Nov 9, 2015 - of the article's assumption. Hence, the scale's internal and external validities for indicating driving impairment are highly ..... and cell phone conversations in simulated driving. Journal of ... Research Part F: Traffic Psychology and Behaviour, 28, 55â64. ... Mobile phone use by young drivers: Effects on traffic.
585053
HFSXXX10.1177/0018720815585053Human FactorsDriver Compensation
SPECIAL SECTION: Peer Commentary on Assessing Cognitive Distraction in Driving
Driver Compensation: Impairment or Improvement? Richard A. Young, Wayne State University, Detroit, Michigan
Strayer et al.’s conclusion that their “cognitive distraction scale” for auditory-vocal tasks indicates “significant impairments to driving” is not supported by their data. Additional analysis demonstrates that slower brake reaction times during auditory-vocal tasks were fully compensated for by longer following distances to the lead car. Naturalistic driving data demonstrate that cellular conversation decreases crash risk, the opposite of the article’s assumption. Hence, the scale’s internal and external validities for indicating driving impairment are highly questionable. Keywords: risk assessment, accidents, human error, attentional processes, cognition, distractions and interruptions, dual task, time sharing, task switching, mental workload, speech user interfaces (sui), displays and controls, interface evaluation, human-computer interaction, computer systems, usability/acceptance measurement and research, distraction, surface transportation, aggressive and risky driving
Address correspondence to Richard A. Young, Wayne State University, School of Medicine, WSU Tolan Park Medical Building, Detroit, MI 48201, USA; email: richardyoung9@ gmail.com. HUMAN FACTORS Vol. 57, No. 8, December 2015, pp. 1334–1338 DOI: 10.1177/0018720815585053 Copyright © 2015, Human Factors and Ergonomics Society.
Because a secondary task is “distracting” does not necessarily mean that it impairs driving. For example, whereas a reaction time (RT) increase due to distraction may be evidence of potential impairment, a longer following distance (FD) to a lead vehicle is likely evidence of improvement (e.g., Angell et al., 2006; Kircher et al. 2004). At a given speed, a longer FD indicates a longer following time (FT) to a lead vehicle. Evans and Wasielewski (1982) found that drivers with >1 s FTs in high-flow freeway traffic are relatively crash and violation free compared to drivers with shorter FTs. Young (2014b), using Naturalistic Driving Study (NDS) data, found that drivers lengthen FT during cellular conversations. Victor et al. (2015) found that increasing FT reduces rear-end crash risk. Rather, it is shorter headways that indicate driving safety impairment, for example, from alcohol (Strayer, Drews, & Crouch, 2006) or aggression (Tasca, 2000). Drivers can increase FT by reducing speed (Yannis, Papadimitriou, Karekla, & Kontodima, 2010, Figure 2). Kircher, Ahlstrom, Palmqvist, and Adell (2015) found that bicyclists proactively reduce speed before calling or texting. Whether such compensations are proactive, reactive, or both, they improve safety margins (see Ahlstrom, Kircher, Thorslund, & Adell, 2015; Platten, Schwalm, Hülsmann, & Krems, 2014). However, Strayer et al. (2015) claim that “brake RT increased as a function of condition over and above any compensatory effects associated with following distance” (p. 1311). To test this claim, the original mean data were reanalyzed. Table 1 gives the mean brake RT (bRT) and FD data provided by D. Strayer (personal communication, September 25, 2013), in bold, with their calculated changes from Single (defined as “just driving” with no secondary task). To test compensation in the provided data, FD is divided by speed to convert to the bRT
Downloaded from hfs.sagepub.com at HFES-Human Factors and Ergonomics Society on November 9, 2015
Downloaded from hfs.sagepub.com at HFES-Human Factors and Ergonomics Society on November 9, 2015
1335
0.0 –37.9 20.0 55.7 25.6 46.0 116.5 171.2
920.8 882.9 940.8 976.4 946.4 966.7 1037.2 1092.0
Task Condition
Single Radio listen Book on tape Passenger Handheld Hands free Speech to text OSPAN
0.0 –4.1 2.2 6.0 2.8 5.0 12.6 18.6
ΔbRTb (%) 23.27 23.00 24.59 24.24 23.44 24.09 26.14 27.76
FDa (m) 0.00 –0.27 1.32 0.97 0.17 0.82 2.87 4.49
∆FDb (m)
ΔFDb (%) 0.0 –1.2 5.7 4.2 0.7 3.5 12.3 19.3
Following Distance (FD) ∆FTb (ms)
2000.0 0.0 1976.8 –23.2 2113.5 113.5 2083.4 83.4 2014.6 14.6 2070.5 70.5 2246.7 246.7 2385.9 385.9
FTc (ms)
Following Time (FT)
0.0 –1.2 5.7 4.2 0.7 3.5 12.3 19.3
ΔFTb (%) 1079.2 1093.9 1172.7 1106.9 1068.2 1103.7 1209.4 1293.9
SMd (ms) 0.0 14.7 93.5 27.7 –11.0 24.5 130.2 214.7
∆SMb (ms)
Safety Margin (SM)
0.0 1.4 8.7 2.6 –1.0 2.3 12.1 19.9
ΔSMb (%)
1.00 1.21 1.75 2.33 2.45 2.27 3.06 5.00
Cognitive Distraction Scalee
Note. OSPAN = Operation Span task. a Bold data provided by D. Strayer (personal communication). b Change from Single. c The trained and instructed mean FT of 2,000 ms is assumed for the Single condition. The other FTs are calculated from the ∆FT percentage change, which is the same as the ∆FD percentage change, given equal speed across task conditions. d Safety margin defined here as FT minus bRT. e The cognitive distraction scores as given by Strayer et al. (their measure of deficit from the Single “just driving” baseline with score 1).
∆bRTb (ms)
bRTa (ms)
Brake RT (bRT)
Table 1: Mean Brake Reaction Time, Following Distance, Following Time, Safety Margin, and Cognitive Distraction Scale for the Eight Task Conditions
1336 December 2015 - Human Factors
scale. The authors did not provide speed data, but their previous studies with the same simulator protocol (Drews, Pasupathi, & Strayer, 2008; Strayer et al., 2006; Strayer & Drews, 2004) found no significant differences in mean speed between task conditions. Assuming a similar situation here, the percentage changes in FD and FT are equal (Young, 2014a, Section 1.3.3), and FTs can be calculated (Table 1). The safety margin (SM) is the difference between FT and bRT, which was >1 s for all task conditions. Overall, during tasks with a longer bRT, drivers increased FT (r = .95, p = .0002), indicating that changes in FT (∆FT) fully compensated changes in bRT (∆bRT). For example, Table 1 shows that from Single to OSPAN, FD lengthened from 23.27 to 27.76 m, or 19.3%. The corresponding 19.3% increase for FT was from 2,000 to 2,385.9 ms, or a ∆FT of 385.9 ms. The ∆bRT between Single and OSPAN was from 920.8 to 1,092.0 ms, or 171.2 ms. The SM change (∆SM) for OSPAN is then 214.7 ms (∆FT minus ∆bRT, or 385.9 ms − 171.2 ms), the largest SM improvement of any tested task and the opposite of the authors’ claim. Figure 1A illustrates these results. The SMs would be even larger if mean speeds declined rather than stayed equal for the dual-task conditions versus the Single condition. Figure 1B plots the change in safety margin (∆SM, the difference of ∆FT and ∆bRT in Figure 1A) for the task conditions. Figure 1C replots the article’s scale. The SM improves as the “cognitive distraction score” gets “worse” (r = .83, p = .011). That is, the authors’ scale indicates improved and not impaired driving safety margins. The authors assume their scale (Figure 1C) has a positive “monotonic relationship” with relative crash risk (RR). If so, cellular conversation should have an RR >1, but naturalistic studies indicate an RR
