Is psychological science progressing? Explained variance in PsycINFO articles during the period 1956 to 2022

Abstract

We aimed to numerically assess the progress of modern psychological science. Average explained variance in 1565 included articles was 42.8 percent, and this was constant during 1956 to 2022. We explored whether this could be explained by a combination of methodological conventions with the semantic properties of the involved variables. Using latent semantic analysis (LSA) on a random sample of 50 studies from the 1,565, we were able to replicate the possible semantic factor structures of 205 constructs reported in the corresponding articles. We argue that the methodological conventions pertaining to factor structures will lock the possible explained variance within mathematical constraints that will make most statistics cluster around 40 percent explained variance. Hypotheses with close to 100 percent semantic truth value will never be part of any assumed empirical study. Nor will hypotheses approaching zero truth value. Hypotheses with around 40 percent truth value will probably be experienced as empirical and plausible and, consequently, as good candidates for psychological research. Therefore, to the extent that the findings were indeed produced by semantic structures, they could have been known without collecting data. Finally, we try to explain why psychology had to abandon an individual, causal method and switch to studying whether associations among variables at the group level differ from chance. Psychological processes take place in indefinitely complex and irreversibly changing contexts. The prevalent research paradigm seems bound to producing theoretical statements that explain each other to around 40%. Any theoretical progress would need to address and transcend this barrier.