This English laboratory study addressed a type of outdoor advertising that is not directly related to the DBBs that are the subject of the present study. Specifically,
Crundall and his colleagues looked at fixed posters mounted either at street level (“street- level advertisements,” or SLAs) such as those on bus shelters, newsstands, or telephone kiosks, and posters located above ground on poles or streetlights (“raised level
advertisements,” or RLAs). The size of the advertising posters studied was 1.8m x 1.2m (approximately 6ft. x 4 ft.) in a vertical format. As such, these advertising signs were more representative of signs that might be seen in urban environments in the U.S., rather than the typical 14ft. x 48ft. size digital billboards that are the subject of the present study. Nonetheless, the hypotheses made by these authors offer a different perspective than those that have generally been adopted by researchers in this field, and their conclusions shed new light on the issue of roadside advertising and driver distraction. The authors discuss the potentially detrimental effects of roadside advertising in a manner similar to other researchers. As they describe it, in undemanding situations drivers have “spare attentional capacity” that they can use to permit their eyes to wander to objects in the visual field, including those, such as advertisements, that are irrelevant to their driving task; however when the cognitive demands imposed on the driver (such as from traffic, weather, roadway geometry, vehicle performance or personal factors such as fatigue) become greater, this spare capacity diminishes, and eye movements must focus on the task at hand. If an advertisement within the driver’s visual field attracts visual fixations under these conditions, sufficient spare capacity may not be available to attend to it, and thus the advertisement draws from the limited attentional capacity that is needed to safely perform the task. Thus, although the authors initially suggest that roadside advertisements are intended to attract a driver’s spare capacity, they go on to describe the interest that advertisers have in placing their signs in locations where the driving task demands may be high. They cite (as have others) the 1967 before-and-after study by Ady, who found that an “eye-catching” billboard at the apex of a curve led to more accidents than similar signs in control locations.
The authors suggest that, because it is possible to identify fixed roadside “hazards” (such as dangerous curves, complex interchanges, etc.), it is therefore possible to ensure that roadside advertisements are not located in such areas. Their greater concern, however, is
with what they call transient hazards, such as changes in traffic density, path intrusion from another vehicle, or a pedestrian crossing the driver’s path from between parked cars. Transient hazards cannot be predicted in time or location. Because such unforeseen events can directly influence a driver’s probability of an accident, “if attention is
distracted by an advertisement during the onset of a sudden (transient) hazard, the chance of an accident occurring will increase” (p.672). Knowing that roadside advertisements do attract driver’s attention (as per Hughes and Cole, 1986, and others) and that drivers’ glances at such advertisements may be made under unsafe conditions such as short headways (as per Smiley at al., 2004), the authors set out to determine whether SLAs or RLAs tend to attract more attention when drivers are looking for hazards.
The most relevant environmental (including traffic and roadway) information important to hazard detection is distributed primarily along a horizontal plane, with the straight- ahead view (the focus of expansion) at the center of this distribution. As a result, as the authors have demonstrated in prior research (Chapman and Underwood, 1998), the majority of visual fixations will fall within a horizontal window when the driver is looking for driving-relevant information, including potential hazards.
These earlier findings lead to their belief that, if an advertisement is located within this “horizontal window of inspection” it will receive more fixations than will other
advertisements. Although such fixations on the advertisement may be immaterial to safety when the driver has spare attentional capacity, those fixations that occur during a visual search for hazards and other driving-relevant information are likely to be
unintentional and may distract the driver and serve to interrupt this critical visual search activity.
The principal research hypotheses tested, therefore, were that, during high demand conditions, when drivers were primed to look for hazards, SLAs would receive the most attention, whereas during periods of reduced demands, when spare capacity was greater, the attention given to RLAs would increase.
The study was conducted in a laboratory, where participants viewed video clips that had been previously recorded from the dashboard of a moving car. Of 34 clips created, half included SLAs and half depicted RLAs. All were essentially equal in size (1.8m x 1.2m), and all were filmed during daylight. The clips ranged from 42 to 61 seconds in duration, and the time when an advertisement first appeared within each clip was randomized. The clips were projected onto a screen 2m in front of the participant and subtended a visual angle of 33º x 27º horizontal. Participants’ eye movements were recorded and
superimposed on the video for analysis. Two different test conditions were established via the instructions given to the participants. In the “hazard group” the participants were instructed to concentrate on the hazardous nature of each video clip. In the
“advertisement group” participants had less emphasis placed on the hazard perception task and, in addition, were told to watch out for advertisements that they might pass. The intent of the instructional set was to create differences in the task demands during visual search – high demand when scanning for hazards; lower demand when still looking for hazards but also attending to irrelevant stimuli.
Results showed significant differences between the two groups in several areas. SLAs were fixated earlier, received more fixations, and received a greater total gaze duration compared to RLAs. In addition, the mean length of advertisement fixations was greater than the mean length of fixations for the entire clip, with one exception. Fixations on the RLAs were lower than the clip averages for the hazard group, suggesting that, as had been found previously, the scanning for hazards takes place essentially within the horizontal plane in front of the driver. A post-drive hazard rating showed that clips with SLAs were judged more hazardous than clips with RLAs.
Our review raised a number of questions about the methods and protocols used in this study, and about their possible effects on the findings. For example, the authors do not provide the text of the actual instructions given to the participants; as a result it is unclear just what the task was for those in the “advertisement” group. There is no description of any of the visual information (except the advertisements) within any of the clips shown, and thus one does not know the implications of the finding that the SLAs were fixated to a greater degree than the clip average, a potentially important observation. Further, with clip durations of one minute or less, the presence of advertisements within the scene may have become expected during the course of the trials, despite their randomized placement within each clip. Finally, as discussed elsewhere in the present report, it might have been useful to have comparisons between values in the tails of the distribution (e.g. the longest glances) in addition to the means.
Despite our uncertainty about some of the details of this study, one relevant finding in particular is a cause for concern regarding the potential effect of roadside advertising on traffic safety. The authors describe, based on their prior research (Chapman and
Underwood, 1998, Crundall et al, 1999) hazard perception searches in visually cluttered environments as displaying higher sampling rates and shorter fixation durations than in less complex environments, until a hazard is detected, at which point the fixation durations of the hazard itself increase. The findings of this study suggest that the SLAs showed “similar effects on fixation durations as an actual hazard, stopping search for other hazards, and potentially reducing peripheral attention, as increased resources are devoted to the fixated stimulus” (p.675). In other words, when scanning the environment for hazards, drivers in this study unintentionally attended to a roadside advertisement that was within their scanning window, and then increased the duration of their glances at the advertisement to the same extent that they would have done to an actual hazard, and at the expense of their continued scanning for hazards, even when they were instructed to search for the hazards. This finding is quite similar to that expressed by Beijer (2002), who reported that, although higher levels of task demand were associated with a
reduction in the number of glances made to the signs, the average and maximum duration of these glances was not reduced as task demands increased. As Beijer states: “This would seem to indicate that drivers are comfortable turning their attention away from the road for a set period of time, regardless of the demands of the driving task” (p. 76). Another finding from Crundall, et al. also raises concern. The authors cite a study by Boersma (1989) that suggests that visual clutter in the observed environment tends to
increase the visual search time for a target of interest, and studies by Eriksen and Eriksen (1974) and Logan (1996) that demonstrated that the proximity of distracters to a target increases the amount of time required to respond to the target. Crundall, et al. conclude that the embedded nature of SLAs within a complex scene may produce the same result, i.e. increasing the time required for a driver engaged in proper scanning behavior to locate and respond to a real hazard that may be present.
If the two findings of this study can be replicated in other research more germane to the U.S. roadway network and to the type, size, and location of typical DBBs, then the implication is that such signs can attract and hold drivers’ attention, even unintentionally, at the expense of their need to scan the environment for immediately relevant hazards, and that the mere presence of a DBB in the visual environment can increase the time required to identify and respond to a present hazard.