Assessment of Driver Distraction at Work Zones with Portable Rumble Strip Deployments in Texas

Study Objective

The objective of this research was to assess the extent of distracted driving occurring on roadways in Texas, U.S., as drivers approach a work zone, and the extent to which PRS may influence those rates.

Study Methodology

The Texas Department of Transportation (TxDOT) has published temporary traffic control (TTC) standard sheets that identify the prescribed location of PRS (when used) within the overall TTC plan for lane closures on 2-lane, 2-way highways and on multi-lane facilities ( 5 ). On 2-lane, 2-way highways with posted speed limits between 55 mph and 75 mph, an upstream set of PRS is shown positioned in the travel lane 1,500 to 2,700 ft upstream of the flagger station, depending on the posted speed limit. A second set of PRS are then shown 750 to 1,350 ft from the flagger station. In addition to the standard set of advance warning signs used for the flagging operation, a “Rumble Strips Ahead” warning sign is positioned 500 to 900 ft upstream of the first set of PRS, again depending on the posted speed limit. For lane closures on multi-lane roadways with posted speed limits between 55 mph and 75 mph, upstream sets of PRS are positioned in the closed lane and lane adjacent 1,000 to 1,800 ft upstream of the merging taper, depending on the posted speed limit on the roadway. The “Rumble Strips Ahead” warning sign is also deployed 500 to 900 ft upstream of the PRS. A second set of PRS is positioned only in the closed lane 750 to 1,350 ft upstream of the merging taper, depending on the posted speed limit.

The research team utilized direct field observations of drivers approaching work zones at several Texas work zones. Data collection was hampered by the COVID-19 pandemic, but data were eventually collected from six different work zones. At each work zone, team members positioned themselves near the location of the upstream set of PRS. Team members also collected observational data at a second location downstream from that initial PRS location. If the work crew deployed only the upstream set of rumble strips, the second data collection point was positioned near the beginning of the lane closure to maximize the distance between the PRS and the downstream data collection location. If two PRS arrays were deployed, downstream data were collected just before the second PRS array. Team member documented the distances from the PRS to these second data collection locations for use in the analyses. It was hoped that data could be collected at multiple downstream locations from the PRS at each study site; unfortunately, traffic volumes at most of the available study sites were too low to allow multiple downstream data collection points. In fact, multiple downstream data collection points could only be collected at one site.

Data were collected at both multi-lane and 2-lane, 2-way roadways to capture a cross-section of roadway types and approach volumes. Field crews installed either one or two sets of PRS upstream of each work zone at the beginning of a work shift. Table 1 summarizes the data collection locations and characteristics used in the analysis. Data were collected February–July, 2020. Only limited data could be collected on February 18 and therefore that data were combined with the data collected on February 17. On May 7, the work crew positioned the upstream PRS almost 1 mi upstream of the work zone. Therefore, the research team collected downstream data at both 1,500 ft and 4,000 ft downstream of that PRS location.

OPEN IN VIEWER

Table 1.

Field Data Collection Sites

Site	Date	Facility type	Distance from portable rumble strips (PRS) to work zone (ft)	PRS to 2nd data collection location (ft)
1	Feb 17–18, 2020	2-lane, 1-way frontage road	1,500	1,300
2	May 7, 2020	4-lane undivided	4,500	a) 1,500
				b) 4,000
3	May 20, 2020	4-lane divided	5,280	3,000
4	June 25, 2020	2-lane, 2-way	3,200	1,800
5	July 9, 2020	2-lane, 2-way	3,200	1,600
6	July 9, 2020	2-lane, 2-way	3,200	1,600

At each data collection location, team members were situated in individual vehicles positioned beyond the shoulder in a parking lot, driveway, or drainage area where they could view the approaching drivers. Observers looked for examples of visual distraction of those approaching drivers (looking down, looking at an electronic device, looking at the radio, head turned looking out the side window or talking to a passenger, etc.) and manually documented on a data collection tally sheet. It is likely that some of the drivers were also experiencing cognitive distraction whereby drivers are looking out the windshield but are not truly perceiving the roadway environment because they are thinking about other things ( 6 ). However, it was not possible for the team members to consistently assess visually whether a driver was cognitively distracted as they passed. As a result, only visual distraction was evaluated in this study.

Results

Table 2 presents the results of the data collection efforts, identifying:

OPEN IN VIEWER

Table 2.

Data Summary

Site	# Vehicles observed	Equivalent approach volume (vph)	# Drivers distracted at PRS	% Distracted drivers at PRS	# Drivers distracted downstream	% Distracted drivers downstream
1	280	44	27	9.6	17	6.1
2a	1,406	803	240	17.1	71	5.0
2b	1,411	705	201	14.2	99	7.0
Total	2,817	751	441	15.7	170	6.0
3	1,214	347	303	25.0	144	11.9
4	126	23	36	28.6	11	8.7
5	79	63	12	15.2	6	7.6
6	218	109	44	20.2	38	17.4
ALL	4,734	na	863	18.3	386	8.2

Note: data were collected at two locations at site 2. Data for each data collection location are shown as 2a and 2b, while the Total row below the underlined values of 2b represent the combination of those two data collection locations; na = not applicable.

As the table illustrates, more than 4,700 vehicles were observed across the six study sites. Averaged across all sites, data collected by the research team indicates that 18.3% of drivers (almost one in five) approaching the work zones were observed to be visually distracted just before reaching the PRS. This percentage did vary significantly from site to site, ranging from a low of 9.6% at the first site to a high of 28.6% at site 4. Recall that site 1 was on a frontage road in a more urban location where attentional demands on drivers are typically greater, whereas site 4 was on a very low-volume 2-lane, 2-way highway in a very rural area where attentional demands tend to be lower. Such differences in attentional demands on drivers may partially explain the range of distraction observed in a “normal” driving environment before encountering a work zone or the PRS. In addition, differences in data collector viewing positions (and different data collectors themselves) from site to site may have contributed to this variability in observed distraction percentages.

Overall, no clear trends are evident in pre-PRS distraction rates as a function of roadway type or approach volume (see Figure 1). Plotted as a function of roadway type and approach volume, the 2-lane, 2-way sites themselves are all clustered toward the low-volume end of the graph such that no clear trends are evident. Meanwhile, the three 4-lane roadway sites imply that distracted driving may be lower as traffic volumes increase, perhaps because of driver perceptions of higher frequencies of potential interactions with other vehicles on the roadway that limit their ability to tend to those distractions. However, if one looks at the data in relation to rural versus urban/suburban land use around the study sites (as depicted by the ovals in the figure), one also sees a difference between rural locations and those in or near urban/suburban areas. A similar hypothesis about distracted driving could be made about these classifications. Specifically, overall attentional demands of driving tend to be less in rural areas (fewer driveways and vehicles accessing the few driveways do exist, few if any traffic signals or stop-controlled intersections, etc.) than at sites in and around urban/suburban areas. Since attentional demands for driving tend to be higher in those urban/suburban areas, drivers may be less inclined to attempt to attend to distraction tasks while driving.

OPEN IN VIEWER

Figure 1.

Plot of approach volume, roadway type, and land use versus driver distraction at portable rumble strip (PRS) locations.

Figure 2 illustrates that that the percentage of drivers who were visually distracted was lower closer to the work zones at each site than immediately before passing over the PRS. Whether the reductions in distracted driving are entirely the result of the presence of the PRS or would have occurred even if PRS were not present cannot be absolutely ascertained from these data, as it was not possible to remove the PRS at these sites for any period of time to allow non-PRS data to be collected. However, data collectors at a few of the sites were able to see at least a portion of the distracted drivers look up after passing over the PRS, indicating that the devices were being successful in reducing some distracted driving behavior. It is possible that, once driver attention was returned to the roadway because of the PRS, drivers noticed the advance warning signs, arrow board, and channelizing devices ahead of them and therefore maintained attention as they approached the work zone rather than returning to a distracted state.

OPEN IN VIEWER

Figure 2.

Change in percentage of drivers distracted from before portable rumble strip (PRS) to downstream of PRS.

As to whether the effect of the PRS on distracted driving degrades the farther one gets from the devices, site 2 was the only location where traffic volumes were sufficient to gather data at multiple locations downstream of the PRS. At that site, a small (but not statistically significant) uptick in the percentage of drivers distracted did occur from the location that was 1,500 ft downstream of the PRS to the location that was 4,000 ft downstream of the PRS. Although the difference was not significant, data does suggest that distracted driving did not continue to decrease at locations closer to the lane closure itself.

Summary

The results of the data collection and analysis show visual distracted driving to be commonplace on Texas roadways. The extent of such distraction appears to depend on:

As expected, driving over the PRS did cause some of the distracted motorists to return their attention to the roadway (based on visual observations by the data collectors of drivers looking up out of the windshield after passing over the strips). Furthermore, the observational data collected downstream of the PRS showed a declining rate of distracted driving at locations closer to the work zones. One possible explanation for this finding is that, as the distracted drivers glance up and see the work zone devices they are approaching, they curtail their distracted driving behavior while they negotiate the work zone.

Finally, while the data did not conclusively demonstrate that the distraction-reducing effect of PRS degrades as one travels farther downstream, data from site 2, where it was possible to collect data at multiple locations downstream, hints at such an effect. Although the distracted driving rate did not increase significantly at the data collection location 4,000 ft downstream of the PRS than was observed 1,500 ft downstream of the PRS, it did not reduce further despite being much closer to the work zone itself. Thus, it appears that most of the effect of the PRS occurs within the 1,500 ft immediately downstream of the devices themselves.