Skip to main content

An Open Access Journal

Table 3 Selected models for intersections

From: The variability of urban safety performance functions for different road elements: an Italian case study

Predictor

Estimate

Std. Error

z- value

p-value

All intersections

 (Intercept)

0.930

1.524

0.610

0.542

Ln (Total AADT)

0.285

0.106

2.675

0.007

Main AADT/Total AADT

−6.870

2.021

−3.399

0.001

Type: Four-legged (reference: Three-legged)

0.443

0.105

4.210

<  0.001

Critical sight distance

0.005

0.002

2.762

0.006

Specialized turning lane (Yes – ref.: no)

0.496

0.114

4.362

<  0.001

Poor maintenance of pavements (Yes - ref.: No)

−0.397

0.115

−3.444

<  0.001

Cycle path crossing (Yes – ref.: No)

0.327

0.122

2.677

0.007

Goodness-of-fit measures

 LRT (reference: null model) = 5.63 (df = 12), p = .934; LRT (reference: full model) = 161.11 (df = 12), p < .001

 Nagelkerke R2 = 0.283, AIC = 1877.6

 Over-dispersion parameter (theta) = 2.84 (std. error: 0.52)

Interval of continuous predictors

 Total AADT: 4175–41,560 vehicles/day, Main/Total AADT: 0.50–0.62, Critical sight distance: 0.0–139.5 m

Three-legged intersections

 (Intercept)

4.696

2.895

1.609

0.108

Ln (Total AADT)

0.366

0.220

1.661

0.097

Main AADT/Total AADT

−16.376

3.711

−4.413

< 0.001

Number of entering lanes on the main road

−0.211

0.119

−1.768

0.077

Critical sight distance

0.006

0.003

2.009

< 0.001

Specialized turning lane (Yes – ref.: no)

0.927

0.228

4.061

< 0.001

Traffic control 1: give-way/stop (ref.: no control)a

0.359

0.214

1.675

0.094

 Traffic control 2: traffic lights (ref.: no control)a

0.377

0.289

1.306

0.192

Traffic control 3: main + turning lights (ref.: no control)a

0.556

0.313

1.777

0.076

Cycle path crossing (Yes – ref.: No)

0.900

0.235

3.830

< 0.001

Goodness-of-fit measures

 LRT (reference: null model) = 63.39 (df = 9), p < .001; LRT (reference: full model) = 5.79 (df = 9), p = .761

 Nagelkerke R2 = 0.292, AIC = 610.4

 Over-dispersion parameter (theta) = 5.26 (std. error: 2.75)

Interval of continuous predictors

 Total AADT: 5420–32,300 vehicles/day, Main /Total AADT: 0.50–0.62, Critical sight distance: 0.0–139.5 m, Number of entering lanes on the main road: 0–4

Four-legged intersections

 (Intercept)

−0.127

1.623

−0.078

0.938

Ln (Total AADT)

0.372

0.108

3.450

0.006

Main AADT/Total AADT

−5.427

2.624

−2.068

0.039

Critical sight distance

0.004

0.002

2.102

0.036

Specialized turning lane (Yes – ref.: no)

0.410

0.134

3.060

0.002

Poor maintenance of pavements (Yes - ref.: No)

−0.446

0.151

−2.961

0.003

 LRT (reference: null model) = 77.98 (df = 5), p < .001; LRT (reference: full model) = 7.55 (df = 12), p = .819

 Nagelkerke R2 = 0.225, AIC = 1263.6

 Over-dispersion parameter (theta) = 2.79 (std. error: 0.59)

Interval of continuous predictors

 Total AADT: 4175–41,560 vehicles/day, Main /Total AADT: 0.50–0.61, Critical sight distance: 1.6–100.0 m

All intersections (alternative)

 (Intercept)

1.713

1.513

1.132

0.257

Ln (Total AADT)

0.371

0.103

3.613

<  0.001

Main AADT/Total AADT

−9.502

2.161

−4.397

<  0.001

Median on the main road (Yes – reference: No)

0.226

0.117

1.930

0.054

Critical sight distance

0.004

0.002

2.359

0.018

Type of intersection (Signalized – ref.: Unsignalized)

0.268

0.104

2.580

0.010

Poor maintenance of pavements (Yes - ref.: No)

−0.321

0.118

−2.730

0.006

Cycle path crossing (Yes - ref.: No)

0.253

0.123

2.058

0.040

Goodness-of-fit measures

 LRT (reference: null model) = 124.59 (df = 7), p < .001; LRT (reference: full model) = 8.29 (df = 8), p = .406

 Nagelkerke R2 = 0.222, AIC = 1914.1

 Over-dispersion parameter (theta) = 2.40 (std. error: 0.41)

Interval of continuous predictors

 Total AADT: 4175–41,560 vehicles/day, Main/Total AADT: 0.50–0.62, Critical sight distance: 0.0–139.5 m

Signalized intersections

 (Intercept)

2.390

2.293

1.043

0.297

Ln (Total AADT)

0.407

0.142

2.863

0.004

Main AADT/Total AADT

−11.302

3.586

−3.152

0.002

Type: Four-legged (reference: Three-legged)

0.452

0.182

2.479

0.013

Critical sight distance

0.004

0.002

1.986

0.047

Specialized turning lane (Yes – ref.: no)

0.307

0.148

2.076

0.038

Poor maintenance of pavements (Yes - ref.: No)

−0.630

0.148

−4.243

<  0.001

Presence of bus stops (Yes – ref.: No)

−0.914

0.357

−2.559

0.010

Goodness-of-fit measures

 LRT (reference: null model) = 95.60 (df = 7), p <  0.001; LRT (reference: full model) = 4.39 (df = 9), p = 0.884

 Nagelkerke R2 = 0.320, AIC = 1078.2

 Over-dispersion parameter (theta) = 3.08 (std. error: 0.71)

Interval of continuous predictors

 Total AADT: 5800–41,560 vehicles/day, Main/Total AADT: 0.50–0.61, Critical sight distance: 1.6–139.5 m

Unsignalized intersections

 (Intercept)

−2.604

2.185

−1.192

0.233

Ln (Total AADT)

0.759

0.163

6.343

<  0.001

Main AADT/Total AADT

−7.868

0.173

4.389

0.001

Type: Four-legged (reference: Three-legged)

1.036

2.406

−3.270

<  0.001

Median on the main road (Yes – reference: No)

−0.998

0.329

−3.031

0.002

Specialized turning lane (Yes – ref.: No)

0.516

0.204

2.523

0.012

Cycle path crossing (Yes - ref.: No)

0.997

0.193

5.156

< 0.001

Presence of sidewalks (Yes - ref.: No)

−1.008

0.331

−3.042

0.002

 LRT (reference: null model) = 65.07 (df = 7), p < .001; LRT (reference: full model) = 11.96 (df = 11), p = .367

 Nagelkerke R2 = 0.238, AIC = 769.8

 Over-dispersion parameter (theta) = 8.11 (std. error: 6.14)

Interval of continuous predictors

 Total AADT: 4175–31,500 vehicles/day, Main/Total AADT: 0.50–0.62, Critical sight distance: 0.0–100.0 m

  1. aNote: The variable traffic control device was further disaggregated into: no control, give-way, stop control for unsignalized intersections, and into: traffic lights, traffic lights with dedicated left turning lights, traffic lights with dedicated right turning lights for signalized intersections