Is a hands-free phone safer than a handheld phone?

Abstract

Introduction

Although it is becoming more and more accepted that driving while talking on a cell phone can be hazardous, most jurisdictions are making handheld phone use illegal while allowing hands-free phone use.

Methods

The scientific literature exploring the effects of these two types of cell phone use on driving and driving-related performance is reviewed here.

Results

Our review shows that talking on the phone, regardless of phone type, has negative impacts on performance especially in detecting and identifying events. Performance while using a hands-free phone was rarely found to be better than when using a handheld phone. Some studies found that drivers compensate for the deleterious effects of cell phone use when using a handheld phone but neglect to do so when using a hands-free phone.

Impact on Industry

Current research does not support the decision to allow hands-free phone use while driving.

Introduction

Today, there are more than 253 million cell phone subscribers in the United States (Cellular Telecommunications Industry Association), and more than 19.3 million cell phone subscribers in Canada (Canadian Wireless Telecommunications Association). These figures are more than half of the total population in both countries, suggesting that cell phones are becoming ubiquitous in our lives. It is not surprising, then, that there is an increase in the number of individuals who talk on their cell phones while driving. In fact, the percentage of drivers reporting using a cell phone “in the past days” rose from 20.5% in 2001 to 37% in 2006 (Vanlaar, Simpson, Mayhew, & Robertson, 2007). Although we use our hands and feet when we drive, driving is controlled by the mind. As William James (1890) noted, paying attention “...implies withdrawal from some things in order to deal effectively with others” (p. 403-404). Hence, using our mind to carry on a conversation or perform any other attention-demanding activity (e.g., eating, problem solving) will render it less available for processing the signals and performing the actions necessary to minimize accidents when driving.

If a cell phone conversation distracts the driver, then a cell phone conversation can be a possible safety hazard for driving. In fact, Goodman, Tijerina, Bents, and Wierwille (1999) noted a potential causal relation between the increasing number of cell phones and an increasing frequency of cell phone-related car accidents. Supporting this relation Laberge-Nadeau et al. (2003) found that the relative risk of being involved in accidents was 38% higher for cell phone users than for non-users and more than 100% greater for frequent cell phone users. Yet, in a survey of crashes reported to the police, Stutts, Reinfurt, Steplin, and Rodgman (2001) found that only 1.5% of self-reported distractions were linked to using or dialing a cell phone. For several reasons, this very low estimate from Stutts et al. (2001) must be regarded with caution. First, by definition, distraction means that the mind, which is needed to record things into memory, may be unavailable to do so; consequently, self-reported distraction should be particularly unreliable. Second, as noted by Stutts et al. (2001) in their conclusion, in post-crash situations drivers will be biased against reporting any self-imposed distractions, perhaps especially those involving cell phone use. Finally, and perhaps most importantly, Stutts and colleagues specifically note that “estimating the true percentage of crashes attributable to various distracting events was not the goal” of their study (p. 36).

Horrey and Wickens (2006) recently reported a 23 study meta-analysis of the costs associated with cell phone use while driving. Strongly supporting the proposal that talking on a cell phone can distract the driver in a way that increases response time to unexpected hazards, they found that using a cell phone while driving had substantial negative effects on driving-related performance in reaction time (RT), but not in lane maintenance. Moreover, evidence concerning the association between cell phone use and traffic accidents in the real world was found through epidemiological reports by Redelmeier and Tibshirani (1997) and McEvoy et al. (2005). In spite of the common understanding that a hands-free (HF) cell phone is a safer option than a handheld (HH) cell phone while talking and driving at the same time, Horrey and Wickens (2006) found that the deleterious effects of cell phone use were similar for the HH and HF phones. The purpose of their study was to provide a comprehensive picture of cell phone use in general, and consequently the difference between the two phone types, while mentioned, was neither emphasized nor covered in depth. The major purpose of the present review is to use the literature (Abdel-Aty, 2003, Burns et al., 2002, Consiglio et al., 2003, Haigney et al., 2000, McEvoy et al., 2005, Patten et al., 2004, Redelmeier and Tibshirani, 1997, Strayer and Johnston, 2001, Strayer et al., 2006, Törnros and Bolling, 2005, Törnros and Bolling, 2006) to determine:

• whether talking on a HF cell phone while driving is safe, and

• whether there is a difference in safety between talking on a HH phone and talking on a HF phone while driving.

Across these studies of cell phone use, the degree to which the task performed resembles driving varies greatly. We will use the term “fidelity” to refer to variation along this dimension. Reed and Green (1999) suggested that driving-related performance suffered more from dialing a phone, in simulated driving, than in field driving tasks. It is possible that not only the manual operation of dialing (Reed & Green), but also the cognitive demands of conversing on a HH phone or a HF phone, may affect driving-related performance differently according to the fidelity of the driving situation being tested. Another purpose of this review is to examine this possibility.

We used the reference sections of papers (e.g., Horrey & Wickens, 2006) and online databases (e.g., PsycINFO, Web of Science Citation) to identify candidate papers for this review. The most important criterion for selection was that the relationship between driving-related performance and phone type (both HH and HF) was examined within a single study. We did not set any restrictions on the year of publication nor did we enforce any methodologically-based exclusions. Thus, there is variability in the methods used to simulate driving tasks, manipulate cell phone conversations, manipulate variables other than the telephone conversations, and measure effects of the cell phone conversation across the studies. Although this variability makes it difficult to combine the results across studies in a quantitative fashion, there is, nevertheless a significant benefit: when patterns of results are similar across studies with variable methods, the common pattern is consequently demonstrated to be robust and generalizable. In total we found 10 studies (eight experimental and two epidemiological¹ studies) that compared driving performance (experimental studies) and real accidents (epidemiological studies) while people were using HH and HF phones (Table 1).

Experimental studies examine the effect of the phone type on the driving-related performance. The advantage of experimental studies is that cell phone use and the type of cell phone can be manipulated, which makes it possible to establish a causal relationship between the cell phone type and driving-related performance. However, there are different degrees of fidelity across the studies reviewed here. At low levels of fidelity, there are non-driving studies that do not involve actual driving, but instead require participants to perform tasks, whose component operations (e.g., tracking and RT tasks) are akin to driving operations (e.g., keeping the car on the road, braking to avoid a collision; Consiglio et al., 2003, Strayer and Johnston, 2001). Simulated driving studies have an intermediate degree of fidelity. In such studies, usually employing a driving simulator, participants drive on simulated roads (Abdel-Aty, 2003, Burns et al., 2002, Haigney et al., 2000, Strayer et al., 2006, Törnros and Bolling, 2005, Törnros and Bolling, 2006) while talking on a HH or a HF phone. Simulated driving studies, which are the most common among studies in general and also in this review, have advantages in terms of safety, cost, and experimental control (Reed & Green, 1999). Finally, a field driving study is the most realistic of all. In a field driving study, performance in an actual driving task is examined (Patten et al., 2004). Generally speaking, there is a tradeoff between fidelity and experimental control. Laboratory studies gain control while sacrificing fidelity; the reverse is true for field studies.

The general approach used in the experimental studies includes measuring driving-related variables while manipulating the phone type and other factors. Driving-related measures differ greatly across the studies. Most studies used the time to respond to targets or events as a measure of attention or degree of distraction (Young, Regan, & Lee, 2008). Because increases in the time to respond to a potential hazard by a braking or steering response can have disastrous consequences, great weight is placed on these data. With varying frequency the studies reviewed also included measures of vehicle control such as lane maintenance (lane deviation, collision, off-road excursion (OFF)) and vehicle speed. The focus here will be on these relevant variables from each study in which they were reported rather than upon all the (unique) variables reported in each study. In some studies drivers were instructed to maintain vehicle speed. In these cases, going too fast or too slow would be considered a failure to follow instructions. Because slowing down can be a compensatory strategy when driving becomes more difficult, and because going too fast increases the risk of accidents, our analysis of speed will focus on studies that allow us to determine the actual driving speed.

Epidemiological studies examine the relationship between cell phone conversation and motor-vehicle accidents in the real world. For example, Redelmeier and Tibshirani (1997) and McEvoy et al. (2005) evaluated cell phone records of individuals involved in motor-vehicle accidents. One advantage of an epidemiological approach is that there are fewer experimental artifacts that can contaminate the data (Goodman et al., 1997). Another advantage is that it is necessarily more realistic than the experimental studies. The main disadvantage is that, unlike experimental studies, epidemiological studies cannot firmly establish a causal relationship between the variables (e.g., cell phone use and occurrence of accidents).