Effects of Human-Centered Factors on Crash Injury Severities
Table 2
Model estimation results.
Variable
MNL
RPL
LC 1
LC 2
Serious injury
Minor injury
Serious injury
Minor injury
Serious injury
Minor injury
Serious injury
Minor injury
Constant
−0.26 (−13.23)
−0.12 (−4.45)
0.69 (11.94)
−20.67 (−0.06)
Driver error
0.95 (50.05)
2.19 (7.61)
1.64 (23.24)
−17.71 (−0.05)
DUI
0.26 (8.99)
0.43 (9.43)
1.26 (8.40)
−19.83 (−0.06)
Speed
0.39 (14.93)
0.94 (4.13)
0.83 (11.60)
−16.92 (−0.05)
Distracted
1.26 (41.57)
6.10 (5.74)
2.48 (29.16)
−9.81 (−0.53)
Fatigue
0.14 (4.35)
0.02 (0.47)
−2.19 (−12.81)
47.11 (0.00)
Invalid license
0.10 (3.44)
0.13 (2.71)
0.09 (1.41)
0.35 (1.21)
No seatbelt
1.01 (41.70)
2.46 (8.71)
1.59 (21.19)
−1.64 (−3.52)
Female
0.17 (10.90)
0.26 (9.09)
0.26 (8.10)
−0.28 (−1.50)
Black
0.12 (6.77)
0.17 (5.89)
0.18 (5.29)
−0.04 (−0.21)
Young
0.23 (13.38)
0.55 (6.58)
0.43 (10.21)
−1.16 (−4.19)
Old
0.11 (4.77)
0.11 (3.42)
0.11 (2.11)
0.50 (1.34)
Unemployed
0.22 (12.74)
0.38 (11.56)
0.40 (10.85)
−0.60 (−2.55)
Latent class probability
0.72 (74.94)
0.28 (29.71)
Log-likelihood at zero
−60529.77
−60529.77
Log-likelihood at convergence
−49584.82
−49451.9
−48868.27
McFadden pseudo
0.069
0.183
0.193
BIC
98032
Halton draw of 200 was used for the RPL model [4]. The random parameters found in the RPL model (indicated by rp superscripts) were assumed to be normally distributed [see [5]] and had statistically significant standard deviations at 0.05 significance levels.