Teen Driving Performance Associated with Distraction, ADHD, and Other Risk Factors (2025)
Chapter: 3 Research Findings
CHAPTER 3
Research Findings
The results are organized by prevalence calculations (percentage of baseline events or percentage of CNC events) followed by a mixed-effects logistic regression model to calculate ORs to estimate CNC risk using two EOR metrics: total time EOR and single longest EOR. Total EOR shows the accumulation of time drivers were willing to look away. Single longest EOR shows how long drivers were willing to look away for a single instance. Both metrics provide a complementary assessment of a teen driverʼs willingness to look away from the forward roadway. Teen driver distraction prevalence and risk estimates are presented first, followed by more detailed analyses of adolescents with ADHD.
Objective 1: Assess the Relationship Between the Crash Risk of Various Glance Durations and Driving Phases (Phase 1 Versus Phase 2 of Independent Driving in SPDS NDS)
For the total EOR, the prevalence of EOR for various durations during the 6 seconds prior to each SCE was higher than the prevalence of EOR for those durations during the baseline events (see Table 2). The prevalence prior to an SCE and during baselines decreased monotonically as eye-glance duration increased. As compared to the first 6 months, the prevalence of EOR during baselines was higher in the second 6 months. This result is reversed for SCEs, with the prevalence of EOR nominally lower during the second 6 months. The prevalence of total EOR and single longest EOR are listed in Table 2 and Table 3.
The prevalence of single longest EOR during baselines for various durations was lower than that of total EOR. The prevalence prior to SCEs was at most 10% and decreased monotonically as eye-glance duration increased. As compared to the first 6 months, the prevalence during baselines was higher in the second 6 months. Additionally, few SPDS participants, especially in the second 6 months, looked away for longer than 3 seconds. Given this, it may be more relevant to focus on the results for duration of ≥ 2 seconds versus < 2 seconds.
In the first 6 months, the OR of an SCE was 2.51 times higher when total EOR was longer than 1 second as compared to when it was less than 1 second. This increased monotonically as eye-glance duration increased, such that the odds of an SCE were 11 times higher for a duration greater than or equal to 4 seconds. That is, the SCE likelihood increased consistently with glance duration from 1 to 4 seconds during the first 6 months after licensure.
In the second 6 months, the odds of an SCE were statistically significant with total EOR longer than 2 seconds in duration, as compared to significant ORs with total EOR longer than 1 second in the first 6 months. ORs in the second 6 months were lower than in the first 6 months
Total EOR |
First 6 Months |
Second 6 Months |
||
|---|---|---|---|---|
Baseline (%) |
SCE (%) |
Baseline (%) |
SCE (%) |
|
≥ 1 s |
5.97 |
14.15 |
9.07 |
13.10 |
≥ 2 s |
2.39 |
8.49 |
3.20 |
8.33 |
≥ 3 s |
1.19 |
6.60 |
1.96 |
3.57 |
≥ 4 s |
0.34 |
3.77 |
0.71 |
1.19 |
Single Longest |
First 6 Months |
Second 6 Months |
||
|---|---|---|---|---|
Baseline (%) |
SCE (%) |
Baseline (%) |
SCE (%) |
|
≥ 1 s |
3.75 |
10.38 |
5.87 |
10.71 |
≥ 2 s |
0.51 |
0.943 |
0.88 |
3.571 |
≥ 3 s |
0.0 |
0.0 |
0.0 |
1.190 |
≥ 4 s |
0.0 |
0.0 |
0.0 |
1.190 |
(see Table 4). In the first 6 months, ORs of SCEs for various durations were significant. However, in the second 6 months, only the OR of an SCE with total EOR longer than 2 seconds was significant.
In the first 6 months, the OR of an SCE was 2.84 times higher when the single longest glance off the road was longer than 1 second as compared to when it was less than 1 second (see Table 5). In the second 6 months, the OR of an SCE for glances greater than or equal to 2 seconds was significant and much higher than for glances greater than or equal to 1 second. The likelihood of an SCE when the single longest EOR exceeded 2 seconds was 4.96 times higher than when it was less than 2 seconds (p < 0.05). However, when the single longest EOR was longer than 1 second, the ORs in the second 6 months were insignificant and lower than those in the first 6 months. Additionally, EOR by specific secondary task was examined, but no significant results were found.
|
Total EOR |
OR |
CI_Lower |
CI_Upper |
P_Value |
|---|---|---|---|---|---|
First 6 Months |
≥ 1 s |
2.51 |
1.29 |
4.89 |
0.007 |
≥ 2 s |
3.59 |
1.46 |
8.81 |
0.005 |
|
≥ 3 s |
5.71 |
1.87 |
17.41 |
0.002 |
|
≥ 4 s |
11.65 |
1.96 |
69.13 |
0.007 |
|
Second 6 Months |
≥ 1 s |
1.51 |
0.72 |
3.19 |
0.275 |
≥ 2 s |
2.91 |
1.08 |
7.83 |
0.034 |
|
≥ 3 s |
1.75 |
0.43 |
7.09 |
0.434 |
|
≥ 4 s |
1.49 |
0.14 |
16.21 |
0.742 |
|
|
CI = confidence interval. |
|||||
|
Single Longest EOR |
OR |
CI_Lower |
CI_Upper |
P_Value |
|---|---|---|---|---|---|
First 6 |
≥ 1 s |
2.84 |
1.29 |
6.20 |
0.009 |
Second 6 |
≥ 1 s |
1.99 |
0.86 |
4.57 |
0.107 |
≥ 2 s |
4.96 |
1.01 |
24.24 |
0.048 |
Objective 2: Determine How the Crash Risk of Various Glance Durations Changes with Age (16- to 17-Year-Olds Versus 18- to 19-Year-Olds in SHRP 2 NDS)
For the total EOR, the prevalence prior to SCEs for all durations was higher than that during baselines. As compared to 16- to 17-year-olds, the prevalence of total EOR during baselines in 18- to 19-year-olds was higher. The prevalence during baselines in both age groups decreased monotonically as glance duration increased. The prevalence of total EOR and single longest EOR are listed in Table 6 and Table 7.
The prevalence of single longest EOR was lower than that of total EOR across all durations. The prevalence prior to SCE and during baselines decreased monotonically as glance duration increased. As compared to the 16 to 17 age group, the prevalence of single longest EOR was higher among those aged 18 to 19 years, but only for glance duration greater than or equal to 1 second.
In the 16 to 17 age group, the ORs of an SCE were 1.89 times higher when total EOR exceeded 1 second as compared to when it was less than 1 second (see Table 8). This increased monotonically as glance duration increased, such that the ORs of an SCE were 11 times higher for duration greater than or equal to 5 seconds.
In the 18 to 19 age group, the ORs of an SCE were 1.66 times higher when total EOR was greater than or equal to 1 second as compared to when it was less than 1 second. As compared to the 16 to 17 age group, the ORs for various durations were lower for both total EOR and single longest EOR.
For both total EOR and single longest EOR, the ORs of SCEs for various durations were significant. In the 16 to 17 age group, SCE likelihood increased monotonically from 2 to 4 seconds. However, in the 18 to 19 age group, the OR remained relatively stable across all durations (see Table 8).
For the 16 to 17 age groupʼs single longest EOR, the OR of an SCE was 2.2 times higher when a single longest glance off the road was longer than 1 second as compared to when it was less than 1 second (see Table 9). The OR of an SCE for an EOR greater than or equal to 2 seconds was 11.03, which was statistically significant and higher than that for an EOR greater than or equal to 1 second. In the 18 to 19 age group, the likelihood of an SCE was 25.36 times higher
Total EOR |
16–17 |
18–19 |
||
|---|---|---|---|---|
Baseline (%) |
SCE (%) |
Baseline (%) |
SCE (%) |
|
≥ 1 s |
9.60 |
17.28 |
10.99 |
18.06 |
≥ 2 s |
3.62 |
9.94 |
5.61 |
12.26 |
≥ 3 s |
1.62 |
6.57 |
3.13 |
7.01 |
≥ 4 s |
0.78 |
3.82 |
1.60 |
3.64 |
≥ 5 s |
0.20 |
1.99 |
0.31 |
0.94 |
Single Longest EOR |
16–17 |
18–19 |
||
|---|---|---|---|---|
Baseline (%) |
SCE (%) |
Baseline (%) |
SCE (%) |
|
≥ 1 s |
6.51 |
13.46 |
8.39 |
14.82 |
≥ 2 s |
0.54 |
4.59 |
1.34 |
3.91 |
≥ 3 s |
0.098 |
1.99 |
0.08 |
1.89 |
≥ 4 s |
0.049 |
1.53 |
0.00 |
0.54 |
≥ 5 s |
0.049 |
0.92 |
0.00 |
0.13 |
|
Total EOR |
OR |
CI_Lower |
CI_Upper |
P_Value |
|---|---|---|---|---|---|
16–17 |
≥ 1 s |
1.89 |
1.89 |
1.90 |
0.0 |
≥ 2 s |
3.05 |
3.05 |
3.05 |
0.0 |
|
≥ 3 |
4.36 |
2.54 |
7.50 |
< 0.01 |
|
≥ 4 s |
5.89 |
2.78 |
12.49 |
< 0.01 |
|
≥ 5 s |
11.00 |
10.99 |
11.02 |
0.0 |
|
18–19 |
≥ 1 s |
1.66 |
1.29 |
2.12 |
0.00 |
≥ 2 s |
2.13 |
1.56 |
2.91 |
0.00 |
|
≥ 3 s |
2.09 |
1.39 |
3.14 |
0.00 |
|
≥ 4 s |
1.95 |
1.11 |
3.42 |
0.02 |
|
≥ 5 s |
2.96 |
0.94 |
9.38 |
0.06 |
|
Single Longest EOR |
OR |
CI_Lower |
CI_Upper |
P_Value |
|---|---|---|---|---|---|
16–17 |
≥ 1 s |
2.20 |
2.20 |
2.20 |
0.0 |
≥ 2 s |
11.03 |
11.02 |
11.04 |
0.0 |
|
≥ 3 s |
29.89 |
5.57 |
160.53 |
0.0 |
|
≥ 4 s |
45.38 |
45.32 |
45.43 |
0.0 |
|
≥ 5 s |
17.51 |
1.60 |
191.24 |
0.01 |
|
18–19 |
≥ 1 s |
1.74 |
1.33 |
2.29 |
0.00 |
≥ 2 s |
2.47 |
1.41 |
4.33 |
0.00 |
|
≥ 3 s |
25.36 |
5.02 |
127.96 |
0.00 |
when the single longest EOR exceeded 3 seconds as compared to when it was less than 3 seconds. As compared to the 16 to 17 age group, the ORs for various durations were lower in the 18 to 19 age group for single longest EOR.
Objective 3: Compare the Crash Risk of Various Glance Durations in Teen Drivers With and Without ADHD (SHRP 2 NDS)
For total EOR, the prevalence prior to SCEs was higher than during baselines. The prevalence prior to SCEs and during baselines decreased monotonically as glance duration increased. As compared to the ADHD group, the prevalence for various glance durations during baselines was lower in non-ADHD groups. The prevalence of total EOR and single longest EOR are listed in Table 10 and Table 11.
The prevalence of single longest EOR was much lower than that of total EOR, regardless of event type and ADHD status. For the single longest EOR, the prevalence during baselines was lower in
Total EOR |
ADHD |
Non-ADHD |
||
|---|---|---|---|---|
Baseline (%) |
SCE (%) |
Baseline (%) |
SCE (%) |
|
≥ 1 s |
12.35 |
14.68 |
10.38 |
17.69 |
≥ 2 s |
7.65 |
6.42 |
4.74 |
11.17 |
≥ 3 s |
3.53 |
2.75 |
2.47 |
6.81 |
≥ 4 s |
1.18 |
2.75 |
1.24 |
3.72 |
≥ 5 s |
0.00 |
0.92 |
0.28 |
1.43 |
Single Longest |
ADHD |
Non-ADHD |
||
|---|---|---|---|---|
Baseline (%) |
SCE (%) |
Baseline (%) |
SCE (%) |
|
≥ 1 s |
10.59 |
10.09 |
7.57 |
14.18 |
≥ 2 s |
1.18 |
2.75 |
0.99 |
4.23 |
≥ 3 s |
0.00 |
0.92 |
0.09 |
1.93 |
≥ 4 s |
0.00 |
0.92 |
0.02 |
1.00 |
≥ 5 s |
0.00 |
0.92 |
0.02 |
0.50 |
the non-ADHD group than in the ADHD group (see Table 11). The prevalence decreased monotonically as glance duration increased. In the ADHD group, the prevalence of glances greater than or equal to 3 and 4 seconds prior to baselines was zero due to unavailable data.
In the ADHD group, none of the ORs reached statistical significance, and total time EOR was not associated with CNC occurrence (see Table 12). The ORs for various durations were insignificant for both total EOR and single longest EOR. They did not increase as the duration of total EOR increased.
In the non-ADHD group, the odds of an SCE were 1.74 times higher when total EOR was greater than or equal to 1 second as compared to when it was less than 1 second. This increased monotonically with glance duration, such that the odds of an SCE were 5.63 times higher for total EOR duration greater than or equal to 5 seconds. As compared to the ADHD groups, the ORs of various durations were higher for both total EOR and single longest EOR in the non-ADHD group. The ORs of SCEs in various durations were significant for total EOR and single longest EOR.
In the non-ADHD group, ORs for single longest EOR were higher than those for total EOR (see Table 13). To further dive into these results, a comparison between the prevalence of total
|
Total EOR |
OR |
CI_Lower |
CI_Upper |
P_Value |
|---|---|---|---|---|---|
ADHD |
≥ 1 s |
1.10 |
0.50 |
2.43 |
0.811 |
≥ 2 s |
0.57 |
0.20 |
1.67 |
0.307 |
|
≥ 3 s |
0.52 |
0.11 |
2.43 |
0.409 |
|
≥ 4 s |
2.02 |
0.28 |
14.87 |
0.488 |
|
Non-ADHD |
≥ 1 s |
1.74 |
1.44 |
2.11 |
0.00 |
≥ 2 s |
2.42 |
1.89 |
3.09 |
< 0.01 |
|
≥ 3 s |
2.73 |
1.98 |
3.77 |
< 0.01 |
|
≥ 4 s |
2.92 |
1.87 |
4.54 |
< 0.01 |
|
≥ 5 s |
5.63 |
2.48 |
12.80 |
< 0.01 |
|
Single Longest EOR |
OR |
CI_Lower |
CI_Upper |
P_Value |
|---|---|---|---|---|---|
ADHD |
≥ 1 s |
0.72 |
0.30 |
1.76 |
0.47 |
≥ 2 s |
2.08 |
0.28 |
15.49 |
0.47 |
|
Non-ADHD |
≥ 1 s |
1.90 |
1.54 |
2.35 |
0.00 |
≥ 2 s |
4.15 |
2.64 |
6.51 |
0.00 |
|
≥ 3 s |
27.16 |
8.47 |
87.14 |
0.00 |
|
≥ 4 s |
55.04 |
6.08 |
497.90 |
0.0003 |
|
≥ 5 s |
22.66 |
2.23 |
229.74 |
0.008 |
time EOR for non-ADHD and ADHD teen drivers was performed. A chi-square test was conducted on the total EOR (< 2 s and ≥ 2 s) that showed no significant difference between those with and without ADHD (χ2 = 2.41, p = 0.12).
Comparison of SPDS NDS Teens with SHRP 2 NDS Teens
The analysis compared the proportion of total EOR greater than or equal to 2 seconds across different groups, categorized by age, ADHD status, and driving experience based on the SHRP 2 and SPDS datasets:
SHRP 2:
- Non-ADHD (16 to 17 years old)
- Non-ADHD (18 to 19 years old)
- ADHD (16 to 19 years old)
SPDS:
- First 6 months after licensure
The chi-squared test showed a significant overall difference in the proportion of total EOR greater than or equal to 2 seconds among these groups (χ2 = 23.44, p = 0.000278). Pairwise comparison with Bonferroni adjustment shows that the following pairs are significantly different regarding the proportion of total EOR ≥ 2 seconds:
- SHRP 2 non-ADHD (16 to 17) versus SHRP 2 non-ADHD (18 to 19) (p = 0.030)
- SHRP 2 non-ADHD (18 to 19) versus SPDS (first 6 months) (p = 0.027)
- SHRP 2 ADHD (16 to 19) versus SPDS (first 6 months) (p = 0.038)
Non-ADHD drivers aged 18 to 19 years in SHRP 2 exhibited a significantly different proportion of long EOR durations as compared to non-ADHD drivers aged 16 to 17 years in SHRP 2. Newly licensed teen drivers in SPDS had significantly different total EOR proportion as compared to those with ADHD or older non-ADHD drivers in SHRP 2.
