Overview

Earlier topics chose a representation, a document, and a labelling method for disclosure tone. This topic asks whether the label is a measurement, and whether a measurement is a tradable feature. Agreement among coders, validity against the construct, and the point-in-time and multiple-testing checks are three different deliverables. Passing one does not pass the others.

By the end of this note you should be able to:

• describe chance-corrected agreement in words, using Gwet’s numerical example, and flag a garbled equation for the PDF;
• apply the five measurement principles and a held-out confusion matrix to a tone score;
• state why a text score still has to face point-in-time availability and the research search before anyone trades it.

1. Contents
2. The big picture