Comparing Network Community Detection and Factor Analysis by Analyzing Big Five Personality Traits

• In Event: AERA Roundtable Session Sunday 11:40 am Four Seasons Ballroom 2-3
  In Roundtable Session: Measurement Issues (Table 9)

Abstract

Network Community Detection (NCD) and Exploratory Factor Analysis (EFA) both identify latent structures in data but differ in approach. NCD analyzes network graphs to find densely connected communities, while EFA uses covariance/correlation matrices to identify latent variables explaining correlations between observed variables. When a correlation matrix is converted into a network, NCD may replace EFA. This project evaluates the performance of NCD and EFA. As a first step, we analyzed a Big-Five personality traits dataset, finding NCD preferable for clustering. Next, a simulation study will explore these methods under various conditions, such as number of factors, distances between clusters, thresholds and evaluation metrics. This study aims to compare NCD and EFA, helping researchers achieve more accurate and efficient results.

Authors

• Jingyang Li, University of Georgia

• Zhenqiu Laura Lu, University of Georgia

• Pengsheng Ji, University of Georgia