Empirics vs. art theory: Exploring a factor structure of pictorial expression based on contemporary artworks

ABSTRACTThe RizbA scale combines psychometrics and art theory and enables a measurement of pictorial expression. This study explores its factor structure and a potential gap between theory and empirics. A sample of 275 pictorial works by artists and nonprofessionals was rated by 179 art experts. Three CFA path models were specified: models A and B based on the empirical results of previous studies, C on the theory of the initial study. Model C was additionally tested on a combined dataset. A and B did not converge, C was associated with fit indices as follows: RSMEA = .122, CFI = .712, TLI = .679, SRMR = .135, for the combined dataset: RSMEA = .086, CFI = .740, TLI = .696, SRMR = .084. Only model C partly suggests an acceptable fit. The results speak to a methodological gap between empirics and theory, that might be solved by a postdisciplinary measurement model.KEYWORDS: art theoryconfirmatory factor analysisformal picture analysispictorial expressionvisual art AcknowledgmentsThe authors gratefully acknowledge Thomas Gengenbach for providing theoretical and practical support regarding IT. Thanks also to Rebecca Kahn for the feedback on the manuscript.Disclosure statementNo potential conflict of interest was reported by the author(s).Ethics approvalThe authors received a positive vote of the Ethics Committee for Creative Arts Therapies at the Nürtingen-Geislingen University, Germany (Reference number 19C27J101W).Abbreviations API=Application programming interfaceCFA=Confirmatory factor analysisCFI=Comparative fit indexdf=Degrees of freedomMLE=Maximum likelihood estimateMLR=Robust maximum likelihood estimatorPAFA=Principal axis factor analysisPCA=Principal component analysisPHP=PHP: Hypertext PreprocessorRizbA=(Rating instrument for two-dimensional pictorial works) Ratinginstrument für zweidimensionale bildnerische ArbeitenRMSEA=Root mean square error of approximationSRMR=Standardized root-mean-square residualTLI=Tucker-Lewis Index