Page 40 - tmp_Neat

3.3. Results and Discussions

The results of the model are summarized in Table 3.2. In general, the model fits the data well,
based on the very large chi-square statistic, the very small p-value for goodness of fit, and the

relatively high McFadden pseudo R-square value. In addition to the variables shown in Table
3.2, several other variables were included in the preliminary analyses but were found to be

statistically insignificant. The insignificant variables were restraint use, license state, age and

gender of the driver, road surface and weather conditions, season, time of day, day of week,
crash severity, police attendance at crash scene, number of people involved in the collision,

light conditions and type of truck involved. Therefore, these factors were not associated with
the location (intersection or mid-block) of a single heavy-vehicle crash, although they may

have a significant effect on the frequency or severity of heavy-vehicle crashes.

Table 3.2: Estimates of binary logit model for intersection and mid-block crashes

Dependent variable: y = 1 for intersection and y= 0 for mid-block

Number of observations: 566

Log likelihood: -278.2

Restricted log likelihood: -383.1

Chi-square statistic: 209.84

Significance level: <0.00001

McFadden pseudo R-squared: 0.2738

Variable Coefficient Std Err p-value OR

Vehicle manoeuver

Turn left 2.003*** 0.366 0.000 7.409

Turn right 2.137*** 0.396 0.000 8.474

Collision classification

Angle crash 0.826*** 0.243 0.001 2.283

Type of collision

29