Validity model proposal applied to human sensors in quality engineering planning: A psychometric approach
Abstract
This dissertation offers a comprehensive mathematical model to conceptualize the interaction between soft measurements and latent traits. Traditionally, engineers divide the measurement process into two groups. The first comprises hard measurements, or concrete variables such as length and velocity; whereas the second comprises soft measurements, or nonconcrete variables such as customer satisfaction and job performance. Hard measurement instruments are widely recognized and accepted by engineers. However, soft measurement instruments, generally questionnaires, are often mistrusted due to their perceived lack of reliability and validity. A proposed model--called the $\nu$-function--employs psychometric principles, latent variables, and the concept of human sensors to enable quality engineers to estimate not only reliability coefficients but also a novel validity index, $\gamma\sp*$, for soft measurement instruments. One of the main advantages of using quantitative parameters to control the precision and the exactness of the measurement process is the rise in the confidence with which inferences are made to and predictions are performed from a particular domain. A case study is also conducted to collect evidence to discuss the efficiency and efficacy of the $\nu$-function according to empirical data.
This paper has been withdrawn.