That's still focused exclusively on the time spent doing the task. Maybe a more extreme example will clarify why the methodology was seriously flawed. Suppose t

That's still focused exclusively on the time spent doing the task. Maybe a more extreme example will clarify why the methodology was seriously flawed. Suppose there's a bug in the app such that it frequently crashes shortly after using the mouse. You have to restart the app and reload the data after every crash. We can all agree that using the mouse is not faster than the keyboard if you have a bug like that. 100% of users would say that using the keyboard is faster. Yet their methodology would lead to exactly the same conclusion that users are wrong and $50 million of research has shown it's faster to use the mouse. Assuming "all else equal" when it's not can lead to really bad decisions.