Psychometrics builds on the fundamental premise that psychological attributes are unobservable and need to be inferred from observable behavior. Consequently, psychometric procedures consist primarily in applying latent variable modeling, which statistically relates latent variables to manifest variables. However, latent variable modeling falls short of providing a theoretically sound definition of psychological attributes. Whereas in a pragmatic interpretation of latent variable modeling latent variables cannot represent psychological attributes at all, a realist interpretation of latent variable modeling implies that latent variables are empty placeholders for unknown attributes. The authors argue that psychological attributes can only be identified if they are defined within the context of substantive formal theory. Building on the structuralist view of scientific theories, they show that any successful application of such a theory necessarily produces specific values for the theoretical terms that are defined within the theory. Therefore, substantive formal theory is both necessary and sufficient for psychological measurement.
Numeric ratings of agreement (also known as Likert-scales) are one of the most common assessment tools in psychology. However, little is known about the measurement theoretic properties of such ratings. Most prominently, the question whether such ratings yield interval scaled measurements has been questioned on theoretical grounds, the key issue being that interval scales do not only require ratings to be order preserving, but also additive. In this article, I apply a representational measurement model to investigate intra-individual response structures with regard to Likert-scale items. The model is borrowed from the psychophysical paradigm of cross-modality comparisons and allows for an explicit empirical test of additivity on the level of individual responses. A within-subjects experiment with N = 140 individuals was conducted, where subjects were instructed to rate their agreement to a set of items from a standardized personality questionnaire. The ratings were performed repeatedly with varying reference stimuli to evoke difference ratings for agreement. These difference ratings were then used to test the conditions of the measurement model. The results indicate that 51% of the individuals showed additivity in their rating behavior. Therefore, an interval scale representation of subjective agreement was possible for about one half of the individuals, whereas the other half did not produce agreement ratings on an interval scale. While the results clearly demonstrate that interval scale measurements using numeric ratings are possible, they also indicate that such quantitative representations should not be taken at face value but need to be tested empirically.
The Trivers-Willard hypothesis (TWH) states that individuals in good condition favor offspring of the sex that has a higher variance in reproductive value. Empirical studies with historical human populations suggest that the TWH might explain biased birth-ratios as well as biased parental investment in male or female offspring. However, empirical tests of the TWH in modern human populations are less conclusive.In this study, we investigate whether parental investment in education might be skewed according to the TWH in an African sample (N = 314) that includes students from 8 different countries. The data show that male students who rate their family’s wealth high tend to report more parental involvement in their own education, whereas the opposite is true for female students. This pattern is in accordance with the TWH for parental investment. The results support the validity of evolutionary explanations of behavioral bias in the context of parental investment in offspring education.
In this paper, we provide a re-interpretation of qualitative and quantitative modeling from a representationalist perspective.In this view, both approaches attempt to construct abstract representations of empirical relational structures.Whereas quantitative research uses variable-based models that abstract from individual cases, qualitative research favors case-based models that abstract from individual characteristics.Variable-based models are usually stated in the form of quantified sentences (scientific laws).This syntactic structure implies that sentences about individual cases are derived using deductive reasoning.In contrast, case-based models are usually stated using context-dependent existential sentences (qualitative statements).This syntactic structure implies that sentences about other cases are justifiable by inductive reasoning.We apply this representationalist perspective to the problems of generalization and replication.Using the analytical framework of modal logic, we argue that the modes of reasoning are often not only applied to the context that has been studied empirically, but also on a between-contexts level.Consequently, quantitative researchers mostly adhere to a top-down strategy of generalization, whereas qualitative researchers usually follow a bottom-up strategy of generalization.Depending on which strategy is employed, the role of replication attempts is very different.In deductive reasoning, replication attempts serve as empirical tests of the underlying theory.Therefore, failed replications imply a faulty theory.From an inductive perspective, however, replication attempts serve to explore the scope of the theory.Consequently, failed replications do not question the theory per se, but help to shape its boundary conditions.We conclude that quantitative research may benefit from a bottom-up generalization strategy as it is employed in most qualitative research programs.Inductive reasoning forces us to think about the boundary conditions of our theories and provides a framework for generalization beyond statistical testing.In this perspective, failed replications are just as informative as successful replications, because they help to explore the scope of our theories.
