Table 1. Road Safety Development Index (RSDI) Factors (10).
RSDI components
|
Factors
|
Human performance
|
- Safer road user’s “behavior”
|
Product performance
|
- Percentage change of death trend
- Personal risk “death per population”
- Traffic risk “death per vehicle”
|
System performance
|
- Safer roads
- Safer vehicles
- Enforcement performance
- Organizational performance
- Socioeconomic performance
|
Table 2. RSDI components, Indices and data resources.
RSDI components
|
Selected factors
|
Indices
|
Data sources
|
Human performance
|
Safer road user’s “behavior”
|
Happiness
|
World happiness: Trends, explanations and distribution (2013) and (2016) (17, 19)
|
Homicides (per 100,000 people)
|
World Bank (18)
|
Product performance
|
Personal risk “death per population”
|
Mortality caused by road traffic injury (per 100,000 people)
|
Global status report on road safety 2015 and 2018 (2, 8)
|
System performance
|
Safer roads
|
|
Global status report on road safety 2015 and 2018 (2, 8)
|
Safer vehicles
|
Enforcement performance
|
Socioeconomic performance
|
HDI
|
UNDP (20)
|
Urban population (% of total)
|
World Bank (15)
|
GINI index (World Bank estimate)
|
World Bank (14)
|
Unemployment, total (% of total labor force) (modeled ILO estimate)
|
World Bank (16)
|
RSDI, Road safety development index, ILO International labour organization, HDI Human development index, UNDP, United Nations Development Programme
Table 3. Common descriptive statistics of the variables.
variable
|
Minimum
|
Maximum
|
Range
|
Mean
|
Standard Deviation
|
2013
|
2016
|
2013
|
2016
|
2013
|
2016
|
2013
|
2016
|
2013
|
2016
|
Mortality ratea
|
2.8
|
2.7
|
36.2
|
35.9
|
33.4
|
33.2
|
16.561
|
16.122
|
9.219
|
9.2045
|
HDI
|
.340
|
.351
|
.946
|
.951
|
.606
|
.600
|
.7152
|
.7252
|
.156
|
.157
|
GINI index
|
25.4
|
25.0
|
63.4
|
63.0
|
38.0
|
38.0
|
37.457
|
37.078
|
7.902
|
7.4880
|
Homicidesb
|
.183
|
2.905
|
74.28
|
7.526
|
74.096
|
82.559
|
6.489
|
5.351
|
9.944
|
1.1829
|
Happiness
|
2.936
|
.2835
|
7.693
|
82.842
|
4.757
|
4.621
|
5.460
|
6.217
|
1.105
|
10.684
|
Urban populationc
|
15.437
|
12.388
|
97.776
|
97.919
|
82.339
|
85.531
|
58.908
|
59.635
|
20.949
|
21.485
|
Unemploymentd
|
.3192
|
.524
|
28.996
|
26.55
|
28.677
|
26.027
|
8.425
|
7.629
|
6.410
|
5.747
|
Safer roads and mobility
|
0
|
.50
|
5
|
5.00
|
5
|
4.5
|
3.16
|
3.668
|
1.405
|
1.131
|
Safer vehicles
|
0
|
3.00
|
3
|
7.00
|
3
|
4
|
1.07
|
6.257
|
1.387
|
.998
|
safer road users
|
2
|
0.00
|
7
|
4.00
|
5
|
4
|
6.33
|
1.442
|
.915
|
1.817
|
Education
|
.204
|
.212
|
.941
|
.940
|
.737
|
.728
|
.66396
|
.67466
|
.178088
|
.181785
|
Income
|
.287
|
.287
|
.975
|
.984
|
.688
|
.697
|
.69570
|
.70135
|
.174642
|
.176655
|
Life expectancy
|
.468
|
.514
|
.975
|
.981
|
.507
|
.467
|
.80230
|
.81301
|
.125288
|
.117707
|
- road traffic injury per 100,000 people
- Homicides per 100,000 people
- % of total
- % of total labor force
HDI Human development index
Table 4. Frequency of countries based on Human Development analytical category.
|
2013
|
2016
|
Frequency
|
Percent
|
Mean of mortality rate
|
Minimum mortality rate
|
Maximum mortality rate
|
Frequency
|
Percent
|
Mean of mortality rate
|
Minimum mortality rate
|
Maximum mortality rate
|
Very Higha
|
39
|
33.9
|
6.751
|
2.8
|
13.7
|
42
|
37.2
|
7.000
|
2.7
|
18.0
|
Highb
|
28
|
24.3
|
18.614
|
7.7
|
36.2
|
25
|
22.1
|
18.064
|
6.4
|
34.6
|
Mediumc
|
26
|
22.6
|
19.908
|
10.5
|
29.1
|
25
|
22.1
|
19.628
|
9.7
|
30.4
|
Lowd
|
22
|
19.1
|
27.382
|
14.2
|
35.0
|
21
|
18.6
|
27.881
|
21.4
|
35.9
|
Total
|
115
|
100.0
|
16.561
|
2.8
|
36.2
|
113
|
100.0
|
16.122
|
2.7
|
35.9
|
- Very High: HDI>=0.8
- High: 0.7<=HDI<0.8
- Medium: 0.55<=HDI<0.7
- Low: HDI<0.55
HDI Human development index
Table 5. Frequency of countries based on income analytical category.
|
2013
|
2016
|
Frequency
|
Percent
|
Mean of mortality rate
|
Minimum mortality rate
|
Maximum mortality rate
|
Frequency
|
Percent
|
Mean of mortality rate
|
Minimum mortality rate
|
Maximum mortality rate
|
Very Higha
|
38
|
33.0
|
7.079
|
2.8
|
23.4
|
34
|
30.1
|
6.003
|
2.7
|
13.4
|
Upper‐ Middleb
|
31
|
27.0
|
18.387
|
7.7
|
36.2
|
30
|
26.5
|
17.220
|
6.4
|
34.6
|
Lower‐ Middlec
|
30
|
26.1
|
20.813
|
10.5
|
29.1
|
34
|
30.1
|
19.874
|
9.7
|
30.1
|
Lowd
|
16
|
13.9
|
27.569
|
13.6
|
35.0
|
15
|
13.3
|
28.360
|
15.9
|
35.9
|
Total
|
115
|
100.0
|
16.561
|
2.8
|
36.2
|
113
|
100.0
|
16.122
|
2.7
|
35.9
|
- Very High: GNI>= $12,736
- Upper‐ Middle: $4,125<=GNI<$12,736
- Lower‐ Middle: $1,045<=GNI<$4,125
- Low: GNI<$1,045
GNI Gross National Income
Table 6. Stepwise Multivariate Linear Regression: model summary (2013).
Model
|
R
|
R square
|
Adjusted R square
|
R square change
|
1
|
.804a
|
.646
|
.643
|
.646
|
2
|
.832b
|
.693
|
.688
|
.047
|
3
|
.849c
|
.720
|
.713
|
.027
|
4
|
.857d
|
.734
|
.724
|
.013
|
5
|
.864e
|
.746
|
.734
|
.012
|
a. Predictors: Constant, income 2013
|
b. Predictors: Constant, income 2013, safer vehicles 2013
|
c. Predictors: Constant, income 2013, safer vehicles 2013, GINI index 2013
|
d. Predictors: Constant, income 2013, safer vehicles 2013, GINI index 2013, life expectancy 2013
|
e. Predictors: Constant, income 2013, safer vehicles 2013, GINI index 2013, life expectancy 2013, safeuser2013
|
Table 7. Stepwise Multivariate Linear Regression: model summary (2016).
Model
|
R
|
R square
|
Adjusted R square
|
R square change
|
1
|
.821a
|
.674
|
.671
|
.674
|
2
|
.877b
|
.769
|
.765
|
.095
|
3
|
.885c
|
.784
|
.778
|
.015
|
a. Predictors: Constant, income 2016
|
b. Predictors: Constant, income 2016, GINI index 2016
|
c. Predictors: Constant, income 2016, GINI index 2016, life expectancy 2016
|
Table 8. Basis functions of the MARS and their coefficients (2013).
Variables
|
Basis Function
|
coefficients
|
(Intercept)
|
|
16.99
|
BF1
|
safer vehicles 2013
|
-1.2
|
BF2
|
h(urban population 2013 - 38.979)
|
0.09
|
BF3
|
h(education 2013 - 0.583)
|
191.51209
|
BF4
|
h(education 2013 - 0.623)
|
-437.70114
|
BF5
|
h(education 2013 - 0.654)
|
225.58727
|
BF6
|
h(income 2013 - 0.59)
|
47.72427
|
BF7
|
h(0.745 - income 2013)
|
36.05110
|
BF8
|
h(income 2013 - 0.745)
|
-66.43674
|
BF9
|
h(Life Expectancy 2013 - 0.613)
|
-36.36855
|
BF Basis function, h Hinge function
Table 9. Basis functions of the MARS and their coefficients (2016).
Variables
|
Basis Function
|
coefficients
|
(Intercept)
|
|
20.9
|
BF1
|
h (45 - GINI index 2016)
|
-0.29
|
BF2
|
h(5.121 - happiness 2016)
|
3.68
|
BF3
|
h(education 2016 - 0.631)
|
-30.82
|
BF4
|
h(0.549 – income 2016)
|
29.53
|
BF5
|
h(life expectancy 2016 - 0.865)
|
-52.24
|
BF Basis function, h Hinge function
Table 10. Countries with more than 30% change in RTF with correspondent growth rate in GINI index and HDI.
Country
|
RTF rate
2013
|
RTF rate
2016
|
RTF growth
rate
|
GINI index 2013
|
GINI index 2016
|
GINI index growth
rate
|
HDI
2013
|
HDI
2016
|
HDI
growth
Rate
|
Iran
|
32.1
|
20.5
|
-36.14
|
37.4
|
40.0
|
6.95
|
0.784
|
0.796
|
1.53
|
Belarus
|
13.7
|
8.9
|
-35.04
|
26.6
|
25.3
|
-4.89
|
0.804
|
0.805
|
0.12
|
Bolivia
|
23.2
|
15.5
|
-33.19
|
47.6
|
44.6
|
-6.30
|
0.668
|
0.689
|
3.14
|
Macedonia
|
9.4
|
6.4
|
-31.91
|
36.2
|
35.6
|
-1.66
|
0.743
|
0.756
|
1.75
|
Kyrgyzstan
|
22.0
|
15.4
|
-30.00
|
28.8
|
26.8
|
-6.94
|
0.658
|
0.669
|
1.67
|
India
|
16.6
|
22.6
|
36.14
|
35.7
|
35.7
|
0.00
|
0.607
|
0.636
|
4.78
|
Turkey
|
8.9
|
12.3
|
38.20
|
40.2
|
41.9
|
4.23
|
0.771
|
0.787
|
2.07
|
Iceland
|
4.6
|
6.6
|
43.48
|
25.4
|
27.8
|
9.45
|
0.920
|
0.933
|
1.41
|
RTF Road traffic fatalities, HDI Human development index
Table 11. Prediction performance measures of the models.
Model
|
r
|
RMSE
|
MAE
|
RAE
|
R2
|
2013
|
2016
|
2013
|
2016
|
2013
|
2016
|
2013
|
2016
|
2013
|
2016
|
SMLR
|
0.86
|
0.88
|
4.62
|
4.27
|
3.21
|
3.11
|
0.40
|
0.39
|
0.75
|
0.78
|
CART
|
0.91
|
0.91
|
3.80
|
3.71
|
2.75
|
2.71
|
0.34
|
0.34
|
0.82
|
0.84
|
MARS
|
0.90
|
0.90
|
4.02
|
3.90
|
2.93
|
2.80
|
0.36
|
0.35
|
0.81
|
0.82
|
r correlation coefficient, RMSE Root mean squared error, MAE Mean absolute error, RAE Relative absolute error, SMLR Stepwise multivariate linear regression, CART Classification and regression trees, MARS Multivariate adaptive regression splines
Table 12. Importance of variables included in the CART and MARS model.
Variable
|
Importance in CART
|
Importance in MARS
|
2013
|
2016
|
2013
|
2016
|
Education
|
21
|
21
|
100
|
100
|
Income
|
19
|
18
|
19.6
|
10.6
|
Life expectancy
|
17
|
17
|
38.7
|
15
|
Safer vehicles
|
14
|
13
|
5
|
unused
|
Happiness
|
12
|
5
|
unused
|
38.5
|
Homicide
|
10
|
12
|
unused
|
unused
|
Urban population
|
3
|
2
|
16.5
|
unused
|
Unemployment
|
2
|
1
|
unused
|
unused
|
GINI index
|
1
|
unused
|
unused
|
19.5
|
Safer road users
|
unused
|
11
|
unused
|
unused
|
Safer roads and mobility
|
unused
|
unused
|
unused
|
unused
|
CART Classification and regression trees, MARS Multivariate adaptive regression splines