Addressing traceability in social measurement: establishing a common metric for dependence


Authored by  T Salzberger, S Cano, L Abetz-Webb*, E Afolalu, C Chrea, R Weitkunat, K Fagerström*, Jed E Rose PhD*

Published in Journal of Physics: Conference Series    
* This author is not affiliated with PMI.

Abstract

Measurement in the social sciences is characterised by a multitude of incompatible paradigms, most of which fail to adhere to fundamental principles of measurement explicated in metrology. This has led to a fragmentation of instruments that are hard to interpret and lack a common reference. Rasch measurement theory, the metrological framework of the Rasch model, combined with strong substantive theories of the measurand, has the potential to advance measurement in the social sciences significantly. The Rasch model establishes a reference standard with a common unit of measurement against which different instruments can be calibrated ensuring traceability in social measurement. Uncertainty in the calibration process is addressed by a standard error of measurement for a specific estimate taking into account that precision varies over the continuum of the measurand. A strong substantive theory of the measurand allows for concrete predictions that can be tested empirically with positive evidence supporting accuracy. We exemplify the potential of the Rasch model by developing the ABOUT-Dependence, a new instrument for the measurement of dependence on tobacco- and nicotine-containing products (TNP). The instrument provides comparable measurements of dependence on different TNPs as well as dependence on multiple TNPs used concurrently. A co-calibration of the new instrument and existing, product-specific instruments allows for a conversion of scores of existing instruments into scores on the new instrument, and vice versa, via the established equal-unit reference scale. The score comparability adds to traceability in the measurement of dependence, is likely to facilitate the acceptance of the new instrument, and helps avoid a discontinuity in the research of dependence.