You are on page 1of 9

lnt. J.

Man-Machine Studies (1988) 28, 1-9

Reading from screen versus paper" there is no


difference
DAVID J. OBORNE AND DOREEN HOLTON
Department of Psychology, University College of Swansea, Singleton Park, Swansea,
West Gtamorgan, UK

(Received 17 September 1986 and in revised form 12 July 1987)


This paper considers the effect of presentation medium on reading speed and
comprehension. By directly comparing performance using screen and paper presen-
tations, it examines the argument that it takes longer to read from a screen-based
display than from paper, and that comprehension will be lower. The hypothesis is
also tested that it takes longer to read light characters on a dark background
compared with dark characters on a light background, and that comprehension will
be lower with light-character displays. Altogether four conditions were used, with
two passages read in each condition: screen with dark characters, screen with light
characters, paper with dark characters, and paper with light characters. Subjects also
ranked the four conditions for preference. No significant difference was found in
either reading speed or comprehension between screen and paper, or between dark
and light character displays. Some preference differences were found, however.
Reasons for the lack of reading and comprehension differences are discussed, and it
is argued that this reflects the close attention to experimental detail paid in the
present experiment, which has often been missing in past studies.

Introduction
Despite the fact that a quick glance around any modern office will suggest the
contrary, it has often been said that we are rapidly entering the paperless society. It
has been argued that, by using visual display units to present written material,
modern technology should soon make obsolete the need for information to be
written onto paper before it can be read. Material can be presented quicker and
with greater flexibility via VDUs than ever it can by using outdated printing systems.
Certainly, the same glance around any m o d e m office will substantiate the
observed trend towards using computerized systems to store, manipulate and
present information. However, it will also be apparent that the amount of paper
consumed in that office will be unlikely to have fallen as a result of the introduction
of the information technology. Indeed, subjective impressions would suggest that
more, not less, paper is consumed in the modern high-technology environment. This
obviously begs the question as to why a paper-based medium is still so popular.
Many reasons could be advanced: paper is more permanent, it can be recalled
without recourse to high technology, it is easily transportable, and so on. However,
these kinds of reasons are less important to the theme of this paper, which is
concerned more with the ease of using the various presentation media. If the
advantages offered by new technology are to be accepted, its operation must present
no additional obstacles to the operator than the technology it replaces. To ensure
that a screen-based medium is accepted, then, it is important that reading and
1
0020-7373/88/010001 + 09503.00/0 9 1988 Academic Press Limited
2 D. I. O B O R N E A N D D. H O L T O N

comprehending information from a VDU is as easy and as efficient as it presently is


using paper.
Initially one would argue that, as long as the information is presented in the same
fashion--line length, level of detail, etc.--then reading and comprehending from
paper and VDU must be equivalent. However, sufficient comparative studies have
now been performed to lead one to question such a proposition--although the
evidence is far from conclusive. This is because a simplistic comparison between
VDUs and paper does not hold: operator posture may be different when using the
two media (VDUs are generally vertical, whereas paper is often laid horizontally);
reading distances are often different; the contrast between the characters and their
background is likely to be different, as well as the contrast direction (paper generally
has dark characters on a white background, VDUs generally have the opposite).
The importance of a number of features like these have been discussed by Waern &
Rollenhagen (1983), and it is probably because such variables are generally not
completely controlled that previous studies have led to varying conclusions.
Perhaps the earliest study to investigate this question indicated a difference
between the two media, with a slight advantage for the paper-based display. Muter,
Latrrouille, Teurniet & Beam (1982) compared reading speed and comprehension
and found that subjects using the screen read 28.5% more slowly than book-based
subjects. Reasons advanced for the difference in reading speed were: (1) un-
familiarity with reading from a screen compared with a book; (2) differing page
layouts between the two media; (3) variable postures between subjects; and (4) the
time taken to refresh the screen between presentations.
There have been several more recent studies published that bear on this question,
although the conclusions to be reached still remain equivocal. Four studies (Kak,
1981; Kruk & Muter, 1984; Wright & Lickorish, 1983, 1984) support the screen
disadvantage; three (Switchenko, 1984; Askwall, 1985; Cushman, 1986) do not;
while one (Newsted, 1985) indicates a screen advantage.
Wright & Lickorish (1983) compared the speed and accuracy of proof reading a
piece of text presented on a screen with print on paper. Their results showed that,
although the method of recording errors on the screen was quickly learned,
screen-based presentations impaired both speed and accuracy. In a subsequent
investigation (1984) they considered further performance differences between the
two media by examining the behaviour of referees when refereeing scientific reports.
Results from eight subjects showed that the articles displayed on screen slowed
referees down (dealing with printed texts took 30 minutes, and screen texts took 36
minutes).
Newsted (1985) also demonstrated differences of a more qualitative nature using
screen and paper presentations--in this case, however, the advantage was towards
using the computer. His respondents to questionnaires presented via the computer
were more accurate than when the questions were presented in a pencil-and-paper
format. However, it should be noted that in some respects the computer group
selected themselves for the study insofar as they were able to choose whether or not
to sit at the terminal. Subjects in the pencil-and-paper group, however, were
approached "randomly" by the experimenter.
Some of the reasons for reported differences between screen and paper suggested
by Switchenko (1984) include a failure to control for a number of critical variables
S C R E E N VERSUS P A P E R 3

that normally exist between screen and paper presentation. In particular, these
relate to differences in presentation format (e.g. number of characters per line and
lines per page) that, cumulatively, bias the data in favour of a paper format. In
addition, Switchenko suggests that the use of between-subject designs fails to
control for what might be important individual differences in reading ability.
When such variables were controlled for, Switchenko (1984) demonstrated no real
difference between screen and paper. Her study was designed to incorporate the
following: (1) more detailed controls of presentation format; (2) varying levels of
difficulties of the material to be read (based on the observation that the material
read by subjects in previous experiments had been, according to Switchenko, unduly
difficult); and (3) two readings of the same materials, in order to evaluate the
potential effect of experience on V D U reading. However, she did not attempt to
investigate the effects of possible postural differences between screen and paper
readers. Nevertheless, her results demonstrated that, although reading time was
longer for the difficult than for the easy passage (as would be expected), there was
no significant difference in reading time between the two presentation modes.
Kruk & Muter (1984) discussed other reasons for the reported screen disadvan-
tage, including page format and the time taken to refresh screens with information.
In addition, they considered the contrast difference that exists between screen and
paper. (Timmers, Van Nes & Blommaert (1980), for example, have demonstrated
increased word-recognition speed with reduced contrast ratios between characters
and their backgrounds.) However, in their study neither the time taken to fill the
screen nor the variations in contrast ratios caused significant differences in reading
speed on the screen. Varying the distance from the screen (40 cm, 80 cm, 120 cm)
was also found to produce non significant differences in reading speed.
These results suggested to Kruk and Muter that, although there is no single cause
of slower reading in the screen condition, at least two factors play important roles.
These are the format (numbers of characters per line and lines per page) and
interline spacing. Nevertheless, they suggest that the slowing down of reading speed
due to these two factors was not enough to account for all differences between the
two media, and that further research is needed to explore several other factors that
may contribute to the difference in reading speed between paper and screen. As
mentioned earlier, these include posture, character set, resolution, justification (left
and full), familiarity with the medium and image polarity (dark characters on light
backgrounds and vice versa).
The last of these factors is particularly interesting because in this case the
available evidence appears to present fairly clear-cut results: dark characters on a
light background (a positive image) are read quicker (Radl, 1980), more accurately
and are more acceptable (Bauer & Cavonius, 1980) than are the light characters on
a dark background generally seen on V D U screens. However, they cause slightly
more visual fatigue (Cushman, 1986), possibly because of increased flicker (Isensee,
1982). It could well be argued, therefore, that the screen disadvantage obtained in
previous experiments reflects simply the fact that, by using negative-appearing
characters, the screen was placed at a disadvantage before the experiments began.
Some support for this contention has been provided by Cushman (1986), who
demonstrated no significant difference in reading speed when the image polarity was
the same in the paper and VDU presentations.
4 D , J , O B O R N E A N D D. H O L T O N

The results from the various studies, then, leave the reader in some confusion: it
is difficult to extract any indication of (a) whether screens and paper create a
difference in visual performance in the operator and (b) if they do, in which
direction the difference occurs. These problems have been created because of a
general lack of consistency within the experimental methods and stimuli used. For
various reasons, like has not been compared with like.
The purpose of the study reported here was again to consider the question of any
screen/paper difference, but, in this case, to attempt to control as many variables as
possible within the experimental situation. This includes subject's posture, distance
from the screen, image polarity, line length and page layout.

Method

SUBJECTS
Sixteen subjects--six males and ten females--with ages ranging from 16 to 40 years,
took part in the experiment. Only three out of the sixteen were familiar with
computers--and then only with video games; none were used to reading text from
screens. All subjects were volunteers with no payment given. They all reported good
vision, corrected or uncorrected.

MATERIAL AND APPARATUS


The material presented to all subjects consisted of eight short passages taken from
randomly chosen short stories of H. H. Munro (also known as Saki). The contents
of the passage were always taken from the beginning of the story, and there were
approximately 380 words per passage.
For both screen and paper conditions the page layout was precisely the same and
was produced in the following way: the stories were first entered into the computer
using a word processor with the line length set to 80 characters per line,
non-right-justified and word-wrapped. This produced the screen-based stimuli. The
paper-based stimuli was produced by typing exactly the screen presentation onto
paper. Thus the position of the words on each line and the line breaks were the
same in both conditions. The text-based paper was then photographed and enlarged
to create a piece of text that was exactly the same size as that which appeared on the
screen. Thus character size between the two presentation modes remained the same,
although it is accepted that very slight differences in interline spacing might have
occurred,
In the screen condition, a BBC "B" microcomputer with a Sanyo monitor was
used. The text was black characters on a bright green background in one condition
and bright green characters on a black background in the other. In the paper
condition subjects were presented with matt photographs of the text with either
black characters on a white background or white characters on a black background.
Ten comprehension questions (multiple-choice) were given on the first four
passages in each condition. The presentation order of the correct answers was
randomized between the three possible answers for each question,
At the end of the experiment a questionnaire was given that asked for the
subject's, age, sex and familiarity with computers. It also asked for the subject's
SCREEN VERSUS PAPER 5

views of the four conditions experienced (screen, paper, and dark and light
character displays). The subjects were asked to rank the conditions according to (1)
ease of reading, (2) preference and (3) pleasantness.
Both the screen and the paper were consistent with regard to the vertical angle
presented to the subject. The computer character heights were 4 mm and the paper
character heights were 3 mm. Because of the difference, the distance from the retina
was adjusted to the paper to give the same visual angle as in the screen condition.
Retinal distance was set to 40cm in the screen condition (recommended by
Helander & Rupp, 1984), which provided a visual angle for full depth of text of
0-33 tad. The distance from retina to paper text was set to 37.5 cm to produce the
same visual angle. In the paper condition the text was propped up on a book stand,
and its angle to the vertical (10~ from the vertical) replicated by altering the angle of
the computer screen.
Illumination from both the paper and the screen was measured 1 m from the
stimulus using a Megatron light meter. Levels were 120 lux for both media with dark
backgrounds and 180 lux with light backgrounds. Room illumination was provided
by one overhead strip light.
The computer and the paper format were positioned on a table side by side, with
the subject's chair facing either paper or screen. A timer was placed behind the
computer in view of the experimenter. The subjects had access to a button that
stopped the timer, recording how long it had taken to read:

PROCEDURE
The experiment used a within-subject design, with each subject experiencing all four
conditions of screen and paper, and light and dark characters. Two passages were
presented in each condition, with comprehension tests being completed on one
passage in each condition. Text medium and character polarity were balanced over
all sixteen subjects. The eight tests were varied in a latin-square design.
Subjects were seated at the pre-determined distance in front of either the screen
or paper. They were then told that they would be given eight short passages to read;
four of the passages would be written on the computer screen and four would be
written on paper. They were asked to read as if for pleasure and that the time taken
to read would be noted. They were also told that after some of the passages they
might be given a set of questions to measure comprehension. Both media stimuli
were presented by uncovering the prepared text so that no time would be lost in the
screen condition when the text was produced.

Results
The time taken to read the passages was analysed using a two-way analysis of
variance with repeated measures over both factors (Table 1). As can be seen, no
significant difference (p > 0.1) was obtained between either stimulus medium (mean
response time for paper = 88.92s and for screen = 88.32s; SD = 22-92 and 22.36
respectively) or character polarity (mean response time for light on dark (negative
image) = 90.2 s and for positive image = 87.03 s; SD = 24.24 and 20-8 respectively).
Comprehension scores were analysed using the same method. Again, as shown in
6 D.J. O B O R N E A N D D. H O L T O N

TABLE 1
Two-way analysis of variance of the effects of
presentation medium and character polarity on
reading speed
Source SS DF MS F
A (screen vs 5.802 1 5-802 0.047
paper)
B (light vs 160.6873 1 160.687 1.303
dark)
AB 119.6016 1 119.601 0.969
Subjects 26991.1410 15
Residual 5549.9731 45 123.333
Total 32827.4744 63

Table 2, no significant differences were obtained between the number of questions


answered correctly for the variable "media" (mean V D U = 131-5; paper = 134.5) or
between character polarity (mean negative = 133-5; mean positive = 133.5).
When subject preferences obtained from the questionnaires were examined,
however, a clear preference was provided for the normal paper type of presentation
with dark characters on a light background. A Friedman two-way analysis of
variance by ranks indicated significantly higher rankings for this type of presentation
for "Most Pleasing" ( X 2 = 13.12), "Easiest to R e a d " (X 2= 11.55) and "Preferred
M o d e " (X 2 = 8.63). Each of these statistics are significant at the 5% level.
Taken together, these results show that in this experiment the subjects were able
to read and comprehend text on both screen and paper equally well, and there is no
performance difference between character polarity. However, subjective preferences
still remained for the normal presentation of dark characters on white paper.

TABLE 2
Two-way analysis of variance of the effects
of presentation medium and character po-
larity on comprehension scores
Source SS DF MS F
A (screen vs 0.5625 1 0.5625 0-34
paper)
B (light vs 6.2500 1 6-2500 3-77
dark)
AB 1.5625 1 1.5625 0-94
Subjects 42.7500 15
Residual 74.6250 45 1-6583
Total 125.75 63
SCREEN VERSUS PAPER 7

Discussion

By returning to a basic comparison, the present experiment has demonstrated that


when all variables remain constant there is no difference in reading speed or
comprehension between screen and paper. This is a particularly important finding
both in the extent to which it places into context previous studies in this area, and
the way in which it allows future developments of electronic media to progress
without the fear that reading performance may be degraded.
To reiterate, there were several major differences between the studies that found
a screen disadvantage (e.g. Muter et al., 1982; Kruk & Muter, 1984) and the
present study. Thus in studies that had obtained a screen disadvantage the subjects
were allowed to hold the paper-format presentation, whereas the screen was
generally in a fixed position. This meant, of course, that subjects were given
freedom to change the viewing position of the paper presentation. They could vary
the distance between eyes and texts, vary the angle of the paper, use fingers to keep
their place, and possibly adjust the amount of light falling on the text by turning the
paper towards or away from the light source. That this represents a disadvantage to
the screen was demonstrated by Bhatneger, Drury & Schira (1985), who found that
performance of workers on VDUs changed if posture was altered. Errors increased
from about 4% to about 6% between best and worst conditions. In the present study
the paper was positioned on a stand in front of the subjects at the same reading
distance, inclination and conditions as the screen.
Time allowed to view the text presents another difference between previous
studies and the present one. Thus Muter et al. (1982) required subjects to read
continuously for 2 hours, much longer than in the present study. In the Kruk &
Muter (1984) study the subjects read continuously for twenty minutes and were then
given comprehension tests. In the present study the subjects read for an average
of 12 minutes, but reading time was interspersed with comprehension question-
naires, which were given immediately after the subjects had read the particular
passage. This resulted in the subjects in the present study reading for much shorter
periods and with breaks in between. It may be the case therefore, that any
screen-paper difference occurs only with longer duration reading periods, a
suggestion supported by the observation that other studies that have demonstrated
no screen disadvantage also employed short reading periods (e.g. Switchenko, 1984;
Askwall, 1985). This would be an important question to test empirically since the
range of tasks that might be performed in the electronic office of the future could
include both long- and short-duration reading periods.
Although the present study has shown quite definitely that there is no inherent
performance difference between computer screen and paper, it might, finally be
asked whether this should be the end of the story. Two questions emerge. First, are
reading speed and comprehension appropriate measures for this type of task?
Second, is it really sensible to compare directly material presented on paper with
exactly the same type of material presented on a screen?
Regarding the first question, it has been suggested that most of the usual methods
of assessing readability lack the ability of fine discrimination. For example, Kolers,
Duchnicky & Ferguson, (1981) suggest that both reading speed and comprehen-
sibility suffer from the major fault that all information regarding perceptual
8 D.J. OBORNE AND D. HOLTON

constituents of the task are lost, and that the only available information that remains
is total time on task. As for other measures, such as identifiability or discriminability
of individual characters, Kolers et al. suggest that clarity and discriminability of
individual letters are only one component in overall reading. After all, Huey (1908)
demonstrated empirically that people can often read, and understand quite well,
texts in which individual characters have been made unrecognizable by smudges,
distortions and other permutations. Hence a text's readability and legibility cannot
always be predicted accurately from confusion matrices or discriminability texts of
individual letters.
For the problem posed by this study, then, Kolers et al. (1981) would suggest that
by far the best method of measuring readability would be to monitor eye
movements. The suggestion is that difficulty, discriminability, comprehensibility and
related feature of processing text reveal themselves by changes in frequency,
duration and location of fixations.
The second question posed relates more to the future. What sort of text is likely
to be presented on VDU screens in the future? Is it likely, at all, to be of the same
form or of the same format as text presently presented on paper. Such questions are
beginning to be asked as the electronic presentation of information becomes more
pervasive. With the more "intelligent" facilities offered by computers, it is
questionable whether future electronic communication of text is likely to be in
exactly the same form, format, amount and type as its paper counterpart. This being
so, the important question might not be as posed in the present paper--does a
screen presentation produce different (worse or better) performance than its paper
counterpart?--but rather: do individuals perform better using screen or paper when
the material is presented in the most optimum format for that medium? This
question, of course, begs a whole series of others, which need to be tested
empirically.

References
ASKWALL, S. (1985). Computer supported reading vs reading text on paper: A comparison of
two reading situations. International Journal of Man-Machine Studies, 22, 425-439.
BAUER, D. CAVONIUS,C. R. (1980). Improving legibility of visual display units through
contrast reversal. In GRANDJEAN, E. & VIGLIANI, E. Eds, Ergonomic Aspects of Visual
Display Terminals. London: Taylor & Francis,
BHATNEGAR,V., DRURY, G. C. & SCIIIRA, S. G. (1985). Posture, postural discomfort, and
performance. Human Factors, 27, 189-200.
CUSHMAN, W. H. (1986). Reading from microfiche, a VDT, and the printed page: subjective
fatique and performance. Human Factors, 28, 63-73.
HELANDER,M. G. & RuPP, B. A. (1984). An overview of standards and guidelines for visual
display terminals. Applied Ergonomics, 15, 185-195.
HUEY, E. B. (1908). The Psychology and Pedagogy of Reading. New York: Macmillan
(Reprinted by MIT Press, 1968).
ISENSEE, S. (1982). The perception of flicker and glare on computer CRT displays.
Proceedings of the 26th Annual Meeting of the Human Factors Society. Baltimore: HFS.
KAK, A. U. (1981). Relationships between readability of printed and CRT-displayed text.
Proceedings of the 25th annual meeting of the Human Factors Society. Baltimore: HFS.
KOLERS, P. A., DUCHNICKY,R. L. & FERGUSON,D. C. (1981). Eye movement measure-
ment of readability of CRT displays. Human Factors, 23, 517-527.
SCREEN VERSUS PAPER 9

KRUK, R. S. & MUTER, P. (1984). Reading of continuous text on video screens. Human
Factors, 26, 339-345.
MUTER, P., LATR~MOUILLE, S. A., TREURNIET, W. C. & BEAM, P. (1982). Extended
reading of continuous text on television screens. Human Factors, 24, 501-508.
NEWSTED, P. R. (1985). Paper versus online presentations of subjective questionnaires.
International Journal of Man-Machine Studies, 23, 231-247.
RADL, G. W. (1980). Experimental investigations for optimal presentation mode and colours
of symbols on the CRT screen. In GRANDJEAN,E. t~ VIGLIANI, E. Eds, Ergonomic
Aspects of Visual Display Terminals. London: Taylor & Francis.
SwrrcrlENKO, D. M. (1984). Reading from CRT versus paper: the CRT-disadvantage
re-examined. Proceedings of the 25th Annual Meeting of the Human Factors Society.
Baltimore: HFS.
TIMMERS, H., VAN NES, F. L. BLOMMAERT, F. J. J. (1980). Visual word recognition as a
function of contrast. In GRANDJEAN, E. & VIGLIANI, E. Eds, Ergonomic Aspects of
Visual Display Terminals. London: Taylor & Francis.
WAERN, Y. & ROLI,ENItAGEN, C. (1983). Reading text from visual display units (VDUs).
International Journal of Man-Machine Studies, 18, 441-465.
WRIGHT, P. t~ LICKORISH,A. (1983). Proof reading texts on screen and on paper. Behaviour
and Information Technology, 2, 227-235.
WRIGHT, P. & LICKORISH, A. (1984). Investigating referee's requirements in an electronic
medium. Visible Language, 18, 186-205.

You might also like