Monash University Accident Research Centre - Report #201 - 2003
Authors: M. Cameron, S. Newstead, K. Diamantopoulou & P. Oxley
Full report in .pdf format [760KB]
The objective of this study was to measure the presence of any interaction between the effect of mobile covert speed camera enforcement and the effect of intensive mass media road safety publicity with speed-related themes. Previous research had shown the individual effect of each of these powerful road safety programs in Victoria, but had not been able to conclusively assess their interaction effect.
During 1999, the Victoria Police varied the levels of speed camera activity substantially in four Melbourne police districts according to a systematic plan. Camera hours were increased or reduced by 50% or 100% in respective districts for a month at a time, during two separate months when speed-related mass media advertising was present and during two months when it was absent. Other Melbourne police districts remained unchanged. Monthly frequencies of casualty crashes, and their severe injury outcome, in each district during 1996-2000 were analysed using Poisson Regression Modelling and Logistic Regression, respectively, to test statistically the effects of the enforcement, publicity and their interaction. In these analyses, the enforcement was represented by five levels of speeding tickets emanating from camera activity during the previous month in the same district, and the publicity by a measure of retained awareness of past and current television advertising intensities.
Reductions in crash frequency were associated monotonically with increasing levels of speed camera ticketing, and there was a statistically significant 41% reduction in fatal crash outcome associated with very high camera activity. High publicity awareness was associated with 12% reduction in crash frequency, but publicity appeared to be unrelated to crash severity. The interaction between the enforcement and publicity was not statistically significant. When two styles of speed-related publicity were analysed separately, the above results were seen to be due to emotive-style rather than enforcement-style advertising.
It was concluded that there was no evidence of an interaction between the effects of speed camera ticketing and speed-related publicity awareness on the frequency of casualty crashes. The effect of speed-related publicity during 1996-2000 was due to advertising with emotive styles.
This research questions strategic principles suggesting that speed camera enforcement and speed-related mass media publicity should operate together to produce maximum effect. However, the study considered changes in speed camera hours of only one month duration. Longer periods of increased speed camera activity may produce greater effects and different interactions with mass media publicity.
The Transport Accident Commission and the Victoria Police work together to ensure that advertising campaigns and enforcement programs are coordinated, often with enforcement and advertising programs targeting similar high-risk behaviours. In the speed domain, the combined effect of enforcement and publicity may be simply additive, or potentially synergistic where the combined effect may be larger than expected given the effect of either program on its own.
It was considered important to understand how speed camera enforcement and speed-related publicity interact, specifically in relation to their effect on the risk of casualty crashes and the injury severity of the crash outcomes. The study also aimed to determine whether varying the levels and co-presentation of publicity and enforcement resulted in a change in perception of the level of speed enforcement and TAC advertising, and whether there was a change in self reported driving behaviour and perceived risk of being caught when speeding.
Previous research conducted by MUARC indicates that the speed camera program conducted by the Police and the road safety publicity program conducted by the TAC are individually linked to reductions in crashes, however, little is understood about the interaction between these two programs.
During 1999, speed camera operations were planned to increase in two Melbourne Police Districts, by either 50% or 100% for a month at a time, during two selected months when TAC speed-related publicity was present and two months when it was absent. One of this project's objectives was to examine the effects on crashes in the two Districts, and also the effects in Districts in which camera operations were reduced, since this had also occurred. In addition, the project also included a component that surveyed drivers' perceptions and reported driver behaviour in response to the increased speed enforcement and the related publicity.
A preliminary analysis of the crash data was undertaken to examine the direct effects of the changed levels of speed camera hours, during the presence or absence of speed-related publicity, in the Districts and during the months in 1999 when these changes occurred. The preliminary analysis did not hypothesise nor examine any delayed or residual effects of the enforcement or publicity. The preliminary analysis was also limited in that only the months when speed camera enforcement conditions had been changed were considered, i.e. only crashes during April, June, August and November, thereby omitting the other months of the year. This analysis ignored substantial crash data that would increase the statistical power of the tests, particularly in relation to measuring the publicity effects.
The principal crash analysis took into account crash data from all of the months of the period 1996-2000, thus providing a more sensitive test of the presence of enforcement and publicity effects and their interaction. This analysis also included seasonality and year-level components, thereby offering a better explanation of the crash variations and increasing the sensitivity of the tests. The analysis represented the enforcement impact level by the number of speeding Traffic Infringement Notices (TINs) from offences detected in the same District during the previous month. It represented the advertising impact level by the Adstock of television TARPs (Target Audience Rating Points) in previous weeks. Adstock is a measure of retained awareness of past and current levels of TARPs, based on an exponential decay function (using a half-life of 5 weeks). The influence of TAC advertising (measured by Adstock) was considered in three different ways, namely:
The analysis was designed to be conservative, making as few assumptions about the relationships between crash outcomes and the explanatory factors as possible, especially the functional forms of the relationships. The aim was to develop models of monthly casualty crash frequencies and monthly crash severity levels for each Melbourne Police District as functions of the level of speed camera enforcement and the level of speed-related TAC advertising achieved per month, and the interaction between these two input factors. By doing this, the statistical significance of the interaction, and each of the main factors, was tested. Two statistical techniques belonging to the Generalised Linear Modelling family were used to analyse the effect of the enforcement and advertising on crash outcomes during 1996-2000. The first was Poisson Regression Modelling, to determine the effect on casualty crash frequency, and the second was Logistic Regression, to determine the effect on the severity of the crashes.
Surveys of Driver Perceptions and Reported Behaviour
Telephone surveys of drivers' awareness of the enforcement increases and publicity presence were conducted in each of the Police Districts in which speed camera activity had been increased in 1999, during the month of increase. There were five survey waves, each conducted over five to eight working days. A survey conducted in February (prior to the first enforcement period) acted as a baseline comparison to allow measurement of the relative effects on the perceived risk of detection of the increased enforcement in the context of the presence or absence of publicity relating to speeding. The surveys were conducted at the end of each relevant month to ensure that the enforcement and publicity had an opportunity to be observed.
The surveys of driver perception and self-reported behaviour were analysed and reported before information had been obtained on the actual enforcement and publicity levels and before any crash analysis had been conducted. The survey analysis had assumed that the effects of the increased enforcement and presence of speed-related publicity could be observed in surveys conducted at the end of months in which these changes occurred. Subsequently available data and reconsideration of likely effect mechanisms cast doubt on these expectations. The timing of the surveys was such that drivers were unlikely to have perceived the increase in speed camera activity, by either the receipt of one or more speeding TINs or knowledge of other drivers who had. The covert nature of speed camera operations would have minimised the perception through direct observation. For related reasons, the surveys were unable to provide evidence of the presence or absence of an interaction between the effects of the enforcement increase and the publicity presence. The survey results were considered inconclusive regarding the presence or absence of interaction effects. However, the survey results were conclusive about the effect of the speed-related publicity on the ratings of the perceived risk of detection for speeding.
Results and Conclusions
This study of the interaction of the effects of mobile covert speed camera enforcement and intensive speed-related mass media publicity in Victoria during 1996-2000 found the following results and reached the following conclusions:
1. There was no evidence of an interaction in the effects of the enforcement and the publicity on casualty crash frequency.
2. The number of speeding tickets detected by speed cameras in Melbourne Police Districts influenced the casualty crash frequency in the same district during the following month. Casualty crashes were reduced by 3.0% following months with very high levels of speeding tickets (more than 30% greater than average) and increased by 6.8% following months with very low levels of speeding tickets (less than 30% lower than average).
3. The risk of fatal outcome of the casualty crashes was also related to the number of speeding tickets detected in the district during the previous month. The fatality risk was reduced by 41% following months with very high levels of speeding tickets and increased by 44% following months with very low levels of speeding tickets.
4. High levels of awareness of TAC speed-related publicity with emotive styles produced casualty crash reductions in Melbourne during the months in which it occurred. Casualty crashes were reduced by 12-13% when awareness, measured by the Adstock of television advertising levels, of emotive-style speed-related publicity exceeded 500 Adstock units, compared with effects during lower levels of awareness of the publicity.
5. There was no evidence of an effect of the emotive-style speed-related publicity on the injury severity outcome of the casualty crashes.
6. Drivers' perceptions of the risk of detection when speeding was increased by high levels of awareness of the speed-related publicity, compared with the perception when the awareness was at medium levels.
The study also reached the following tentative conclusions:
7. There was an interaction effect on fatal casualty crash outcome when there were very high levels of speeding tickets in the previous month and high levels of awareness of enforcement-style speed-related publicity. The reduction in risk of fatal outcome was greater than expected from effects estimated when the enforcement and publicity operated alone at these levels.
8. There was no evidence that awareness of the speed-related publicity with enforcement styles contributed to casualty crash reductions during 1996-2000. This was not inconsistent with the apparent interaction of this type of publicity with the speed camera enforcement in terms of the effect on the severity outcome of casualty crashes (conclusion 7 above).
This study is limited in its ability to generalise to all circumstances of speed camera operations in conjunction with speed-related publicity because of the short (one-month) period during which speed camera hours were increased (and decreased) in selected Police Districts during 1999. The 1999 extreme variations in enforcement activity played an important role in the sensitivity of the crash analysis (though it should be noted that there were other months during 1996-2000 when TINs detected by speed cameras achieved very high and very low levels in individual Districts, thus also contributing to the sensitivity of the analysis).
It is possible that, with longer periods of change in speed camera levels, drivers may have developed stronger perceptions of the increased enforcement through personal experience of receiving TINs, knowledge of other drivers who had, and perhaps even direct observation of camera operations. A longer period of change may have led to stronger effects of the presence (or absence) of the enforcement on speed behaviour and hence on crashes. The interaction of these enforcement changes with speed-related publicity may also have been different, and the interaction may have been related to the style of publicity in different ways.
Sponsoring Organisation: Baseline Research Program - Department of Justice, Transport Accident Commission, Royal Automobile Club of Victoria (RACV) Ltd, VicRoads