Accelerat ing t he world's research.

Online newspapers and ad banners:

an eye tracking study on the effects of
congruity
Marco Porta, Francesco Spaghi

Online Information Review

Cite this paper Downloaded from Academia.edu 

Get the citation in MLA, APA, or Chicago styles

Related papers Download a PDF Pack of t he best relat ed papers 

Reint egrat ing t he Ad

Diana Rieger

On-line newspapers and mult imedia cont ent

Marco Port a

T he Effect s of Animat ion and Format on t he Percept ion and Memory of Online Advert ising
Anssi Oorni, Jaana Simola
 The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1468-4527.htm

Online
Online newspapers and ad newspapers and
banners: an eye tracking study on ad banners
the effects of congruity
405
Marco Porta, Alice Ravarelli and Francesco Spaghi
Dipartimento di Informatica e Sistemistica, Università di Pavia, Pavia, Italy Received 3 January 2012
First revision approved
19 May 2012
Abstract
Purpose – The purpose of this paper is to study the possible influence of congruity on the way ad
banners are watched and remembered in online newspapers.
Design/methodology/approach – The effects of the thematic connection between banner subject
and article content are examined at the level of gaze behaviour using eye tracking technology, while
questionnaires are exploited to assess both free and helped recall. Two “kinds of reading” are also
considered, namely “imposed” (where the tester has to read a specific article) and “free” (where the
tester can choose the article’s topic from a set).
Findings – Results indicate that congruity matters for the way banners are looked at, although it
seems to have a limited influence on memory. Moreover, the kind of reading is significant in a few
cases.
Practical implications – The design of ad banners for web content is very critical, due to the
so-called “banner blindness” phenomenon. At the same time, ad banners are crucial for online
newspapers, which rely on them as the main source of revenues. Understanding the potential of
congruity is therefore an essential point.
Originality/value – Results to date obtained from previous research about congruity between online
content and ad banners are contrasting, and further investigations are thus necessary. Moreover, in the
context of online newspapers, congruity has been considered in extremely few cases and, to the best of
the authors’ knowledge, without using eye tracking technology. The outcomes of this study can be
exploited to drive further research in the field.
Keywords Internet, Web sites, Advertisements, Online newspapers, Ad banners, Congruity,
Eye tracking, Newspapers, Advertising effectiveness
Paper type Research paper

Introduction
Online newspapers are increasingly becoming one of the major sources of news for
many people. As the number of readers increases – also thanks to the rising popularity
of tablets – companies invest considerable amounts of money to advertise their
products within information webpages, primarily in the form of ad banners. Despite
intensive research arguing that this kind of promotion is hampered by several issues
that limit its effectiveness, banners are widely exploited in practically all electronic
newspapers, and come in a variety of styles and formats.
The term “banner blindness” refers to the well-known phenomenon in which web
surfers tend to ignore page elements that look like ad banners (e.g. Drèze and Hussherr,
Online Information Review
2003; Nielsen, 2007), thus making most of the efforts of advertisers and graphic Vol. 37 No. 3, 2013
designers pointless. To improve the likelihood of banners being noticed, several pp. 405-423
q Emerald Group Publishing Limited
solutions have been proposed and tested, with variations on message components 1468-4527
(e.g. with or without graphic elements, static, animated, etc.), dominant colours, size, DOI 10.1108/OIR-01-2012-0001
OIR position and other factors. In this context the thematic congruity between banner
subject and webpage content has also been considered, as a potential way to overcome
37,3 users’ distrust of banner-like elements.
In general, “congruity” refers to the relationship between the nature of the
content of an ad and its surrounding context, and to the association that the user
can produce between them. Some studies (e.g. Anderson, 1983; Heckler and Childers,
406 1992) have demonstrated that congruent stimuli can create an enhanced memory
trace, making it easier for people to recall things. In the marketing context,
considering the relation between products and brand names, it has been shown that
congruity can generate a comfortable situation where the individual finds expected
and repeated things according to a mental schema he or she already has in mind,
thus generating a positive and focused attitude (Mandler, 1982). Conversely some
findings also reveal different perspectives, showing that incongruent information
may enhance the consumer’s cognitive arousal (e.g. Mandler, 1982; Meyers-Levy
et al., 1994).
The study we present in this paper is primarily focused on the possible influence of
congruity (between banner subject and article content) on the way readers relate to ad
banners, both in terms of “how” they look at them and with regard to the memorability
of the message. Moreover our work introduces a further aspect that represents a
novelty in this kind of investigation, namely the “reading condition”: imposed or free.
Usually, experiments aimed at exploring user behaviour while interacting with web
content involve specific pages, prepared on purpose. The tester is asked to perform
some tasks on some pages, which may vary among testers, but are typically imposed
on the single participant in the experiment. Considering our online newspaper setting,
such a situation does not correspond to what happens in a real scenario, where it is the
reader who chooses what to read. In our study we have therefore considered both the
common experimental case in which the tester is asked to read a specific article, and
the more “natural” condition in which the tester can choose among a set of topics,
according to his or her preference. Our assumption is that free reading can better reflect
what occurs in a real usage scenario, possibly producing different results compared to
imposed reading. The choice to include this further experimental variable in our study
was indirectly inspired by the work of Pagendarm and Schaumburg (2001), who
argued that differences in navigation style may influence the way users relate to
banners.
Our investigation exploits eye tracking, a technology that allows the user’s gaze
to be monitored while interacting with a computer (Duchowski, 2007). Eye
movements occur as sudden, almost instantaneous (,100 ms) “saccades”, followed
by fixation periods (, 100-600 ms) during which the eye is almost still. Fixations are
the basis for the analysis of user activity and performance, because they can be
considered directly connected to the concept of “attention”. In fact, eye tracking
studies of user behaviour while performing a task are essentially based on the
“eye-mind hypothesis” ( Just and Carpenter, 1976, 1980), which claims the existence
of a direct correspondence between the user’s gaze and his or her point of attention.
Some experiments (Poole and Ball, 2005) have shown that while it is possible to
move one’s attention without shifting the gaze, it is difficult to move one’s gaze
without shifting the attention, which can thus be considered closely correlated to
eye behaviour. As will be discussed later, however, this theory cannot be considered
separately from “inattentional blindness” (Mack, 2003), according to which
sometimes we look at things without being fully aware of them.
 Our study stems from three main considerations: Online
(1) results obtained to date from previous research on congruity between online newspapers and
content and ad banners are inconsistent, and further investigations are thus
necessary;
ad banners
(2) in the context of online newspapers, congruity has been considered in extremely
few cases; and
(3) to our knowledge no work has really studied congruity in online newspapers
407
using eye tracking technology.

Previous work
Issues related to the thematic congruity between editorial content (both paper- and
web-based) and ads have received some interest to date. However, although most
studies have mainly focused on discovering possible effects of congruity on user
attention and recall, currently there is not a unique interpretation of findings, and
several points still remain unclear.
For example Johar and Pham (1999) noted that, in sponsor identification tasks
related to an event, sponsor recall is better in the case of brands that are semantically
connected with the event itself. Moorman et al. (2002) studied context-induced
psychological responses in print magazine ads, as well as the effect of thematic
congruity on memory. Besides noticing that liking and positive feelings positively
affected the attitude towards ads, thematically congruent ads turned out to be better
remembered than incongruent ones. The “affective context” and its relationship with
congruity in print advertising were also considered in previous works (e.g. Yi, 1990).
In the web context a recent study by Lee et al. (2010) dealt with banner advertising
and congruity in e-commerce sport websites. The experimental design took into
account two variables, namely “type of congruity” (high, medium and low) and
“endorsement” (with or without sport celebrities in the ad). The results proved a strong
effect of congruity, with users who saw highly related content showing more positive
attitudes and intentions.
Previous research by McCoy et al. (2004), however, supported the opposite thesis,
showing that banner incongruity can lead to higher attention and recall. In the study,
focused on pop-up, pop-under and in-line ads, testers were asked to search for
information about some products within an e-commerce website, designed on purpose.
Similarly, studying the effect of ads and website congruity using a three-level
congruity scale, Moore et al. (2005) noticed that users paid more attention and showed
higher recall when there was an incongruent banner subject. The authors suggest that
the contrasting results obtained in congruity investigations may depend on the specific
product category and website content considered.
A study on goal-directed navigation and web advertising (Danaher and Mullarkey,
2003) showed that highly task-oriented consumers exhibit more ad avoidance, and that
ad avoidance is related negatively to ad recall. Wang et al. (2008) focused on the
importance of congruity in moderating the relationship between forced ad exposure
(e.g. pop-up ads) and memory effects, with the relationship being intensified by
congruity and reduced by incongruity. Another investigation, by Hershberger and
Costea (2009), considered the connection between ad recall, congruity and task
orientation, showing that task-oriented navigation leads to a more marked avoidance
of ads, but that the effect can be mitigated by congruity.
OIR Other research has been aimed at examining the effects of banner ads and website
congruity on users’ attitudes towards the website of a brand, discovering that high
37,3 congruity is positively correlated with attitude (Newman et al., 2004).
In the context of online publishing Choi and Rifon (2002) examined two online
magazines and found that congruity is positively related to both attitude toward the
brand and purchase intent. With regard to online newspapers Rodgers’ (2007) study
408 reveals a significant relationship between sponsor congruity and memory. The
investigation was mainly aimed at testing whether low/high congruity yields better or
worse memory of the ad banner. Memory was measured both as free recall, in which
users were asked to list brands and products as they remembered them, and as helped
recall, in which they could choose from a list.
Eye tracking has rarely been used to study thematic congruity in online content.
Janssens et al. (2007) carried out an experiment in which they combined an ad with
congruent and incongruent webpages, measuring the impact by using attitude towards
the ad and click intention. The consequences of displaying the ad before, along with, or
after a webpage were examined. Eye tracking results, although very preliminary,
suggested that the way the medium context and the ad are watched may affect the
impact that the ad has on people. Moreover congruity effects seem to be more marked
when the number of switches between context (webpage) and ad is low, while they
become less relevant as the number of switches increases. More recently, in a study of
banner blindness, Hervet et al. (2011) found that, although congruity between ads and
webpage content had no appreciable effects on fixation duration on banners, congruent
ads were better memorised.
Moreover outside the congruity matter, there are few cases of eye tracking studies
conducted on print and online newspapers. Among these important investigations
have been performed by the Poynter Institute (Garcia and Stark, 1991; Adam et al.,
2007). In one experiment some testers were asked to surf different online newspapers
designed according to five different “styles”. Users were assigned three kinds of tasks:
(1) free surfing the sites;
(2) reading a piece of news presented either in textual or multimedia form; and
(3) choosing and using an example of multimedia content within a proposed list.
Among other results, the study provided some findings about advertising. An
important lesson learned was that the position of ad banners affects the way users
look at them (the best location being the left column). Other interesting research
includes the works by Holmqvist et al. (2003) and Zambarbieri et al. (2008), both
focused on the way online newspapers are read. Recently Ferreira et al. (2011)
examined the impact of verbal emotional cues in capturing attention on advertising
regions of a news portal.

Our study
Our experiments were aimed at studying congruity in online newspapers and finding
possible relationships between banner subjects and article content. In particular,
considering the typical case in which a single main banner is displayed along with the
article, we identified five main hypotheses (H1-H5):
H1. When the tester is explicitly asked to read a specific article from an online
newspaper, the ad banner receives more fixations (sub-hypothesis H11), with
a higher total duration (sum of all fixation durations, sub-hypothesis H12), if
 the subject of the banner is congruent with the content of the article, rather Online
than if the subject of the banner is incongruent.
newspapers and
H2. When the tester is asked to read an online newspaper article whose topic is ad banners
chosen by him or herself (from a set), the ad banner receives more fixations
(sub-hypothesis H21), with a higher total duration, if the subject of the banner
is congruent with the content of the article, rather than if the subject of the
banner is incongruent. 409
H3. When the tester is explicitly asked to read a specific article from an online
newspaper, the ad banner is better remembered (sub-hypothesis H31) and
recognised (sub-hypothesis H32) if the subject of the banner is congruent
with the content of the article, rather than if the subject of the banner is
incongruent.
H4. When the tester is asked to read an online newspaper article whose topic is
chosen by him or herself (from a set), the ad banner is better remembered
(sub-hypothesis H41) and recognised (sub-hypothesis H42) if the subject of
the banner is congruent with the content of the article, rather than if the
subject of the banner is incongruent.
H5. Hypotheses H1 and H3 (“imposed” article) are confirmed or not confirmed to a
different extent with respect to hypotheses H2 and H4 (“free choice” article).
The above hypotheses were selected according to the main aim of our investigation:
studying the possible influence of congruity on the way ad banners are watched and
remembered in online newspapers, also taking into account the “imposed” and “free”
reading conditions. The number of fixations and their total duration are the typical
metrics used in eye tracking analyses for assessing ocular activity (Duchowski, 2007),
while memory and recognition are the usual quality measures utilised to evaluate the
effectiveness of advertising solutions (e.g. Moore et al., 2005; Rodgers, 2007; Wang et al.,
2008). Besides hypotheses H1-H5, other related facts were tested and will be discussed
in the following.

Apparatus
As an eye tracker we used the Tobii 1750 (Tobii Technology AB, 2006), which
integrates all the necessary components (camera, infrared lighting, etc.) into a 17-inch
LCD monitor. With an accuracy of 0.5 degrees and relatively high freedom of
movement, the system is ideal for real-use settings, where it would be awkward to
constrain users too much in their activities.
The sampling frequency of the device is 50 Hz (i.e. the user’s gaze coordinates on the
screen are acquired 50 times a second). For the purposes of our experiments a fixation
was considered a sequence of successive samplings detected within a circle with a 30
pixel radius (,1 visual degree), for a minimum duration of 100 ms. These settings are
recommended by the eye tracker’s manufacturer for studies on media with mixed
content and are widely used for web surfing analyses (e.g. Yang, 2011).

Independent and controlled variables

Independent variables used in the experiments were:
.
congruity/incongruity between banner subject and article content (binary
variable); and
OIR .
imposed/free-choice article (i.e. the tester was asked to read a specific article or
37,3 could choose among a set of topics; binary variable).

Controlled variables were:

.
banner size: 300 £ 250 pixels, constant size, “medium rectangle” (Interactive
410 Advertising Bureau, 2011);
.
banner position: at the upper right of the online newspaper’s article, constant
position; and
.
banner and article subjects/contents.

Dependent variables
Primary dependent variables were:
.
number of fixations on the banner area;
.
total duration of fixations on the banner area;
.
correctly remembered banners (“free recall”, obtained through a questionnaire
after the eye tracking test); and
.
correctly recognised banners from a set (“helped recall”, obtained through a
questionnaire after the eye tracking test).

Further (secondary) dependent variables, not necessary to test the main hypotheses
H1-H5 but nevertheless considered to investigate other potential effects of independent
variables, were:
.
average duration of fixations on the banner area; and
.
time intervening between page load and the first fixation (if any) on the banner
area (“time to first fixation”).

Participants
A total of 35 volunteer testers aged between 20 and 60 (mean ¼ 34, 21 males and 14
females) participated in the experiments, of whom five were subsequently excluded
from data analysis for the reasons that will be explained in the procedure section. All
participants reported normal or corrected-to-normal vision and were regular internet
users. Only few of them had been previously involved in other (different) eye tracking
tests. Of the 30 valid testers, 15 were students, while the others were selected from a
variegated population composed of three unemployed graduates, three retired people,
two office workers, two retail accounts executives, a teacher, a researcher, a real estate
agent, a press attaché and a warehouse worker. The majority (19) of participants were
between 20 and 29 years old, three were between 30 and 39, four were between 40 and
49 and four were ebetween 50 and 59. All findings discussed in the following were
checked against possible gender and age biases: the same results regarding hypotheses
H1-H5 were assessed both for the male and female populations separately, as well as
for the age ranges 20-29 and 30-59. A potential age selection bias may characterise our
study, due to the predominance of testers in the range 20-29 (who, however, represent
an audience particularly engaged in web activities).
 Online
newspapers and
ad banners

411

Figure 1.
Two examples of pages
reworked from Corriere
della Sera

Experiment design
Each tester was asked to read four articles on four different pages, which were the
result of a reworking of real pages from the online version of Corriere della Sera
(Figure 1), one of the most well-known Italian newspapers. Everything was left
unchanged, apart from the ad banner placed at the upper right of the article and few
other details. This position for banners is common not only in Italy, but in many online
newspapers worldwide as well. The article also contained a picture in the upper left
corner, within the text.
The material for the experiments was prepared as follows. Two articles about
cinema were used to test hypotheses H1 and H3 (i.e. imposed reading). The subjects of
the corresponding banners were cinema (congruity between banner subject and article
content) and cars (incongruity). Five further articles about books, cooking, music, the
web and sport were used to test hypotheses H2 and H4 (free reading). In five versions
of these articles the banner subjects were the same as the articles’ topics (i.e. books,
cooking, music, the web and sport); in another five versions, the banner subjects were
different, namely radio, pubs, theatre, travel and telephones). A total of 12 different
pages/articles were therefore created and employed to form the various test
combinations. All articles were about the same length (approximately 2,600 characters,
including spaces).
We used a within-subjects design, in which each tester dealt with both values of the
two conditions set by independent variables. Therefore each participant read (not
necessarily in this order):
.
one imposed article with a congruent banner; one imposed article with an
incongruent banner;
. one free-choice article with a congruent banner; and
.
one free-choice article with an incongruent banner.
OIR Considering the presentation order of these four cases, 24 combinations were obtained.
At the end of each page a link to the next page in the sequence was added, in order for
37,3 the experiment to be carried out without interruptions. The other links present in the
original pages from Corriere della Sera were disabled (but not removed), to prevent
testers from loading pages extraneous to the study.
A total of 12 banners, moderately animated, were created using Adobe Flash CS4.
412 To avoid any kind of influence from known brands, the banners advertised imaginary
products or services. Apart from the specific textual and visual content, all banners
had a similar structure (Figure 2), and were based on the typical ad model which starts
with a question (e.g. “Are you looking for a different night out?”) and ends with an
answer (e.g. “Don’t miss this occasion”) and with an invented product/brand
(e.g. “Dream café, more than 10 cocktails to try”). Figure 2 shows these final parts of
banner display.
Free-choice articles were preceded by a “topic selection page” displaying five
buttons that listed five available topics (Figure 3): by clicking on them, the
corresponding articles were loaded. The second time the topic selection page was
shown, a warning advised the tester not to choose the topic they picked the first time.
After reading the four articles the tester was presented with an online questionnaire
composed of three questions:
(1) Did you notice any advertising in the pages you visited? (Available answers:
“Yes”, “No”, “I don’t remember”);
(2) (Only if the answer to the previous question was “Yes“) Do you remember
where the advertising element was placed with respect to the article? (Available
answers: “Upper Left”, “Upper Right”, “Lower Left”, “Lower Right”); and
(3) (Only if the answer to the first question was “Yes”) Do you remember the
theme/brand of any ad banners you noticed in the page? (The tester had to write
his or her answers within a text area in the page).
Last, the tester was presented with a page containing all the 12 banners created for the
experiments and was asked to select the possible banners he or she had noticed while
reading the four articles (although the tester could have seen only four banners out of
the 12, as only one banner was displayed along with each article). Of course a
participant’s interest in the specific content of an incongruent banner may have
influenced results. We have tried to reduce this problem, affecting all studies on
congruity, by choosing very common banner subjects, uniformly distributed among
testers.
All pages employed in the experiments were published on a web server and were
accessible online, with server-side data management implemented in PHP.

Procedure
Prior to each test session, participants were instructed to read the articles they would
be presented with. Testers were also explicitly told that, although reading the articles
was a mandatory task, they would not be questioned about their contents. This
clarification was necessary – suggested by some preliminary trials – in order to
prevent participants from paying too much attention to the articles’ content, thinking
that it was their memory that was being investigated. Such behaviour was to be
avoided, as it would have affected the testers’ attitude towards the trial’s tasks: in fact
 Online
newspapers and
ad banners

413

Figure 2.
Banners used in the
experiments
OIR
37,3

414

Figure 3.
“Topic selection page”
listing the available article
themes: libri (books),
cucina (cooking), musica
(music), the web and sport

in normal situations users read online newspapers for their “pleasure”, without putting
any excessive effort into the activity.
After the experiments we explicitly checked the presence of fixations throughout the
rows of all articles read by testers, and had to reject three participants who did not fulfil the
“whole reading” requirement. Two further testers had to be excluded because of poor eye
tracking data. Therefore we considered 30 valid tests out of 35 (18 males and 12 females).
As mentioned above, each tester read four articles in sequence. The 24 combinations
given by the conditions “imposed/free-choice article” and “congruent/incongruent
banner”, as well as by the presentation order of these four cases, were randomly
assigned to 24 testers. For the remaining 11 participants, 11 combinations (still
randomly chosen from the available 24) were replicated. After reading the articles the
testers answered Questions 1, 2 and 3 and selected the banners they remembered
seeing in the visited pages.
On average a test session, including task explanation, device calibration and any
other practical tasks, lasted about 25 minutes.

Results
It is interesting to note that the banner blindness phenomenon was definitely evident in
our experiments too, affecting all the four test cases but not showing significant
differences among them. In particular there were 12 occurrences of banner blindness
(with not even one fixation on the banner area) in the case of imposed reading and
congruity between banner and article, 14 occurrences in the case of imposed reading
and incongruity, 11 occurrences in the case of free reading and congruity, and
13 occurrences in the case of free reading and incongruity. In the following statistical
analysis we will consider both the whole group of 30 testers and the subsets of
participants for whom at least one fixation was detected on the banner area in at least
one of the two cases being compared (therefore excluding total banner blindness
instances).
 The analysis was carried out using non-parametric statistics, since data populations Online
did not satisfy the normality assumption (assessed both by looking at histograms
and, more formally, through the Kolmogorov-Smirnov test). The significance level was
newspapers and
5 per cent. ad banners
As regards the average duration of fixations on the banner area and the time
intervening between page load and the first fixation – if any – on the banner, both
these secondary dependent variables did not exhibit any relation either with 415
congruity/incongruity between article and banner or with imposed/free reading; this
was true both considering all testers and excluding total banner blindness cases.
Therefore such variables will not be taken into account in the following, apart from
some considerations about the correlation between number of fixations and time to
first fixation.

H1
H1 states that when the tester is explicitly asked to read a specific article from an
online newspaper (i.e. imposed reading), the ad banner receives more fixations
(sub-hypothesis H11), with a higher total duration (sub-hypothesis H12), if the subject
of the banner is congruent with the content of the article.
Wilcoxon signed ranks tests indicated that:
.
Number of fixations: banners congruent with their article received more fixations
(median ¼ 2:0) than incongruent banners (median ¼ 1:0): Z ¼ 2:02, p ¼ 0:044,
r ¼ 0:37. This was also true considering only the 22 cases without total banner
blindness: with congruity the median score was 8.0, while with incongruity
median ¼ 1:5: Z ¼ 2:02, p ¼ 0:044, r ¼ 0:43 (of course Z and p do not change,
since only couples of identical values have been removed from both populations).
Sub-hypothesis H11 is therefore verified.
.
Fixation duration: with congruent banners median ¼ 338:5 milliseconds, while
with incongruent banners median ¼ 169:0, Z ¼ 1:87, p ¼ 0:062, r ¼ 0:34.
Considering only the 22 cases without total banner blindness: median ¼ 1; 038:5
with congruence and median ¼ 408:5 with incongruence: Z ¼ 1:87, p ¼ 0:062,
r ¼ 0:4. Thus, even if narrowly, sub-hypothesis H12 seems not to be verified.
Figure 4 shows the boxplots for number of fixations and fixation duration, referring to
cases without total banner blindness only. The bottom and top of the box are the 25th
and 75th percentiles, while the inner band indicates the 50th percentile; the ends of the
whiskers represent the smallest and largest non-outlier values; circles denote outliers
standing more than 1.5 box-lengths above the box; and stars indicate extreme values,
standing more than three box-lengths above the box.

H2
H2 states that when the tester is asked to read an online newspaper article whose topic
they have chosen from a set (i.e. free reading), the ad banner receives more fixations
(sub-hypothesis H21), with a higher total duration (sub-hypothesis H22), if the subject
of the banner is congruent with the content of the article.
Wilcoxon signed ranks tests showed that:
.
Number of fixations: congruent banners received more fixations (median ¼ 3.0)
than incongruent banners (median ¼ 1:0): Z ¼ 2:4, p ¼ 0:016, r ¼ 0:44. This
was also true considering only the 21 cases without total banner blindness: with
OIR
37,3

416

Figure 4.
Imposed reading: (a)
boxplots for number of
fixations, and (b) fixation
duration (only cases
without total banner
blindness)

congruity the median score was 11.0, while with incongruity median ¼ 4:0:
Z ¼ 2:4, p ¼ 0:016, r ¼ 0:52. Sub-hypothesis H21 is therefore verified.
.
Fixation duration: congruent banners were watched for longer (median ¼ 568:0)
than incongruent banners (median ¼ 179:5): Z ¼ 2:45, p ¼ 0:014, r ¼ 0:45. This
was also true considering only the 21 cases without total banner blindness: with
congruity median ¼ 1; 996:0, while with incongruity median ¼ 798:0: Z ¼ 2:45,
p ¼ 0:014, r ¼ 0:53. Thus sub-hypothesis H22 is also verified.
Figure 5 shows the boxplots for the number of fixations and fixation duration,
referring to cases without total banner blindness only.

H3
H3 states that when the tester is explicitly asked to read a specific article from an
online newspaper (imposed reading), the ad banner is better remembered (free recall,
sub-hypothesis H31) and recognised (helped recall, sub-hypothesis H32) if the subject
of the banner is congruent with the content of the article. “Remembered” means that
the tester correctly recalled the theme/brand of the banner when answering Question 3
(see experiment design section). “Recognised” means that the tester correctly identified
the banner among the 12 presented.
 Online
newspapers and
ad banners

417

Figure 5.
Free reading: (a) boxplots
for number of fixations,
and (b) fixation duration
(only cases without total
banner blindness)

Cochran’s tests showed that:

.
Memory: with congruent banner and article, nine banners were remembered (and
therefore 21 were not); with incongruent banner and article, seven banners were
remembered: Q ¼ 0:3, p ¼ 0:564. Thus no relation can be claimed between
banner/article congruity and remembered banners.
.
Recognition: with congruent banner and article 15 banners were recognised; with
incongruent banner and article nine banners were recognised: Q ¼ 2:0,
p ¼ 0:157. In this case too it is not possible to state any connection between
banner/article congruity and recognition rate.
Therefore both H31 and H32 are not verified.

H4
H4 states that when the tester is asked to read an online newspaper article whose topic
they have chosen from a set (i.e. free reading), the ad banner is better remembered (free
recall, sub-hypothesis H41) and recognised (helped recall, sub-hypothesis H42) if the
subject of the banner is congruent with the content of the article.
Cochran’s tests showed that:
.
Memory: with congruent banner and article seven banners were remembered,
while with incongruent banner and article only one banner was remembered:
OIR Q ¼ 6:0, p ¼ 0:014. Thus there is a relation between banner/article congruity
and remembered banners.
37,3 .
Recognition: with congruent banner and article 16 banners were recognised,
while with incongruent banner and article 10 banners were recognised: Q ¼ 2:57,
p ¼ 0:109. No plain relation can in this case be claimed between banner/article
congruity and recognised banners.
418
Hence only sub-hypothesis H41 is verified.

H5
H5 states that H1 and H3 (imposed reading) are confirmed or not confirmed to a
different extent with respect to H2 and H4 (free reading). In other words, H5 aims at
finding possible effects of the imposed/free-choice article conditions on the way the
banner is watched, remembered and recognised. Of course, H5 could be indirectly
tested through H1-H4, but we prefer to clearly explicate its focal points.
Considering only articles congruent with the banner, Wilcoxon signed rank tests
indicated that:
.
Number of fixations: there was no significant difference between imposed
(median ¼ 2) and free (median ¼ 3) reading: Z ¼ 0:02, p ¼ 0:984.
.
Fixation duration: there was no significant difference between imposed
(median ¼ 338:5) and free (median ¼ 568) reading: Z ¼ 1:12, p ¼ 0:263.
Analogously, considering only articles incongruent with the banner, Wilcoxon signed
rank tests showed that:
.
Number of fixations: there was no significant difference between imposed
(median ¼ 1) and free (median ¼ 1) reading: Z ¼ 1:1, p ¼ 0:277.
.
Fixation duration: there was no significant difference between imposed
(median ¼ 169) and free (median ¼ 179:5) reading: Z ¼ 1:2, p ¼ 0:23.
Regarding memory, from H3 and H4 we draw that, considering the cases of congruity
and incongruity together, 16 banners were remembered with imposed reading and only
eight with free reading. Thus, in general, the kind of reading seems to matter for
memory. More precisely, taking into account only the case of congruity, nine banners
were remembered with imposed reading and seven banners were remembered with
free reading: Q ¼ 1:0, p ¼ 0:0317. Since the difference is not significant with congruity,
it must be in the case of incongruent article and banner that the kind of reading has an
effect on memory. Under this condition seven banners were remembered with imposed
reading and only one was remembered in case of free reading: Q ¼ 6:0, p ¼ 0:014.
As regards recognition a Cochran’s test showed that taking into account both the
cases of congruity and incongruity between article and banner, 21 banners were
recognised with imposed reading and 20 with free reading: Q ¼ 0:1, p ¼ 0:739. Thus,
in general, there seems not to be any influence of the reading condition on the
recognition rate – more specifically this was also confirmed considering the cases of
congruity and incongruity separately.

Further results
A majority (27) of the testers remembered having seen some advertising in the pages
and 22 remembered the right position of the banner (upper right).
 Globally, 24 banners were remembered (mean per participant ¼ 0:8, SD ¼ 0:96): 16 Online
testers remembered no banners, five testers remembered only one banner, eight testers
remembered two banners, one tester remembered three banners and no tester
newspapers and
remembered all the four banners. ad banners
Globally, 50 banners were recognised (mean per participant ¼ 1:7, SD ¼ 1:2): five
testers did not recognise any banner, nine testers recognised one banner, nine testers
recognised two banners, five testers recognised three banners and two testers 419
recognised four banners.
When at least two fixations (single fixations may be due to unintentional gaze shifts
to the banner) were detected on the banner, the banner was not remembered seven
times in the case of imposed reading and congruity, six times with imposed reading
and incongruity, 11 times with free reading and congruity and 14 times with free
reading and incongruity. Analogously, still with at least two fixations in its area, the
banner was not recognised four times in the case of imposed reading and congruity,
five times with imposed reading and incongruity, three times with free reading and
congruity and eight times with free reading and incongruity. When no fixations were
detected on the banner, it was not remembered, while it was recognised twice in the
case of imposed reading and congruity, once with imposed reading and incongruity,
one time with free reading and congruity and three times with free reading and
incongruity.
In the cases of congruity significant correlations between the number of fixations
and the time to first fixation were noticed (calculated only for instances without total
banner blindness): with imposed reading Spearman’s r ¼ 20:518, p ¼ 0:028, and with
free reading Spearman’s r ¼ 20:457, p ¼ 0:049. In both cases, lower values of the time
to first fixation tended to correspond to higher numbers of fixations.

Discussion and concluding remarks

The results of our experiments can be summarised as follows:
.
With imposed reading congruity increases the number of fixations on the banner,
although the total fixation duration is not markedly affected. Moreover congruity
affected neither memory nor recognition.
.
With free reading congruity increases both the number of fixations on the banner
and the total fixation duration. Congruity also enhances memory, but seems not
to noticeably affect recognition.
.
Although no appreciable differences can be claimed between imposed and free
reading for either the number of fixations or the total fixation duration on the
banner, the incongruity case banners turn out to be better remembered with
imposed reading. As for recognition there is no substantial divergence between
the imposed and free reading conditions.
.
Most (90 per cent) testers said they had noticed at least one advertisement during
the trials (53 per cent in the correct position). Globally 20 per cent of banners
were remembered (with about 47 per cent of testers recalling at least one banner),
while 42 per cent of banners were recognised (with 83 per cent of testers
identifying at least one banner).
.
Globally 32 per cent of banners were not remembered even if there were at least
two fixations on them, and 17 per cent were not recognised. With no fixations at
all, no banner was remembered, while 6 per cent were recognised.
OIR .
With congruity the number of fixations on the banner was inversely correlated
37,3 with the time to first fixation on the banner itself.

According to our results congruity does have a positive effect on the number of
fixations on the banner and on their total duration. Free reading seems to favour this,
since both the number of fixations and their duration increased. In this respect it is
420 important to observe that even if no substantial differences were found between
imposed and free reading in relation to the number of fixations and to the total fixation
duration on the banner, our assumption is that free reading can better reflect what
occurs in a real usage scenario. Thus results obtained for the free reading condition
should be regarded as a better representation of “reality”. While for imposed reading
only sub-hypothesis H11 (more fixations with congruity) was verified, both
sub-hypotheses H21 (more fixations with congruity) and H22 (higher fixation total
duration with congruity) were confirmed for free reading, as well as sub-hypothesis
H41 (better memory with congruity).
The fact that the banner is seen more does not necessarily imply that it is also more
remembered and recognised. In our experiments memory was improved by congruity
in the case of free reading, while there were no clear effects on recognition. A partial
explanation of this may be that testers answered the questionnaires only after reading
all the four articles, which was the only way for participants not to discover the trials’
purpose in advance in a within-subjects design. Another aspect to be considered is
“inattentional blindness” (Mack, 2003), which states that sometimes we do not pay
attention to what we are looking at. Therefore it may be that the tester’s gaze was
captured by the banner, but it did not hold his or her attention.
It is interesting to observe that, in the case of free reading, memory is enhanced by
congruity: when the user can select a topic of his or her interest, the banner is better
remembered if its subject is similar to the topic. This condition of free choice and
congruity is in a sense the most “natural” one, being closer to real navigation scenarios.
Another result worthy of note is that, in the case of incongruity, memory is better with
imposed reading. A possible interpretation of such behaviour is that since the tester
reads something he or she is not really interested in, he or she gets “distracted” by the
banner, which is therefore more noticed.
The fact that 32 per cent of banners were not remembered and 17 per cent were not
recognised, although at least two fixations were detected on them, may again be
explained by the inattentional blindness phenomenon. “Peripheral vision” (Greenwald
and Leavitt, 1984; Ryu et al., 2007) can instead account for 6 per cent of banners being
correctly recognised even if they received no fixations, an occurrence also reported in
other studies (e.g. Burke et al., 2005). Finally the fact that with congruity a higher
number of fixations corresponds to a shorter time to first fixation seems to indicate that
what catches the attention more (many fixations) is also watched sooner. Congruity
favours such a phenomenon, since it increases the number of fixations on the banner
(sub-hypotheses H11 and H21).
Ad banners are vital for online newspapers, which rely on them as the main source
of revenue. The banner blindness phenomenon, however, strongly reduces their
efficacy, and advertisers are striving to find effective strategies for online marketing.
Understanding more about banners and the way users relate to them is therefore an
essential requirement for website designers.
 Our study has provided a significant contribution to our understanding of the Online
effects of banner/article congruity. In particular our findings highlight that congruity
increases the number of fixations that a banner receives, as well as their total duration.
newspapers and
Results from past research are inconsistent, and suggest that congruity may have a ad banners
positive or negative effect depending on the context and other factors. An advantage of
our investigation is that, besides analysing typical quality measures such as free and
helped recall, eye tracking is used to gain insights about users’ eye behaviour. 421
Knowing that a congruent banner attracts more fixations than an incongruent one has
important consequences, even if in our experiments memory and recall do not seem to
be greatly affected. The first and most significant implication is that banner blindness
can be partially overcome (if a banner is not seen at all, it will hardly have any effect on
the user’s mind). Second our findings prompt researchers to continue studies on
congruity, which might seem of little interest because of the divergent results produced
to date. Web advertising agencies and online newspapers are encouraged to collaborate
more closely to design banner campaigns where the subject of the ad is connected with
the topic of its article, in order to potentially increase their success. We think that in a
period like the present, in which several newspapers are debating whether to continue
offering free access to their online content, the problem is particularly relevant.
Our results can be used to drive further research in the field, for example taking into
account additional independent variables such as banner position, style and size.

References
Adam, P.S., Quinn, S. and Edmonds, R. (2007), Eyetracking the News, The Poynter Institute for
Media Studies, St Petersburg, FL.
Anderson, J.R. (1983), “A spreading activation theory of memory”, Journal of Verbal Learning
and Verbal Behavior, Vol. 22 No. 3, pp. 261-295.
Burke, M., Hornof, A., Nilsen, E. and Gorman, N. (2005), “High-cost banner blindness: ads
increase perceived workload, hinder visual search, and are forgotten”, ACM Transactions
on Computer-Human Interaction, Vol. 12 No. 4, pp. 423-445.
Choi, S.M. and Rifon, N.J. (2002), “Antecedents and consequences of web advertising credibility:
a study of consumer response to banner ads”, Journal of Interactive Advertising, Vol. 3
No. 1, p. 12-24.
Danaher, P.J. and Mullarkey, G.W. (2003), “Factors affecting online advertising recall: a study of
students”, Journal of Advertising Research, Vol. 43 No. 3, pp. 252-267.
Drèze, X. and Hussherr, F. (2003), “Internet advertising: is anybody watching?”, Journal of
Interactive Marketing, Vol. 17 No. 4, pp. 8-23.
Duchowski, A.T. (2007), Eye Tracking Methodology – Theory and Practice, 2nd ed.,
Springer-Verlag, London.
Ferreira, P., Rita, P., Morais, D., Rosa, P., Oliveira, J., Gamito, P., Santos, N., Soares, F. and
Sottomayor, C. (2011), “Grabbing attention while reading website pages: the influence of
verbal emotional cues in advertising”, Journal of Eyetracking, Visual Cognition and
Emotion, Vol. 1 No. 1, pp. 64-68.
Garcia, M.R. and Stark, P. (1991), Eyes on the News, The Poynter Institute for Media Studies,
St Petersburg, FL.
Greenwald, A.G. and Leavitt, C. (1984), “Audience involvement in advertising: four levels”, The
Journal of Consumer Research, Vol. 11 No. 1, pp. 581-592.
OIR Heckler, S.E. and Childers, T.L. (1992), “The role of expectancy and relevancy in memory for
verbal and visual information: what is incongruency?”, The Journal of Consumer Research,
37,3 Vol. 18 No. 4, pp. 475-492.
Hershberger, E. and Costea, A.E. (2009), “Task orientation and ad recall: an exploratory study”,
Proceedings of the 3rd International Conference on Management, Marketing and Finances,
30 April-2 May, WSEAS, Houston, TX, pp. 82-87, available at: www.wseas.us/e-library/
422 conferences/2009/houston/AMMF/AMMF11.pdf (accessed 14 October 2012).
Hervet, G., Guérard, K., Tremblay, S. and Chtourou, M.S. (2011), “Is banner blindness genuine?
Eye tracking internet text advertising”, Applied Cognitive Psychology, Vol. 25 No. 5,
pp. 708-716.
Holmqvist, K., Holsanova, J., Barthelson, M. and Lundqvist, D. (2003), “Reading or scanning?
A study of newspaper and net paper reading”, in Hyönä, J.R. and Deube, H. (Eds),
The Mind’s Eye: Cognitive and Applied Aspects of Eye Movement Research, Elsevier,
Amsterdam, pp. 657-670.
Interactive Advertising Bureau (2011), “Ad unit guidelines”, available at: www.iab.net/iab_
products_and_industry_services/1421/1443/1452 (accessed 24 June 2011).
Janssens, W., De Pelsmacker, P. and Geuens, M. (2007), “Does a medium context have a priming
or an interference effect? It depends on how you look at it”, Advances in Consumer
Research, Vol. 34, pp. 684-685.
Johar, G.V. and Pham, M.T. (1999), “Relatedness, prominence, and constructive sponsor
identification”, Journal of Marketing Research, Vol. 36 No. 3, pp. 299-312.
Just, M.A. and Carpenter, P.A. (1976), “Eye fixations and cognitive processes”, Cognitive
Psychology, Vol. 8, pp. 441-480.
Just, M.A. and Carpenter, P.A. (1980), “A theory of reading: from eye fixation to comprehension”,
Psychology Review, Vol. 87 No. 4, pp. 329-354.
Lee, W., Hur, Y. and Lim, C. (2010), “Sport website advertising: the impact of congruity and
endorsement on the effectiveness of banner advertising”, Proceedings of the 18th
European Association for Sport Management Conference, 15-18 September, Prague,
available at: http://goo.gl/juNC6 (accessed 3 July 2011).
McCoy, S., Everard, A., Galletta, D. and Polak, P. (2004), “The effects of online advertising”,
Proceedings of the Third Annual Workshop on HCI Research in MIS, Washington DC,
10-11 December, pp. 50-54.
Mack, A. (2003), “Inattentional blindness: looking without seeing”, Current Directions in
Psychological Science, Vol. 12 No. 5, pp. 180-184.
Mandler, G. (1982), “The structure of value: accounting for taste”, in Clark, M.S. and Fiske, S.T.
(Eds), Proceedings of Affect and Cognition: The 17th Annual Carnegie Symposium on
Cognition, Lawrence Erlbaum Associates, Hillsdale, NJ, pp. 3-36.
Meyers-Levy, J., Louie, T.A. and Curren, M.T. (1994), “How does the congruity of brand names
affect evaluations of brand name extensions?”, Journal of Applied Psychology, Vol. 79 No. 1,
pp. 46-53.
Moore, R.S., Stammerjohan, C.A. and Coulter, R.A. (2005), “Banner advertiser – web site context
congruity and color effects on attention and attitudes”, Journal of Advertising, Vol. 34 No. 2,
pp. 71-84.
Moorman, M., Neijens, P.C. and Smit, E.G. (2002), “The effects of magazine-inducted
psychological responses and thematic congruence on memory”, Journal of Advertising,
Vol. 31 No. 4, pp. 27-40.
Newman, E.J., Stem, D.E. and Sprott, D.E. (2004), “Banner advertisement and web site congruity Online
effects on consumer web site perceptions”, Industrial Management & Data System, Vol. 104
No. 3, pp. 273-281. newspapers and
Nielsen, J. (2007), “Banner blindness: old and new findings”, Jakob Nielsen’s Alertbox, available ad banners
at: www.useit.com/alertbox/banner-blindness.html (accessed 14 June 2011).
Pagendarm, M. and Schaumburg, H. (2001), “Why are users banner-blind? The impact of
navigation style on the perception of web banners”, Journal of Digital Information, Vol. 2 423
No. 1, available at: http://journals.tdl.org/jodi/article/viewarticle/36/38 (accessed 12 July
2011).
Poole, A. and Ball, I.J. (2005), “Eye tracking in human-computer interaction and usability
research: current status and future prospects”, in Gahoui, C. (Ed.), Encyclopedia of
Human-Computer Interaction, Idea Group, Hershey, PA, available at: www.alexpoole.info/
blog/wp-content/uploads/2010/02/PooleBall-EyeTracking.pdf (accessed 22 May 2011).
Rodgers, S. (2007), “Effect of sponsorship congruity on e-sponsorship and e-newspapers”,
Journalism & Mass Communication Quarterly, Vol. 84 No. 1, pp. 24-39.
Ryu, G., Lim, E.A.C., Tan, L.T.L. and Han, Y.J. (2007), “Preattentive processing of banner
advertisements: the role of modality, location, and interference”, Electronic Commerce
Research and Application, Vol. 6 No. 1, pp. 6-18.
Tobii Technology AB (2006), Tobii 1750 Eye Tracker, Tobii Technology AB, Danderyd.
Wang, K., Chen, S. and Chang, H. (2008), “The effects of forced ad exposure on the web”, Journal
of Informatics & Electronics, Vol. 3 No. 1, pp. 27-38.
Yang, C. (2011), “Exploring cultural variation in eye movements on a web page between
Americans and Koreans”, Design, User Experience, and Usability. Theory, Methods, Tools
and Practice (Proceedings of the First International Conference, DUXU 2011, held as Part
of HCI International 2011, Orlando, FL, July 9-14, Part II), Springer, Berlin, pp. 68-76.
Yi, Y. (1990), “Cognitive and affective priming effects on the context for print advertisements”,
Journal of Advertising, Vol. 19 No. 2, pp. 40-48.
Zambarbieri, D., Carniglia, E. and Robino, C. (2008), “Eye tracking analysis in reading
online newspapers”, Journal of Eye Movement Research, Vol. 2 No. 4, pp. 1-8, available at:
http://e-tracking.unipv.it/eye/publication/JEMR%202008.pdf (accessed 14 October 2012).

About the authors

Marco Porta is an Assistant Professor at the University of Pavia, specialising in
Human-Computer Interaction with particular reference to user-centred interface design,
eye-tracking applications, visual languages, e-learning technology and multimedia interfaces
(including web design, usability and accessibility). He holds a PhD degree in Electronic and
Computer Engineering from the University of Pavia and has published more than 50 articles in
international journals, conference proceedings and books. Marco Porta is the corresponding
author and can be contacted at: marco.porta@unipv.it
Alice Ravarelli is a Postdoc Researcher at the University of Pavia, working in the field of
Human-Computer Interaction and specialising in interface evaluation based on eye tracking. She
holds a PhD in Electronic, Electric and Computer Engineering from the University of Pavia.
Francesco Spaghi, after receiving his Master’s degree in Multimedia Communication and
Publishing from the University of Pavia, is currently working as a Web designer (with
significant experience in ad banners design and implementation).

To purchase reprints of this article please e-mail: reprints@emeraldinsight.com

Or visit our web site for further details: www.emeraldinsight.com/reprints