John Martin

Psychology's crises (e.g., replicability, generalisability) are currently believed to derive from Questionable Research Practices (QRPs), thus scientific misconduct. Just improving the same practices, however, cannot tackle the root causes of psychology's problems—the Questionable Research Fundamentals (QRFs) of many of its theories, concepts, approaches and methods (e.g., psychometrics), which are grounded in their insufficiently elaborated underlying philosophies of science. Key problems of psychological measurement are critically explored from independent perspectives involving various fields of expertise and lines of research that are well established but still hardly known in mainstream psychology. This comprehensive multi-perspectival review presents diverse philosophies of science that are used in quantitative psychology and pinpoints four major areas of development. (1) Psychology must advance its general philosophy of science (esp. ontology, epistemology, methodology) and elaborate coherent paradigms. (2) Quantitative psychologists must elaborate the philosophy-of-science fundamentals of specific theories, approaches and methods that are appropriate for enabling quantitative research and for implementing genuine analogues of measurement in psychology, considering its study phenomena's peculiarities (e.g., higher-order complexity, non-ergodicity). (3) Psychologists must heed the epistemic necessity to logically distinguish between the study phenomena (e.g., participants' beliefs) and the means used for their exploration (e.g., descriptions of beliefs in items) to avoid confusing ontological with epistemological concepts—psychologists' cardinal error. This requires an increased awareness of the complexities of human language (e.g., inbuilt semantics) and of the intricacies that these entail for scientific inquiry. (4) Epistemically justified strategies for generalising findings across unique individuals must be established using case-by-case based (not sample-based) nomothetic approaches, implemented through individual-/person-oriented (not variable-oriented) analyses. This is crucial to avoid the mathematical-statistical errors that are inherent to quantitative psychologists' common sample-to-individual inferences (e.g., ergodic fallacy) as well as to enable causal analyses of possibly underlying structures and processes. Concluding, just minimising scientific misconduct, as currently believed, and exploiting language-based algorithms (NLP, LLMs) without considering the intricacies of human language will only perpetuate psychology's crises. Rethinking psychology as a science and advancing its philosophy-of-science theories as necessary fundamentals to integrate its fragmented empirical database and lines of research requires open, honest and self-critical debates that prioritise scientific integrity over expediency.