Scottish Government

THE SPEEDING DRIVER: WHO, HOW AND WHY?

CHAPTER FIVE POLICE RECORDS OF DRIVERS REPORTED FOR SPEEDING OFFENCES

INTRODUCTION

5.1 In 1996 the Scottish Office Development Department commissioned research to establish the personal profiles of those groups of people most likely to commit speeding offences on Scottish Roads. The research involved analysing over 2000 actual police records of speeding offences. The work was carried out and reported by Colin Buchanan and Partners (1996). Part of the present study involves carrying out an analysis of police records of speeding offences in 2001. This Chapter presents the results of this analysis.

5.2 Recent research (Taylor et al, 2000) has shown that a small reduction in speed can result in marked reductions in accidents, particularly in urban areas. In order to produce a reduction in speed it is necessary to use a number of countermeasures including Engineering, Education and Enforcement (commonly known as the three 'Es'). To enable limited resources to be effectively targeted, it is important to understand which types of driver are likely to speed and to be caught speeding. This profile can then inform suitable countermeasures.

5.3 This Chapter reviews the results of the study undertaken in 1996, comparing them with the results of other research undertaken within this area, describes the methodology used here, presents the results of the analysis of police records of speeding offences for 2001 from a sample of Scottish police forces, and compares the findings with those of the 1996 study.

RESULTS OF PREVIOUS RESEARCH

The Speeding Driver 1996

5.4 Colin Buchanan and Partners (1996) carried out an initial study into the profile of speeding drivers in Scotland. The main conclusions of the study were:

• A male is twice as likely to commit a minor speeding offence than a female.
• A male is four times as likely to commit a more serious speeding offence than a female.
• Males are more prone to speeding than females regardless of road type, with the most marked difference being on rural roads.
• Males are almost twice as likely to speed as females during leisure periods, and are even more likely to speed during peak periods and working hours.
• The likelihood of an individual committing a minor speeding offence is inversely related to age, with the exception of the youngest age group (16 to 20 years), who are less likely to speed overall than 21 to 29 year olds.
• The 21 to 25 age range is the most prone to speeding on all roads with the exception of motorways where the 26 to 29 age range are the worst offenders.
• The 21 to 25 age range has the greatest propensity to speed during all periods except during daytime hours at weekends when the 26 to 29 age range are more likely to speed.
• The 16 to 20 age group displays the greatest variation in propensity to speed by time period, with evenings and nights the most likely period for them to commit an offence.
• About 40 years old is the transition age when drivers become "less likely" to speed.
• Non-manual workers display the greatest propensity to commit speeding offences and manual-unskilled the least of the work force, regardless of severity.
• Employees with "high mileage" occupations are more prone to speeding than others.
• Retired persons show the least inclination to speed of all occupations.

Other Research

5.5 The results of the 1996 study, and other relevant research, were brought together by Webster and Wells (2000). Their work indicated that many different types of people are speeders and a majority of drivers admit to speeding at some time. With regard to age and sex, one study (Quimby et al, 1999) indicated that age remained a strong predictor of relative speed even when other variables were included in the analysis. This study found that the overall difference between the sexes was not statistically significant. However, another similar study (Maycock et al, 1998) found that age and sex were both important variables in distinguishing speeders.

5.6 Quimby et al (1999) showed that occupational group was a predictor of speeding behaviour. Senior managers drove 1.4% faster than junior managers or manual workers and 2.8% faster than housewives/husbands, students or the unemployed. In addition Maycock et al (1998) found that managerial drivers were the fastest drivers and retired drivers were the slowest. Company car drivers were 2.7% faster than private car drivers. It was also found that low mileage drivers (under 5,000 miles/year) were more likely to be in the slow speed band and high mileage drivers (25,000 miles/year) were more likely to be found in the fast speed band.

Methodology

5.7 As one of the main purposes of this study is to determine whether the profile of the speeding driver has changed since the 1996 study, it was decided to use a similar methodology for this study, that is extracting data from police records of drivers reported for speeding offences.

Police Records

5.8 Police Records of Speeding offences are categorised as follows:

5.9 For the 1996 study, data were collected from the Strathclyde, Grampian, and Dumfries and Galloway police forces. The data set used for the 1996 study is shown in Table 5.1.

Table 5.1: Data set for 1996 study

5.10 The data set for the current study is shown in Table 5.2.

Table 5.2: Data set for current study

5.11 For the current study, the same three police forces were asked to supply a total of 2000 records for analysis: 200 Police Reports from each force and the remaining 1400 records split between the forces according to level of traffic within each area. In total 1885 records were supplied (Dumfries and Galloway were not in a position to supply as many records as originally requested). Although, the number of records supplied was slightly less than envisaged, it was of a level comparable with the 1996 study.

5.12 The data from the three forces was collated into a single database for analysis.

Dealing with Possible bias In the Sample

5.13 By taking a sample of reports, there is always the possibility that the sample could be biased in someway. For example, there is no guarantee that police enforcement within an area will be distributed in the same way as traffic is distributed.

5.14 Another factor, which needs to be accounted for, is different levels of usage by different driver groups. For example, analysis of the data might indicate that two-thirds of the offenders were male, implying that males are more likely to offend than females. However, this result is meaningless unless the overall proportion of male and female drivers is also known.

5.15 The 1996 study used National Travel Survey data from 1992 to 94 to provide a control for the speeding data obtained from the police records. The National Travel Survey data were used as a basis for estimating the proportion of males and females using the network at different times and the proportion of males and females on different road types. In building the model a number of assumptions were made. The most significant of these was that trip length is a proxy for the road type used, will all trips using a minor road first up to a maximum distance, and a progressively higher class of road being used as the trip distance increases.

5.16 Whilst it is recognised that there are no lengths of motorway in the Grampian Police area the proportion of journeys in excess of 50 miles is relatively small and the model structure was driven by the need to provide robust comparison with previous studies.

5.17 For the current study consideration was made to creating a new model using data from the most recent National Travel Survey which covers the period 1998 to 2000. The main difficulty in doing this was that there was no way of knowing what assumptions were made when the first model was built. Therefore, there was no way of ensuring that the new model would be sufficiently similar to the old model to enable a fair comparison to be made.

5.18 The decision was therefore taken to use the model produced for the 1996 study. This would allow direct comparison of the results from both studies. This approach is acceptable as long as driver profiles have not changed significantly between the two National Travel Surveys.

5.19 Table 5.3 compares the relevant data from both surveys.

Table 5.3: Comparison of results of National Travel Surveys 1992-94 and 1998-2000

For 1992 - 94 the average distance travelled represents only the distance travelled as a car driver by adults. The male and female figures relate only to adults.

5.20 The data shown in Table 5.3 indicate that for the purposes of this study, the profile of car drivers has not changed significantly between the two surveys. For example, the percentage of trips of different length being undertaken has remained virtually the same. The proportion of males and females undertaking car trips (as driver) is also similar.

5.21 The above data seem to indicate that there will be no difficulty in using the model derived for the 1996 study. However, it should be noted that there is no way of verifying that the assumptions made about the proportion of male and female drivers on different types of road at different times of the day are correct.

5.22 For this reason a second set of analyses were carried out based on data presented in the most recent travel survey which covers data collected between 1998 and 2000. Using these data alone, it was possible to workout that, for example, the overall proportion of males on the road is 71%. These data were then used as a basis for analysing the police records on the assumption that the proportion of males and females on the road did not vary with respect to road type, driver sex or driver age.

5.23 The National Travel Survey covers travel undertaken within the UK as a whole, and therefore may not accurately reflect travel patterns in Scotland. However, in April 2002 the Scottish Executive released a statistical bulletin which presented data about travel within Great Britain by Scottish residents extracted from the 1998 to 2000 National Travel Survey (Scottish Executive 2002). As this bulletin better represents travel patterns in Scotland, it was used as the primary source of information for these analyses. However, it should be noted that it was necessary to use national data to complete the picture. For example, the bulletin did not present data by time of day.

Analysis

5.24 As noted, the decision was taken to carry out two sets of analyses. The first compares the data in the 2002 police reports with control data derived from the National Travel Survey conducted in the period 1992 to 1994 as used in the 1996 report. The second compares the police records with control data taken directly from the Scottish elements of the National Travel Survey conducted in 1998 to 2000. For brevity, the former is referred to as '1996 model' and the later 'Scottish NTS'.

Effects of sex

5.25 A total of 1885 offence records were used in this data sample. Of these 422 (22%) related to females and 1463 (78%) related to males.

Severity of offence

5.26 Six sets of results have been calculated to show the effect of gender on propensity to speed. In this report the propensity to speed is defined as the observed number of offenders in a particular category divided by the expected number of offenders in that category. Therefore, if the observed number of offenders equals the expected number the propensity equals 1.

5.27 The result categories are shown in Table 5.4. The corresponding tables are given in Appendix B.

Table 5.4: Severity of offence result categories

5.28 The results are presented with respect to the 1996 model and the Scottish NTS and have been divided into All Vehicles and Cars Only. (The NTS only covers 'personal' travel and the figures for miles driven do not include the use of motorbikes. Therefore, figures based only on offences committed by car drivers were also calculated.) Each set of results includes tables based on each of Conditional Offers (COs), Police Reports (PRs) and All Offences.

5.29 The 1996 model showed no significant difference in the number of COs per mile driven received by men and women. (From Table 1a, Appendix B, it can be seen that the observed number of male offenders was 1037, where as the expected number was 1015. Therefore, the propensity to speed is 1037 divided by 1015, which equals 1.02.)

5.30 This was true for offences committed by all drivers as well as only those committed by car drivers (see Figure 5.1).

Figure 5.1: Expected and Observed numbers of Conditional offers using the 1996 model.

5.31 The picture is somewhat different when Police Reports are considered. In absolute terms, male drivers received about seven times as many PRs as female drivers, whereas it would be expected that males would only receive 2.5 times as many if their propensity to speed were equal (Figure 5.2).

Figure 5.2: Expected and Observed numbers of Police Reports using the 1996 model

5.32 Overall male drivers were 34% more likely to be charged with speeding than females per mile driven. Male car drivers were 18% more likely to be charged than females per mile driven (Figure 5.3).

Figure 5.3: Expected and Observed numbers of all offences using the 1996 model

5.33 Figures based on mileage data taken from the Scottish NTS indicated that the propensity for drivers of all vehicles to receive Conditional Offers is 1.09 (compared with 1.02 for the 1996 model), and for car drivers it is 1.06 (compared with 0.99 for the 1996 model). For Police Reports a similar picture emerges. For all offences the propensity for males to offend is 1.15 for drivers of all vehicles and 1.11 for car drivers (Figure 5.4).

Figure 5.4: Expected and Observed numbers of all offences using Scottish NTS data

5.34 The differences between the results of the two models arise because of a significant difference in the miles travelled according to each model.