Throughput across tasks and postures cannot be compared directly

What it is

The context dependence of pointing throughput means the same device can yield different throughput under different target tasks, display configurations, postures, mappings, and practice. A value describes the human-device-task system under its measurement conditions, not a device's universal performance rank.

Why it happens

Throughput comes from a regression relating effective difficulty to movement time, and the regression itself — its intercept and slope — is a product of the context, not a fixed property of the device. Sitting versus standing changes bodily support and fatigue; display size, gain, and target direction change the range of motion and which muscle groups are engaged; discrete click, drag, and touch rely on entirely different sensory feedback during the terminal correction phase. Even with an identical formula, the control task the input channel is performing is no longer the same task: the slope measured for a stylus in a single-direction repeated-tapping task and the slope measured for the same stylus in the sixteen-direction ring task specified by ISO 9241-9 describe two different bundles of movement strategy, and calling both "throughput" does not make them the same number. This is exactly why standardized task paradigms such as the multi-directional tapping task exist — to hold task geometry constant as one source of context — while posture and display size remain outside that standard.

Studying it

Report the task paradigm, the distance-width combinations, device, gain, display specification, posture, participant expertise, practice amount, error-handling rule, and the exact throughput calculation (nominal or effective width, how the regression coefficients were fit) in full. To compare multiple contexts, treat context itself as an independent variable and analyse its interaction within one experimental design and one participant pool, rather than pulling a single mean throughput figure from each of several unrelated papers and settings and assembling a ranking — that move effectively subtracts the slopes of two different regression lines while hiding the confound between task geometry, posture, and device.

Where it stops holding

Context dependence does not make cross-device comparison useless: under strictly matched task paradigm, display configuration, and posture, a throughput difference is strong evidence of a real difference in device control quality. The actual error is extrapolating a value measured seated, in 2D, at one display size, straight onto mobile use, freehand drawing, public displays, or sustained long-session work — those settings change not just the difficulty number but the movement strategy itself. Meta-analyses across studies face the same requirement: comparable task paradigm and calculation convention must be established first; numbers that happen to be numerically close are not automatically methodologically comparable.

Related

• Same group: C1.15.1 Throughput captures a pointing device's speed-accuracy tradeoff, not speed alone · C1.15.2 Fitts' law models movement time as a logarithmic relation to target distance and width · C1.15.3 Effective width is recalculated from endpoint distribution rather than nominal width · C1.15.5 Mean completion time alone hides speed-accuracy differences
• Nearby: Q1 Research methods and evaluation · C1.01 Types and properties of pointing devices
• Search terms: pointing throughput · experimental context · posture