Texting Bans, a Possibly Low-Cost and Effective Means to Help Improve Motor Vehicle Safety

In this issue of AJPH, an article by Ferdinand et al. (p. 748) looks at the question of whether statewide texting-while-driving bans are effective at reducing crash-related emergency department (ED) visits. Their results suggest that the answer is a cautious yes. Given the ever-growing body of evidence that texting while driving increases crash risk,¹ it is important to assess whether interventions such as texting bans are having the effect we hope they are having. So, should states all rush to pass texting-while-driving bans? The answer to this is less clear.

TEXTING AND DRIVING

A meta-analysis by Caird et al.¹ included 28 studies of texting and driving, all of which were experimental, conducted either in a simulator or on a test track. All estimated mean effects were positive, showing worse performance while texting than at baseline, and all but two of these effects were significant. However, these effects may or may not translate to increases in crashes in the field. Driving is undertaken in a complex environment, and drivers may alter their risk by engaging in secondary tasks in particular situations (e.g., low speeds, lighter traffic) or by adjusting their speeds and following distances.²

One reason that so few studies have been done with field data is that texting while driving has been relatively rare until recent years. Pickrell and Li³ reported results of annual national roadside surveys that showed that texting while driving (defined as visibly manipulating an electronic device) was done by just 0.4% of drivers in 2006 but rose steadily to 2.1% in 2016. Moreover, data from police-reported crashes have not included reliable information on whether the driver was texting before the crash, making texting-while-driving risk difficult to evaluate with crash data.

From 2012 to 2015, the Transportation Research Board–sponsored second Strategic Highway Research Program (SHRP2) Naturalistic Driving Study collected data from more than 3000 drivers’ own vehicles while they were driving. The resulting data set includes more than 35 million miles of driving and 1000 crashes. In-vehicle cameras recorded drivers’ actions including texting, and a random sample of this video has been coded for a variety of secondary tasks.⁴

Using the SHRP2 data set, Dingus et al.⁵ reported that texting occurred in 1.91% of randomly sampled 6-second driving epochs, and the estimated crash incidence rate ratio (IRR) for texting was 6.1 (95% confidence interval = 4.5, 8.2). Although the Dingus et al.⁵ study provided strong evidence for the high risk associated with texting, more recent work has suggested that their approach may overestimate the potential risk reduction that would come from stopping drivers from texting. First, the SHRP2 sample drastically overrepresents younger and, to a lesser degree, older drivers.⁶ Moreover, Guo et al.⁷ showed in another analysis of SHRP2 data that age is an effect modifier for cell phone–related distraction, such that both younger and older drivers have higher estimated IRRs for crashing when involved in cell phone–related tasks including texting, compared with middle-age drivers. These results, combined with young drivers’ high rate of texting compared with other age groups,³ suggest that the Dingus et al.⁵ IRR estimate is probably an overestimate of the driver-population average IRR.

Finally, Dingus et al.⁵ and Guo et al.⁷ both compared the crash risk of texting while driving to a baseline of idealized driving in which the driver is engaging in no secondary tasks and is not impaired, fatigued, or visibly emotionally upset. Flannagan et al.⁶ reanalyzed the SHRP2 data using propensity scoring to develop a baseline comparison of non–cell phone use that better represents what drivers are likely to replace cell phone use with if a ban were effective. Although that study did not directly report an IRR for texting, the estimated IRR for all forms of cell phone use was 3.56 with the Dingus et al.⁵ method, but dropped to 1.98 (still significantly greater than 1) when compared with activities (including undistracted driving) that would be likely to replace cell phone use if a ban were effective.

These points are not meant to suggest that texting while driving is not risky. The evidence on that is clear. However, the magnitude of the effect and, most importantly, the likely benefit of reducing or eliminating texting and other cell phone use while driving may be less than the large IRRs might suggest.

DO BANS WORK?

As a starting point, research cited in the introduction to Ferdinand et al. generally indicates that handheld phone bans reduce the amount of phone use while driving. Although reducing texting while driving is clearly a prerequisite for bans to be effective, the real measure is whether it reduces crashes and injuries. The Ferdinand et al. analysis suggests yes.

The analysis done by Ferdinand et al. was one of the most rigorous and well designed in the literature. Thus, we can believe that their results show reliable patterns in the data. However, interpretation of those results as evidence of the causal relationship between texting bans and ED-visit reductions is less certain. Ferdinand et al. do not make causal claims, but, in principle, a causal relationship is important if these results are used to justify texting bans. Specific results give indications that texting bans alone could not have produced all of the differences attributed to them. First, the significant 40% decrease in ED visits resulting from secondary novice driver–only texting bans is implausibly large. Novice drivers are not responsible for 40% of crash-related ED visits, so they could not be responsible for the 40% reduction alone.

A second indication that the results might not reflect the causal action of texting bans is the lack of differences in IRRs across age groups. Given the large differences in texting rates by age (in 2016, 4.5% of drivers aged 16–24 years, 2.0% of drivers aged 25–69 years, and 0.3% of drivers aged 70 years or older were texting³), the estimated IRRs should have decreased in magnitude with age.

SHOULD STATES ADOPT TEXTING BANS?

Despite these uncertainties, there may be reasons for states to enact texting bans even if the Ferdinand et al. results do not arise (fully) from a causal link between texting bans and ED visit reductions. First, as mentioned, numerous studies indicate that texting increases crash risk, especially for younger (and older) drivers. Second, additional research shows that handheld cell phone bans dramatically reduced handheld cell phone use in states that enacted them, and that the reductions in use persisted over time. If such bans are relatively low cost to implement, then these two facts alone may justify them.

Regardless, the literature on texting and driving, along with the hope of the Ferdinand et al. results, underscores the need to promote attentive, eyes-on-road driving practices. Although advances in vehicle automation hold the promise of eliminating human-driver error, that solution is still decades from widespread deployment. In the meantime, it is critical to take measures to help drivers avoid crashes and thereby reduce injuries and deaths associated with motor vehicle crashes.