ra g re p la c e m e n ts
D
W
H
(a) A target of widthW and
heightH
ra g re p la c e m e n ts
D
W\u2192\u221e
H
(b) When the widthW tends
to in\ufb01nity, the task becomes
anDPtask.
P S fra g re p la c e m e n ts
D
W
H\u2192\u221e
(c) When the heightH tends
to in\ufb01nity, the task becomes
aAPtask (with the differ-
ence that the movement am-
plitude is measured to the
edge of the target instead of
its center).
Figure 4: Bivariate pointing and its two limit tasks
width, which is counter-intuitive. In the case of theIDmin
model, the interaction between width and height exists but is
very crude: it predicts that the movement time does not de-
pend onH as soon as it is greater thanW ; vice versa, it is
not affected byW as soon asW >H . This is dif\ufb01cult to be-
lieve and actually inconsistent with available empirical data.
For example, as shown in Table 3 in [5], for the same height
(e.g.10), the movement time did increase asW changed (e.g.
from40 to10). Sheikh & Hoffmann [11] also showed that
the dif\ufb01culty for acquiring a square is indeed greater than for
a rectangle, which is not captured by theIDmin model.
MOTIVATION AND DIRECTION OF THE CURRENT STUDY
The unsatisfactory state of affairs in 2D pointing motivated
us to conduct a new study to provide a more complete under-
standing of it. Our study will depart from the prior art in two
ways: experimental manipulation and theoretical modeling.
Experimental manipulation
In terms of experimental manipulation, three factors differ-
entiate our study from the prior art. First, knowing thatDP
tasks also follow Fitts\u2019 law, we viewW-dominant(W <H )
andH-dominant(H < W ) cases symmetrically and study
both as mirror situations. Second, unlike previous works that
manipulate absoluteH andW values and leave their ratio
as a dependent variable, we control theH andW ratio and
view it as a basic independent variable. The ratio, which de-
termines the shape of a rectangular target and is hence the
fundamental factor to the interaction betweenH andW , has
been an oversight in the literature. Third, we are interested in
the impact of the secondary constraint (the greater ofH and
W) as it approaches the primary constraint. The literature
never studiedH: W ratios between1 and2, and was hence
more likely to \ufb01nd only binary all-or-nothing impact. For
example, Hoffmann & Sheikh [5] tested the ratios1,2,4,5,
8, 10, 20, and 40. Our hypothesis is that large ratios will be
equivalent to no constraint at all on the low-constraint axis.
Desirable model properties
In terms of quantitative modeling, we start with a set of de-
sirable properties for a bivariate pointing model:
\u2022Scale independency: multiplyingD,W andH by a same
constant should leave movement time unchanged.
\u2022Limit tasks: the model should regress toward 1D Fitts\u2019 law
when eitherH orW tends to in\ufb01nity.
\u2022Dominance effect: the smaller ofH orW should dominate
theID value, with little or smaller impact of the other.
\u2022Duality ofH andW : bothH andW effects should be
contained in the same model and be of similar nature.
\u2022Continuity: the effect ofH andW should be continuously
represented, rather than stepwise or segmented.
New model candidates
In order to quantify a bivariate task with an univariate index
of dif\ufb01culty, we need to de\ufb01ne the appropriate \u201cdistance\u201d in a
two dimensional space, with the two dimensions determined
byW andH. The general mathematical notion of distance
we will use for that purpose is the weighted\ue001p-norm.
The weighted\ue001p-normGiven a vectorx= (x1 ,...,xn),
a real numberp\u22651 and a set of positive weightsw=
(w1 ,...,wn), thew-weighted\ue001p-norm ofx is de\ufb01ned by:
\ue001x\ue001p,w=
\ue004
n
\ue00a
i=1
wi|xi|p
\ue005
1/p
(7)
The weighted\ue001p-norm generalizes common norms, such as:
\u2022the\ue0011-norm with unitary weights:
\ue001x\ue0011=
n
\ue00a
i=1
|xi|= |x1|+... + |xn|
(8)
also known as the city-block or Manhattan norm.
\u2022the\ue0012-norm with unitary weights:
\ue001x\ue0012=
\ue004
n
\ue00a
i=1
|xi|2
\ue005
1/2
=
\ue00b|x1|2+... + |xn|2
(9)
which is the well-known Euclidean norm.
\u2022the generic\ue001p-norm(p > 0) with unitary weights:
\ue001x\ue001p=
\ue004
n
\ue00a
i=1
|xi|p
\ue005
1/p
(10)
also known as the H\u00f6lder or Minkowski norm.
\u2022Whenp tends to in\ufb01nity, the value of \ue001x\ue001preaches asymp-
totically the maximum value ofxi. The\ue001\u221e-norm with uni-
tary weights is hence de\ufb01ned by:
\ue001x\ue001\u221e= max
1\u2264i\u2264n{|xi|}= max {|x1|,..., |xn|}
(11)
ans is also called the Chebyshev, uniform or max norm.
\ue001p-norm models for bivariate pointingUsing the previous
notations, we investigate models in the following form:
T= a+ blog2 (\ue001X\ue001p,w+ 1)
(12)
wherea andb are constants, andX is the constraint vector:
X=
\ue000
D
W,D
H
\ue001
(13)
Leave a Comment