DCSIMG

CHAPTER 3: OBJECTIVE 2, WHAT ARE THE ENVIRONMENTAL CONDITIONS ASSOCIATED WITH DRIVER CHOICE OF ENGAGEMENT IN SECONDARY TASKS OR DRIVING WHILE DROWSY? WHAT ARE THE RELATIVE RISKS OF A CRASH OR NEAR-CRASH WHEN ENGAGING IN DRIVING INATTENTION WHILE ENCOUNTERING THESE ENVIRONMENTAL CONDITIONS?

This research objective used large-scale naturalistic driving data to determine the environmental conditions in which drivers choose to engage in secondary tasks or to drive while drowsy. The associated relative near-crash/crash risks of either engaging in complex or moderate secondary tasks or driving drowsy during poor environmental conditions was also assessed. Several types of environmental variables were recorded during the data reduction process for both the 100-Car Study event database and the baseline database. A list of these variables, the respective levels of each, and a definition of each variable is presented in Table 3.1. Please note that all of these variables were recorded based solely upon the video observed at the time of the event or epoch. For lighting levels, the corresponding time stamp was also used to distinguish between dawn and dusk.

Table 3.1. A detailed list of the environmental variable names, levels of each,
and operational definition.

Variable Name

Levels of Variable

Definition of Variable

Lighting

Daylight
Darkness, lighted
Darkness, non lighted
Dawn
Dusk

Ambient lighting levels to denote the time of day.

Weather

Clear
Raining
Sleeting
Snowing
Foggy
Misty
Other

Description of the presence of ambient precipitation and type of precipitation occurring.

Road Type

Divided
Not divided
One-way Traffic
No lanes

Description of the type of roadway and how traffic is separated.

Road Alignment/Road Profile

Straight, level
Straight, grade
Curve, level
Curve, grade

Description of the road profile at the onset of the conflict.

Traffic Density

Free flow
Stable flow, speed restricted
Unstable flow, temporary restrictions
Unstable flow, temporary stoppages
Restricted Flow
Forced flow with low speeds and traffic volumes

Level of service definitions (NHTSA) to define six levels of traffic density ranging from free flow to stop-and-go traffic.

Surface Condition

Dry
Wet
Snowy
Icy
Other

Description of the resulting condition of the roadway in the presence of precipitation.

Traffic Control Device

Traffic signal
Stop sign
Yield sign
Slow, warning sign
Traffic lanes marked
Officer/watchman
Other
Unknown
None

Denotes the presence of a traffic signal near the onset of the conflict.

Relation to Junction

Intersection
Intersection-related
Interchange area
Entrance/exit ramp
Driveway/alley access
Parking lot
Non-junction
Other

Description of the road and whether a junction was present.


DATA INCLUDED IN THESE ANALYSES

Two databases were used for this analysis. The first was the event database , which consisted of all the crashes, near-crashes, and incidents identified and reduced as part of the 100-Car Study. Only the crashes and near-crashes were used in these analyses (for a discussion of the reasons for this, please refer to Chapter 2, Objective 1 ). Recall that this data is referred to as event data for this report. The second was the baseline database , which consisted of 20,000 randomly selected 6-second segments of video that were viewed by trained data reductionists. The random sample was stratified to produce a case-control data set which increased power for odds ratio calculations. For a complete description of the variables that were recorded for the baseline database, please refer to Chapter 1: Introduction and Method.

For the following analyses, the term inattention-related event refers only to complex- and moderate-secondary-task engagement. Simple secondary task engagement and driving-related inattention to the forward roadway were not used in these analysis; as shown in the previous chapter, these two types of inattention were either not significantly different than normal, baseline driving or provided a protective effect. Also, non-specific eyeglance was not considered, since its inclusion would have reduced the number of baseline epochs available for analysis, and because it was found to be a relatively redundant source of inattention for the baseline epochs (as shown in the previous chapter).

As the effect of risk factors were to be compared across levels of environmental variables, a different analysis method was used. The odds ratio estimates in the chapter were obtained using maximum likelihood estimates obtained from logistic regression models. The stratified analysis or logistic regression allows for comparable evaluation of risk factors across the levels or strata of an environmental variable of interest. To ascertain whether it is more risky to engage in complex tasks on a dark roadway or to drive while alert on a dark roadway, the interaction of both complex-secondary-task engagement ( inattentive or attentive driver) and ambient light levels ( daylight , dusk , dawn , darkness - lighted , darkness - not - lighted ) must be assessed. Logistic regression models provide a point estimate for the odds of a crash or near-crash based upon the driver engaging in a secondary task (or driving attentively) and driving environment.

Three independent odds ratio calculations were conducted to assess the relative near-crash/crash risk in various weather, roadway, and traffic environments. These three odds ratio calculations assess the following:

  1. Is driving drowsy during < environmental variable level > riskier than driving alert in < environmental level >?

  2. Is engaging in complex secondary tasks during < environmental variable level > riskier than driving alert in < environment level >?

  3. Is engaging in moderate secondary tasks during < environmental variable level > riskier than driving alert in < environment level >?

Only drowsiness , complex , and moderate secondary tasks were used in the following odds ratio calculations. Recall from the previous chapter that complex and moderate secondary task engagements were operationally defined based upon the frequency of eyeglances away from the forward roadway and/or button presses that were necessary to complete the task. Complex secondary tasks required more than three button presses and/or eyeglances away from the forward roadway to complete the task, while moderate secondary tasks required two eyeglances or button presses. It was also demonstrated in the previous chapter that these two types of secondary tasks , as well as drowsiness , had higher relative near-crash/crash risks than normal, baseline driving, whereas simple secondary tasks were found to not be significantly riskier than normal, baseline driving. Therefore, only drowsiness , complex , and moderate secondary tasks were used in these calculations.

AMBIENT LIGHT/WEATHER CONDITIONS

Lighting Level

To record light levels for this analysis, data reductionists used the video footage and the time stamp corresponding to the epochs or events to make determinations of the ambient lighting levels. Table 3.2 presents the number of drowsiness - and secondary - task -related crashes, near-crashes, and baseline epochs observed for each of these lighting levels.

Table 3.2 The frequency of drowsiness- and secondary-task-related events and epochs that were recorded for each type of lighting level.

Lighting Level

Frequency of Drowsiness- Related Crash and Near-Crash Events

Frequency of Secondary-Task-Related Crash and Near-Crash Events

Frequency of Drowsiness- Related Baseline Epochs

Frequency of Secondary-Task-Related Baseline Epochs

Darkness-Lighted

27

42

2

13

Darkness- Not Lighted

18

17

279

3021

Dawn

2

5

51

205

Daylight

52

143

240

571

Dusk

13

20

183

305

Total

308

277

755

4115


Using only the baseline data, the percent of inattention-related epochs and the percent of the total number of baseline epochs were used to determine: (1) the percentage of baseline epochs that drivers engaged in secondary tasks or drove while drowsy during each of these lighting conditions, and (2) whether these percentages differed from the total number of baseline epochs that drivers encountered or were exposed to for each of these lighting conditions. These percentages were calculated by dividing the number of baseline epochs where drivers were engaging in a secondary task at a particular lighting level by the total number of epochs where the drivers engaged in a secondary task. For example, the number of baseline epochs where the driver was engaging in a complex or moderate secondary task during daylight was divided by the total number of baseline epochs where the driver was engaging in a complex or moderate secondary task.

Figure 3.1 presents the baseline data percentages for secondary-task-related epochs (N = 4,115), drowsiness-related epochs (N = 755), and total number of epochs (N = 19,467) for each level of lighting. The majority of complex- and moderate-secondary-task-related events and total baseline epochs occurred during daylight hours; this replicates findings from many previous instrumented-vehicle studies (e.g., Lee, Olsen, and Wierwille, 2003; Dingus et al., 2001). The percentages are very similar for the secondary-task-related epochs and the total number of epochs, suggesting that drivers are not selecting to engage in secondary tasks differently based on ambient lighting conditions. Drivers are experiencing drowsiness differently across the ambient lighting conditions, which is to be expected as ambient lighting levels are associated with time of day and daily wake/sleep cycles. Lower percentages of drowsiness were observed during the day, whereas higher percentages of drowsiness were observed at night compared to the total baseline epochs.

Figure 3.1. Percentage of secondary-task-related, drowsiness-related, and total baseline epochs for the different lighting levels observed.

click for long description

As shown in Table 3.3, driving drowsy in any of the ambient lighting levels is riskier than driving while alert during similar lighting levels. However, it appears that driving drowsy during the daylight may be slightly riskier than driving drowsy in the dark. While it is commonly thought that most drowsiness-related crashes occur at night, a majority of the drowsiness-related crashes in this study occurred during the daytime in heavy traffic (during morning and evening commutes). Thus, the risks of driving drowsy during the day may be slightly higher than at night due to higher traffic density.

Table 3.3. Odds ratio point estimates and 95 percent confidence intervals for the interaction of drowsiness by type of lighting.

Type of Lighting

Odds Ratio

Lower CL

Upper CL

Dawn

2.43

0.96

6.17

Daylight

5.27

3.55

7.82

Dusk

6.99

3.82

12.80

Darkness-Lighted

3.24

1.92

5.47

Darkness-Not Lighted

3.26

1.82

5.86


Note: numbers in bold font indicate that the point estimate is significantly different than normal, baseline driving (or an odds ratio of 1.0).


Relative near-crash/crash risks for the complex- and moderate-secondary-task engagement showed that engaging in complex tasks for all levels of ambient lighting were significantly more risky than driving alert at the same lighting levels (Tables 3.4 and 3.5). This was especially true for engaging in complex tasks at night, as these relative near-crash/crash risks were higher than during dawn , dusk , or daylight . The relative near-crash/crash risks for engaging in moderate secondary tasks were all near 1.0, but not significantly different than 1.0, which suggests that engaging in these tasks is not nearly as risky as engaging in complex tasks or driving while drowsy.

Table 3.4. Odds ratio point estimates and 95 percent confidence intervals for the interaction of complex secondary tasks by type of lighting.

Type of Lighting

Odds Ratio

Lower CL

Upper CL

Dawn

N/A

N/A

N/A

Daylight

3.06

1.84

5.06

Dusk

8.91

4.41

18.03

Darkness-Lighted

4.58

2.46

8.52

Darkness-Not Lighted

24.43

12.40

48.10

Note: numbers in bold font indicate that the point estimate is significantly different than normal, baseline driving (or an odds ratio of 1.0).

Table 3.5. Odds ratio point estimates and 95 percent confidence intervals for the interaction of moderate secondary tasks by type of lighting.

Type of Lighting

Odds Ratio

Lower CL

Upper CL

Dawn

0.71

0.21

2.39

Daylight

0.80

0.59

1.08

Dusk

1.55

0.87

2.76

Darkness-Lighted

0.98

0.61

1.56

Darkness-Not Lighted

0.98

0.61

1.56

Note: numbers in bold font indicate that the point estimate is significantly different than normal, baseline driving (or an odds ratio of 1.0).

Weather

Reductionists used the video to assess the weather conditions outside the vehicle. Table 3.6 presents the frequency counts of the number of drowsiness- and secondary-task-related events and baseline epochs that occurred during the different weather conditions. A majority of events and epochs occurred during clear weather.

Table 3.6. The frequency of drowsiness-related and secondary-task-related events and epochs that were recorded for each type of weather.

 

Type of Weather

Frequency of Drowsiness-Related Crash and Near-Crash Events

Frequency of Secondary-Task- Related Crash and Near-Crash Events

Frequency of Drowsiness-Related Baseline Epochs

Frequency of Secondary-Task-Related Baseline Epochs

1.

Clear

92

181

669

3,624

3.

Rain

20

45

79

462

4.

Sleet

0

0

1

4

5.

Snow

0

0

3

12

6.

Fog

0

0

2

6

7.

Mist

0

0

1

5

8.

Other

0

0

0

2

 

Total

112

226

755

4,115


Figure 3.2 presents the percent of drowsiness-related, secondary-task-related, and total baseline epochs for each weather type. Nearly all of the epochs occurred during clear weather , with 11 percent occurring during rainy weather . The percentages are nearly identical for secondary-task-related, drowsiness-related, and total baseline epochs for all weather conditions, indicating that drivers were not engaging in secondary tasks or driving drowsy substantially more often during any particular type of weather. The total number of events and epochs that occurred during sleet , snow , fog, mist , and other weather conditions was very small (the sample size was perhaps not large enough to adequately address the issue of secondary-task engagement during these types of weather).

Figure 3.2. Percentage of secondary-task-related, drowsiness-related, and total baseline epochs for each type of weather.

click for long description

Table 3.7 presents the odds ratio calculations for the different types of weather. Driving while drowsy during both rainy and clear weather is significantly more risky than driving alert during the same conditions. Interestingly, the elevated near-crash/crash risk is the same for both, suggesting that driving drowsy is very dangerous, regardless of roadway conditions. Unfortunately, the other weather conditions could not be assessed due to low statistical power.

Table 3.7. Odds ratio point estimates and 95 percent confidence intervals for the interaction of drowsiness by type of weather.

Type of Weather

Odds Ratio

Lower CL

Upper CL

Clear

4.34

3.22

5.86

Rain

4.41

2.41

8.08

Note: numbers in bold font indicate that the point estimate is significantly different than normal, baseline driving (or an odds ratio of 1.0).

The relative risk calculations for a crash or near-crash for complex secondary tasks also suggest that engaging in complex secondary tasks is significantly more risky than driving alert in similar conditions (Table 3.8). The relative near-crash/crash risk estimate is higher for rain, suggesting that it may be riskier to engage in complex secondary tasks during the rain than in clear weather. Some caution is urged in this interpretation because the confidence limit surrounding the odds ratio for engaging in a complex task during the rain is also larger than it is for clear weather.

Table 3.8. Odds ratio point estimates and 95 percent confidence intervals for the interaction of complex secondary tasks by type of weather.

Type of Weather

Odds Ratio

Lower CL

Upper CL

Clear

3.68

2.29

5.92

Rain

5.11

1.86

14.07

Note: numbers in bold font indicate that the point estimate is significantly different than normal, baseline driving (or an odds ratio of 1.0).

The odds ratio for engaging in moderate secondary tasks indicates that it may be safer to engage in moderate secondary tasks than complex secondary tasks (Table 3.9). Most of the odds ratios for moderate secondary tasks were not significantly different than 1.0 suggesting that engaging in moderate secondary tasks are not protective but rather are simply not riskier than driving while drowsy or engaging in complex secondary tasks.

Table 3.9. Odds ratio point estimates and 95 percent confidence limits for the interaction of moderate secondary tasks by type of weather.

Type of Weather

Odds Ratio

Lower CL

Upper CL

Clear

0.86

0.65

1.13

Rain

0.65

0.37

1.15

Note: numbers in bold font indicate that the point estimate is significantly different than normal, baseline driving (or an odds ratio of 1.0).

ROADWAY AND SURFACE CONDITIONS

Road Type

Road Type (called “Traffic Flow” in the GES Database) primarily refers to whether there is a physical barrier between traffic. The No Lanes category was added for parking lots and should be interpreted as “no barrier.” One-way streets possess a barrier since all traffic is flowing in one direction. Table 3.10 shows the distribution of drowsiness- and secondary-task-related events and epochs that occurred on each type of traffic-flow roadway. Most secondary-task-related events and epochs occurred on divided roadways.

Table 3.10. The frequency of secondary-task-related events and epochs that were recorded for each road type.

Road Type

Frequency of Drowsiness-Related Crash and Near-Crash Events

Frequency of Secondary Task-Related Crash and Near-Crash Events

Frequency of Drowsiness-Related Baseline Epochs

Frequency of Secondary-Task-Related Baseline Epochs

Divided

64

118

530

2,612

Undivided

43

95

199

1248

One-way

4

11

17

114

No Lanes

1

2

9

141

Total

112

226

755

4,115


Figure 3.3 presents the percent of total drowsiness-related epochs, secondary-task-related epochs, and total baseline epochs for the various road types. While divided roadways were most frequent for all categories, a substantial number of epochs also occurred on undivided roadways as well. One-way roadways and/or parking lots were represented in a smaller percentage of epochs. There were no practical differences between the percent of secondary task or drowsiness epochs as compared to total baseline epochs, which suggests that drivers are engaging in secondary tasks regardless of type of roadway that they happen to be navigating at the time. There was a slightly higher percent of occurrence for drowsiness-related epochs on divided roadways than on undivided roadways. One possible hypothesis for this result is that drivers are more relaxed and less active on divided roadways (i.e., interstates) because they do not have to monitor cross traffic as frequently as on undivided roadways. This feeling of relaxation may result in higher occurrence of drowsiness.

Figure 3.3. Percentage of secondary-task-related, drowsiness-related, and total baseline epochs by type of roadway.

click for long description

Even though drivers appear to be engaging in secondary tasks or driving drowsy on these types of roadways equally, that does not necessarily mean that it is equally safe to do so. Odds ratios for drowsiness, complex-secondary-task and moderate-secondary-task engagement were calculated for each road type and are presented in Tables 3.11 through 3.13. All of the odds ratios for the interaction of drowsiness and road type were greater than 3.0, suggesting that driving while drowsy on any of these road types increases near-crash/crash risk by at least three times that of driving alert on the same types of roadways with the highest risk associated with undivided roadways.

Engaging in complex secondary tasks while driving on undivided roadways was slightly less dangerous than engaging in complex secondary tasks while driving on a divided roadway. While this may not make intuitive sense, this result may be an artifact of the higher percentage of driving on divided roadways and the higher traffic densities occurring on these roadways given the metropolitan environment where these data were collected. The odds ratios for engaging in moderate secondary tasks were not significantly different from 1.0 indicating that engaging in moderate secondary tasks is less risky than engaging in complex secondary tasks or driving drowsy.

Table 3.11. Odds ratio point estimates and 95 percent confidence intervals for the interaction of drowsiness by road type.

Road Type

Odds Ratio

Lower CL

Upper CL

Divided

3.73

2.61

5.34

Undivided

5.54

3.47

8.84

One-Way

3.40

1.76

6.59

Parking Lots

N/A

N/A

N/A

Note: numbers in bold font indicate that the point estimate is significantly different than normal, baseline driving (or an odds ratio of 1.0).

Table 3.12. Odds ratio point estimates and 95 percent confidence intervals for the interaction of complex secondary tasks by road type.

Road Type

Odds Ratio

Lower CL

Upper CL

Divided

4.20

2.40

7.33

Undivided

3.60

1.89

6.79

One-Way

3.66

1.63

8.18

Parking Lots

N/A

N/A

N/A

Note: numbers in bold font indicate that the point estimate is significantly different than normal, baseline driving (or an odds ratio of 1.0).

Table 3.13. Odds ratio point estimates and 95 percent confidence intervals for the interaction of moderate secondary tasks by road type.

Road Type

Odds Ratio

Lower CL

Upper CL

Divided

0.79

0.57

1.10

Undivided

0.85

0.54

1.35

One-Way

0.94

0.48

1.84

Parking Lots

0.68

0.25

1.85

Note: numbers in bold font indicate that the point estimate is significantly different than normal, baseline driving (or an odds ratio of 1.0).

Roadway Alignment

Roadway alignment is a GES Crash Database variable that refers to both the curvature and percent grade of the roadway. Both curvature and percent grade can dramatically shorten the driver’s sight distance of the roadway and traffic patterns in front of them. Coupled with driver inattention or drowsiness, specific types of roadway alignment may increase near-crash/crash risk. Given reduced sight distance, do drivers tend not to engage in secondary tasks or attempt to become more alert, if even for a brief time?

Table 3.14 presents the frequency of secondary-task-related events and baseline epochs that were observed for each type of roadway alignment. Most events and epochs occurred on straight and level roadways. This is most likely an artifact of the geographic location where the data were collected (Northern Virginia/Washington, DC, metro area).

Table 3.14. The frequency of drowsiness and secondary-task-related events and epochs that were recorded for each type of roadway alignment.

Type of Roadway Alignment

Frequency of Drowsiness-Related Crash and Near-Crash Events

Frequency of Secondary-Task-Related Crash and Near-Crash Events

Frequency of Drowsiness-Related Baseline Epochs

Frequency of Secondary-Task-Related Baseline Epochs

Curve Grade

0

6

7

41

Curve Level

20

31

73

387

Straight Grade

1

4

15

95

Straight Level

90

184

659

3,587

Straight Hill Crest

0

0

0

1

Curve Hill Crest

0

0

0

0

Other

0

0

0

1

Total

111

225

754

4,112


Figure 3.4 compares the percentage of drowsiness-related, secondary-task-related, and total baseline epochs for different levels of roadway alignment. While 90 percent of drowsiness-, secondary-task-related, and total baseline epochs occur on straight and level roadways, other roadway alignments did occur in the dataset. The percentages for each type of alignment were nearly identical for all three groups. This suggests that drivers are not selecting to engage in secondary-task-related activities based upon the alignment of the roadway, nor are there differences in driver drowsiness on these different roadway alignments.

Figure 3.4. Percentage of secondary-task-related, drowsiness-related, and total baseline epochs by type of roadway alignment.

click for long description

To determine whether there is increased individual near-crash/crash risk for driving drowsy or engaging in secondary-task-related activities for particular types of roadway alignment, odds ratios were calculated and are presented in Tables 3.15 through 3.17. The odds ratio calculation for straight, grade had the highest near-crash/crash risk, suggesting that drowsy drivers are over six times as likely to be involved in a crash or near-crash as an alert driver on a straight, grade roadway (Table 3.15). The odds ratio for the straight, grade was not significantly higher than for curve, level or straight, level (since the confidence limits of all three roadway alignments overlap).

Engaging in complex secondary tasks on these four roadway alignments was also shown to be riskier than driving alert on the same roadway types (Table 3.16). The odds ratio for curve, level was nearly the same as the odds ratio for straight, level, suggesting that these two are equally riskier than driving while alert. The odds ratios for straight, grade was significantly higher than the other road alignments (except for straight, grade), suggesting that this road alignment is a riskier road environment for engaging in complex secondary tasks. The odds ratio for curve, grade was not significantly different than curve, level and straight, level. Driving while performing complex secondary tasks was at least three times riskier than driving while alert for all of these road alignments.

The odds ratios for moderate secondary tasks indicate that these types of tasks are not as risky as engaging in complex secondary tasks or driving drowsy on these road alignments.

Table 3.15. Odds ratio point estimates and 95 percent confidence intervals for the interaction of drowsiness and roadway alignment.

Type of Roadway Alignment

Odds Ratio

Lower CL

Upper CL

Straight, Level

3.96

2.93

5.34

Curve, Level

5.81

3.66

9.21

Straight, Grade

6.29

2.20

17.96

Note: numbers in bold font indicate that the point estimate is significantly different than normal, baseline driving (or an odds ratio of 1.0).

Table 3.16. Odds ratio point estimates and 95 percent confidence intervals for the interaction of complex secondary tasks and roadway alignment.

Type of Roadway Alignment

Odds Ratio

Lower CL

Upper CL

Straight, Level

3.59

2.20

5.84

Curve, Level

3.58

1.95

6.60

Straight, Grade

26.00

7.31

92.53

Curve, Grade

6.75

2.08

21.89

Note: numbers in bold font indicate that the point estimate is significantly different than normal, baseline driving (or an odds ratio of 1.0).

Table 3.17. Odds ratio point estimates and 95 percent confidence intervals for the interaction of moderate secondary tasks and roadway alignment.

Type of Roadway Alignment

Odds Ratio

Lower CL

Upper CL

Straight, Level

0.79

0.60

1.03

Curve, Grade

1.69

0.56

5.09

Curve, Level

0.88

0.56

1.39

Straight, Grade

1.86

0.56

6.19

Note: numbers in bold font indicate that the point estimate is significantly different than normal, baseline driving (or an odds ratio of 1.0).

Traffic Density

Traffic density was recorded by the data reductionists using the Transportation Research Board's (TRB) Level of Service (LOS) Definitions ( Highway Capacity Manual , 2000). The LOS is a scale from 1 to 6 of increasing traffic density with 1 being free-flow traffic and 6 being stop-and-go traffic with extended stoppages. The six levels of traffic density are listed in Table 3.18 along with the frequency of drowsiness- and secondary-task-related events and epochs that were recorded at each level of traffic density.

Table 3.18. The frequency of secondary-task-related events and epochs that were recorded at each level of traffic density.

Traffic Density

Frequency of Drowsiness-Related Crash and Near-Crash Events

Frequency of Secondary Task-Related Crash and Near-Crash Events

Frequency of Drowsiness-Related Baseline Epochs

Frequency of Secondary-Task-Related Baseline Epochs

LOS A: Free Flow

44

84

430

2,013

LOS B: Flow with Some Restrictions

31

73

237

1,529

LOS C: Stable Flow Maneuverability and Speed are more Restricted

20

43

56

391

LOS D: Flow is Unstable Vehicles are unable to pass with temporary stoppages.

10

19

14

84

LOS E: Unstable Flow- Temporary restrictions, substantially slow drivers

5

7

10

55

LOS F: Forced Traffic Flow Conditions with Low Speeds and Traffic Volumes Below Capacity

2

0

8

43

Total

112

226

755

4,115

Note: inattention is defined as only those events where drivers were involved in secondary tasks or were severely drowsy.

Figure 3.5 presents the percentage of drowsiness-related, secondary-task-related, and total baseline epochs that occurred at each level of traffic density. As traffic density increased, the frequency of drowsiness- and secondary-task-related epochs decreased. The percentage for secondary-task-related epochs and total epochs did not differ, indicating that drivers are not choosing to engage in complex or moderate secondary tasks differently for these traffic densities. The drowsiness-related epochs were slightly different, with more drowsiness-related events occurring during free-flow and fewer occurring during flow with restrictions and stable traffic flow. One hypothesis for this result is that driving in free-flow traffic is less interesting and requires less activity by the driver. Therefore, these types of traffic flow may help induce drowsiness because the driver is under-stimulated. \

Figure 3.5. Percentage of secondary-task-related, drowsiness-related, and total baseline epochs by type of traffic density.

click for long description

Odds ratios were calculated to determine if any of these traffic densities present greater individual near-crash/crash risk. Tables 3.19 through 3.21 present the odds ratio calculations for each level of density for drowsiness. The odds ratio calculations for driving drowsy at each level of traffic density suggest that driving drowsy is at least three times riskier than driving while alert during the same level of traffic density. None of the traffic densities were significantly riskier than any another level of traffic density.

Similar results were found for engaging in complex secondary tasks where this activity was found to increase near-crash/crash risk by at least three times that of alert driving during the same traffic density. Again, engaging in complex secondary tasks was equally risky at all levels of traffic density, except for LOS D.

The odds ratios for moderate secondary tasks did not demonstrate similar risk levels and thus engaging in moderate secondary tasks during these traffic levels is not as risky and does not elevate near-crash/crash risk to the extent as driving drowsy or engaging in complex secondary tasks. This result was found to be true across all levels of traffic density for moderate-secondary-task engagement.