e.g., moral attitudes, human values, and trust) within 6 commonly used social categories: gender, age, education, income, nation of residence, and religious denomination (N = 86,272). On average, the amount of similarity between 2 groups (e.g., high vs. low educated or different countries) was greater than 90%. Even large effect sizes revealed more similarities than differences between groups. Studies 2–5 demonstrated the importance of presenting information about similarity in research reports. Compared with the typical presentation of differences (e.g., barplots with confidence intervals), similarity information led to more accurate lay perceptions and to more positive attitudes toward an outgroup. Barplots with a restricted y-axis led to a gross underestimation of similarities (i.e., a gross overestimation of the differences), and information about similarities was rated as more comprehensible. Overall, the presentation of similarity information achieves more balanced scientific communication and may help address the file drawer problem.