Summary: For four decades the Big Five model has been the dominant framework for describing personality, but new research suggests that it does not capture the full complexity of human personality. Using a data-driven method called taxonomic graph analysis (TGA), researchers reconstructed the IPIP-NEO personality inventory from the ground up and identified additional traits and meta-traits that expand and refine the traditional Big Five taxonomy.
By building structure from item-level relationships instead of imposing a top-down model, this bottom-up approach reveals richer connections among survey items, facets, traits, and higher-order dimensions. The findings not only propose a more nuanced hierarchy for personality assessment, they also point to potential implications for how clinicians conceptualize and classify psychopathology.
Key Facts:
- Beyond the Big Five: The TGA-derived hierarchy identifies six second-level traits—neuroticism, conscientiousness, openness, and three additional traits labeled sociability, integrity, and impulsivity—expanding the traditional five-factor framework.
- Bottom-up methodology: Taxonomic graph analysis maps statistical relationships among individual survey items to form facets and traits, allowing empirical structure to emerge rather than forcing data into a predefined model.
- Psychopathology implications: Applying TGA to mental health symptom data could lead to more precise diagnostic categories—for example, by revealing subtypes of depression in which anxiety symptoms are a component rather than a fully separate diagnosis.
Source: Vanderbilt University
Alexander Christensen’s recent study does not claim to overturn 40 years of personality research, but it offers a rigorous, quantitative framework intended to provoke discussion about how personality is defined and measured.
Christensen, an assistant professor of psychology and human development at Vanderbilt Peabody College, and his colleagues published their work, “Revisiting the IPIP-NEO personality hierarchy with taxonomic graph analysis,” in the European Journal of Personality. Their article presents a network-based strategy for reconstructing hierarchical personality structure using empirical relationships among questionnaire items.
Traditional personality inventories typically follow a top-down design: researchers assume broad traits such as the Big Five at the top of a hierarchy and then link narrower facets and items beneath those traits. While this approach has produced robust and predictive measures, it can obscure statistical relationships between individual items that might point to alternative or additional trait groupings.
A new and improved approach to personality structures
In their study, Christensen and colleagues applied Taxonomic Graph Analysis (TGA) to a large, open-source IPIP-NEO dataset. TGA is a network psychometrics tool that identifies clusters of items that cohere empirically, then aggregates those clusters into facets, traits, and higher-order meta-traits. By letting connections emerge from the data rather than enforcing a predefined topology, TGA produces a hierarchy that better reflects the observed structure of responses.
Using this bottom-up method, the research team recovered a three-tier hierarchy consisting of 28 first-level facets, 6 second-level traits, and 3 third-level meta-traits. The three meta-traits were stability, plasticity, and a novel disinhibition dimension. At the trait level, the analysis reproduced established dimensions such as neuroticism, conscientiousness, and openness, while also identifying sociability, integrity, and impulsivity as distinct second-level traits.
Because TGA is driven by item-level patterns, it can reveal meaningful distinctions that are often hidden when items are aggregated prematurely. The resulting structure provides a more detailed and potentially more valid taxonomy for describing personality variation in large samples.
Application to psychopathology
Christensen emphasizes that the methodological advantages of TGA extend beyond personality assessment. Structural models of psychopathology commonly adopt top-down frameworks similar to traditional personality taxonomies. By applying TGA to symptom-level data, researchers might discover new clusters or subtypes that better represent the empirical relationships among symptoms.
For example, symptoms of internalizing disorders—such as depression and anxiety—often co-occur and are diagnosed as separate conditions. A TGA-based analysis could indicate that anxiety is an element of a particular depression subtype rather than an independent disorder in some cases. That reclassification could guide clinicians toward more tailored diagnoses and treatment plans focused on specific symptom clusters.
Ultimately, Christensen notes, the paper’s enduring contribution will likely be methodological: demonstrating the power of Taxonomic Graph Analysis and the suite of tools the team developed to probe hierarchical psychological constructs in a transparent, data-driven way.
The importance of team science
The study reflects a collaborative “team-science” approach that combined substantive theory with advanced quantitative methods. Christensen credits lead author Andrew Samo and the interdisciplinary team for bringing together theoretical expertise and computational tools necessary to carry out a large-scale, item-level investigation of personality structure.
The study also points readers to a shorter version of the IPIP-NEO inventory used in their work for those interested in exploring the measures firsthand.
Christensen teaches Behavioral Data Science in the Psychological Sciences Ph.D. program and Quantitative Methods in the M.Ed. program, and he contributes courses in data science and advanced statistics across Vanderbilt’s undergraduate and graduate programs.
About this personality and psychology research news
Author: Jenna Somers
Source: Vanderbilt University
Contact: Jenna Somers – Vanderbilt University
Image: The image is credited to Neuroscience News
Original Research: Open access. “Revisiting the IPIP-NEO personality hierarchy with taxonomic graph analysis” by Alexander Christensen et al., European Journal of Personality.
Abstract
Revisiting the IPIP-NEO personality hierarchy with taxonomic graph analysis
Understanding the structure of personality is essential for predicting and explaining human behavior. Recent research advocates for analyzing large item pools at the item level, because aggregation can obscure meaningful distinctions. This study introduces Taxonomic Graph Analysis (TGA), a network-based psychometrics framework designed to recover hierarchical structures from the bottom-up. We applied TGA to an open-source 300-item IPIP-NEO dataset (N = 149,337).
TGA addresses methodological challenges that have limited accurate recovery of hierarchies—such as violations of local independence, wording effects, dimensionality assessment, and structural robustness. The analysis revealed a three-level hierarchy with 28 first-level facets, 6 second-level traits, and 3 third-level meta-traits. Although several dimensions aligned with the theoretical IPIP-NEO model, notable deviations emerged, including Sociability, Integrity, and Impulsivity at the trait level and a Disinhibition meta-trait at the highest level. These results integrate empirical and theoretical findings scattered across the literature and demonstrate TGA’s utility for investigating hierarchical psychological constructs. This study contributes a rigorous, data-driven perspective to ongoing discussions about personality taxonomy.