device: MyGaze eye tracker. First, the effects of banner format
(horizontal banners on the top of a webpage vs. skyscraper banners
at the side) were tested. Second, the effects of animation (static
banners vs. animated banners) as well as animation speed (high
speed vs. low speed) were tested. Third, the interaction effects of
banner formats and animation on banner effectiveness were
investigated.
Given the low temporal resolution of the MyGaze eye tracker,
instead of measuring fixations, we focused on the number and the
average duration of contacts with the AoIs as rough measures of
attention distribution. Moreover, we investigated attention in two
time spans (i.e., the full experimental time span of 180 s, and the
first 30 s) in order to explore whether or not viewers' attentionchanges during the time they view a webpage.
The findings of the banner format's effects on attention were
mixed. We found that banner formats had significant effects on the
number of contacts. That is, no matter how long the time span was,
the first 30 s or the full 180 s, horizontal banners were found to grab
significantly more contacts than skyscraper banners. However, the
average duration of contacts on banners indicated different results
on attention. During the first 30 s, no significant effects of banner
format were found on the average duration of contacts; there were
no significant differences on contact durations at the beginning
when users were surfing internet. But during the full 180 s,
skyscraper banners were found to grab significantly longer contact
durations than horizontal banners. This result indicates that
different banner formats may attract attention in different ways.
Horizontal banners keep grabbing greater number of contacts than
skyscraper banners during the time when users surf on the
internet, which indicates users are more frequently to be attracted
by horizontal banners. However, the different results of contact
durations show that, even though a horizontal banner can grab
more contacts; it cannot hold people's contacts longer than
skyscraper banners. And with time passed, it even holds people's
contacts shorter than skyscraper banners. An explanation for this
might be that horizontal banners at the top of a webpage are more
obvious and salient to users so that users' attention can be caught
by it more easily and frequently. Whereas after a short time of
viewing the webpage freely, people start getting into a more stable
status and start reading the information-rich area, and then, during
the longer viewing time, skyscrapers attract longer attention
because they are close to the information-rich area.
This finding was not the same with the finding of Kuisma et al.’s
(2010) eye-tracking study, which indicated that skyscraper banners
attracted more visual attention than horizontal banners based on
the number of fixations. The reasons for the different findings between
their study and the current study may mainly exist in a
couple of aspects. First, the measures between these two studies
were not exactly same. Kuisma et al. (2010) measured attention by
fixation frequency, and they only used this variable. While in the
present study, we measured attention by both contact frequency
and contact duration, which may cause the result differences and
also enable this study to obtain more elaborate findings. Second, we
analyzed the data in different time spans, which also provided us
more elaborate findings. Third, the opposite attention results of
their fixation frequency and our contact frequency may be because
these two studies were conducted in different experimental
modes: Kuisma et al. (2010) used task mode, while we used surfer
mode. According to the previous research, visual selection attention
is influenced by intentions of the action (Hannus, Cornelissen,
Lindemann, & Bekkering, 2005). A specific action intention is
more likely to lead users to focus attention on the action-relevant
information (Bekkering & Neggers, 2002). This case suggests a
top-down process, in which visual attention is user-driven
(Bekkering & Neggers, 2002; Hamborg et al., 2012). In contrast to
Kuisma et al.'s (2010) study, which instructed participants to do a
reading task, the present study conducted a task-free experiment in
a surfer mode (i.e., the participants were not required to do reading
tasks or search for specific information). The surfer mode follows a
bottom-up process, in which visual attention is stimulus-driven. In
this case, users browse information without a specific task intention,
so that the salient stimuli are much easier to grab their
attention (Bekkering & Neggers, 2002; Hamborg et al., 2012).
Therefore, based on the results, we suggest that when users are
surfing on the internet, horizontal banners may be able to attract
more attention with regard to the number of contacts than skyscrapers,
as skyscrapers do not have the advantages of being close
to the task area during surfer mode. However, to test whether ornot this result difference comes from the search or browsing mode
difference, further research with a rigid experimental design is
needed.
As for the effects of animation on attention, we found that both
the presence of animation and animation speed mattered, which
was consistent with previous studies (e.g. Sundar et al., 1998; Borse
& Lang, 2000; Yoo et al., 2004; Yoo & Kim, 2005). Nevertheless, our
findings showed that whether or not animated banners attracted
significantly more attention than static banners depended on the
banners' animation speeds and the time span.
During the first 30 s, the banners with a high animation speed
grabbed more contacts than static ones, but there was no difference
between the effects of low-speed animated banners and static
banners. This may be because that at the beginning when a user
was surfing the internet, highly dynamic content attracted people's
eyes more frequently. However, a longer contact duration on the
banners with a low animation speed was found during the first 30 s,
which indicated that even though at the beginning of Internet
surfing, a high-speed animation attracted contacts more frequently,
it still could not hold attention longer than a low-speed animated
banner. This may be because comparing to a moderate low-speed
animated banner, a high-speed one looks too flashing and
workload-intensive, so that people try to avoid looking at it (Burke
et al., 2005; Hamborg et al., 2012). For the full 180-s duration, the
banners with a low animation speed grabbed more contacts than
the static ones; and both speeds of animated banners grabbed
longer contact duration than static banners. This also confirmed
that compared to animated banners, static banners were relatively
less effective at catching people's attention. Nevertheless, no significant
difference was found between the effects of high-speed
animation and low-speed animation on contact frequency. This
finding was not consistent with Yoo and Kim's (2005) study, which
showed there was a positive linear relationship between animation
speed and the amount of attention. Since there were few other
studies on the effects of banner animation speed, especially for
studies using eye-tracking systems, more research on this area is
needed in the future. In addition, because we only included two
animation speed rates, the effects of animated banners with more
rates could be investigated in future research. Besides the main
effects of both banner format and animation, we explored the
interaction effects between them on attention. With regard to the
number of contacts, the interaction effects between banner format
and animation were significant no matter what time span was
examined. In terms of contact duration, however, the interaction
effects depended on the time span that users spent on the webpage.
Specifically, during the first 30 s, horizontal banners with a high
animation speed and static skyscraper ads were more effective at
attracting greater number of contacts than other types of banners
but there were no interaction effects between banner format and
animation on contact duration. This indicated that at the beginning
of surfing the internet, people may aimlessly look around on a
webpage; even high-speed animated horizontal banners and static
skyscraper banners can frequently attract users' contacts, but
neither of them are effective in terms of holding people's attention.
During the longer time span of 180 s, besides high-speed animated
horizontal banners and static skyscrapers, low-speed animated
horizontal banners were also effective at grabbing greater number
of contacts. Moreover, high-speed animated horizontal banners
and low-speed animated skyscraper banners held longer contact
durations than other banner types. Thus, horizontal banners are
more effective to catch attention when they are with animation.
And high-speed animation may make a horizontal banner look
more salient and enable it to hold longer attention during a longer
time span. In addition, at the beginning of surfing, skyscrapers
catch contacts more frequently when they are static. This may bebecause skyscrapers are close to the information-rich area; and
people may perceive the text areas should be static, so they tend to
avoid any animated skyscraper ads when they just begin surfing a
webpage. But after a short time of viewing the webpage, people
start getting into a more stable status and start reading the
information-rich area, and then, during the longer viewing time,
skyscrapers with a moderate low-animation speed, which is not
exactly static or too flashing, can attract longer contact durations.
The above interaction effects findings were partially consistent
with Kuisma et al.'s (2010) study. Both studies found significant
interaction effects on attention between banner format and animation;
however, the specific findings were different because
Kuisma et al. (2010) found that static horizontal banners and
animated skyscrapers were th