The four models developed are presented in the following sections. The presentation progresses from roads with higher speed limits (motorways and main roads) to roads with lower speed limits (country and urban roads). Numbers connecting the various boxes of the diagrams stand for the z-values of each parameter (parameter is a part of the log-linear equation). All of the variables were recoded and they contain two values (1, 2), except from the age groups, which contain three values (1, 2, 3). Age groups were categorized into three groups, so in the models where this variable appears, there are two z-values standing for the respective parameters.
Motorways
For those roads with the highest speed limits, a general model for exceeding the speed limit was developed, which includes the following set of variables.
Variables in the model:
Figure 1 shows the five main effects on exceeding speed limits on motorways and the five interactions between independent variables, which are also relevant to exceeding speed limits on motorways.
The model achieves a considerable fit with significance of 0,940, which indicates that this model describes the large majority of the sample.
Direct relations between independent variables and exceeding speed limits on motorways: Results indicates the following:
-
Drivers, who declare that they enjoy driving fast, exceed speed limits on motorways.
-
Drivers with an annual kilometrage of more than 15.000 km exceed speed limits on motorways frequently.
-
The vehicle’s engine capacity is related to exceeding speed limits on motorways. Drivers whose vehicle’s engine capacity is more than 1,300 cc exceed speed limits on motorways more frequently than those whose vehicle’s engine capacity is less than 1,300 cc.
-
Male drivers exceed speed limits on motorways more often than female drivers.
-
Drivers, who believe that other drivers exceed speed limits, exceed speed limits themselves on motorways often.
Interactions between several independent variables and exceeding speed limits on motorways: Interactions between independent variables (ranked by importance) and their relation to exceeding speed limits on motorways are to be understood as follows:
-
Drivers who enjoy driving fast, never or sometimes believe that other drivers exceed speed limits.
-
Male drivers, who believe that other drivers exceed speed limits, drive cars with low engine capacity.
-
Drivers of cars with low engine capacity and with an annual kilometrage of less than 15.000 km, never or rarely exceed speed limits on motorways.
Main roads
Similarly to the analysis for exceeding speed limits on motorways, the model for exceeding the speed limits on main roads includes the following set of variables.
Variables in the model:
Comparing to the model for exceeding speed limits on motorways, it can be observed that the age of the drivers is relevant to their self-reported behaviour towards exceeding speed limits on main roads but it is not relevant to exceeding speed limits on motorways. It is worth noting that driver gender has not a significant direct relation to the dependent variable (self-reported exceeding speed limits on main roads), but its significant interactions with the annual kilometrage and the age groups provide a more detailed relation of those independent variables with exceeding speed limits on main roads.
Figure 2 shows the five main effects on exceeding speed limits on main roads and the five interactions between independent variables, which are also relevant for exceeding speed limits on motorways.
The model attains a considerable fit with significance of 0,844.
Direct relations between independent variables and exceeding speed limits on main roads: Results here indicate that:
-
Younger drivers exceed speed limits on main roads more often.
-
Drivers with an annual kilometrage of more than 15.000 km exceed speed limits on main roads more often.
-
Drivers, whose vehicle’s engine capacity is more than 1,300 cc, exceed speed limits on main roads more often.
-
Drivers, who believe that other drivers exceed speed limits, exceed speed limits themselves more often on main roads.
Interactions between several independent variables and exceeding speed limits on main roads: Interactions between independent variables (ranked by importance) and their relationship to exceeding speed limits on main roads are to be understood as follows:
-
Drivers, who believe that other drivers exceed speed limits, are younger drivers with an annual kilometrage of less than 15.000 km.
-
Younger female drivers drive less than 15.000 km annually.
Country roads
Similarly to the analysis for exceeding speed limits on motorways and main roads, the model for exceeding the speed limits on country roads includes the following set of variables.
Variables in the model:
The final model has very good GoF-statistics and it includes variables associated with the drivers’ self-reported speed behaviour and their experience in sanctions for speeding.
The interactions between the independent variables are complicated. Experience in sanctions for speeding, as well as the expectation of speed enforcement, is included in the model giving high goodness of fit statistics. Therefore there is no significant direct relationship between expectation of speed enforcement and exceeding speed limits on country roads.
Figure 3 shows the four main effects of exceeding speed limits on country roads and the six interactions between independent variables.
The model attains a considerable fit with significance of 0,932, which represents a high fidelity model.
Direct relations between independent variables and exceeding speed limits on country roads: Results here show that:
-
Drivers, who have been fined for exceeding speed limits on country roads, are more likely to exceed speed limits again on this type of roads.
-
Younger drivers exceed speed limits on country roads more frequently.
-
Drivers, who signal others for police traps, exceed speed limits on country roads more frequently.
-
Drivers of vehicles with an engine capacity of more than 1,300 cc exceed speed limits on country roads more often.
Interactions between several independent variables and exceeding speed limits on country roads: Interactions between independent variables (ranked by importance) and their relation to exceeding speed limits on country roads are to be understood as follows:
-
The fact that younger drivers exceed the speed limit on country roads is more pronounced in the group of younger drivers who believe that other drivers often or always exceed speed limits.
-
Younger drivers, who signal other drivers for police traps, are more likely to exceed speed limits on country roads than the rest of the drivers in the group.
-
Regarding the self-reported penalty for exceeding speed limits, drivers who have been fined before, usually signal other drivers for police traps.
-
Although drivers of vehicles with high engine capacity expect speed enforcement, they continue to exceed speed limits on country roads.
-
Drivers of cars with high engine capacity, who believe that other drivers exceed speed limits, are more likely to exceed speed limits on country roads.
Built-up area roads
Finally, regarding exceeding speed limits in built up areas, the same procedure was followed and the final model includes the following set of variables.
Variables in the model:
The final model has very good GoF-statistics and it includes variables regarding the drivers’ self-reported speed behaviour and their experience in sanctions for speeding.
It is important to note that the variable of wishing for different speed limits in built up areas is included in the model while the respective variable did not show to be good in models concerning the other types of road.
Another issue to consider is that the opinion that other drivers exceed speed limits and the expectation of speed enforcement are not directly related to exceeding speed limits, in built up areas, but have interactions with other variables as shown in this section.
Figure 4 shows the four main effects of exceeding speed limits in built up areas and the five interactions between independent variables. The model attains a considerable fit with significance of 0,841, which means a very high quality of model.
Direct relations between independent variables and exceeding speed limits in built up areas: The direct relationship between independent variables and exceeding speed limits in built up areas ranked by importance are given as follows.
-
People who wished for higher limits in built up areas, exceed speed limits from sometimes to always in those areas.
-
As found for the other types of roads, younger drivers exceed speed limits more often than the other age groups, in built up areas too.
-
Regarding the drivers’ experience in penalties for exceeding speed limits, people who have been fined, insist in exceeding speed limits in built up areas.
-
Finally, people who signal others for police controls, exceed speed limits themselves.
Interactions between several independent variables and exceeding speed limits in built up areas: Interactions between independent variables (ranked by importance) and their relation to exceeding speed limits in built up areas are to be understood as follows:
-
The finding that younger drivers exceed the speed limit in built up areas is more pronounced in the group of younger drivers who signal other drivers for police traps.
-
In addition, people who wished for higher speed limits and exceed the existing limits, usually warn other drivers for police traps.
-
Drivers, who exceed the speed limits in built up areas and also warn other drivers for police controls, are more pronounced to speed if they also believe that other drivers exceed speed limits, too. However, this belief is not directly related to exceeding speed limits in built up areas.
-
Younger drivers, who do not expect speed enforcement, are more likely to exceed speed limits in built up areas.
-
The same appears to happen with older drivers who have not been fined for speeding.
Validity of the exceeding speed limits model for different types of road with respect to the different countries
It is important to note that, in spite of this general model, there is still a range of differences concerning exceeding speed limits in different types of road among the countries, which have participated in the survey. Applying the method of log-linear modelling to samples of countries individually tended, in some cases, to yield different results for each type of road. This means that, in some countries, specific circumstances tended to change the model in the sense that it would be improved according to the special conditions in that given country. These conditions (traffic density and composition, pedestrian traffic density, level of enforcement, etc.) are specifically related to certain distribution patterns of the variables, which have been included in the models.
For example, Fig. 5 shows on the y-axis that the percentage of people who say that they exceed the speed limits on motorways ranges from about 51 % in Hungary to about 60 % in Denmark and Estonia; the average in the total sample being about 56 %. Those percentages concern drivers who fulfil the conditions described by the log-linear model. More specifically, the parameters that affect the proportion of drivers exceeding the speed limits on motorways in this model are the annual kilometrage, the engine capacity, the drivers’ gender, and whether they enjoy driving fast and/or believe that other drivers exceed speed limits. The results of the principal analysis indicated that the differences among the participating countries tend to be associated with the above-mentioned variables. On the x-axis the residuals of the independent variables for the Netherlands, Portugal, Sweden, Croatia, Finland, Greece, Switzerland and United Kingdom are quite close to the total, while Estonia, Hungary, Denmark, Germany and Cyprus have the greatest distance.
The closer the dot of the country to the “Total” the better the described model will fit to the data of that country. Regarding motorways, it is noted that models specific to the countries Estonia, Italy and Cyprus differ from the described model.
Similarly to the above, Table 3 shows the average, the minimum and maximum percentage of drivers exceeding speed limits on different types of road accordingly to their country. It also shows the countries with the best and the worst fitting to the model each time.
Table 3 Distance from average distribution from set of independent variables and from distribution of dependent variable The proportions in Table 3 refer to drivers who fulfil the conditions described in the log-linear model. So, in a principal analysis within the whole sample of the participant countries, the proportions of drivers exceeding speed limits on different types of road might be different.