The user’s aim point is systematically biased above the touch point

What it is

When people subjectively aim at a target, the intended point is often systematically above the sensor-reported touch point: they use the visible fingertip edge while the contact centroid lies lower, under occlusion. A seemingly accurate tap can consequently fall below target or on a neighbour.

Why it happens

The direction of this bias is not arbitrary — it is geometrically inevitable. When aiming, a person's eyes track the visible tip of the finger pointed at the target, but the fingertip is a soft, curved pad; pressed against flat glass at an angle, the region that actually forms contact and gets reduced to a centroid by the sensor sits below that visible tip, closer to the first knuckle. The point the eyes are aimed at and the point the sensor reports were never the same geometric location to begin with — vision fixates on the former while the sensor reports the latter. This also explains why a stylus rarely shows this problem: its tip is hard and thin, and the point a person's gaze focuses on nearly coincides with where the tip actually touches the screen, so there is no split between a "visible leading edge" and a "soft contact patch" the way there is with a finger.

Studying it

Ask people to mark intended targets, then record reported contacts and hits across fingers, grips, target positions, and orientation. Analyse mean bias and individual variation rather than letting aggregate hit rate conceal directional error. When measuring this bias, prefer a precise crosshair-style small target over a large, diffuse region — only a small, precise target reveals the distance between the aim reference point and the contact centroid consistently; a large target's own tolerance will absorb and hide this systematic bias.

Where it stops holding

Upward bias is a common tendency, not a fixed constant for every person and posture. Apps should not add a uniform empirical offset; target size, spacing, and recoverability are usually more robust than self-correction of coordinates.

Applying it

• Absorb bias with adequate hit areas and spacing rather than asking people to tap lower.
• Use visible offset cursors or magnified confirmation for precision work.
• How to check: measure the bias between the aim point and the reported contact point using a precise crosshair target rather than a large diffuse one — only this isolates the geometric systematic bias from the target's own tolerance for error.

Related

• Same group: C2.06.2 Touch bias varies with finger angle and screen position · C2.06.3 Offset compensation belongs at the system layer and should not be repeated by applications
• Nearby: C2.01 Finger contact area and touch centroid · C2.05 Finger occlusion
• Search terms: touch offset · aim bias · finger angle