Katrina Deane

Glance time at billboards for skilled versus unskilled drivers ​Deane​ 1 Glance time at billboards for skilled versus unskilled drivers Katrina Deane 4/16/13 Glance time at billboards for skilled versus unskilled drivers ​Deane​ 2 Abstract Advertising on the side of the road is always in risk of being distracting to the point of dangerous. In this paper, I propose to research just how long skilled drivers spend looking at signs per glance and in total and how many glances they give each sign against unskilled drivers. I expect skilled drivers to spend less time looking at signs and to remain at a constant speed and lane keep variation while looking. The experiment will be set in a very controlled environment, something that has been lacking with previous research on the topic. When the research is conducted, a series of t-tests will be performed on the subsequent data and correlations will be drawn to hopefully draw conclusions. Improvements that can be made on the subject would be looking further into what types of ads are the most informative versus distracting. Glance time at billboards for skilled versus unskilled drivers ​Deane​ 3 Introduction When it comes to driving skill and safety, there is a fine line between distracting and entertaining. At the same time, marketers and companies are constantly seeking to advertise with the use of billboards on the side of the road. As with all marketing strategies, the point is to communicate with the public about the service or product that they offer. The question we wish to eventually answer through research is how to optimize the ability to communicate with drivers without becoming dangerously distracting. There is very little research in this area pertaining to normal drivers, as opposed to brain injury patients. Beyond that, there is even less research having to do with the skill of the driver in correlation with time spent observing billboards. This study proposes to rectify that situation. It is the intention of this study to specifically observe time spent per glance at a single sign, amount of glances at that sign, and skill of driver looking at the sign. Beijer, Smiley, and Eizenman (2007) conducted a study on glance behavior at billboards. They tested 25 participants, observing how long and often they glanced at various signs along the Expressway in Canada. They found that the average duration of glances was 0.57 seconds (standard deviation (SD) = 0.41). Altogether there was a total of 35.6 glances per subject (SD = 26.4). In general, active signs (signs with movement) attracted more attention than passive signs, drawing more frequent and longer glances. This study could be improved with a greater number of participants and a higher amount of control. The small group offered an average of glances with a standard deviation almost as high, leading to the conclusion that the high variability of participant behavior swayed the results. Although this was a skilled starting point for research in Glance time at billboards for skilled versus unskilled drivers ​Deane​ 4 this topic, the test was performed under varying road conditions and with varying signs. Different signs draw different amounts of attention by their nature of being more or less cluttered, colorful, and/or wordy. Therefore these results must be retested in a more controlled way and with more participants to be reliable. Another study having to do with distraction with advertising was performed by Crundall, Van Loon, and Underwood (2006). They focused more on the position of signs, whether street-level, or raised-level signs are more effective. They had two groups of participants watch videos of driving, and then asked them to rate the clips on hazardousness. Their eye movements recording. One group was primed to attend advertisements. This group fixated mostly on raised-level ads and had better memory recall later of these advertisements. The other group was told simple to watch for hazardousness. They focused most attention on street-level ads but had poor recognition of these ads later. It was concluded that street-level ads draw attention at more inappropriate times as opposed to raised-level ads. This was an important study on assessing safe placement of ads. Based on this study, the research proposed in this paper will test participants on raised-level ads for marketing purposes. This is because these were the ads with better recall later. Noy et al. (2004) studied interruptability as a measure of visual distraction. The basic premise was that if a task was easily interrupted, it was less dangerously distracting to the driver in that the cognitive load could be shared. Twenty-four participants, between the ages of 21 and 34, completed two separate experimental sessions. In the first session they performed two simulated visual search tasks in the vehicle, one under occlusions, one not. In the second session, these participants were asked to do the same tasks in a driving simulator. Although tasks time Glance time at billboards for skilled versus unskilled drivers ​Deane​ 5 remained roughly the same, differences were found using occlusion paradigm and subjective workload ratings. This is an important thing to note; the interruptability of a task is positively correlated with shared attention to driving, leading to safer driving. For this purpose, it can be inferred that signs that have smaller clusters of information that are spread out are safer forms of advertisement. The separation of information allows easy interruption in reading the sign and the smaller amounts of information given in each cluster reduces time needed to read that section before moving on. Olsson and Burns (n.d.) performed an experiment in Sweden assessing how much peripheral information people could work with in order to create safer cars. Thirteen participants drove on a motorway and country road, performing different tasks in the vehicle. Such tasks included changing a CD, tuning the radio and backward counting. Reaction time, hit rate, subjective mental workload, and heart rate variability were measured. Although the purpose of the experiment was mainly to test whether the tool they were using, Peripheral Detection Task (PDT), was accurate, it was noted that mean reaction times were slowest for the backward counting task on the country road and hit rates were best for driving on both roads for the CD changing task. Also, there was no significant difference found in task performance on the country road versus the motorway. This is substantial to our research because advertisement is peripheral information when we are driving until we focus directly on it. There is no significant difference on a lighter traffic road (country road) versus a heavier traffic road (motorway) so there will be no specific measurement for this in the experiment, although it will be controlled for. Glance time at billboards for skilled versus unskilled drivers ​Deane​ 6 Another experiment with a similar field of research was conducted by Engstroem, Johansson, Oestlund (2005). The focus of this study was the affects of visual and cognitive demand on driving performance. Both simulated and real motorway driving were used, collecting data on speed, lane keeping performance, steering wheel movements, eye movements, physiological signals, and self-reported driving performance. Higher cognitive load increased the tendency to look only in the center of the lane and ignore peripheral vision. Higher visual distraction led to decrease in speed and increase in lane keeping variation. The purpose of this study was for the development of a generic safety test regime for In-vehicle Information Systems. This experiment correlated visual distraction to driving performance in two segments: speed and lane keeping variation. These are the two most important to look at when testing for glance time at advertisements because they are the most likely to show variations in different driving skill levels. Based on the experiments conducted above, there is a lot of room for further research on this topic. As mentioned before, there are not reliable studies that focus on advertising based on driver skill level. The reason it is important to take skill level into account for advertisement purposes is so that marketers can make advertisements that can be appreciated at a deeper level by the drivers that are able to look at them more closely while being safe, without overly distracting drivers that are unsafe. This can lead to answers for questions like how much art should be used, with what kinds of color schemes, how much information can be put up in what format, etc. For instance, all the information should be able to be read in the average time an unskilled driver looks at the sign, without demanding more time from that driver. However, there should be more context at a second glance by the driver that can afford it, making a more Glance time at billboards for skilled versus unskilled drivers ​Deane​ 7 meaningful communicative piece. All of this can be taken into consideration when we know how long and how many times drivers look at billboards based on their skill level. In the research proposed here, the independent variables will be the driving assessment test and the simulated environment, including 3 billboards as visual distractors with no opposing traffic. The dependant variables will be driving skill level, time spent looking at each sign per glance, time spent looking at each sign in total, and times glancing at each sign, as well as speed fluctuation and lane keeping variation in the simulated environment. Ideally there will be roughly 100 participants. My hypothesis is that skilled drivers will glance at signs an average of less than 1 second, a total of 3 times (making a total of less than 3 seconds spent on the sign) with very little speed fluctuation or lane keep variation. Bad drivers will glance at signs an average of between 1 and 2 seconds, a total of 5 times (making a total of between 5 and 10 seconds spent on the sign) with significantly greater speed fluctuation and lane keep variation. Another possible hypothesis is that skilled drivers will spend less time per glance at the sign but glance at it more times than unskilled drivers, creating an equal total of time spent looking at the sign. For this experiment, participants will take part in two sessions. The first session will be a driving skill assessment test on the road; the Advanced Driver Improvement assessment tool (described below) will be used. From there, drivers will either be placed in one of two categories: Skilled Drivers or Unskilled Drivers. Despite category differences, each participant will then use a driving simulator for the next session. The environment will be controlled to be the same for each subject and there will be 3 billboards per experiment. The signs will remain consistent across experiments but will contain different messages within. The participant’s eye movements will be measured by an eye tracker to assess glance time and frequency. The driving simulator Glance time at billboards for skilled versus unskilled drivers ​Deane​ 8 will keep a running measurement of speed and lane keep variation throughout. It will be assumed that more skilled drivers will perform better on speed and lane keep variation and will spend less time per glance looking at the signs. However, biases will not be taken into account do to precision measuring instruments and a third party driving instructor to perform the driving skill assessment portion of the experiment. This will be a very controlled experiment across participants. Method This experiment will be using top of the line, precision equipment. The driving assessment test below is used to assess driving instructors at The Big Red L Company. It will be slightly modified wording to fit the purposes of this test. Participants will be assigned an ID number to protect identity and encourage truthful results of driving skill level and time spend on advertisements (it is thought that since the participants know what exactly is being tested they might try to alter the results to make themselves look better if their actual names are being used). All participants will sign a consent form before taking part in any part of the experiment and all experimenters and driving instructors will be signing a code of conduct form. Participants For adequate reliability, there will be ideally at least 100 participants, averaging out to at least 50 skilled versus 50 unskilled drivers. The more participants that can be gathered, the better. Participants will be incentivized with cash and the experiment will be advertized. Neither gender nor age will not be taken under consideration for choosing participants. Gender makes no Glance time at billboards for skilled versus unskilled drivers ​Deane​ 9 difference for the purposes of the test and age will not be controlled for because it is likely that a lot of the unskilled drivers will come from very young or older drivers for various reasons. This test will be taking place in northern Michigan, USA so the population will be drawn from there and will have cultural biases associated with that. Therefore the ethnicity will be mostly Caucasian because that is most common. The only thing we will assess about any participant we receive is their driving ability. People without any ability to drive, due to physical or mental health problems will not be able to participate. Design, Apparatus, and Procedure In the first session, a modified version of the ADI will be used. The original version had 6 categories, ranking 1 the lowest and 6 the highest skill level (Simcock, 2010). For the purposes of this test, skill levels 1 through 3 will be considered unskilled and levels 4 through 6 will be considered skilled. In merging the categories it will look like Image 1 in the Appendix. Participants will drive with a real driving instructor for 20 minutes, following the route that the instructor prescribes them. The route that they take will be chosen by the instructor in order to reduce bias on the part of the experimenter and to defer to the greater knowledge in this area to the instructor. However, the instructor will be asked to use the same route with the same traffic conditions (as much as possible) with each participant. At the end of the test, the instructor will send the participant on his or her way, giving the grade that they assigned to the experimenter in conjunction with the ID number of the participant tested. In the following week, whenever the participant is available after taking the adapted ADI, they will come into a laboratory setting with a driving simulator hooked up to an eye tracker. The Glance time at billboards for skilled versus unskilled drivers ​Deane​ 10 setup should look something like Image 2. The environment that the simulator produces will be the same for every experiment. It will consist of a 25 mile stretch with a speed limit of 75 miles per hour, based on Michigan highways. The simulator will measure speed and placement in lane throughout the entire experiment. The road will have no opposing or coinciding traffic. Both sides will have forest landscape observably distant from the road so as not to cause extra distraction. There will be three billboards place on the road at the standard size of 14’X48’ (Highway Billboards, 2013). The first will be at the 10 mile mark, the second at the 15 mile mark and the third at the 20 mile mark. Each billboard will have a different kind of advertisement on it, none will be actual ads or actual companies. The first billboard will have a very simply worded message with an uncomplicated image (see Image 3 for example). The second will have a very simply worded message with a complicated image (see Image 4 for example). The last one will have a lengthier worded message with an uncomplicated image (see Image 5 for example). All three signs will be placed on the left side of the road, as a raised-level advertisement. Eye trackers will be measuring eye movement and fixation time on signs during their presentation on screen and the road. The time spent per glance at all of the signs together will be averaged. There will also be an average accounting for the total time a sign was fixated upon. Finally, the amount of times each sign is looked at will be averaged. Analysis When the experiment has been completed, all of the scores within the Skilled and Glance time at billboards for skilled versus unskilled drivers ​Deane​ 11 Unskilled Drivers’ categories will be averaged separately from each other, creating two sets of data. Scores will be denoted as follows: skilled drivers average glance time = SAGT; skilled drivers total average glance time = STGT; skilled drivers average number of glances = SANG. Unskilled drivers’ categories will simply replace the S with a U respectively: UAGT, UTGT, and UANG. Each category will be T-tested to find standard deviation. A correlation chart will be grafted for all scores. STGT and UTGT will also be correlated to speed variation and lane keep variation, creating two more graphs. Expected Results If the hypothesis is correct SAGT will be less than 1s, STGT will be less than 3s, and SANG will be around 3. UAGT will be around 1.5s, UTGT will be around 8.5s, and UANG will be around 5. UTGT will have a higher correlation with speed variation and lane keep variation. An example of how the graphs are expected to look are as follows: Glance time at billboards for skilled versus unskilled drivers ​Deane​ 12 Figure 1 shows the difference between the average glance times between unskilled and skilled drivers. Figure 2 shows the difference between total average glance times of skilled and unskilled Glance time at billboards for skilled versus unskilled drivers ​Deane​ 13 drivers. Figure 3 shows the difference between the average number of glances given per billboard by skilled and unskilled drivers. Glance time at billboards for skilled versus unskilled drivers ​Deane​ 14 Figure 4 shows how skilled and unskilled drivers compare with speed variation during observation of billboards. Figure 5 shows how skilled and unskilled drivers compare with lane keep variation during observation of billboards. Discussion What we hope to see come from this study is a better understanding of how long safer drivers view billboards as opposed to less safe drivers. The idea is to be able to cater to drivers who are not as skilled so that they do not have to spend any extra effort being distracted from the road. However, we want something with higher context for safe drivers who are being responsible to be able to get more out of the advertising experience. In Figure 1 through 3 (above) it is demonstrated, albeit hypothetically, that unskilled drivers spend more time looking at billboards than skilled drivers. From this we might deduce Glance time at billboards for skilled versus unskilled drivers ​Deane​ 15 that higher skill at driving is positively correlated with better media comprehension skills, thereby allowing the skilled drivers to understand the ad just as well as the unskilled driver that is spending more time looking at it. It should be noted that although it is expected that the unskilled drivers look longer than the skilled driver, they also look a greater number of times. This might be because the unskilled driver is not as well able to be interrupted from the distraction task, which is slightly more dangerous. However, this is highly speculative and should be researched further as well as reformulated after the actual scores are in. Figure 4 and 5 are another test of driving skill with distraction to reinforce the placement that the participants scored in their respective categories. If all goes as it should, speed and lane keep variation should remain largely constant for skilled drivers and should variate more for unskilled drivers. If this measure is true it will make the test more reliable over time and across different demographics. This research will surely raise more questions, as all research does. Possible directions to take from here involve finding out how much information skilled versus unskilled drivers are actually ascertaining from billboards and how much of that information they retain afterwards. Other possible experiments include testing what color/color schemes are optimal for driver attention without being distracting, how far away from the roads signs should be for best information gathering, or how interruptible is the task of looking at billboards while driving. Conclusion This experiment won’t be enough to go on by itself, but it will pave the way for further research on this topic. Completing this data set will give companies and marketers new Glance time at billboards for skilled versus unskilled drivers ​Deane​ 16 information to draw assumptions from and better place to continue building from. This will allow them to be safe about printed media on roads, while at the same time creating a better advertising experience for drivers who don’t need the precaution. In the end, they should be able to better optimize their marketing strategies. References Beijer, D., Smiley, A., & Eizenman, M. (2007, January 29). Observed driver glance behavior at roadside advertising signs [Electronic version]. ​Transportation Research Record: Journal of the Transportation Research Board, ​1899(2004), 96-103. doi:10.3141/1899-13 Keywords: N/A Summary: A study on glance time at signs on the side of the road in Canada. Crundall, D., Van Loon, E., & Underwood, G. (2006, July). Attraction and distraction of Glance time at billboards for skilled versus unskilled drivers ​Deane​ 17 attention with roadside advertisements. ​Accident Analysis and Prevention, 38(4), 671-677. Retrieved April 13, 2013 Keywords: Advertisements; Attention; Distraction; Driving; Eye movements Summary: A study on optimum positioning of roadside advertisements. Engtroem, J., Johansson, E., & Oestlund, J. (2005, March). Effects of visual and cognitive load in real and simulated motorway driving [Electronic version]. Psychology and Behavior, 8(2), 97-120. Keywords: Driver distraction; Driver workload; Road safety; Evaluation methodology; IVIS Summary: A study to test affects of cognitive load and visual distraction of driving performance. Highway Billboards. Advertisement. Billboards. CBS Outdoor, 2013. Web. 12 Apr. 2013. . Keywords: N/A Summary: An advertisement selling billboard space. Glance time at billboards for skilled versus unskilled drivers ​Deane​ 18 Noy, Y. I., Lemoine, T. L., Klachan, C., & Burns, P. C. (2004, May). Task interruptability and duration as measures of visual distraction [Electronic version]. ​Applied Ergonomics, 35(3), 207-213. Keywords: Driver distraction; Occlusion; Road safety Summary: A study on interruptability of advertisement and its affect on driving. Olsson, S., & Burns, P. C. (n.d.). Measuring visual distraction with a peripheral detection task (Master's thesis). Retrieved April 12, 2013 Keywords: Cognitive Tunneling, Driver Workload, Secondary Task, and Visual Distraction Summary: A study on peripheral distraction and driving to improve safety of cars. Simcock, T. (2010). Instructor Grades. In High Grade Driving Instructors . Retrieved April 13, 2013 Keywords: N/A Summary: A breakdown of how driving instructors are graded on performance. Glance time at billboards for skilled versus unskilled drivers ​Deane​ 19 Appendix Grade Number of errors Description of Driving Unskilled > 10 Extremely poor to inadequate performance. Dangerous to careless mistakes. Lack of to gaps in knowledge of road rules. Skilled