Content uploaded by Sho Sato
Author content
All content in this area was uploaded by Sho Sato on Jul 28, 2020
Content may be subject to copyright.
Impact of Bookshelf Locations Using Eye-Tracking Methodology
Sho Sato1, Yukari Eto2, Kotomi Iwaki3, Tadashi Oyanagi4, and Yu Yasuma5
1Center for License and Qualification, Doshisha University
Contact: min2fly@slis.doshisha.ac.jp
2Faculty of Letters, Doshisha University
Contact: YukariEto@slis.doshisha.ac.jp
3Faculty of Letters, Doshisha University
Contact:KotomiIwaki@slis.doshisha.ac.jp
4Faculty of Letters, Doshisha University
Contact:TadashiOyanagi@slis.doshisha.ac.jp
5Faculty of Letters, Doshisha University
Contact:YuYasuma@slis.doshisha.ac.jp
IMPACT OF VERTICAL 2
Corresponding author:
Sho Sato
Center for License and Qualification, Doshisha University
Karasuma-higashi-iru, Imadegawa-dori, Kamigyo-ku, Kyoto-shi, Japan
Phone No.: +81 75 251 3454
Fax No.:+81 75 251 3229
Contact: min2fly@slis.doshisha.ac.jp
Abstract
Purpose: This study aimed to understand better the user gaze behavior on bookshelves
using eye-tracking technology.
Methodology: An eye-tracking experiment in a public library with 11 participants was
performed. The impact of vertical shelf location of books on the number of times the
books are looked at, the impact of horizontal location, and the relationship between user
behavior and location impact were examined by the findings.
Findings: The results showed that vertical location of books has a significant impact on
the number of times the books are looked at. More than 80% of time spent looking at
bookshelves was spent on books on the top to fourth rows. It was also revealed that
horizontal location of books has a little impact. Books located on the left side of shelves
will be looked at significantly more often than those on the right side. No significant
relationships between type of user behaviors and location impact were observed.
IMPACT OF VERTICAL 3
Originality/Value: The study explored the impact of vertical location of books on time
spent looking at bookshelves using eye-tracking methodology. Few published studies do
such experiments to address user gaze behavior on bookshelves. The study explored that
vertical location of books has a great impact and horizontal location has a little impact on
user gaze behavior.
Keywords: Public library, Bookshelves, Eye-tracking, User gaze behavior, Library
users, Library use studies, Browsing
Article Classification: Research Paper
IMPACT OF VERTICAL 4
1. Introduction
This study reveals the effect of location of books on the number of times the
books are looked at in public libraries. Although user behavior in libraries has been one
of the most researched topics in library and information science (LIS), surprisingly
some basic facts have not been revealed. User gaze behavior on library bookshelves is
one of them. Of course, it has been empirically known that there is a difference in the
conspicuousness of the book depending on the location on the shelves. For example,
library architects suggest you should not use the lowest rows of shelves because the
rows are hard to see and access for users (Uematsu, 2014). Nevertheless, some libraries
fill the shelves top to bottom because of limited space.
Exactly how much harder is it for patrons to see the books on the bottom rows?
In other words, how well can the upper rows be seen by users compared to the lower
rows? In addition, does horizontal location have any effect on the number of times the
books are looked at? To address these questions will not only give us basic knowledge
to manage library bookshelves but also important knowledge to manage library
displays. Besides that, it may have an effect on volume of library circulation. Despite
user gaze behavior on shelves being important in these ways, only a few LIS studies
have addressed these topics because of the high cost of mobile eye-tracking
technologies.
However, recent development of inexpensive mobile eye-trackers makes it
easier to use the machines in real library environments. Thus, we performed an eye-
tracking experiment in a public library to gain more understanding of user gaze
behavior on bookshelves. We especially focused on the impact of vertical and horizontal
location of books on the number of times the books are looked at. Our research question
IMPACT OF VERTICAL 5
and hypotheses were:
RQ1: Does vertical location of books have an impact on the number of times the books
are looked at?
H1: Books located on higher shelves are looked at significantly longer than those
located on lower shelves.
RQ2: Does horizontal location of books have an impact on the number of times the
books are looked at?
H2: Books located on the center of shelves are looked at significantly longer than
those located to the right or left.
RQ3: Are the location impacts different between when users search one or more books
and when they browse bookshelves?
H3: Book location only has an impact during browsing time. During searching one or
more books, users tend to look at broad areas of bookshelves.
2. Literature Review
2.1 Library User Behavior
Some studies approached user’s browsing behavior without using eye-tracking
methodology. Ross (1983) found that academic library users spent an average of 6.94
minutes when browsing. Kitaoka et al. (1996) reported that Japanese public library
users spent an average of about 11 minutes in browsing.
Matsuda (2003) conducted observational research on library users and bookstore
customers in two public libraries and two bookstores. She found that some users
checked bookshelves row by row when they could not use Online Public Accessible
Catalog (OPAC) machines. This type of behavior was observed only in libraries and
bookstores without OPACs or search machines. Users are considered to have been
IMPACT OF VERTICAL 6
searching for one or more books. In the libraries that have OPACs, no users did this type
of behavior because they could find the location of the books they searched for with
OPACs and went straight to the shelves. In other words, while users do browse (i.e.,
they do not search for one or more specific books), they do not check bookshelves row
by row and just look at bookshelves in a disorganized way. This result suggests that,
while searching, users tend to look at broad areas (each row) and there will be little
location impact on the number of times the books will be looked at. From this result, we
devised the aforementioned RQ 3.
2. 2 Eye-Tracking Studies in LIS Fields
Lund (2016) presented a systematic literature review of the application of eye-
tracking technology within the field of LIS. He found that the amount of published
research utilizing eye-tracking technologies within LIS was still limited, although an
increase in the use of eye-tracking technologies has been observed during recent years.
In addition, most of the LIS research utilizing eye-tracking technologies was interested
in user search behavior with computer systems and thus gathered data of eye
movements on screens.
Among such studies of eye movements on screens, studies about library users’
eye movements in selecting books are particularly salient to our study. Pöntinen and
Vakkari (2013) recorded and analyzed the eye movements of 30 library users selecting
fiction in online public library catalogs. The results indicated that participants paid most
attention in book pages to content description and keywords, while author and title
information received less attention. Nevertheless, as a predictor of user selection of
interesting books, the visit duration of author and title information was significant and
that of content description and keywords had no impact. The same data was analyzed in
IMPACT OF VERTICAL 7
Vakkari, Luoma and Pöntinen (2014) to explore whether books’ interest grades
differentiate the time devoted to various metadata types in book pages. The results
showed that most time was spent assessing a somewhat interesting novel compared to a
non-interesting or very interesting one. Gudinavičius and Šuminas (2018) did
experiment to explore the relationship between the users’ preference of book cover
colors and their choice of book in an online bookstore by its cover. The study found a
significant positive correlation between the women’s preferences in selecting covers and
the time women spent looking at them; however, there was no such correlation in the
case of the men’s data.
While there is no research utilizing eye-trackers in the real library environment
in Lund’s review, there have been a few LIS studies that used eye-trackers to collect
data of user eye movements on bookshelves or in libraries. Nakahata et al. (2016)
explored the effect of book cover color, title typeface, and author name on human gaze
behavior in laboratory environments. They found that book cover color had no effect on
gaze duration and user choice. On the other hand, book title typeface and author name
had an effect. Sato et al. (2014) performed a laboratory experiment to understand users’
browsing behaviors on bookshelves. They found that if subjects have enough time to
choose a book from a bookshelf, the subjects would take the time to look at each book
in each row from the top to the bottom of the first stack of shelves. In addition, the
direction of user eye movement reverses with each row scanned. Ambiru and Itsumura
(2013) used an eye-tracker in an academic library to explore the features of student
search behavior. They found that when searching for books on bookshelves, graduate
students and third-grade undergraduate students look at books horizontally because they
only check call number labels. First-grade undergraduate students look at books
IMPACT OF VERTICAL 8
vertically because they read book titles to find their target book. However, no study used
eye-tracking technologies as a data collection method on bookshelves in public
libraries.
2.3 Eye-Tracking Studies in Non-LIS Fields
As research related to our study other than LIS fields, in the marketing area, eye-
trackers have been used to understand user gaze behavior on supermarket shelves.
Chandon et al. (2009) examined the factors on consumer attention to and evaluation of
brands displayed on supermarket shelves using an eye-tracking experiment. They found
that top (vertically) and middle (horizontally) shelf positions gain more attention.
Chynal et al. (2016) studied customer behavior during the selection of products from
the store shelf with eye-tracking glasses and an emotional information recorder. They
also revealed that higher shelf positions gain more attention. From these results, we
devised the aforementioned RQ 1 and RQ 2.
Both Chandon et al. (2009) and Chynal et al. (2016) were conducted in laboratory
settings; there are some studies conducted in real supermarkets, such as Gidlöf et al.
(2013) and Hendrickson and Ailawadi (2014). In particular, Hendrickson and Ailawadi
(2014) found some important facts that are relevant to our research: 1) shoppers look in
a narrow window above and below eye level, 2) shoppers read signs for an average of
one second, and 3) shoppers first read horizontally in one direction and then vertically
in one direction.
3. Methods
3. 1 Location and Situation
The experiment was conducted at the Ebina City Main Library in Kanagawa,
Japan, in September 2016. The library has four floors and a basement. Figure 1 shows
IMPACT OF VERTICAL 9
the floor plan of the library. The basement houses a literature collection. Three areas—
café, bookstore, and a library with business books—occupy the first floor. On the
second and third floors, there are other general monographs. The fourth floor is a space
for children.
Most of the bookshelves in the library comprise six or seven rows with dummy
zones. Figure 2 shows a typical bookshelf in the Ebina City Main Library. Dummy
zones are filled with old books or mock books. The top to seventh or sixth rows are
filled with books for reading. Some bookshelves have only five rows for bigger books.
Even in that case, the height of the top to fourth rows is the same as shelves with seven
rows.
3.2 Subjects
A total of 11 people ranging in age from 21 to 61 years participated in our
experiment. Table 1 shows the subjects’ demographics. Six were male and five were
female. The height of the subjects ranged from 1.50 m to 1.80 m (mean = 1.64 m). Four
lived near the library and seven came from far away for the experiment. Three have
used the library often, four used it sometimes, and four visited the library for the first
time. None of the participants had a visual impairment or impaired color vision.
<Table 1 here>
3.3 Procedure
The participants were given 10 minutes and were required to choose more than
one book of their interest in the library. Although the time limit was set for limiting the
data capacity of the experimental settings, it was considered enough for book choice
based on browsing studies such as Ross (1983) and Kitaoka et al. (1996). Each
participant started the experiment at the entrance of the library on the first (ground)
IMPACT OF VERTICAL 10
floor. Subjects were permitted to go anywhere in the library buildings. However, none
went to the fourth (i.e., children’s) floor. They were also permitted to use OPAC
machines, read books, or talk with librarians. They were forbidden to use their
smartphones or computers.
3.4 Data Collection
The subjects’ eyesight and their points of gaze were recorded with a wearable
eye-tracker named “ViewTracker2” (VT2) Version 1.0.1, developed by DITECT Co.
Ltd. The eye-gaze measurement system makes use of infrared and visual camera
technologies to determine the eye-gaze locations. Figure 3 shows a subject wearing
VT2. VT2 outputs WMV moving image data of user range of view with their point of
gaze. Figure 4 shows one picture of eye-tracking data. The red cross is the subject’s
point of gaze. The data acquisition rate was at the level of 15 frames per second.
3.5 Data Analysis
Based on the data, we analyzed gaze duration time on each of the objects (e.g.,
books on shelves, displayed books, and OPAC machines) and vertical and horizontal
locations of gazed-at books. For analysis of gazed-at objects, we checked the moving
images frame-by-frame and counted how many times each category of object was
looked at. Table 1 shows the categories of objects.
For vertical location analysis, we checked the height of rows on which gazed-at
books were placed.
In horizontal location analysis, we separated all rows into three areas—left,
central, and right—and checked where gazed-at books were placed.
IMPACT OF VERTICAL 11
4. Findings
4.1 Task Summary and Gazed Objects
In the experiments, all participants succeeded in choosing one or more books of
their interested. Table 2 summarizes user behavior in the task. Six chose only one book
and five chose more than two books. Six participants used OPAC machines to search for
a book or a subject they were interested in. Five did not use OPACs and only browsed
bookshelves. Subject No. S2 used an OPAC machine and chose a book in the stack
room. No. S6 chose the book in the bookstore space. Others chose books on the shelves
of the library.
<Table 2 here>
Books located on top rows were chosen most often: five times. Books located on
third and fifth rows were chosen second most often: three times. Books on second and
fourth rows were chosen two times. This result suggests that the vertical location of
books does not have a big impact in regards to book-choosing behavior.
Table 3 shows the gaze duration ratio of each object. Gaze duration ratio of
objects was obtained by dividing the number of frames each type of object was looked
at by the number of valid frames in the experiment. It is not surprising that most
subjects (except “Others”) looked at books on shelves longest. Nine of eleven looked at
bookshelves longest and two (S2 and S9) looked at OPAC machines longest.
Participants who used OPACs to search for one or more books (S1–2, S4, and S9–11)
looked at OPACs first or second longest. Participants who did not use OPACs and only
browsed bookshelves (S3 and S5–8) tended to look at displayed books longer. Signs and
floor plans were not looked at as long.
<Table 3 here>
IMPACT OF VERTICAL 12
4.2 Vertical Location of Gazed-At Books
Table 4 shows the relationships between vertical location of gazed-at books and
gaze duration ratio. These ratios were obtained by dividing the number of frames a book
at a location was looked at by the number of frames bookshelves were looked at.
Participants tended to look at the third rows longest. On average, about 26% of time
looking at bookshelves was spent on books on the third rows. About 21% was spent on
the second rows, 19% on the top rows, and 16% on the fourth rows. More than 80% of
time looking at bookshelves was spent on these higher rows. Participants spent only
about 14% of time on the fifth or lower rows. On the other hand, although the dummy
rows are higher than the top rows, participants spent only a small amount of time (about
3%) on books on the dummy rows.
<Table 4 here>
Table 5 shows the result of one-factor within-subject ANOVA for relationships
between vertical location of gazed-at books and times gazed at. It revealed that vertical
location of books has a significant impact on the number of times the books are looked
at (p < 0.01).
<Table 5 here>
Table 6 shows the results of pairwise comparisons of times gazed at between
rows with Bonferroni correction. There were no significant differences among top,
second, third, and fourth rows. Books on these four rows were gazed at significantly
more than books on dummy, sixth, and seventh rows (p < 0.05). In addition, books on
second and third rows were gazed at significantly more than books on fifth rows (p <
0.05).
<Table 6 here>
IMPACT OF VERTICAL 13
A correlation was expected between vertical location of gazed-at books and
height of subjects. However, this expectation did not prove to be true. There was no
significant correlation between the percentage of time looking at books on the top to
fourth rows and height of subjects (r = 0.06, df = 9, p > 0.05).
4.3 Horizontal Location of Gazed-At Books
Table 7 shows the relationships between horizontal location of gazed-at books
and gaze duration ratio. Participants tended to looked at the left area of bookshelves
longest and right area shortest. On average, about 39% of time looking at shelves was
spent on the left area, 34% on the central area, and 27% on the right area. The result of
one-factor within-subject ANOVA shows that these differences were significant at a 1%
level (Table 8). The results of pairwise comparisons of times gazed at between areas
with Bonferroni correction shows that books on the left areas of shelves were gazed at
significantly more than books on the right areas (p < 0.05). There were no significant
differences between books on the left area and the central area, and between the central
area and the right area.
<Table 7 here>
<Table 8 here>
4.4 Relationship between User Behavior and Location of Gazed-At Books
As mentioned above, five of 11 participants never used OPAC machines during
the experiments. Among six OPAC users, one (S9) went to OPAC directly and searched
for some books throughout the experiment and never did browsing. Two (S4 and S10)
used the OPAC to get the location of some subjects and never searched for a book. Only
three participants (S1, S10, S11) did both search and browsing.
Table 9 shows the relationship between user behavior and vertical location of
IMPACT OF VERTICAL 14
gazed-at books among three participants who exhibited both behaviors. S1 and S11
spent more time looking at higher rows (top to fourth) during browsing than during
searching. In contrast, S2 spent less time looking at higher rows during browsing. The
result of two-factor within-subject ANOVA (Table 10) shows that user behavior had no
effect on gaze duration ratio. On the other hand, vertical location of books had
significant effect. Interaction effect was not significant.
<Table 9 here>
<Table 10 here>
Table 11 shows the relationship between user behavior and horizontal location of
gazed-at books among three participants who exhibited both behaviors. S1 spent the
most time looking at left areas during browsing and looking at right areas during
searching. In contrast, S2 and S11 spent the most time looking at left areas during both
browsing and searching. The result of two-factor within-subject ANOVA (Table 12)
shows that there was no significant effect of user behavior, horizontal location, and
interaction.
<Table 11 here>
<Table 12 here>
5. Discussion
5.1 The Impact of Vertical Location
RQ1 (Does vertical location of books have an impact on the number of times the
books are looked at?) and H1 (Books located on higher rows of shelves are looked at
significantly longer than books located on lower rows.) examine the impact of vertical
location of books on the number of times the books are looked at. The findings
supported H1. In the experiments, more than 80% of time looking at bookshelves was
IMPACT OF VERTICAL 15
spent on higher (top to third) rows. Participants spent only about 14% of time on the
fifth or lower rows. One-factor within-subject ANOVA revealed this impact of vertical
location was significant.
As expected, subjects looked at books located on higher rows of shelves more
often. However, among higher (top to fourth) rows, there were no significant
differences. Higher rows did not necessarily mean better rows. On the other hand, lower
rows did mean worse rows. Participants spent 7.5% of time looking at bookshelves on
fifth rows, 5.3% on sixth rows, and only 1.6% on seventh rows.
5.2 The Impact of Horizontal Location
RQ2 (Does horizontal location of books have an impact on the number of times
the books are looked at?) and H2 (Books located in the center of shelves are looked at
significantly longer than those located to the right or left) examine the impact of
horizontal location of books on the number of times the books are looked at. Our results
showed that horizontal location of books has a significant impact on the number of
times books are looked at, but they did not support H2. Books located on the left area of
shelves—not those in the center—were looked at most in the experiments. On average,
about 39% of time looking at shelves was spent on the left area, 34% on the central
area, and 27% on the right area. One-factor within-subject ANOVA revealed that books
on the left areas of shelves were gazed at significantly more than books on the right
areas.
Our findings did not tell us the reason why books on the left areas were looked
at more often. In the supermarket studies, it was found that the center position of
shelves gains more attention (Chandon et al., 2009). Our results were contrary to this
previous study. One possible explanation is difference between supermarket shelves and
IMPACT OF VERTICAL 16
library shelves. In supermarkets, horizontal shelf location of items has no meaning. On
the other hand, in libraries, books must be ordered from the left to the right of the
shelves. This library order rule may have an effect on user gaze behavior.
5.3 The Relationship between Type of User Behavior and the Location Impact
RQ3 (Are the location impacts different between when users search one or more
books and when they browse bookshelves?) and H3 (Book location only has an impact
during browsing time. During searching one or more books, users tend to look at broad
areas of bookshelves) examine the relationship between type of user behavior and the
location impact. Unfortunately, our findings did not give enough evidence on this
question and hypothesis. We could not observe any significant relationships between
type of user behaviors and the location impact because only a few participants did both
search and browsing.
6. Conclusion and Future Research
The purpose of this study is to reveal the effect of location of books on the
number of times the books are looked at in public libraries. The most meaningful
finding of this research was that vertical location of books has a great impact on the
number of times the books are looked at. More than 80% of time spent looking at
bookshelves was spent on books on the top to fourth rows. In addition, it was revealed
that horizontal location of books also has an impact. Books located on the left area of
shelves will be looked at more often than those on the right area.
While there were some useful findings, the results of this research also had
important limitations. The findings were derived from only 11 participants and only one
library setting in a country. The limitation of number of participants made it difficult to
reveal the relationship between type of user behavior and the location impact. As for the
IMPACT OF VERTICAL 17
library setting, Ebina City Main Library differs greatly from other public libraries in
Japan in various respects. Most of the bookshelves in the library are very tall and spaces
between the shelves tend to be narrow. These characteristics of the library make it like a
book labyrinth. This tendency may have an impact on human gaze behavior. Cultural
behavior, including the reading manner of Japanese people, can also have an impact on
user gaze behavior. As in China, Japanese people traditionally write and read text
vertically, from right to left. Conversely, books in libraries are lined up horizontally,
from left to right. This conflict between the manner of reading and the manner of
arranging books on a shelf could affect the impact of the horizontal location of books on
gaze behavior. It could be thought that people who read text horizontally, from left to
right, would look at books placed to the left more often than Japanese people.
Thus, we need to perform the same experiment with more participants, in other
public libraries and other cultural settings. Research on user gaze behavior in libraries
has only just begun, and to make the findings of our research more reliable, it is
expected that many studies will be conducted in the future.
Acknowledgements
We wish to thank Ebina City Main Library for their cooperation in our experiment. This
work was supported by JSPS KAKENHI [grant numbers JP 26730164, JP 25330389, JP
17H02026, JP 18K18335, and JP 18K02915]. A part of this study was presented at the
9th International Conference on Qualitative and Quantitative Methods in Libraries, 23–
26 May 2017 in Limerick, Ireland.
IMPACT OF VERTICAL 18
References
Ambiru, T. and Itsumura, H. (2013), “Student information-seeking behavior in an
academic library”, Chubu Tsohokan Joho Gakkai Shi (Library and Information
Science Journal in Central Japan), Vol. 53, pp. 17-34. (In Japanese)
Chandon, P., Hutchinson, J.W., Bradlow, E.T. and Young, S.H. (2009), “Does in-store
marketing work? Effects of the number and position of shelf facings on brand
attention and evaluation at the point of purchase”, Journal of Marketing, Vol. 73
No. 6, pp. 1-17.
Chynal, P., Janusz, S., Rymarz, M. and Kilijańska, B. (2016), “Shopping behaviour
analysis using eyetracking and EEG”, in Proceedings of 2016 9th International
Conference on Human System Interactions, pp. 458-464.
Gidlöf, K., Wallin, A., Dewhurst, R. and Holmqvist, K. (2013), “Using eye tracking to
trace a cognitive process: Gaze behaviour during decision making in a natural
environment”, Journal of Eye Movement Research, Vol. 6 No. 1, pp. 1-14.
Gudinavičius, A. and Šuminas, A. (2018), “Choosing a book by its cover: Analysis of a
reader’s choice”, Journal of Documentation, Vol. 74 No. 2, pp. 430-446.
Hendrickson, K. and Ailawadi, K.L. (2014), “Six lessons for in-store marketing from
six years of mobile eye-tracking research”, in Grewal, D., Roggeveen, A.L. and
NordfÄlt, J. (Eds.), Shopper Marketing and the Role of In-Store Marketing.
Emerald, Bingley, UK, pp. 57-74.
Kitaoka, T., Aoki, M. and Takeshita, T. (1996), “The relation of each corner and the
location of browsing corner for adults’ multipurpose uses at public libraries”,
Journal of Architecture and Planning (Transactions of AIJ), Vol. 61 No. 486, pp.
61-68. (In Japanese)
IMPACT OF VERTICAL 19
Lund, H. (2016), “Eye tracking in library and information science: A literature review”,
Library Hi Tech, Vol. 34 No. 4, pp. 585-614.
Matsuda, C. (2003), “Browsing behavior in information seeking process: On the basis
of observation of information-seeking behavior in libraries and bookstores”,
Library and Information Science, Vo. 49, pp. 1-31. (In Japanese)
Nakahata, S., Sakamoto, E., Oda, A., Kobata, N. and Sato, S. (2016), “Effects of color
of book cover and typeface of title and author name on gaze duration and choice
behavior for books: Evidence from an eye-tracking experiment”, in Proceedings
of the Annual Meeting of the Association for Information Science and
Technology, pp. 1-4.
Pöntinen, J.and Vakkari, P. (2013) “Selecting Fiction in Library Catalogs: A Gaze
Tracking Study”, in Aalberg T., Papatheodorou C., Dobreva M., Tsakonas G.,
Farrugia C.J. (Eds.), Research and Advanced Technology for Digital Libraries.
TPDL 2013, Lecture Notes in Computer Science, Vol. 8092, Springer, Berlin,
Germany, pp. 72-83.
Ross, J. (1983), “Observations of browsing behavior in an academic library (research
note)”, College & Research Libraries, Vol. 44 No. 4, pp. 269-276.
Sato, S., Ambiru, T., Terai, H., Harada, T. and Itsumura, H. (2014), “How people scan
bookshelves in browsing behavior”, in Proceedings of the 62nd Annual Meeting
of Japan Society of Library and Information Science. Japan Society of Library
and Information Science, Hirakata, pp. 37-40. (In Japanese)
Uematsu, S. (2014), Theory of library facility. Jusonbo, Tokyo. (In Japanese)
Vakkari, P., Luoma, A. and Pöntinen, J. (2014), “Books' interest grading and dwell time
in metadata in selecting fiction”, in Proceedings of the 5th Information
IMPACT OF VERTICAL 20
Interaction in Context Symposium, pp. 28-37.
IMPACT OF VERTICAL 21
Figures
Figure 1. The floor plan of the Ebina City Main Library.
Figure 2. Sample picture of typical bookshelf in the Ebina City Main Library.
IMPACT OF VERTICAL 22
Figure 3. A subject wearing ViewTracker2.
Figure 4. Sample picture of eye-tracking data.
IMPACT OF VERTICAL 23
Tables
Table 1. Subjects’ demographics
No.
Height of subject (m)
Age
Sex
Address
Ebina City Library experience
S1
1.76
30+
Male
Far from the library
First time
S2
1.80
20+
Make
Far from the library
Sometimes
S3
1.51
20+
Female
Far from the library
First time
S4
1.62
40+
Female
Far from the library
Sometimes
S5
1.69
20+
Male
Far from the library
First time
S6
1.66
20+
Male
Near the library
Sometimes
S7
1.55
20+
Female
Near the library
Often
S8
1.62
20+
Female
Far from the library
First time
S9
1.65
30+
Male
Near the library
Often
S10
1.67
60+
Male
Far from the library
Sometimes
S11
1.50
40+
Female
Near the library
Often
IMPACT OF VERTICAL 24
Table 2. Chosen books and OPAC machine usage
No.
Number of books
chosen
Height of books chosen
OPAC machine usage
S1
1
Top
Used
S2
1
-
Used
S3
1
3rd
Never used
S4
1
2nd
Used
S5
1
Top
Never used
S6
1
-
Never used
S7
2
3rd
Never used
S8
2
Top, 2nd
Never used
S9
2
5th
Used
S10
4
4th, 5th
Used
S11
4
Top, 3rd, 4th, 5th
Used
IMPACT OF VERTICAL 25
Table 3. Gaze duration ratio of each object (%)
Category name
S1
S2
S3
S4
S5
S6
S7
S8
S9
S10
S11
Avg.
Books on shelves
38.6
23.7
69.9
36.8
29.0
54.8
40.4
19.6
23.1
32.6
69.2
39.8
Displayed books
3.5
7.5
3.5
8.3
4.9
22.7
15.3
17.9
0.6
6.1
0.9
8.3
Flat displayed
0.1
1.9
1.3
2.8
0.4
7.6
1.0
5.7
1.2
2.7
9.4
3.1
OPAC machines
18.7
29.6
0.5
21.5
0.0
0.0
0.3
0.1
29.3
14.6
7.7
11.1
Signs on floor
0.2
0.1
0.9
0.1
0.1
0.0
0.5
0.3
0.2
0.1
0.0
0.2
Signs on shelves
0.4
0.2
2.0
2.4
2.4
0.2
1.7
1.6
0.6
0.3
0.7
1.1
Floor plans
4.7
1.1
1.2
1.3
4.1
0.0
0.2
0.1
0.0
0.0
0.0
1.2
People
1.1
4.1
1.0
1.0
1.4
1.5
0.7
2.4
1.6
0.7
1.4
1.5
Stairways
3.0
4.5
5.0
0.5
2.4
0.0
6.5
2.8
3.2
0.0
3.6
2.9
Others
29.7
27.2
14.6
25.4
55.4
13.2
33.5
49.4
40.1
43.0
7.1
30.8
IMPACT OF VERTICAL 26
Table 4. Vertical location and gaze duration ratio (%)
Vertical location
S1
S2
S3
S4
S5
S6
S7
S8
S9
S10
S11
Avg.
Dummy
2.3
3.5
1.6
0.9
2.3
0.3
1.9
1.9
10.0
0.1
6.7
2.9
Top
27.5
10.3
13.7
30.3
35.9
21.9
11.0
12.5
23.4
8.4
17.4
19.3
2nd
21.1
23.5
22.5
18.6
33.1
18.1
27.3
20.7
16.3
9.4
19.6
20.9
3rd
18.5
48.4
29.0
20.3
20.2
11.7
36.3
32.8
17.6
25.4
26.4
26.1
4th
18.0
10.3
21.3
6.1
6.8
16.2
18.1
21.3
18.9
25.3
17.8
16.4
5th
6.3
2.2
7.2
5.6
1.7
14.7
4.0
6.5
5.4
21.4
7.6
7.5
6th
6.3
1.4
4.6
12.0
0.0
12.3
0.7
2.6
6.6
8.8
3.3
5.3
7th
0.0
0.5
0.0
6.2
0.0
4.6
0.6
1.5
1.8
1.2
1.1
1.6
IMPACT OF VERTICAL 27
Table 5. Result of ANOVA for relationships between vertical location of gazed-at books
and times gazed at
SS
df
MS
F
p
Vertical location of books
7522337.08
7
1074619.583
17.06
<0.01
Subjects
2531077.364
10
253107.7364
V x S
4410347.546
70
63004.9649
Total
14463761.99
87
IMPACT OF VERTICAL 28
Table 6. Results of pairwise comparison between rows with Bonferroni correction
Row
Significant differences (p < 0.05)
Dummy
<Top, 2nd, 3rd, 4th
Top
>Dummy, 6th, 7th
2nd
>Dummy, 5th, 6th, 7th
3rd
>Dummy, 5th, 6th, 7th
4th
>Dummy, 6th, 7th
5th
<2nd, 3rd
6th
<Top, 2nd, 3rd, 4th
7th
<Top, 2nd, 3rd, 4th
IMPACT OF VERTICAL 29
Table 7. Horizontal location and gaze duration ratio (%)
Horizontal location
S1
S2
S3
S4
S5
S6
S7
S8
S9
S10
S11
Avg.
Left
36.4
39.4
43.0
35.2
38.0
42.4
41.0
37.6
39.5
33.5
46.0
39.3
Central
32.9
30.5
31.5
35.3
36.0
37.5
32.0
34.3
35.9
33.1
29.9
33.5
Right
30.7
30.1
25.5
29.5
25.9
20.1
27.1
28.1
24.6
33.4
24.1
27.2
IMPACT OF VERTICAL 30
Table 8. Result of ANOVA for relationships between horizontal location of gazed-at
books and times gazed at
SS
df
MS
F
p
Horizontal location of books
947769.8788
2
473884.9394
12.06
<0.01
Subjects
5783680.0606
10
578368.0061
H x S
786146.1212
20
39307.3061
Total
7517596.0606
32
IMPACT OF VERTICAL 31
Table 9. Relationships between user behavior and vertical location of gazed-at books
(%)
S1
S2
S11
Browsing
Searching
Browsing
Searching
Browsing
Searching
Dummy
1.6
7.3
0.0
3.7
5.7
14.3
Top
31.0
5.1
0.0
10.8
16.0
27.5
2nd
21.4
21.2
0.0
24.6
20.5
13.0
3rd
17.8
24.8
44.3
48.6
26.8
24.0
4th
15.7
27.0
36.4
9.0
18.7
11.0
5th
6.0
8.8
17.0
1.5
8.3
2.5
6th
6.4
5.8
2.3
1.4
3.0
5.7
7th
0.0
0.0
0.0
0.5
1.0
2.2
Top–4th
85.9
78.1
80.7
93.0
82.1
75.4
Others
14.1
21.9
19.3
7.0
17.9
24.6
IMPACT OF VERTICAL 32
Table 10. Result of ANOVA for relationships between behavior and gaze duration ratio
of higher and other rows
SS
df
MS
F
p
Subjects
0.0000
2
0.0000
Behavior
0.0000
1
0.0000
−2.00
n.s.
B × S
0.0000
2
0.0000
Vertical location of books
1.2691
1
1.2691
191.48
<0.01
V × S
0.0133
2
0.0066
B × V
0.0002
1
0.0002
0.01
n.s.
B × V × S
0.0255
2
0.0218
Total
1.3081
11
IMPACT OF VERTICAL 33
Table 11. Relationships between user behavior and vertical location of gazed-at books
(%)
S1
S2
S11
Browsing
Searching
Browsing
Searching
Browsing
Searching
Left
36.1
32.2
37.0
39.5
46.5
42.9
Central
33.9
28.8
31.5
30.5
30.6
25.2
Right
30.1
39.0
31.5
30.0
22.9
31.9
IMPACT OF VERTICAL 34
Table 12. Result of ANOVA for relationships between behavior, horizontal location, and
gaze duration ratio
SS
df
MS
F
p
Subjects
0.0000
2
0.0000
Behavior
0.0000
1
0.0000
0.00
n.s.
B × S
0.0000
2
0.0000
Horizontal location of books
0.0293
2
0.0147
3.30
n.s.
H × S
0.0177
4
0.0044
B × H
0.0072
2
0.0036
2.62
n.s.
B × H × S
0.0055
4
0.0014
Total
0.0597
17
