Predictive validity of manually coded handwriting features for questionnaire derived personality scores
Year: 2026
Type: article
Abstract: Claims that handwriting can reveal personality are common, but their predictive validity remains disputed. This study evaluates whether manually coded handwriting features predict continuous questionnaire-derived scores associated with the Big Five personality domains. The source data comprised five trait-specific workbooks containing questionnaire responses and graphologically coded handwriting features from a study conducted in Gujarat, India. Participant identifiers 1–62 were excluded as a data-collection pilot. After domain-specific screening of incomplete, invalid and invariant questionnaire responses, analytic samples ranged from 320 to 330. Each outcome was calculated as the mean of 12 five-point Likert items; no binary personality labels were created. Handwriting was represented by 16–26 binary features per domain. Inter-rater agreement was recalculated from 26 common samples independently coded by two external graphologists. A mean-prediction baseline, elastic-net regression and histogram-based gradient boosting were evaluated using repeated five-fold cross-validation with five repeats. Performance was assessed using mean absolute error, root mean squared error, R 2 and out-of-fold correlations, supplemented by baseline-comparison and permutation tests. Neither machine-learning model consistently improved on the mean baseline, and mean cross-validated R 2 was negative for every domain and predictive model. Although the two external raters generally agreed when identifying handwriting features, the feature matrices coded by the first author did not predict the questionnaire-derived scores in this sample. These findings do not establish that handwriting contains no personality-related information under all conditions; they show that this specific manual feature-coding and questionnaire pathway lacked criterion validity and should not be used in high-stakes settings without independent validation.
Source: Discover Artificial Intelligence
Authors: Sheetal Thomas, Manjeevan Seera, Mridula Goel
Language: English
Subfield: Clinical Psychology
Field: Psychology
Domain: Social Sciences
Open Access status: gold
Estimated APC paid: $1,590
OpenAlex ID: W7221756715