580 California St., Suite 400
San Francisco, CA, 94104
Key research themes
1. How can changes in response processes affect the validity and interpretation of longitudinal patient-reported outcome measurements?
This research area investigates the phenomenon of response shift, which occurs when the meaning of a respondent’s self-evaluation changes over time, particularly in the context of patient-reported outcomes (PROs). The focus lies on defining response shift, developing theoretical frameworks, and refining methods to detect its occurrence in longitudinal studies. Accounting for response shift is critical because unrecognized changes in response processes can bias inferences about true health changes and affect the validity of clinical and policy decisions based on PROM scores.
Key finding: This paper advances the conceptualization of response shift by refining its formal definition to emphasize that it is an effect on observed change not attributable to true changes in the target construct but rather due to... Read more
Key finding: Although primarily focused on mediation analysis, this study provides theoretical and simulation evidence relevant to response shift research by demonstrating how adjustment for covariates predictive of outcomes and mediators... Read more
Key finding: This theoretical paper extends causal regression models by incorporating covariates to estimate both conditional and unconditional average causal effects. It elucidates how conditional adjustment can account for confounding... Read more
2. What are the mechanisms, types, and statistical interrelations of response effects such as mediation, confounding, suppression, and response biases in psychological measurement?
This theme encompasses the conceptual and statistical distinctions among mediation, confounding, and suppression effects, exploring their mathematical equivalences and differences in interpretation. It also covers systematic response biases present in self- and other-report questionnaires, including acquiescence and extreme responding, as well as social desirability biases. Understanding these biases and effects is fundamental for designing valid instruments, controlling for measurement error, and interpreting observed relationships in psychological and behavioral data.
Key finding: This paper rigorously shows that mediation, confounding, and suppression effects share the same statistical quantification—the change in the relationship between an independent and dependent variable upon inclusion of a third... Read more
Key finding: This comprehensive review identifies key response biases affecting questionnaire data, including response styles (e.g., extreme response style, acquiescence), rater biases (e.g., halo effect, leniency/severity bias), and... Read more
Key finding: Through simulation, this study compared three approaches to correcting extreme response style (ERS) effects in trait estimation—mixed Rasch models, multidimensional item response models, and regression residual correction.... Read more
3. How can response time and accuracy interdependence be modeled to better understand cognitive and behavioral processes?
This research area focuses on jointly modeling response time and accuracy data to infer underlying mental process architectures and to capture conditional dependencies between speed and correctness beyond overarching latent traits like ability and processing speed. By extending traditional hierarchical IRT frameworks and incorporating residual dependencies, these approaches reveal nuanced insights into item properties and respondent behaviors, advancing precision in psychological and educational measurement and process modeling.
Key finding: This paper proposes an extension to hierarchical joint models of response time and accuracy by explicitly modeling residual dependence between them after accounting for latent ability and speed. By allowing item parameters... Read more
Response time distributions: Some simple effects of factors selectively influencing mental processes
Key finding: This work advances the methodology of analyzing entire response time distributions (beyond means and variances) to infer mental processing architectures, including serial and parallel processing and logical gating (AND/OR).... Read more
Key finding: By experimentally manipulating blackout durations following responses in pigeons under fixed-ratio and fixed-interval reinforcement schedules, this study independently varied interreinforcement time and number of responses.... Read more