Sei sulla pagina 1di 8

Displays 20 (1999) 3–10

CRT versus LCD: A pilot study on visual performance and suitability of


two display technologies for use in office work
Marino Menozzi*, Urs Näpflin, Helmut Krueger
Institute of Hygiene and Applied Physiology, Swiss Federal Institute of Technology, Clausiusstrasse 25, CH - 8092 Zürich, Switzerland
Received 24 July 1998; received in revised form 22 September 1998; accepted 22 September 1998

Abstract
Cathode ray tube (CRT) display and liquid crystal display (LCD) were compared for their suitability in visual tasks. For this purpose visual
performance was assessed by means of a search task carried out using both displays with different levels of ambient light. In addition,
suitability was rated subjectively by users of visual display units (VDUs). Error frequency for search tasks carried out using LCD were
significantly smaller when compared to error frequency for tasks at CRT. LCD gave rise to 34% less errors than did CRT. Reaction time in
search task was found to be significantly shorter using LCD when tasks were carried out in darkness. Subjective rated suitability of LCD was
scored twice as high as suitability of CRT. Results indicate that LCD used in this experiment may give better viewing conditions in
comparison to CRT display. 䉷 1999 Elsevier Science B.V. All rights reserved.
Keywords: Cathode ray tube (CRT); Liquid crystal display (LCD); Visual performance

1. Introduction requiring a high frequency of changes in location of the


work place as is the case with jobs like those of a represen-
Liquid crystal displays (LCD) have become more and tative or a salesman. Minimal power consumption and light
more popular in visual display units (VDUs) (Cladis [1], weight is a must for displays used in notebooks. Actually
Firester [2]). Size of actual LCD can cover the needs of there are only very few alternatives to LCD for displays for
most applications running on computers. For many reasons use in notebooks. Data compiled by Nelson [4] and by
LCD might become more important and might replace cath- Caladis [1] demonstrate that market volume of flat panel
ode ray tube (CRT) displays in many applications. Weight displays is predominantly controlled by LCD technologies.
and volume of LCD are among most the important advan- As a result of technical progress, physical-optical quality
tages when compared to CRT VDU. A 17 00 CRT typically of LCD has immensly improved (Tannas [5]). Among
occupies an area of 40 cm × 45 cm (width × depth). If others, backlight techniques, thin film super twisted LCD
according to suggested settings for work place (e.g. DIN and new materials enable a better visibility of information
4549 [3]) a desk of 120 cm × 80 cm is considered, a 17 00 displayed on an LCD and make requirements for ambient
CRT may occupy about one-fifth of the surface of the desk. light of VDU less critical. In contrast to CRTs, LCDs have
Dimensions of LCD monitors are smaller, therefore requir- sharp edged pixels being therefore more suited to produce
ing less space and facilitating handling of the monitor. From sharp edged horizontal and vertical lines. Moreover, pixels
an ecological point of view operation of an LCD is more of LCD are not subject to spatial instabilities such as jitter.
advantageous than that of a CRT. Owing to lower power Uncontrolled external electromagnetic radiation may induce
consumption, LCDs emit less heat, therefore causing less jitter at a CRT reducing legibility of characters displayed.
problems in air conditioning at offices where many displays Visibility of flicker may be less at LCD because of a more
run at the same time. Low power consumption also gives favorable time-course of luminance of single pixels. Most
LCD an advantage over CRT with regard to potential of CRT displays are equipped with phosphors with a short
electromagnetic radiation for causing possible effects on persistence-time. Light emitted by pixels of CRT can be
biological matter. compared to series of single flashes causing the perception
Notebook PCs are very popular during travel or at any job of flicker phenomenon which is especially pronounced at
large screens with a low refresh rate (Farrell [6]). In
* Corresponding author. Tel.: ⫹ 41 1 632 39 81; Fax: ⫹ 41 1 632 11 73. contrast, shape of time-course of luminance of a single
0141-9382/99/$ - see front matter 䉷 1999 Elsevier Science B.V. All rights reserved.
PII: S0141-938 2(98)00051-1
4 M. Menozzi et al. / Displays 20 (1999) 3–10

pixel at LCD is square wave-like causing less or no percep- VDU, search tasks may constitute a good starting point. We
tion of flicker. Flickering light is supposed to disturb control investigated visual performance by assessing time for
of eye movements (Neary [7]), possibly a cause for visual completing a search task and the amount of errors occurring
complaints. Drawbacks of use of LCD are, among others, during a search task. Suitability of a display for use in office
reduced brightness and restricted viewing angle (Nelson work is controlled by factors more than only visual perfor-
[4]). mance. Field studies constitute a method accounting for a
VDUs are often used in the presence of ambient light. large quantity of factors. However, a large number of
Ambient light is reflected on the front of the screen thereby subjects must be observed over a long period of time in
reducing contrast of information displayed. Contrast reduc- order to be able to cancel out individual interferences.
tion can be controlled by measures applied on the surface of Long-term observations in the field lack from constancy
the screen. With regards to material used, glass in CRT and of environmental variables confounding possible effects.
organic material in LCD, LCD may have an advantage over Mentioned limitations can partially be overcome, if suitabil-
CRT displays when used in bright ambient light. To our ity is assessed subjectively. Users of VDUs may judge suit-
knowledge, reflectivity of LCD has not yet been assessed ability considering a multitude of factors and their impact on
and compared to reflectivity of CRT. Activation of pixels of stress at long-term exposure. In this study we undertook an
LCD might change reflectivity of the display. To account for attempt to rank suitability of the two displays by interview-
this possibility, reflectivity should also be assessed while ing VDU users.
information is presented on the display.
Insufficient optical quality of displays is a potential candi-
date for causing visual complaints (Cole [8], Jackson [9], 2. Method
Krueger [10], Läubli [11]). The domain of visual complaints
is complex. Inter-individual variations in physiological and 2.1. Procedure
psychological predispositions of VDU users are probably
amongst the most confounding factors in localization of Performance was evaluated by means of a search task in
causes of the complaints. An accurate assessment of optical which we recorded reaction times for detecting targets and
quality of displays may help in identification of causes for amount of errors which occurred during the task. The para-
complaints. In accordance with the concept of strain and digm used in our task was adapted from paradigmata used to
stress (Hacker [12]) we may deduce that complaints are assess visual performance in human factors in lighting (see
closely related to visual performance. Good optical quality Boyce [20]) and from paradigmata used in basic visual
of a display constitutes a low visual strain and facilitates science (e. g. Fiorentini [21], Treisman [22]). Adaptations
reading (Grisham [13], Legge [14]). Scientific literature aimed to consider particular conditions of office work. A
offers some papers on optical quality of VDU and visual two-alternative forced-choice task was set up using an
performance (Edwards [15], Farrell [6], Montegut [16], uppercase letter ‘‘F’’ as target and uppercase letters ‘‘E’’
Roufs [17]). MacKenzie et al. [18] compared user perfor- as distractors. Target and distractors were arranged in a 20
mance in manipulative tasks carried out using CRT or LCD. × 20 matrix with equally spaced horizontal and vertical
In their experiments subjects had to select targets on the gaps. Target was shown in 50% of the displays. The task
display by using a mouse. Time between successive button consisted in scanning the display and pressing either the
down actions of the mouse were recorded and defined as ‘‘yes’’ or the ‘‘no’’ button of an answer box, depending
movement time. MacKenzie et al. [18] found that LCD gave on whether target was seen or not. Subjects were informed
rise to 34% longer movement times than did CRT. Saito et after each trial on the correctness of the given answer. An
al. [19] recorded several visual functions during different acoustic feedback was used for this purpose. The subjects
tasks carried out using different displays, such as LCD, were asked to accomplish the task within a minimum time
CRT and plasma displays. Visual accommodation was avoiding errors.
found to be faster while using CRT when compared to In order to account for the fact that ambient light may
tasks where LCD or plasma display was used. vary depending on location of workplace, experiments were
Based on actual results reported in the literature it is carried out at two levels of ambient light. In one condition,
difficult to draw a conclusion on the suitability of a parti- further on referred to as darkness, horizontal and vertical
cular display technology to improve viewing conditions. We illumination was about 50 lux. In the other condition,
therefore set up an experiment by means of which visual further on referred to as brightness, vertical illumination
performance was assessed using either LCD or CRT at was 250 lux while horizontal illumination was set to
otherwise identical conditions. 550 lux. Only diffused light consisting out of indirect
There are many parameters contributing to an overall daylight or artificial light reflected from the surrounding
visual performance. As a first attempt in investigating suit- walls was used to install described levels of ambient light.
ability of mentioned techniques we were interested in para- Each subject completed the task at both displays and on both
meters related to visual performance at a common task at a conditions of ambient light. The four different settings, i. e. two
VDU. If office work is considered to be a common task at displays used at two different conditions of illumination, were
M. Menozzi et al. / Displays 20 (1999) 3–10 5

administered in random order. The experimental session was 0.28 mm. The CRT display was run at a resolution of
consisted out of four blocks of 40 trials each. A training 800 pixels × 600 pixels at 60 Hz frame rate.
course comprising one block of 15 trials each was carried Black distractors and targets were generated on a white
out in brightness at each of the two displays before the background. The luminance of the background was set at
experimental session. Data obtained from trial session 57 cd/m 2 at both displays, CRT and LCD. In order to avoid
were discarded. At the end of the experiment, subjects changes in light adaptation, background of the display was
were asked to rate subjectively how much they would also set to 57 cd/m 2 during the intermediate period. A large
appreciate completing office work in the four experimental black area served to estimate luminance of the black char-
conditions. acters. Luminance measured within this black field was
Subjects were informed on aim and procedure of the 1.4 cd/m 2 when measured with no ambient light, i. e., in
experiment. Before starting the experiment, subjects gave darkness. In brightness luminance on the screen is increased
their written consent for participating in the experiment and because of reflections on the screen.
for using their data for scientific evaluation. Reflection properties of the two displays were assessed in
accordance with ISO 9241-7 (1998) [24] procedure. By
2.2. Instrumentation means of this procedure diffuse (RD) and specular (Rs)
reflectance properties of a display were evaluated. Specular
A program was developed and run on a notebook reflectance is evaluated using an extended light-source
(Toshiba Satellite 100CS). This hardware and software subtending 15⬚ (Rs(EXT)) and a small light-source subtend-
enabled to display a matrix (20 × 20) consisting of target ing aproximately 0.9⬚ (Rs(SML)). In general, reflectances of
and distractors. The program enabled control of the percen- LCD used in our experiment were much smaller than the
tage of displays containing a target and to present displays ones of CRT. For LCD all three reflectances were found to
with and without targets in random order within a sequence be less than 0.01 (RD ˆ 0.007, Rs(EXT) ⬍ 0.001, Rs(SML)
of 40 displays. As mentioned earlier probability for target ⬍ 0.001). For CRT values of reflectance were RD ˆ 0.035,
was set to 50%. The location of the target was selected Rs(EXT) ˆ 0.037, Rs(SML) ˆ 0.005. We can therefore
randomly to be one of the 400 places of the 20 × 20 matrix. estimate that for CRT, surplus luminance owing to reflec-
Reaction time and answer given by the subjects were tion of ambient light is about 3 cd/m 2 (specular reflection).
recorded using an answer box (Von Buol, submitted [23]) In brightness, contrast is therefore reduced by about 10% if
which was connected to the notebook. The answer box was expressed in term of modulation (Michelson contrast).
equipped with a quartz driven timer. The timer was started Contrast reduction based on diffuse reflections can be
by a photodiode connected to the display on which the neglected in any of the experimental conditions. As illumi-
matrix was presented. This technique enabled synchroniza- nation was carried out using diffuse light, we expect reflec-
tion of the start of the timer with the switching on and off of tions not to be a factor interfering with visual task.
pixels within a particular area of the screen, avoiding there- Viewing distance in LCD-task was set to 50 cm. Equality
fore errors caused by buffering of data occurring in high of viewing angle of characters and matrix in LCD-task and
level programming languages and high level operating in CRT-task was achieved by adjusting settings of CRT
systems. The timer was stopped by the subjects by pressing monitor (magnification of image) and by adjusting viewing
either the ‘‘yes’’ or the ‘‘no’’ button on the answer box distance in CRT-task. As a result of limitation in settings of
depending on whether they detected a target or not. A CRT, viewing distance in CRT-task had to be increased by
new display was presented three seconds after the answer 10 cm compared to viewing distance at LCD-task in order to
had been recorded. During these three seconds a number equalize the sizes of characters and matrix displayed in
indicating the number of trials completed was displayed in LCD-task. The size of the letter E subtended 20.6 0 ×
the center of the display. The number served as a target for 26.8 0 (horizontal × vertical), roughly corresponding to a
fixation for the subjects during the intermediate period 12 point size letter E viewed at a distance of 50 cm. Hori-
between the displays. zontal and vertical spacing of neighboring characters were
The display of the notebook was used as LCD in one approximately equal (about 0.6⬚). The size of the matrix was
condition, further on referred as LCD-task. The LCD was about 10.7⬚ × 12.8⬚ (horizontal × vertical). We did not use
a backlight dual scan STN 10.4 00 color display with a reso- any rest to fix viewing distance. Subjects were asked to keep
lution of 640 pixels × 480 pixels (horizontal × vertical). their viewing distance fixed during the experiment.
The distance between pixels was 0.3 mm in horizontal and Subjective ratings of suitability of each of the four experi-
in vertical direction. Referring to the manual of the display mental settings for office work was accomplished by asking
the visible area of the display was 217.2 mm × 164.4 mm. the subjects ‘‘How much would you like to work using this
A 14 00 CRT (Dell model D1428-HS, shadow mask color screen on this ambient light?’’ (German: ‘‘Wie gerne
display, no surface treatment) was connected to the external würden Sie an diesem Bildschirm bei diesen Beleuchtungs-
display connector of the notebook and used in the task bedingungen arbeiten?’’). Subjects answered by putting a
where the matrix was displayed on a CRT. This task is cross on an interval scale of six steps. They were instructed
further on referred as CRT-task. The dot pitch of the CRT to make the position of the cross coincide with their ratings.
6 M. Menozzi et al. / Displays 20 (1999) 3–10

The scale ranged from ‘‘not at all’’ (‘‘sehr ungerne’’) to display. Correct detection of target occurred in 559 trials. In
‘‘very much (‘‘sehr gerne’’). Subjects were also asked to conditions where displays did not contain the target, a
comment their decision by shorthand notes. correct answer was given 4.14 ^ 2.74 s after onset of
All the measurements were carried out within two days. the display. A correct answer was given at nearly all (795)
Measurements lasted 35–80 min (median less than 50 min) of the 800 displays without targets.
including completion of formalities. In order to assess learning effects, blocks were numbered
according to the order they were administered. We will refer
2.3. Subjects to this number further on as block number. An ANOVA was
run with subject and block number as factors and reaction
A total of 10 subjects, 5 males and 5 females, participated time as independent variable. The results show that between
in this study. All subjects had a good visual acuity of at least subjects variations are significant (F(9) ˆ 15.17, p ⬍
1.0 for near (33 cm, Landolt rings) and 1.25 for far. Subjects 0.001). The same is true for the factor block number (F(3)
age ranged from 26 to 34 years. VDU work constituted an ˆ 3.87, p ⬍ 0.01), indicating that reaction time increases
important part of the daily work of our subjects. Same significantly with block number. However, the model used
subjects were used to assess suitability subjectively. in ANOVA turned out to explain only about 14% of varia-
tions in reaction time. If means for reaction times over 40
2.4. Data analysis trials were used instead of single reaction times in ANOVA,
First, reaction time and error frequency were analyzed in learning effects disappear (F(3) ˆ 0.953, p ⬎ 0.4). Error
order to determine whether effects of learning were present frequency, defined as the number of errors within one block
in the data. For this purpose an analysis of variance of 40 trials, were calculated for each subject and each block
(ANOVA) and Pearson chi-square tests were used. separately. Pearson chi-square test applied on a two way
Second, we investigated whether errors were equally table listing error frequency within each block and for
distributed within the matrix displayed or whether they each subject indicates no (x 2(3, 9) ˆ 28.03, p ⬎ 0.4)
occurred more frequent in a specific direction of gaze, there- systematic effect of subject or block number on error
fore in a particular part of the matrix. For this purpose the 20 frequency. Therefore error frequency are not subject to
× 20 matrix was subdivided into nine sectors (3 × 3) each learning effects.
defining a sub-matrix of either 7 × 7, 6 × 7 and 7 × 6 In order to investigate whether errors preferably occurred
(periphery) or 6 × 6 (central) characters. By means of a within a specific area of the display, error frequency of the
Pearson chi-square test homogeneity of distribution of error nine sectors defined earlier on in the text, one central square
frequency within this sectors was analyzed. sector surrounded by eight square sectors, were normalized.
Third, the central question whether different technologies This procedure consisted of dividing error frequency for
for displays and different ambient light give rise to varia- each sector with the number of targets presented within
tions in visual performance was studied by applying the same sector. Owing to low frequency, errors of all
ANOVA statistics to reaction times. On account of a depar- subjects were pooled. Normalization enabled to account
ture of the distribution of errors from normal distribution for different number of targets presented within a sector.
error frequencies were processed by means of Wilcoxon Differences were because of unequal size of sectors as
rank test. well as the random selection of position of target. Pearson
Fourth, subjective ratings on suitability of settings for chi-square test indicates that frequency of errors did not
office work were compared. For this purpose, position of vary (x 2(2, 2) ˆ 2.66, p ⬎ 0.6) among sectors.
the cross on the rating scale mentioned before was An ANOVA was used to determine the variation of reac-
converted to scores and fed to a Wilcoxon rank test. tion time with experimental conditions, i.e., the display
Keywords were assigned to shorthand comments made by technology and the ambient light. Multivariate ANOVA
the subjects. analysis included factors such as block number, settings
(CRT-task, LCD-task, darkness, brightness), subject,
presence of target (yes, no) and error of target detection.
3. Results In order to account for individual preferences for settings,
interaction between settings and subjects was also consid-
A total of 1600 displays were presented (four sets of 40 ered in statistics. Reaction time was found to be independent
trials, 10 subjects), 800 with and 800 without target. An of block number (F(2) ˆ 1.15, p ⬎ 0.3) and of settings
incorrect answer was given at 246 trials, i.e., at each seventh (F(1) ˆ 1.24, p ⬎ 0.2). Reaction time depends on subject
display or at about 15% of the displays shown. Nearly all (F(9) ˆ 14.72, p ⬍ 0.001), on presence of target (F(1) ˆ
incorrect answers, i.e., 241, resulted from missed detection 36.6, p ⬍ 0.001), on error of detection (F(1) ˆ 19.5, p ⬍
of target. About 30% of the targets shown were missed. In 0.001) as well as on the interaction between subject and
trials where target was present, a correct answer was given settings (F(26) ˆ 4.39, p ⬍ 0.001). We may suppose
after 3.28 ^ 2.20 s (mean ^ standard deviation over all that failure to demonstrate any effects of settings on reaction
subjects and all four experimental settings) after onset of the time may be a result of the strong effect of the factors
M. Menozzi et al. / Displays 20 (1999) 3–10 7

Table 1
Reaction time (RT) and error frequency. Experimental settings: CRT-task, LCD-task, darkness or brightness. First to third row show mean and standard
deviation for reaction time in seconds. Only differences in third row are statistically significant. Bonferroni adjusted Student’s t-test shows significant
difference of mean of CRT-task and LCD-task for dark ambient light. Fourth row denotes median and quartile of errors for each setting. Total number of
errors is reported in fifth row. Median and quartiles of pooled error frequency for CRT-task and for LCD-task are shown in the sixth row (Standard deviations
appear in parantheses)

CRT-task LCD-task
Brightness Darkness Brightness Darkness

RT mean (s) for detecting target 3.02 (1.88) 3.48 (2.30) 3.52 (2.47) 2.91 (2.03)
RT mean (s) for missing target 4.11 (2.69) 4.70 (2.91) 4.07 (3.00) 4.18 (2.87)
RT mean (s) when target present 3.52 (2.23) 3.93 (2.60) 3.64 (2.60) 3.25 (2.34)
Error frequency (median 7 [4;9] 6.5 [5;10] 4 [3.5;7.5] 5 [4;6]
[quartiles])
Total errors 71 74 43 53
Error frequency of pooled data 12.5 [9;19] 8.5 [6;11]
(median [quartiles])

presence of target, error detection of target and their inter- frequency. As can be seen from Table 1, median error
action. We therefore carried out post hoc ANOVA on a frequency varied from 7 errors per block (40 trials) in the
reduced data set in which only trials with targets were CRT-task in brightness to 4 errors per block in the LCD-task
considered. Table 1 shows mean values and standard devia- for same conditions of ambient light level. Significance of
tions for reaction time as well as error frequency assessed difference in error frequency between the four settings is
using the reduced data set. listed in Table 2.
Reaction times assessed for displays without target are In general, error frequency at LCD-tasks was signifi-
not shown. First two rows list mean and standard deviation cantly lower when compared to error frequency at CRT-
of reaction time for trials in which the target was detected task. In brightness error frequency in CRT-task was found
(first row) and in which the target was missed (second row). to be significantly higher (p ⬍ 0.02) than error frequency at
None of the difference in reaction time were significant LCD-task. The same difference applies for darkness (p ⬍
within each of the first two rows. If an ANOVA is run 0.01). Error frequency at CRT-task on brightness are signif-
considering reaction time of all trials at which target was icant (p ⬍ 0.01) higher than the ones assessed at LCD-task
present, independent of whether target was detected or in darkness. In accordance with Table 2 error frequencies
missed, reaction time turns out to depend significantly on within display type do not differ significantly. Errors occur-
settings (F(3) ˆ 2.81, p ⬍ 0.05). A Bonferroni adjusted ring at the same display type can therefore be pooled. If total
matrix of pairwise comparison of data of the third row errors of CRT-task are pooled (145) and compared to pooled
reveals that in dark ambient light a significant (p ⬍ total errors of LCD-task (96), we can conclude that at LCD-
0.05) longer reaction time results for CRT-task (3.93 s) task the total amount of errors is 34% less than it is at CRT-
when compared to LCD-task (3.25 s) at same ambient light. task. A paired t-test reveals that difference in pooled data is
As only few errors (5 out of 800 possible) were done at significant at a level of p ⬍ 0.001 (t(9) ˆ 4.72) for two
displays without target, trials without target were not taken sided probability (p ⬍ 0.005 if non-parametric Wilcoxon
into account in the analysis on error frequency. Settings rank test is used).
were found to exhibit a significant influence on error Distribution of scores for suitability of settings for office
work are plotted in Fig. 1. One subject did not answer the
Table 2 two questions on suitability of settings of LCD-task.
Significance of difference in error frequency. Each entry denotes p value for A Wilcoxon rank test failed (p ⬎ 0.1) to show
difference in error frequency between two settings. Probabilities listed in any significant differences in scores given to the four
first four lines were calculated using Wilcoxon rank test whereas probabil- different settings. A comparison of pooled data
ity listed in the last line of the table refers to two tailed t-test statistics for
paired samples
assessed at CRT-tasks with pooled data assessed at
LCD-tasks also fails to show significant effects
CRT-task LCD-task (Wilcoxon rank, p ⬎ 0.1; two-tailed t-test, p ⬎ 0.1).
Bright Dark Bright Dark Shorthand notes used to comment on given scores
CRT-task Bright 1 were mostly of negative nature. Table 3 summarizes
Dark .587 1 negative remarks which were mentioned at least twice.
LCD-task Bright .015 .070 1 Interestingly two subjects found fault with sharp char-
Dark .007 .005 .147 1 acters of LCD, a characteristic which usually denotes
high quality. Two subjects contested inhomogeneous
CRT versus LCD task (pooled) 0.0011
luminance of LCD.
8 M. Menozzi et al. / Displays 20 (1999) 3–10

possibility of specular reflections of displays based on


their personal experience rather than visible reflections
present during the experiment.
Attention is influenced by behavior, and its focus must
not necessarily be at the location of the fixation point (Treis-
man [22]). Experiments reported by Posner [27] showed a
redirection of focus of attention towards locations in which a
pre-attentive stimulus was shown. We may postulate that
focus of attention can be dislocated by manipulative tasks
as is the case with spatial orientation. Shebilske [25] demon-
strated that a task requiring downward gaze distorted spatial
Fig. 1. Box plot of scores. Scores (median, quartiles, extreme values) were orientation, e.g., that of baseball batters. As there are many
assessed by means of the question ‘‘How much would you like to work tasks in office work requiring downward gaze and down-
using this screen in this ambient light?’’ using a scale of six intervals. * ˆ wards focused attention, we could assume a drift of attention
outlier. towards lower half of visual fields. However, error
frequency was found to be independent of location of target
4. Discussion on display. If it is supposed that variations of visibility did
not cancel out effects of varying attention, we may state that
Reaction time was found to depend on learning effect. attention does not vary with direction of gaze or orientation
Apparently, two training sessions before the experiment in space in a critical manner. We may therefore conclude
were not enough to eliminate the effect. Learning effect that position of information to be displayed on the screen
disappeared after reducing amount of data. Therefore, learn- must not be determined in consideration with relevance of
ing effect can be supposed to be weak and will probably not the information to be displayed. This conclusion seems to be
interfere with conclusions reported at the end of this article. somewhat in contrast to actual recommendations made in
Although time needed to search the display is reduced by style guides (e.g. DIN 66234 [26]).
learning, error frequency is found to be constant throughout Differences between error rates at CRT-task and at LCD-
the experiment. We might therefore postulate that accuracy task are considerable. Total amount of errors assessed at
of detection could not be improved whereas speed of LCD-task was 34% lower than amount of errors assessed
processing result of detection was improved by training. at CRT-task. As mentioned in the introduction several tech-
In our experiment speed of processing depends on motor nical factors may be responsible for causing differences in
as well as on mental skills, both of which can be improved accuracy of detection of target such as sharpness of pixels
by training. Given these circumstances we may conclude and time characteristics of luminance. Given the low
that optimal viewing conditions are imperative for ergo- complexity of presented distractors and targets we suggest
nomics if accuracy is a significant requirement in a task spatial characteristics to be of minor importance in deter-
because no training can make up for poor viewing condi- mining visual performance. Time course of luminance
tions in order to improve accuracy of vision. might have played a decisive role in vision so as to give
Among reasons for involuntary missing a target in our rise to more convenient viewing condition at LCD-task.
task are, low visibility of the target and an insufficient atten- However, LCD based technology is not necessarily required
tion. Visibility may be biased by reflections and by inhomo- to produce a time course of luminance needed for good
geneous emission characteristics of the displays. As was visual accuracy. The same beneficial effects result by
shown above, it is unlikely that reflections interfere with increasing refresh rate or by installing phosphors with
visibility of target. Subjective estimation of reflections long persistence in CRT. Increasing refresh rate has been
(Table 3) deviates slightly from this assumption. However, shown to facilitate reading (Montegut [16]). Technical
deviations are small. Probably some subjects rated reasons limit refresh rate in screens with high spatial resolu-
tion and with high depth of color representation. There are
Table 3 drawbacks in the use of phosphors with long persistence.
Shorthand remarks on suitability of settings for office work (summary)
First, effects like smearing while scrolling and transient
CRT LCD ghost images while changing mask reduce legibility of
Bright Dark Bright Dark displayed information at CRT equipped with long-persis-
tence phosphor. Second, there are no triplets of long-persis-
Size too small 3 3 5 5
Flicker 5 6 0 0 tence phosphors enabling to generate a color space of
Blur 5 4 0 0 similar size as can be done if triplets of short-persistence
Reflections 2 1 0 0 phosphors are combined.
Inhomogeneous luminance 0 0 2 2 Our experiment revealed a weak dependency of reaction
time on settings or on technology of display used. Failure to
Total 14 15 7 7
demonstrate strong effects might be because of low
M. Menozzi et al. / Displays 20 (1999) 3–10 9

sensitivity of our experimental procedure. As mentioned experiment. The computer program used for the search task
before, it is possible that mechanisms other than vision was written by Ara Hagopian. Andreas Hoffmann adapted
such as motor behavior, played a more significant role in the program to the needs of the notebook. Andreas von Buol
determining reaction time and could have cancelled out any provided electronics for precise assessment of reaction time.
effect on vision. Efficiency of target detection should be A special thank is addressed to our subjects who kindly
assessed without manipulatory tasks and without require- participated at the experiment without being paid. Last but
ments in time needed for highly cognitive work for decision not least we thank our reviewers for fruitful comments.
making. We propose to assess detection performance by
means of an experiment in which target is displayed tachis-
toscopically. Visual performance shall thereby be assessed References
in terms of percentage of correct detection of target as a
function of duration of presentation of target. By using [1] Patricia E. Cladis, World view of liquid crystal flat panel displays, in:
this approach it has been shown elsewhere (Fiorentini L.E. Tannas, E. Glenn, J.W. Dorne (Eds.), Flat-panel Display Tech-
nologies, Noyes Publications, Parkridge, NJ, 1995.
[21]) that important changes in correctness of detection [2] A.H. Firester, Active matrix technology, in: L.E. Tannas, E. Glenn,
are found if duration of visibility of display is changed J.W. Dorne (Eds.), Flat-panel Display Technologies, Noyes Publica-
from 100 to 200 ms. tions, Parkridge, NJ, 1995.
No subjective preference of either display technology or [3] DIN 4549 Büromöbel, Schreibtische, Büromaschinentische und
settings could be demonstrated statistically even though Bildschirmtische. Masse. Deutsches Institut für Normung e. V.,
(Office furniture, desks, desks for office machines and desks for
frequency distribution of subjective ratings reported in VDUs.) German Standards Institute, November, 1982.
Fig. 1 suggests a possible advantage of LCD technology [4] T.J. Nelson, J.R. Wullert II, Electronic Information Display Technol-
for office work. We conclude that the number of subjects ogies, World Scientific, Singapore, 1997.
used to carry out the rating is too small to demonstrate [5] L.E. Tannas, E. Glenn, J.W. Dorne (Eds.), Flat-panel Display Tech-
significant effects. Summary of shorthand notes on scoring nologies Noyes Publications, Parkridge, NJ, 1995.
[6] E. Farrell-Joyce, Fitting physical screen parameters to the human eye,
suitability of settings (Table 3) also indicate a possible in: J.A.J. Roufs (Ed.), Vision and Visual Dysfunction. The Man–
advantage of LCD over CRT. Subjects reported only few Machine Interface, Mocmillan Press, Hondmills, 1991.
drawbacks of LCD. Too small size of the screen as well as [7] Catherine Neary, A.J. Wilkins, Effects of phosphor on perception and
an inhomogeneous luminance were complained at LCD. the control of eye movements, Perception 18 (1995) 257–264.
The former complaint can be accommodated by using [8] B.L. Cole, Jennifer D. Maddocks, Effects of VDUs on the eyes:
Report of a 6-year epidemiological study, Optometry and Vision
displays with an adequate size. Screen size plays a role in Science 73 (8) (1996) 512–528.
suitability of a display for office work. However, screen size [9] A.J. Jackson, E.S. Barnett, A.B. Stevens, M. McClure, C. Patterson,
was not subject of our investigation. Findings made here are M.J. McReynolds, Vision screening, eye examination and risk assess-
independent of screen size. Latter factor might have exhib- ment of display screen users in a large regional teaching hospital,
ited some effect on our results, i.e., that errors might appear Ophthalmics and Physiological Optics 17 (3) (1997) 187–195.
[10] H. Krueger, Visual function and monitor use, in: J.A.J. Roufs (Ed.),
more frequent in specific areas within LCD. As we did not Vision and Visual Dysfunction. The Man–Machine Interface,
record variation of luminance across display area we are not Mocmillan Press, Hondmills, 1991.
able to estimate influence of inhomogeneity of luminance [11] T. Läubli, W. Hünting, E. Grandjean, Postural and visual loads at
on error frequency. VDT workplaces. II: Lighting conditions and visual impairments,
Ergonomics 24 (1981) 933–944.
[12] W. Hacker, Allgemeine Arbeits- und Ingenieurpsychologie (General
5. Conclusion Work- and Engineering-Psychology), 2, Huber, Bern, 1978.
[13] J.D. Grisham, Melissa M. Sheppard, Wendy U. Tran, Visual symp-
toms and reading performance, Optometry and Vision Science 70 (5)
The fact that the use of LCD improves accuracy in detect-
(1993) 384–391.
ing targets and might also improve time for visual detection [14] G.E. Legge, G.S. Rubin, A. Luebker, Psychophysics of reading V:
of the target is an advantage of LCD over CRT. It is there- The role of contrast in normal vision, Vision Research 27 (1987)
fore expected that LCD will give rise to lower visual strain 1165–1177.
and therefore cause less visual complaints as CRT will do. [15] C.B. Edwards, V.J. Demczuk, Legibility of visual display terminals
and an investigation of subjective image quality ratings with respect
Given the broad variety in technology of LCD available on
to physical image quality measures, in: H. Luczak, A. Cakir, G. Cakir
the market we may expect that there may be even further (Eds.), Work with Display Units 92, Elsevier Science Publishers,
improvements in visual performance if other LCD are used. Amsterdam, 1993.
Unfortunately subjective ratings did not reflect findings [16] M.J. Montegut, B. Bridgeman, J. Sykes, High refresh rate and oculo-
made on visual performance as was demonstrated by Roufs motor adaptation facilitate reading from video displays, Spatial
Vision 10 (4) (1997) 305–322.
[17] and Edwards [15].
[17] J.A.J. Roufs, M.C. Boschman, Visual comfort and performance, in:
J.A.J. Roufs (Ed.), Vision and Visual Dysfunction. The Man–
Machine Interface, Mocmillan Press, Hondmills, 1991.
Acknowledgements [18] I.S. MacKenzie, S. Riddersma, Effects of output display and control-
display gain on human performance in interactive systems, Behaviour
We thank the following people for supporting us in this and Information Technology 13 (5) (1994) 328–337.
10 M. Menozzi et al. / Displays 20 (1999) 3–10

[19] S. Saito, S. Taptagaporn, G. Salvendy, Visual comfort using different [24] ISO 9241 - 7 (1998): Ergonomic requirements for office work with
VDT screens, International Journal of Human–Computer Interaction visual display terminal (VDTs) – Part 7: Requirements for display
5 (4) (1993) 313–323. with reflections. Genève, Switzerland, 1998.
[20] P.R. Boyce, Human Factors in Lighting, Applied Science Publishers, [25] W.L. Shebilske, Baseball batters support an ecological efference
London, 1981. mediation theory of natural event perception, Acta Psychologica 63
[21] Adriana Fiorentini, Differences between fovea and parafovea in visual (1986) 117–131.
search, Vision Research 29 (9) (1989) 1153–1164. [26] DIN 66234. Bildschirmarbeitsplätze. Gruppierung von Information
[22] Anne Treisman, P. Cavanagh, B. Fischer, V.S. Ramachandran, R. von und Daten. Hinweise und Beispiele. Beiblatt 1 zu DIN 66234, Teil
der Heydt, Form perception and attention. Striate cortex and beyond, 3. Deutsches Institut für Normung e. V., März 1983. (Workplaces
in: L. Spillmann, J.S. Werner (Eds.), Visual Perception: The Neuro- with VDUs. Grouping of Information and Data. Remarks and Exam-
physiological Foundations, Academic Press, San Diego, CA, 1990. ples. German Standards Institute, March 1983).
[23] A. Von Buol, The measurement of reaction time by computers. A [27] M.I. Posner, Chronometric Explorations of Mind, Lawrence Erlbaum
simple hardware solution to provide millisecond resolution time. Associates, Hillsdale, NJ, 1978.
(Submitted)

Potrebbero piacerti anche