As Fast and as Accurately as Possible: Toward a Renewed Understanding of Two-Dimensional Performance Measurement in Human Behavioral Experimentation
June 15th, 14:00
Online Talk: link removed after the event
The presentation will pretty much revolve about pointing. I will present a critique of the somewhat authoritarian ISO standard that has been ruling Fitts' law experimentation for two decades in HCI. To the extent that Fitts' law is a tool for the practical evaluation of interfaces and pointing devices, the standard has been useful. However, HCI happens to have gradually become the primary field for basic research inquiries into the general issue of pointing: from a basic research viewpoint an industrial standard has always been undesirable.
The ISO standard having made Fitts’s (1954) experimental paradigm sacrosanct, it is all the more important to try to remain lucid about the weaknesses of that paradigm. Concerning its independent variables (all about what we are supposed to do as experimenters), it is not true that the so-called "index of difficulty" (ID) controls movement difficulty: what the ID controls is the relative balance of the concurrent speed and accuracy demands; and we have the problem that the recommended factorial design for the manipulation of the ID is seriously flawed. Concerning the dependent measures (all about the performance measures we decide to consider in concertation with our participants), we have the problem that the recommended data-processing options, based on blind multiple-level averaging, are hard to reconcile with the recommended formulation of task instructions, which urge participants to try their best to minimize movement time and endpoint error.
Some alternative options will be presented regarding both the design of experiments and the definition/elaboration of performance data in a basic approach to pointing.