Falsification testing for usability inspection method assessment
Alan Woolrych, Gilbert Cockton, Mark Hindmarch · Northumbria Research Link (Northumbria University) · 2004
We need more reliable usability inspection methods (UIMs), but assessment of UIMs has been unreliable [5]. We can only reliably improve UIMs if we have more reliable assessment. When assessing UIMs, we need to code analysts’ predictions as true or false positives or negatives, or as genuinely missed problems. Defenders of UIMs often claim that false positives cannot be accurately coded, i.e., that a prediction is true but has never shown up through user testing or other validation approaches. We show this and similar claims to be mistaken by briefly reviewing methods for reliable coding of each of five types of prediction outcome. We focus on falsification testing, which allows confident coding of false positives.