The effects of Electronic Stability Control (ESC) on fatal crash rates in the United States.

PROBLEM: Electronic Stability Control (ESC) is believed to be among the most efficient vehicle safety interventions with reported effects around 50% for fatal single and rollover crashes. However, such estimates have used sample data, which have not controlled for the possibilities of self-selection, behavioral adaptation, increased access to the technology by less safe drivers, and the calculation of effects on very specific categories of crashes. Effects of ESC in the population can therefore be expected to be smaller than is currently believed. METHOD: National U.S. data for fatal crashes, driving exposure and other control factors, and market penetration of ESC over 1991-2021 were used to calculate whether the trends in fatalities over time in crash rates for singles, rollovers, and fatal crashes in general matched projections from estimates of effectiveness. RESULTS: It was found that downward trends in the relevant crash types were generally present before ESC was introduced, and that the trends thereafter were weaker. Although some trends were consistent with effects of ESC, they were markedly smaller than the projected ones, and could be explained by other factors such as the number of vehicles per capita. At best, the effect for rollovers could be up to two-thirds of previous estimates, no effect was detected for singles, while for all fatal crashes results depended upon the type of analysis performed. These results conflict with conclusions in all published ESC crash sample studies, which have compared vehicles with and without ESC. This discrepancy can be explained by methodological errors in the previous studies using induced exposure methods and self-selected samples. PRACTICAL APPLICATIONS: Traffic safety may not be as much improved by technological interventions as believed. Alternative approaches to traffic safety are needed, which do not rely on technology that interferes with driver behavior.

The Driver Behaviour Questionnaire as accident predictor; A methodological re-meta-analysis.

INTRODUCTION: The Manchester Driver Behaviour Questionnaire (DBQ) is the most commonly used self-report tool in traffic safety research and applied settings. It has been claimed that the violation factor of this instrument predicts accident involvement, which was supported by a previous meta-analysis. However, that analysis did not test for methodological effects, or include unpublished results. METHOD: The present study re-analysed studies on prediction of accident involvement from DBQ factors, including lapses, and many unpublished effects. Tests of various types of dissemination bias and common method variance were undertaken. RESULTS: Outlier analysis showed that some effects were probably not reliable data, but excluding them did not change the results. For correlations between violations and crashes, tendencies for published effects to be larger than unpublished ones and for effects to decrease over time were observed, but were not significant. Also, using the mean of accidents as proxy for effect indicated that studies where effects for violations are not reported have smaller effect sizes. These differences indicate dissemination bias. Studies using self-reported accidents as dependent variables had much larger effects than those using recorded accident data. Also, zero-order correlations were larger than partial correlations controlled for exposure. Similarly, violations/accidents effects were strong only when there was also a strong correlation between accidents and exposure. Overall, the true effect is probably very close to zero (r<.07) for violations versus traffic accident involvement, depending upon which tendencies are controlled for. CONCLUSIONS: Methodological factors and dissemination bias have inflated the published effect sizes of the DBQ. Strong evidence of various artefactual effects is apparent. PRACTICAL APPLICATIONS: A greater level of care should be taken if the DBQ continues to be used in traffic safety research. Also, validation of self-reports should be more comprehensive in the future, taking into account the possibility of common method variance.

The accident-exposure association: self-reported versus recorded collisions.

PROBLEM: It has been claimed that exposure to risk of road traffic accidents (usually conceptualized as mileage) is curvilinearly associated with crashes (i.e., the increase in number of crashes decreases with increased mileage). However, this effect has been criticized as mainly an artifact of self-reported data. METHOD: To test the proposition that self-reported accidents create part of the curvilinearity in data by under-reporting by high-accident drivers, self-reported and recorded collisions were plotted against hours of driving for bus drivers. RESULTS: It was found that the recorded data differed from self-reported information at the high end of exposure, and had a more linear association with the exposure measure as compared to the self-reported data, thus supporting the hypothesis. DISCUSSION: Part of the previously reported curvilinearity between accidents and exposure is apparently due to biased methods. Also, the interpretation of curvilinearity as an effect of exposure upon accidents was criticized as unfounded, as the causality may just as well go the other way. IMPACT ON INDUSTRY: The question of how exposure associates with crash involvement is far from resolved, and everyone who uses an exposure metric (mileage, time, induced) should be careful to investigate the exact properties of their variable before using it.

Re-education of young driving offenders: effects on self-reports of driver behavior.

INTRODUCTION: Offending drivers are often re-educated, but these courses have seldom been shown to have any safety effects. METHOD: An on-line improvement course for offending drivers below the age of 25 was evaluated with several driver inventories. RESULTS: The drivers reported higher levels of aggression, stress, sensation seeking, drunk driving, and driving violations, six months after the course than before. However, these levels were lower than those of controls, indicating that the initially low levels for the education group were due to socially desirable responding, as measured by a lie scale, an effect that waned after the course. DISCUSSION: The results can be interpreted as a positive effect of the education, although this conclusion is tentative and not in agreement with all effects in the data. IMPACT ON INDUSTRY: The results are in disagreement with previous evaluation studies using the same or similar instruments, and show the need to include controls for social desirability in self-report studies.

A reporting guide for studies on individual differences in traffic safety.

PROBLEM: Studies on individual differences in traffic safety report differently on their methodologies, and use different statistics, and these are therefore difficult to compare and meta-analyze. METHOD: Based upon a previous, extensive review and meta-analysis of the traffic safety literature, several recommendations are made about what features of the methodology of studies on individual differences (including evaluations) in safety need to be reported to facilitate interpretation and meta-analysis. Similarly, some basic types of statistical values are recommended. IMPACT ON INDUSTRY: The accumulation of knowledge about individual differences in traffic safety would be facilitated if scientific authors and journals adhered to these guidelines.

Social desirability effects in driver behavior inventories.

PROBLEM: The use of lie scales to control for common method variance in driver behavior inventories has been very limited. Given that such questionnaires often use self-reported safety variables as criteria, and have social implications, the risk of artefactual associations is high. METHOD: A questionnaire containing scales from several well known driver inventories that have been claimed to predict traffic accident involvement was distributed three times to a group of young drivers in a driver education program, as well as a random group twice. The Driver Impression Management scale (DIM) was used to control for socially desirable responding. RESULTS: For all behavior scales, the correlation with the DIM scale was substantial. If a scale correlated with self-reported crashes, the amount of predictive power was more than halved when social desirability was controlled for. Results were similar for both samples and all waves. The predictive power of the behavior scales was not increased when values were averaged over questionnaire waves, as should have been the case if the measurement and predictive power were valid. Results were similar for self-reported penalty points. The present results indicate that even the most well-known and accepted psychometric scales used in driver research are susceptible to social desirability bias. DISCUSSION: As social desirability is only one of a number of common method variance mechanisms that can create artefactual associations, and the great popularity of the self-report methodology, the problem for traffic research is grave. IMPACT ON INDUSTRY: Organizations that fund traffic safety research need to re-evaluate their policies regarding what methods are acceptable. The use of self-reported independent and dependent variables can lead to directly misleading results, with negative effects on traffic safety.

The relation of non-culpable traffic incidents to bus drivers' celeration behavior.

PROBLEM: The driver celeration behavior theory predicts that celerations are associated with incidents for which the driver has some responsibility in causing, but not other incidents. METHOD: The hypothesis was tested in 25 samples of repeated measurements of bus drivers' celeration behavior against their incidents for two years. RESULTS: The results confirmed the prediction; in 18 samples, the correlation for culpable incidents only was higher than for all incidents, despite the higher means of the latter. Non-culpable incidents had correlations close to zero with celeration. DISCUSSION: It was pointed out that most individual crash prediction studies have not made this differentiation, and thus probably yielded underestimates of the associations sought, although the effect is not strong, due to non-culpable accident involvements being few (less than a third of the total). The methods for correct identification of culpable incident involvements were discussed.

Culpable versus non-culpable traffic accidents; what is wrong with this picture?

INTRODUCTION: It is often implicitly or explicitly assumed in traffic accident research that drivers with accidents designated as non-culpable are a random sample from the population. However, this assumption is dependent upon differences in the criterion used for culpability. If drivers are erroneously categorized by assuming randomness, results could be grossly misleading. METHOD: The assumption of randomness leads to two predictions: first, no correlation should exist between culpable and non-culpable crashes; and second, the accident groups should differ on the variables known to be associated with accidents, such as amount of driving experience. These predictions were tested in two samples of bus drivers. RESULTS: It was found that in a sample with a harsh criterion (70% culpable accidents) for crash responsibility, the drivers with non-culpable accidents had the features expected, namely, they were more experienced for example, while in a sample with a lenient criterion (50 % culpable), this was not so. DISCUSSION: It was concluded that similar studies to the present one would need to be undertaken to establish exactly what percentage of drivers in a given population should be assigned culpable accidents, and construct a criterion that yields this ratio. Otherwise, the theoretical assumptions of randomness and non-responsibility will probably be violated to some degree. IMPACT ON INDUSTRY: Many estimates of risk of crash involvement may have been wrong. Given the potential for erroneous criteria, a number of studies may make invalid assumptions from their data.

Effects of passengers on bus driver celeration behavior and incident prediction.

PROBLEM: Driver celeration (speed change) behavior of bus drivers has previously been found to predict their traffic incident involvement, but it has also been ascertained that the level of celeration is influenced by the number of passengers carried as well as other traffic density variables. This means that the individual level of celeration is not as well estimated as could be the case. Another hypothesized influence of the number of passengers is that of differential quality of measurements, where high passenger density circumstances are supposed to yield better estimates of the individual driver component of celeration behavior. METHOD: Comparisons were made between different variants of the celeration as predictor of traffic incidents of bus drivers. The number of bus passengers was held constant, and cases identified by their number of passengers per kilometer during measurement were excluded (in 12 samples of repeated measurements). RESULTS: After holding passengers constant, the correlations between celeration behavior and incident record increased very slightly. Also, the selective prediction of incident record of those drivers who had had many passengers when measured increased the correlations even more. CONCLUSIONS: The influence of traffic density variables like the number of passengers have little direct influence on the predictive power of celeration behavior, despite the impact upon absolute celeration level. Selective prediction on the other hand increased correlations substantially. This unusual effect was probably due to how the individual propensity for high or low celeration driving was affected by the number of stops made and general traffic density; differences between drivers in this respect were probably enhanced by the denser traffic, thus creating a better estimate of the theoretical celeration behavior parameter C. The new concept of selective prediction was discussed in terms of making estimates of the systematic differences in quality of the individual driver data.

Speed choice versus celeration behavior as traffic accident predictor.

INTRODUCTION: The driver celeration behavior theory predicts that this variable is superior to all other variables as a predictor of individual traffic accident involvement, including the ever-important speed parameter. The study was undertaken to test this prediction. Also, it was expected that most variables would associate fairly strongly. METHOD: The use of speed choice as a predictor of individual traffic accident record was discussed, and four different variants of this variable (maximum, net mean, gross mean, and standard deviation of speed) identified. These variables were then compared to celeration behavior as predictors of accident record of bus drivers in the same set of data. RESULTS: Celeration behavior was found to be slightly superior, in accordance with the prediction made from the driver celeration behavior theory, although the differences were not significant. Furthermore, the predictor variables were found to associate fairly strongly between themselves, with the exception of gross mean speed, and to have fair stability over time, especially when aggregated. CONCLUSIONS: These results tentatively confirm some of the predictions made from the driver celeration behavior theory. As the results for accidents were in the expected direction, but not significant, and the maximum speed variable may have suffered from a ceiling effect, the conclusion is provisional. IMPACT ON INDUSTRY: The correlations found were strong enough to warrant the use of celeration behavior as a predictive variable for transportation companies in their safety work.

Characteristics of low speed accidents with buses in public transport: part II.

A number of accident characteristics of bus crashes are analyzed in relation to each other using data from 2237 accident involvements in the city of Uppsala (Sweden) during the years 1986-2000. The breakdown of accidents into sub-categories show, for example, that injury was common in intersection accidents, that bus stops present large risk for shunts and side contacts, while single vehicle accidents were seldom preceded by the loss of control or a skid. The treatment of accident data is discussed in terms of methodology, statistics and data reduction strategies.

The stability of driver acceleration behavior, and a replication of its relation to bus accidents.

The reliability over time of a method for measuring driver acceleration behavior was tested on bus drivers in regular traffic. Also, a replication of an earlier finding of a correlation between driver acceleration behavior and accident frequency for the individual drivers was made. It was found that the split-half correlation is probably around 0.50 for the mean (of accelerations) of a 30-min drive, and similar for the test-retest of 2.5h measured about a month apart. With such reliability, the sample was probably too small to reliably determine any association with accidents, but some significant correlations were found. Some ways of holding constant the differences in exposure and driving environment were tried with mixed success. Alternate ways of analyzing the data and several methodological problems were briefly discussed. It was concluded that the measurements of acceleration behavior, for bus drivers, are fairly reliable over at least a few months. However, some strange discrepancies between samples make all interpretations concerning the link to accidents tentative.

Some methodological deficiencies in studies on traffic accident predictors.

The paper discusses some methodological problems in (psychological) research on traffic accident predictors and reviews a convenience sample of the literature. Three methodological aspects are identified as being important: reliability of accident predictors, time period for accidents used as dependent variable, and culpability for accidents. Papers are scrutinized and most are found to be wanting in these aspects. Traffic researchers do not adhere to, or hardly even discuss, these basic methodological problems. It is concluded that the current research into (psychological) accident predictors is fraught with methodological deficiencies. Why most studies seem to be deficient in these aspects is not clear, as several researchers have pointed out these problems.

Characteristics of low speed accidents with buses in public transport.

Low-speed accidents with buses in public transport in the city of Uppsala during the years 1986-2000 are coded in 17 variables concerning mainly physical properties of the accident. The taxonomy uses classifications from existing schemes, but some are altered and some new are added to capture common features of reports of bus accidents in this population. It is found that side contacts and singles are the most common accidents, and that more than a quarter of all accident involvements occurs at bus stops. Inter-rater reliability calculations for the categories show that all except one have reliabilities above 80%. The level of internal validity, calculated as agreement of frequencies between time periods, is acceptable, despite many possible sources of change and bias. It is argued that the validity of this database far exceeds that of the, for research purposes normally used, non-company self-reports, state- and police-archives, due to more extensive reporting and corroborating evidence. The practical usefulness of these results and accident taxonomies in general is discussed.