Reaction to Owens, McLaughlin, & Sudweeks (2011) by Richard Thripp
EXP 6506 Section 0002: Fall 2015 – UCF, Dr. Joseph Schmidt
September 22, 2015 [Week 5]
Owens, McLaughlin, and Sudweeks (2011) conducted what was, to their knowledge, the first controlled, real-world study regarding text messages and driving (p. 940). They used a closed, 1.4-mile two-lane road, and the actual trials were conducted on straight uphill and downhill sections (I was surprised that no information was given regarding the steepness of these grades). Participants (n = 20) sent and received text messages using their mobile phones (typed) on some trials, and using the in-vehicle Ford SYNC system and selecting from a pre-programmed list of 15 possible text messages on other trials (p. 940). Overall, sending text messages was more dangerous than receiving them, mobile phone use was more dangerous than the in-vehicle system, and texting in general appeared more dangerous for older participants.
I question the generalizability of this experimental study. Why not do a naturalistic study, where drivers agree to have their cars equipped with audiovisual and kinematic sensors that monitor their texting habits in real-world situations? An example of such a study, funded by the same agency (the National Surface Transportation Safety Center for Excellence), is Distraction in commercial trucks and buses: Assessing prevalence and risk in conjunction with crashes and near-crashes (2010). Instead, we get an experimental study that is simplistic in implementation and hampered by safety concerns. Owens, McLaughlin, and Sudweeks (2011) conducted their trials with no other vehicles on the roadway, at a maximum speed of 35 miles per hour, on straightaways! This is not like actual texting while driving, which may involve congestion, traffic signals, curves, higher speeds, pedestrians, and more. They proceed to make inferences that texting by hand results in greatly degraded control of the vehicle, based on steering velocities (p. 945); however, the possibility that participants might text more cautiously (with more frequent steering corrections) on an actual roadway with other drivers is not explored. We are expected to believe that the conditions are valid because participants did not know if a single confederate vehicle might enter the roadway again, after passing them in the opposite direction on the first practice lap (p. 942)—quite a stretch, to say the least.
While mental demand, glances, and steering was measured, there was no consideration of velocity, following distance, weather conditions, or a host of other factors. To be fair, the authors did conduct a naturalistic study for about an hour with each participant, immediately prior to the 40-minute study in question, the results for which were released in 2010 (p. 942). However, both studies are of limited depth and were conducted with an “in-vehicle experimenter” present (p. 942), possibly influencing behavior. Considering that the experimenters had a control tower and many cameras and sensors (pp. 941–942), they could have eliminated the in-vehicle experimenter if they wanted. As a further point to limit generalizability, the system used does not even exist in the real world—the actual Ford SYNC system had to be modified by the manufacturer to allow texting while driving, since it typically disables texting at speeds over 3 miles per hour (pp. 940, 945).
This study was conducted in Virginia, where texting while driving is illegal—therefore, the researchers did not screen participants on their texting while driving habits (p. 940). Had the researchers conducted the research in a state where texting while driving was “legal,” such as Florida, they could have asked these questions and perhaps gained further insights.
The researchers relied on post-hoc tests to investigate interactions (p. 943), including measuring the baseline duration post hoc (p. 942). Post-hoc analysis should be used with caution and may reveal statistically significant patterns that are of no practical significance. They also assumed normality and homogeneity even though there were deviations in the ANOVA residual plots, and did not show us the plots (p. 943).
Importantly, driving while texting was not measured with respect to where the mobile phone was located—it could be better if the phone was in a cradle on the dashboard or mounted to the windshield, since participants would not have to look down (away from the roadway) to use their phones. I did not see any mention of this possibility, though interior glances were timed and counted. Further, the information in this article is already somewhat dated: only 6 of 20 participants had touch-screen phones (p. 940), while in 2015, this proportion would be much higher. 10 of 20 participants had archaic numeric keypads that require much more typing than a full QWERTY keypad (whether it is on a touch-screen or with physical keys). Many phones have fairly reliable voice recognition systems now, which may be less distracting than typing. It is possible that text messaging could be safer for some drivers than inferred from this study: for example, drivers who use a dashboard cradle and primarily text at red lights. Beneficial factors may even exist, such as reduced speed and increased following distance while texting.
Distraction in commercial trucks and buses: Assessing prevalence and risk in conjunction with crashes and near-crashes. (2010). Washington, DC: U.S. Dept. of Transportation, Federal Motor Carrier Safety Administration, Office of Analysis, Research and Technology, . Retrieved from http://ntl.bts.gov/lib/51000/51200/51287/Distraction-in-Commercial-Trucks-and-Buses-report.pdf
Owens, J. M., McLaughlin, S. B., & Sudweeks, J. (2011). Driver performance while text messaging using handheld and in-vehicle systems. Accident Analysis and Prevention, 43, 939–947. doi:10.1016/j.aap.2010.11.019
Note: Per Florida Statute 316.305, texting while driving became illegal on 10/01/2013, but was “legal” at the time of this study (most states already have laws against driving while encumbered, reckless driving, etc., but they typically go unenforced with respect to texting, necessitating the creation of new laws against texting on a state-by-state basis).