Color and defective color vision as factors in the conspicuity of signs and signals.
Research on factors affecting the conspicuity of signs and signals, especially the conspicuity of road traffic control devices, has been summarized by Jenkins and Cole (1986) and the Commission Internationale de L'Eclairage (CIE, 2000). The principal factors affecting the conspicuity of signs are angular size, the boldness of the internal graphics, the luminance contrast of the sign with its immediate background, the visual complexity of the background, and the angle between the direction of the sign and the direction of gaze. At night the luminance of the sign is also a factor.
The color of a sign has not emerged as an important determinant of conspicuity. The CIE report (2000) observed that the effect of color on sign conspicuity has not been studied rigorously, although some trends have emerged as a by-product of studies of other parameters. The CIE report tentatively concluded that white signs seen at night may need to have a higher brightness than do red, orange, green, and blue signs to achieve the same conspicuity. This suggests that these colors contribute to sign conspicuity.
Odeschalchi (1960) asked a panel of observers to subjectively judge the effect of sign area on conspicuity of road signs in a rural environment in daylight. He found that white, yellow, and red signs need to be about the same size to give what was judged to be adequate conspicuity (although yellow signs may give slightly greater conspicuity than do white signs), whereas green and blue signs need to be larger.
Forbes, Pain, Joyce, and Fry (1968) had observers rate the relative attention-getting property of pairs of differently colored simulated signs seen against four different-colored background scenes. These researchers found that white, yellow, and red signs were rated equally in attracting "first attention," whereas, in decreasing order, green, blue, and black attracted attention less well. However, Jenkins and Cole (1979) found that red, blue, and yellow discs superimposed onto projected colored scenes of the urban road traffic environment were no more conspicuous than grey discs of the same luminance.
It is surprising that color has not been identified as an important determinant of the conspicuity of signs, considering that people's everyday experience is that color -- especially vivid color -- can attract attention. Moreover, it has been shown that search time for colored objects is significantly less than that for the same achromatic objects when the color is unique to the target and the observer knows the color in advance (Christ, 1975). Under some circumstances color has been shown to have a pop-out characteristic, meaning that the colored object is immediately obvious, a quality that is independent of the number of distractor elements in the background. This is evidence of a parallel-processing search mechanism, which D'Zmura (1991) attempted to elucidate. However, target color seems to have little benefit to visual search when the color is not known to the observer in advance and when the proportion of nontargets (distractors) that have the same color as the target exceeds 7. Christ (1975) presented an elegant analysis of data on the role of color in target location and identification.
The advent of color-coded electronic visual displays provoked further research on the role of redundant color coding in information acquisition for complex electronic displays, in which color is not the primary source of the information displayed but is used only to reinforce the message or to provide visual organization. Although Tullis (1981) found no improvement in response time for a redundantly color-coded graphic display, both Luder and Barber (1984) and Macdonald and Cole (1988) found better performance for color-coded electronic aviation displays than for the same display without color.
This paper reports a study that aims to determine whether redundant color coding contributes to the conspicuity of road traffic signs and signals. The experiment does so in a unique manner by comparing the conspicuity of road traffic control devices in typical road scenes for observers with normal color vision with that for observers with a severe loss of color vision.
There were two reasons for taking this approach. One is that a laboratory-based comparison of the conspicuity of colored and achromatic targets can be confounded by changes in luminance contrast and in border sharpness when the colored slides are copied into a black-and-white form. The alternative, a field study, would have the significant logistic problem of replacing conventional signs with replicas lacking the color coding. Comparing the performance of those with normal color vision with that of observers with reduced ability to see colors enabled us to use the same stimulus material. The comparison is valid because (with the exception of people with monochromasy, who are totally color blind and have reduced visual acuity) those with the most common forms of abnormal color vision have normal vision in all other respects (Pokorny, Smith, Verriest, & Pinekers, 1979). The second reason is that a comparison of the performance of observers with normal color vision with that of observers with abnormal color visi on provides additional information about the practical consequences of abnormal color vision.
Defective color vision is common. It affects some 8% of males and 0.4% of females and is often a reason for excluding persons from occupations in which color recognition and discrimination are deemed important. Most research on the occupational handicap of abnormal color vision has been concerned with the problem of recognizing the colors of light signals (see Cole, 1992; Vingrys & Cole, 1988), although there have been some more broad-ranging questionnaire studies of the problems experienced by persons with abnormal color vision at work and in everyday life (Spalding, 1999; Steward & Cole, 1989).
Although color might be important in visual search, few studies have been conducted of the search performance of people with abnormal color vision in which color is one of the features of the target or in which color is used redundantly to provide visual organization of a complex display. Cole and Macdonald (1988) found that persons with abnormal color vision were slower and made more errors in acquiring information from a redundantly color-coded electronic flight navigation display. Scholz, Andresen, Hofmann, and Duncker (1995) found that nearly half of those with abnormal color vision could not find all targets in a color-coded maritime navigation display during a 640-s observation time, a task 98% of observers with normal color vision could do within the allotted time.
In addition to providing information about whether redundant color-coding contributes to conspicuity, we also set out in this investigation to establish whether persons with abnormal color vision are less likely to notice color-coded targets embedded in complex visual scenes than are observers with normal color vision. The outcome is not easily predicted. On one hand, it could be argued that persons with abnormal color vision -- or at least those with one of the more severe types of abnormal color vision -- will be denied some of the advantage of color as a target attribute facilitating search and will therefore be less likely to notice colorcoded targets in time-limited search. On the other hand, color may not be important to the search process, or those with abnormal color vision may make use of another attribute of the target in preattentive visual processing. It is even possible that the conspicuity of targets may be enhanced for people with abnormal color vision, in that their reduced ability to see col ors may simplify the background within which the target is embedded by reducing visual clutter caused by variations in color of the background.
The number of road traffic control devices in road scenes reported by a group of observers with the color vision defect of deuteranopia was compared with the number reported by observers with normal color vision. The experiment was conducted in the laboratory using projected colored slides of road scenes, each of which was shown for 300 ms.
The experiment was conducted twice with two different instructions given to the observers. On the first occasion the instruction was to report "whatever comes to your attention"; on the second occasion the instruction was to report "any road traffic signals or signs seen, including traffic lights." The first instruction provides a measure of what Cole and Hughes (1984) called attention conspicuity. The second instruction asked the observers to search for a particular class of targets and provides a measure of search conspicuity. Theeuwes (1991), who envisaged a similar distinction in the conspicuity of objects depending on the mental set of the observer, used the terms sensory and cognitive conspicuity.
In both instructions observers were required to identify any objects they noticed and to state the quadrant in which it was located. If they could not identify the exact nature of an object, they were asked to state whatever features of the object they had noticed, such as its color or shape. If the object reported was a traffic signal, the observers were asked to state its color if they could. This procedure is similar to that used by Cole and Jenkins (1982) in a study of conspicuity of road signs for observers with normal color vision.
The observers were aware that the experiment was concerned with how deficient color vision might affect the ability to notice things when driving. They were told this in the letter sent to them when they were recruited as observers and again when they attended an experimental session. They were not told that the specific purpose was to study the effect of redundant color coding of road traffic signs, but they would have deduced that there was a special interest in road traffic signs and signals when they were given the second set of instructions to report "any road traffic signals or signs seen, including traffic lights."
The instructions were read to the observers to minimize response variability caused by variations in the words used to explain the task. The observers were not forced to make a response for every slide, but at least one object was normally reported for each slide shown.
There were two groups of observers. One consisted of 20 men with normal color vision, and the other was 11 men with the color vision defect of deuteranopia. The number of observers provided sufficient statistical power to detect differences between the color-normal and color-deficient observers in the number of traffic signs reported, based on assumptions of hit rates and between-subjects variability reported by Hughes and Cole (1984). The greater number of color-normal observers, compared with color-deficient observers, was needed to achieve sufficient statistical power, bearing mind that it is difficult to recruit large numbers of persons with the color vision defect of deuteranopia.
Deuteranopia is one of nine classes of inherited defective color vision, and it affects 1% of men. It is a defect caused by the absence of the middle wavelength (green) retinal receptor, which causes a severe loss of ability to recognize colors (Pokorny et al., 1979). Deuteranopes are unable to distinguish red, yellow, and green when the brightness of all three colors is equal. Deuteranopes perceive all these colors as yellow. They also confuse green, white, and purple but are able to perceive blue and to differentiate it from other colors.
Only one class of defective color vision was studied because the inclusion of all four of the most common types of defective color vision -- protanopia, deuteranopia, protanomaly, and deuteranomaly -- would have increased the complexity of the experiment, especially considering that persons with the latter two vary quite a bit in the extent of their loss of color discrimination.
All observers held a current driver's license. Their ages ranged from 17 to 33 years. The color-vision-normal observers were selected so that two of them had the same age ([+ or -]12 months) as one of the color-vision-deficient observers.
All observers had visual acuity of at least 20/20 (with no more than two errors) binocularly with spectacle correction if needed. They also had normal peripheral vision as tested by a Medmont (Melbourne, Australia) static automated visual field analyzer. Color vision was tested using the Ishihara Test for Color Blindness, the Nagel anomaloscope (Schmidt and Haensch, Berlin, Germany), the Farnsworth Dl 5 test (Psychological Corp., New York), and the Medmont C 100. All color-vision-normal observers passed the Ishihara Test and showed no anomalous behavior on any of the other tests. The diagnosis of deuteranopia was established by failure on the Ishihara Test and by finding a full matching range at the Nagel anomaloscope without reduction of yellow brightness at the red end of the matching range. A full matching range means that the observers were able to match both a spectral red (671 nm) and a spectral green (546 nm) to a spectral yellow (589 nm); in other words, they were unable to distinguish spectral red, orange, yellow, and green. The diagnosis as deutan, rather than protan ("red blind"), was confirmed with the Farnsworth D15 test and the Medmont C100 test.
The slides were photographed in daylight, and each contained at least one road traffic control device located 50 m from the camera. These were the target signs and signals.
The camera was located at the average driver head height of 1200 mm in the left half of the road, or the leftmost lane, because traffic drives on the left in Australia. The camera lens axis was directed horizontally and parallel to the midline of the lane in order to simulate a driver's view of the road.
There were 98 target road traffic signs and signals in the 84 slides. They are grouped into nine classes: fluorescent red-orange warning signs (n = 9), yellow warning signs (n = 17), green direction signs (n = 8), blue parking signs (n = 8), give-way signs (comparable to yield signs in the United States; n = 15), stop signs (n = 8), speed restriction signs (n = 8), traffic lights with the red signal illuminated (n = 12), and traffic lights with the green signal illuminated (n = 13). In many slides there were other incidental road traffic control devices that were nearer or farther, as well as advertising signs, motor vehicles, and buildings, but the observers' responses to these incidental (nontarget) signs and signals were not recorded.
The fluorescent red-orange warning signs were diamond shaped with the words "Children crossing" or "Elderly persons crossing" in black lettering. The yellow warning signs were also diamond shaped with a black border and black graphics or lettering, except for one sign, which was a large rectangular road work sign. The green signs were large rectangular signs with a white border displaying directional information in white letters. Blue parking signs were rectangular with a white border and a white P on the blue background. One had an additional message in white letters. Give-way signs were white inverted triangles with a red border and black letters. Stop signs were red octagons with white letters and border. Speed limit signs were white rectangles displaying a red circle containing a black number on a white background.
All the signs were color coded. However, the blue color-coded parking signs served as a control to test for spurious group differences, because it is known that deuteranopic observers perceive blue normally (Kaiser & Boynton, 1996; Pokorny et al., 1979).
The slides were projected in a darkened room. Two projectors were used. One projected a uniformly white field with a fixation cross at its center, and the other projected the slides of the road scenes. The observers sat 3 m from the screen with their eyes level with the optical axis of the projectors. The main projector was fitted with an adapting lens so that the angular dimensions of the projected scene were identical to those of the real scene if the observer had been located at the same position as the camera on the road.
On the command "ready," observers were instructed to fixate on the cross on the white field. After a brief (~1 s), fixed interval, the slide projecting the road scene was shown in place of the fixation slide. The position of the fixation cross was always at the center of the road scene slide.
A 300-ms exposure time for the slides was chosen so that there could be no more than one exploratory eye movement. Hughes and Cole (1984) found that an average of 1.24 objects were reported using this experimental paradigm. In effect, the method measures what first attracts attention.
All observers were given practice with three slides to ensure that they understood the instructions and to familiarize them with the procedure.
The attention conspicuity trial of 84 slides was always given first; after a break of 10 min, this was followed by the search conspicuity trial, in which the same 84 slides were shown. The slides were shown once for each instruction.
The order of the slides was the same for all participants and both instruction sets. Order effects were not controlled because the sole purpose of the experiment was to test whether or not color-vision-deficient observers notice color-coded traffic signs and signals less often than do color-vision-normal observers. The observers knew that the same slides were to be shown in the second showing with the "report signs and signals" instruction, so there is a practice effect for the second instruction condition. However, this does not invalidate the comparison between the two groups of observers because the practice and order effects were the same for both. The opportunity for observers to learn to anticipate what might be in a slide was limited because there were 84 slides and each slide was shown only once for 300 ms for each instruction.
The number of reports of a target sign or signal made by the observers, expressed as a proportion of the maximum number of possible target reports for each class of road traffic control device, is called the hit rate and is taken as the measure of conspicuity of that class of sign or signal.
No record was made of reports of objects other than the target signs and signals because the purpose of the experiment was to test for differences between the two observer groups in their reporting of color-coded traffic signs and signals. More general studies of what is noticed in a road environment have been reported by Cole and Hughes (1984) and Hughes and Cole (1986a, 1986b).
Hit rates were calculated using two different criteria for a hit. For the first criterion a hit was recorded if the observer was able to state that a target sign or signal was seen and gave its location correctly, whether or not he was able to identify what kind of sign it was or any of its defining characteristics. This is called the detection criterion. For the second criterion a hit was recorded if the observer correctly detected and located the target sign/signal and was able to state what type of sign it was, or could state at least one correct feature of the target sign (its color, its shape, or both). Observers must have correctly identified whether a stop or go signal was displayed by traffic lights in order to gain a hit by this criterion. This is called the recognition criterion.
The data of two deuteranopic observers were excluded from the data presented here because the slides were inadvertently projected on an off-white screen, which slightly altered the color of slides. However, the results of these two observers were not statistically different from those reported for the other 9 deuteranopic observers.
Hit rates for each class of road traffic control device and each instruction were analyzed using nonparametric tests because normality of data was not consistent for all target sign/signal categories. The Mann-Whitney U test was applied to each observer's hit rate for each target sign category to compare conspicuity between the deuteranopic and normal color vision groups. The difference between search and attention conspicuity within groups was analyzed using Wilcoxon's signed ranks test.
Figure 1 shows the mean hit rates for each of the nine classes of road traffic control devices and the two classes of observer using the detection criterion for determining whether the target object was noticed. Figure 2 shows the mean hit rates using the recognition criterion.
Attention conspicuity was significantly less for deuteranopic observers than for color-normal observers, for both detection and recognition criteria, for all classes of road traffic control devices except the yellow warning signs and blue parking signs.
For both criteria for a hit, search conspicuity was also significantly less for the deuteranopic observers for traffic signal lights, stop signs, red-orange warning signs, and give-way signs but not for the yellow warning signs, green direction signs, blue parking signs, and speed restriction signs.
The differences are not trivial. Attention conspicuity for the deuteranopic observers is 10% to 30% of that for the color-normal observers for some classes of road sign.
The criterion for a hit did not affect the result. The correlations between the measures of conspicuity using the two criteria approached 1.00 for both instruction sets and both types of observer. This means that first detection of an object that attracted attention was closely associated with meaningful recognition. There is one exception: The conspicuity of traffic signals was lower for the recognition criterion than for the detection criterion for deuteranopes. This is because the recognition criterion required identification of whether a stop or go traffic light signal was displayed, and deuteranopes are known to have difficulty recognizing the colors of signal lights (Vingrys & Cole, 1988).
The differences in conspicuity for color-normal and deuteranopic observers were of a lesser magnitude for the search conspicuity condition than for the attention conspicuity condition. This is expected because the search instruction simplified the task, allowing the observers to focus their attention on only one class of objects in the visual scene. Search conspicuity for the deuteranopes is about 60% of that found for color-normal observers, except for the yellow warning signs, green direction signs, blue parking signs, and speed restriction signs, for which conspicuity did not differ between the two groups of observers.
Search conspicuity was consistently and significantly higher than attention conspicuity for both observer groups for all target categories, as would be expected and has been reported previously (e.g., Cole & Hughes, 1984; Hughes & Cole, 1986a, 1986b; Johansson & Backlund, 1970). Figure 3 shows the relationship between attention and search conspicuity for the nine classes of road traffic control devices. The search instruction has greater effect for those road traffic devices with low attention conspicuity than it does for those with high attention conspicuity. The deuteranopic observers generally had lower hit rates under the attention conspicuity instruction, so the search instruction afforded them greater potential for improvement. Figure 3 shows that the increase of conspicuity under the search instruction is generally greater for deuteranopic observers than it is for the observers with normal color vision.
The results show that deuteranopes are less likely to notice road traffic control devices than are observers with normal color vision, with the exception of yellow warning signs and blue parking signs, which have the same conspicuity for both observer groups. The explanation for these exceptions is that deuteranopes have a profound loss of ability to perceive color, but they are not color blind. They cannot perceive red and green, but they can perceive yellow and blue. This is because they lack green-absorbing cone receptors in their retinas. The absence of this cone disables their "red-green" neural pathway, which needs inputs from both the red and green absorbing cones to function. As a result, red, orange, yellow, and green colors all appear the same color. However, people with deuteranopia have normal red-absorbing and blue-absorbing cones, which means their blue-yellow neural pathway continues to operate with inputs from the blue- and red-absorbing cones. This means they retain the ability to perceive y ellow and blue (Kaiser & Boynton, 1996; Pokorny et al., 1979).
Longer-wavelength color stimuli (dominant wavelength > 540 nm) -- which arouse the perception of red, orange, yellow, and green colors for a person with normal color vision -- all appear yellow to a deuteranope, although there will be variations in be rightness; for example, reds will appear darker than yellow stimuli. Short-wavelength color stimuli are seen as blue by deuteranopes, as they are for persons with normal color vision. Deuteranopes therefore see the yellow of the yellow warning signs and the blue of the blue parking signs normally. As a consequence, these signs are expected to have the same conspicuity for both colorvision-normal and deuteranopic observers, as was found in this experiment.
However, deuteranopes perceive any red color in road signs as dark yellow. The dark yellow will be less conspicuous to them for several reasons. First, yellow is a less conspicuous color than red because it is less strongly colorful. That is, there is a greater chromatic separation between red and white than there is between yellow and white (Wyszecki & Stiles, 1967). The lower conspicuity of dark yellow, compared with a red, can also be appreciated intuitively. Dark yellow is yellow-brown in appearance, a color that is subjectively less eye catching than a vivid red color. Therefore, deuteranopes are expected to find signs that have red or red-orange color coding to be less conspicuous, as was found.
Green direction signs were found to have lower attention conspicuity to deuteranopic observers than to color-normal observers. This is also an explainable result: Deuteranopes see green as grey, or as a grey with a yellowish tinge. This is because stimuli with a dominant wavelength of about 590 nm (seen as green by those with normal color vision) provide an equal input to a deuteranope's blue-yellow neural system. This system then signals to the brain that the stimulus is neither blue or yellow, and deuteranopes lack a functioning red-green neural system to signal the greenness of the stimulus. A background that appears grey lacks the color contrast that would help make the signs stand out from their backgrounds. Lettering on green direction signs will also have less contrast for deuteranopes than for color-normal observers because the former see the white letters on a grey background. This will also reduce the conspicuity of the green signs, because the boldness of internal graphics on a sign is known to con tribute to conspicuity (Cole & Jenkins, 1982; Hughes & Cole, 1984).
The search instruction increased the hit rates for both color-vision-normal and deuteranopic observers because of a reduction in uncertainty brought about by limiting the range of target objects and by affecting a change of so-called mental set. The change in strategy induced by the search instruction enabled deuteranopes to improve their hit rate performance to a greater extent than did color-normal observers (Figure 3). Under the search instruction, deuteranopes were as successful as color-vision-normal observers in correctly reporting green direction signs and speed restriction signs. However, they continued to have significantly lower hit rates for red and green traffic signals, stop signs, give-way signs, and red-orange warning signs.
The experiment involved only one type of color vision deficiency, deuteranopia. The resuits can probably be safely generalized to the other types. Protanopes (who lack the long-wavelength-sensitive red cone) have a loss of color vision as profound as that of deuteranopes. The more severely affected deuteranomals and protanomals will most likely perform similarly to the deuteranopes in this study because the more severe forms of these defects approach the loss of color perception experienced by deuteranopes. Certainly, in the study by Cole and Macdonald (1988), all participants with any of the four types of color vision deficiency demonstrated significantly diminished performance in searching for information in an electronic flight navigation display that was redundantly color coded. Mildly affected deuteranomals and protanomals may not have a reduced ability to notice color-coded targets, but that would be a matter to be determined through further experimentation.
The reduced ability of deuteranopes (and probably those with other forms of defective color vision) to notice targets in time-limited visual search may have implications for a number of occupations that involve visual search, such as defense, police work, and search and rescue. The reduced ability of these observers to notice important elements in complex visual environments when color is one attribute contributing to the conspicuity of the target might also contribute to the risk of accident in transportation. System designers should be cautious in using color as the sole or principal means of attracting attention to information displays when there are color-deficient persons in the user group. They should ensure that other object attributes -- such as size, edge definition, brightness, and background simplicity -- provide adequate conspicuity. However, it can be noted that road signs using yellow and blue colors have the same conspicuity for deuteranopic observers as for color-normal observers.
The last observation to be made is that despite the equivocal evidence in the prior literature, color in road traffic control devices does contribute to their conspicuity. This is evident from the finding that the conspicuity of road traffic control devices is reduced for people with defective color vision and who are unable to perceive the full gamut of color.
[FIGURE 1 OMITTED]
[FIGURE 2 OMITTED]
[FIGURE 3 OMITTED]
David Goddard, Department of Epidemiology and Preventive Medicine, Monash University, provided helpful guidance to the first author. The technical help of Adam Robertson in the Department of Optometry and Vision Sciences, University of Melbourne, is much appreciated. We are also grateful to Steven Jenkins, who kindly reviewed the manuscript and made a number of important suggestions.
Date received: May 31, 2000
Date accepted: October 15, 2002
Christ, R. E. (1975). Review and analysis of color coding research for visual displays. Human Factors, 17, 542-570.
Cole, B. L. (1992). Does defective colour vision really matter? In B. Drum (Ed.), Colour vision deficiencies XI (pp. 67-86) Dordrecht, Netherlands: Kluwer.
Cole, B. L., & Hughes, P. K. (1984). A field trial of attention and search conspicuity. Human Factors, 26, 299-313.
Cole, B. L., & Jenkins, S. E. (1982). conspicuity of traffic control devices. Australian Road Research, 12, 223-238.
Cole, B. L., & Macdonald, W. A. (1988). Defective colour vision can impede information acquisition from redundantly colour coded video displays. Ophthalmic and Physiological Optics, 8, 198-210.
Commission Internationale de L'Eclairage. (2000), The conspicuity of traffic signs in complex backgrounds (Tech. Report TC 4-18, CIE Publication 137). Vienna: Author.
D'Zmura, M. (1991). Color in visual search. Vision Research, 31, 951-966.
Forbes, T. W., Pain, R. F., Joyce, R. P., & Fry, J. P. (1968). Color and brightness factors in simulated and full scale traffic sign visibility. Highway Research Record, 216, 55-65.
Hughes, P. K., & Cole, B. L. (1984). Search and attention conspicuity of road traffic control devices. Australian Road Research, 14, 1-9.
Hughes, P. K., & Cole, B. L. (1986a). Can the conspicuity of objects be predicted from laboratory experiments? Ergonomics, 29, 1097-1111.
Hughes, P. K., & Cole, B. L. (1986b). What attracts attention when driving? Ergonomics, 29, 377-391.
Jenkins, S. E., & Cole, B. L. (1979). A note on the role of color in conspicuity (ARRB Internal Report AIR 218-4). Melbourne, Victoria: Australian Road Research Board.
Jenkins, S. E., & Cole, B. L. (1986). Daytime conspicuity of road traffic control devices, Transport Research Record, 1093, 74-80.
Johansson, G., & Backlund, F. (1970). Drivers and road signs. Ergonomics, 13, 749-759.
Kaiser, P. K., & Boynton, R. M. (1996). Human color vision (2nd ed.). Washington, DC: Optical Society of America.
Luder, C. B., & Barber, P. J. (1984). Redundant color coding on airborne CRT displays. Human Factors, 26, 19-23.
Macdonald, W. A., & Cole, B. L. (1988). Evaluating the role of colour in a flight information cockpit display. Ergonomics, 31, 13-37.
Moses L. E., Shapiro, D., & Littenberg, B. (1993). Combining independent studies of a diagnostic test into a summary ROC curve: Data analytic approaches and some considerations. Statistics in Medicine, 12, 1293-1316.
Odeschalchi, P. (1960). Conspicuity of signs in rural surroundings. Traffic Engineering and Control, 2, 390-393.
Pokorny, J., Smith, V. C., Verricst, G., & Pinckers, A. J. (Eds.). (1979). Congenital and acquired color vision defects. New York: Grune and Stratton.
Scholz, R., Andresen, S., Hofmann, H., & Duncker, G. (1995). Recognition performance of observers with color-vision deficiencies on a polychromatic sonar screen for ship navigation. German Journal of Ophthalmology, 4, 103-106.
Spalding, A. (1999). Medical students and congenital colour vision deficiency: Unnoticed problems and thc case for screening. Occupational Medicine, 49, 247-252.
Steward, S. M., & Cole. B. L. (1989). What do colour vision defectives say about everyday tasks? Optometry and Vision Science, 66, 288-295.
Theeuwes, J. (1991). Visual selection: Exogeneous and endogeneous control. In A. G. Gale, I. D. Brown, C. Haslegrove, I. Moorhead, & S. P. Taylor (Eds.), Vision in vehicles III (pp. 53-61). Amsterdam: Elsevier Science.
Tullis, T. S. (1981). An evaluation of alphanumeric, graphic and color information displays. Human Factors, 23, 54 1-550.
Vingrys, A. J., & Cole, B. L. (1988). Are colour vision standards justified for the transport industry? Ophthalmic and Physiological Optics, 8, 257-274.
Wyszecki, G., & Stiles, W. S. (1967). Color science. New York: Wiley.
Kylie A. O'Brien is a project officer with the Victorian Department of Human Services in Melbourne. She obtained her M.P.H. in 1997 from Monash University.
Barry L. Cole obtained his Ph.D. in optometry from the University of Melbourne in 1971. He is a professorial fellow in the Department of Optometry and Vision Sciences at the University of Melbourne.
Jennifer D. Maddocks obtained her bachelor's degree in optometry in 1970 at the University of Melbourne. She runs her own optometry practice in Melbourne.
Andrew B. Forbes obtained his Ph.D. in statistics from Cornell University in 1990. He is head of the Biostatistics Unit in the Department of Epidemiology and Preventive Medicine at Monash University.
Address correspondence to Barry L. Cole, Department of Optometry and Vision Sciences, University of Melbourne, Victoria, Australia 3010; b.cole@optometry_unimelb.edu.au.
Copyright [c] 2002, Human Factors and Ergonomics Society. All rights reserved.
|Printer friendly Cite/link Email Feedback|
|Author:||O'Brien, Kylie A.; Cole, Barry L.; Maddocks, Jennifer D.; Forbes, Andrew B.|
|Date:||Dec 22, 2002|
|Previous Article:||Effects of a biocybernetic system on vigilance performance.|
|Next Article:||Returning human factors to an engineering discipline: expanding the science base through a new generation of quantitative methods -- preface to the...|