This paper describes the key fundamentals of road safety. These fundamentals shape the Highway Sa... more This paper describes the key fundamentals of road safety. These fundamentals shape the Highway Safety Manual (HSM), First Edition, expected to be published in 2009. This paper is divided into four main sections: accidents as the basis for estimating safety (Section 1), defining safety (Section 2), fundamentals of safety estimation (Section 3), and safety evaluation, crash causation, and Accident Modification Factors (Section 4). It is noted that the use of the word “accident” does not imply that such an event is unavoidable; it indicates the need to study the randomness that is inherent in accident counts.
When corrective treatment is applied to road sections, intersections, drivers, or vehicles that h... more When corrective treatment is applied to road sections, intersections, drivers, or vehicles that had a poor accident record in the past, the safety effect of the treatment is properly estimated by comparing the number of accidents in a post-treatment period with the number of accidents that would have occurred in this period without the treatment. Earlier papers have shown that simple before-and-after comparisons are consistently biased; that is, treatments appear to be more effective than they really are. Accordingly, two methods--a nonparametric method and a Bayesian method--have been separately proposed for purging this bias. The nature of the bias and the two debiasing methods are reviewed. In the main body of the paper several data sets are used to compare the performance of the methods. In most cases the Bayesian method was found to yield better estimates.
Planning & Transport Res & Comp, Sum Ann Mtg, Proc, 1983
Normally a two-stage process is used to try and identify so-called "blackspots". In the... more Normally a two-stage process is used to try and identify so-called "blackspots". In the first stage a limited number of apparently dangerous locations is selected from all sites on the basis of their accident history. These are then examined in more detail in the second stage of the process. This paper deals with the first stage of the blackspot identification process which is likened to a sieve. A good "sieve" will retain most sites which require detailed examination and allow to pass most sites which need not to be looked at further. Accordingly, a theory of sieve efficiency is formulated in which the number of sites to be inspected, and the expected numbers of "correct positives", "false positives" and "false negatives" serve as measures of performance. This theory is converted into a procedure for a special but common case. This procedure is applied to a population of Ontario highway ramps and to a population of California drivers. In both cases the object of the screening process is to identify deviant units and the efficiency of the sieve is described by the aforementioned measures of performance. (TRRL)
In this paper, the existence in before-and-after studies of a consistent bias resulting from a ph... more In this paper, the existence in before-and-after studies of a consistent bias resulting from a phenomenon known as regression-to-the-mean (bias-by-selection) is explained. Many convincing examples drawn from real world data demonstrate the reality and magnitude of the phenomenon, the importance of which is two-fold. Firstly, it tends to make results of specific projects look better than they are. Secondly, the cumulative effect of several biased before-and-after comparisons is to establish overly optimistic "accident reduction factors" for various treatments which, in turn, can lead to misallocation of resources in designing countermeasures. In this paper, a simple, yet efficient, procedure for the elimination of bias-by-selection is described. Revised estimates of treatment effectiveness for three studies reported in the literature showed that the effectiveness could be significantly overestimated due to bias-by-selection. In this paper, the procedure described is used to imply that good judgement was applied in the definition of the control group - the treatment, indeed, had no effect on the control group. It was concluded that the regression-to-mean phenomenon is both real and important and that the elimination of bias from the results of before-and-after studies should therefore be undertaken by the best method available. Examples in this paper show that, in the absence of control groups, the simple procedure described achieves this end. The limitations of the method were noted. For the covering abstract of the conference see TRIS 378581. (Author/TRRL)
... Title: HOW TO ESTIMATE THE SAFETY OF RAIL-HIGHWAY GRADE CROSSINGS AND THE SAFETY EFFECTS OF W... more ... Title: HOW TO ESTIMATE THE SAFETY OF RAIL-HIGHWAY GRADE CROSSINGS AND THE SAFETY EFFECTS OF WARNING DEVICES. ... Current estimates of the safety effect of warning devices used at crossings are incorrect. This claim is supported in this paper. ...
In earlier work, accident prediction models for the general population of Ontario drivers were de... more In earlier work, accident prediction models for the general population of Ontario drivers were developed. This paper presents an extension of that work to the specific population of older drivers. Two sets of 16 models were calibrated. Each set had different combinations of variables depending on the weightings and groupings assigned to convictions and accidents, and whether or not age and gender were included as variables. The set of models for which all demerit point offenses were placed in one conviction group performed much better than the other set for which the grouping was identical to that used for the general driver population. The results indicate that models using age, gender, accidents and assigning equal weights to all demerit point convictions are somewhat better in identifying older drivers likely to be involved in accidents than the variable weight demerit point system. While it seems possible that one could develop conviction groupings that would be even more appropriate for older drivers, the indication is that the gain in prediction accuracy would be modest and probably not worth the effort required for a separate demerit point system for older drivers. (A)
This project had two main objectives. The first was to assemble existing information on the relat... more This project had two main objectives. The first was to assemble existing information on the relationship between accident frequency and traffic flow (AADT) for various parts of the roadway system. This information could then be used to update the MicroBENCOST software for the evaluation of highway investments when used in Canada. The second objective was to highlight possible areas of future research for improving safety information. This would benefit not only the users of MicroBENCOST, but practitioners of road safety management in general. Part 1 of the report provides a description of MicroBENCOST and the safety-related information used in it. Part 2 reviews the relationship between accident frequency and traffic flow. In addition, a list of Accident Modification Factors (multipliers to estimate the effect of certain changes or interventions) is provided to supplement existing MicroBENCOST data. Lastly, Part 3 discusses how to improve on currently available data. Recommendations are made as to data needed, how to prioritize needs, and implementation of the work.
The reasons for doubting the causal interpretation of single equation regression models based on ... more The reasons for doubting the causal interpretation of single equation regression models based on cross-section data are well known. But hope springs eternal. The hope is founded on the belief that if the function linking the response variable to the predictor variables was known and its parameters estimated from plentiful data then one could predict what change in the response variable is caused by a change in a predictor variable. But what if this foundational belief was incorrect? As this paper shows even perfect models can lead to incorrect conclusions. The problem is that to say what change in the response variable is caused by a change in a predictor variable one must assume that all the other predictor variables remain unchanged. This may not be possible or may require changes to reality that are entirely outside of the model, changes that almost certainly will not exist. Predictions of the effect of change for conditions that do not exist are not trustworthy.
Papers presented at this session include: preliminary location of passing lanes using a simulatio... more Papers presented at this session include: preliminary location of passing lanes using a simulation model (morrall,j); safety effect of resurfacing rural highways (sabo,pa and hauer,e). For the covering abstract of the conference see IRRD 291074. (TRRL)
A sample of 827,955 records of Ontario drivers containing information about age, gender, convicti... more A sample of 827,955 records of Ontario drivers containing information about age, gender, convictions, accidents, demerit points, and suspensions for 1981-1984 has been examined. On this basis 16 alternate models to estimate a driver's accident potential have been formulated. It appears that the currently used demerit point system, wherein the number of points associated with an offense reflects the perceived seriousness of the offense, is not a good predictor of accident potential. One can predict better by relying on the driver's record of accidents and convictions and still better by making use of a model for which the "regression weights" have been rigorously estimated. The performance of alternative models for the estimation of drivers' accident potential is described in terms of "hits" and "false alarms." It is shown, for example, that if the top 10,000 drivers are selected by the best model, 3,757 of these are expected to have an accid...
Most agencies with responsibility for extensive road systems use some variant of the rate-and-num... more Most agencies with responsibility for extensive road systems use some variant of the rate-and-number method to identify hazardous locations or blackspots. Sites so identified are later examined in detail to diagnose deficiencies and to suggest remedial measures. In this paper the degree to which the rate-and-number method is successful in identifying the unidentified, and what proportion of the sites that are subjected to detailed examination are not deviant at all, is examined. The first part of the paper is devoted to the analysis and development of the mathematical machinery. In the second part the use of the analytical results is illustrated by application to two data sets--one dealing with highway ramps in Ontario and the other with California drivers. The main result of this research is the facility to examine the performance of various identification procedures on the basis of measures of performance that are easy to understand. Such an examination should lead to a realistic ...
In earlier work, the authors developed accident prediction models for the general population of O... more In earlier work, the authors developed accident prediction models for the general population of Ontario drivers based on age, gender, conviction record and accident history. This paper presents the results of an extension of that work to the separate populations of truck and car drivers. For each sub-population, a set of linear regression models was calibrated using different combinations of variables depending on the weightings and groupings assigned to convictions and accidents, and whether or not age and gender were included as variables. The models were used to predict which drivers were likely to have accidents in a subsequent two-year period based on their driving record in an initial two-year period.
The current demerit point system in use in Ontario allocates points to offences on the basis of t... more The current demerit point system in use in Ontario allocates points to offences on the basis of the perceived seriousness of the offence. The goal of the work reported here was to allocate points to offences with a different purpose in mind. The purpose here was to use a driver's record of convictions and accidents in order to predict, as well as possible, which drivers, based on their past record, are most likely to have an accident in the near future. A sample of 827,955 records of Ontario drivers containing information about age, gender, convictions, accidents, demerit points and suspensions, for 1981-1984 was examined. On this basis 16 new models to estimate a driver's accident potential were formulated. Some models used age and gender information, others did not. Some models assigned different weights to each of the conviction categories; in other models all convictions were weighted equally (equivalent to 1 point per conviction). In some models accident data were used;...
Data on accidents, road characteristics, and traffic for rural, two-lane roads in seven states ha... more Data on accidents, road characteristics, and traffic for rural, two-lane roads in seven states have been assembled. It was found that, for the same amount of traffic, different states record widely discrepant numbers of accidents. The discrepancy does not disappear even when roads with the same lane width, shoulder type, and terrain are examined. It is concluded that (a) data from different states should not be pooled, (b) warrants and standards based on accidents should be tailored to each state, and (c) the cause of the noted differences should be investigated.
Decisions that highway and traffic engineers make significantly affect the safety of road users. ... more Decisions that highway and traffic engineers make significantly affect the safety of road users. The documents that guide highway and traffic engineering practice suggest that many of these decisions be made by 'engineering judgment'. One would like this judgment to be informed by evidence-based anticipation of their likely safety consequences and by a professional ability to balance safety against mobility and other dimensions of 'utility'. I show that these desiderata are largely unfulfilled. The many implications of this finding are discussed.
... Suppose then that there exists a cause-effect relationship between flow and accident frequenc... more ... Suppose then that there exists a cause-effect relationship between flow and accident frequency µ=µ(q) in which both µ and q pertain to the same short time period. ... Estimation of the Relationship Between Accidents and Traffic Flow ABRAHAM MENSAH AND EZRA HAUER ...
This paper describes the key fundamentals of road safety. These fundamentals shape the Highway Sa... more This paper describes the key fundamentals of road safety. These fundamentals shape the Highway Safety Manual (HSM), First Edition, expected to be published in 2009. This paper is divided into four main sections: accidents as the basis for estimating safety (Section 1), defining safety (Section 2), fundamentals of safety estimation (Section 3), and safety evaluation, crash causation, and Accident Modification Factors (Section 4). It is noted that the use of the word “accident” does not imply that such an event is unavoidable; it indicates the need to study the randomness that is inherent in accident counts.
When corrective treatment is applied to road sections, intersections, drivers, or vehicles that h... more When corrective treatment is applied to road sections, intersections, drivers, or vehicles that had a poor accident record in the past, the safety effect of the treatment is properly estimated by comparing the number of accidents in a post-treatment period with the number of accidents that would have occurred in this period without the treatment. Earlier papers have shown that simple before-and-after comparisons are consistently biased; that is, treatments appear to be more effective than they really are. Accordingly, two methods--a nonparametric method and a Bayesian method--have been separately proposed for purging this bias. The nature of the bias and the two debiasing methods are reviewed. In the main body of the paper several data sets are used to compare the performance of the methods. In most cases the Bayesian method was found to yield better estimates.
Planning & Transport Res & Comp, Sum Ann Mtg, Proc, 1983
Normally a two-stage process is used to try and identify so-called "blackspots". In the... more Normally a two-stage process is used to try and identify so-called "blackspots". In the first stage a limited number of apparently dangerous locations is selected from all sites on the basis of their accident history. These are then examined in more detail in the second stage of the process. This paper deals with the first stage of the blackspot identification process which is likened to a sieve. A good "sieve" will retain most sites which require detailed examination and allow to pass most sites which need not to be looked at further. Accordingly, a theory of sieve efficiency is formulated in which the number of sites to be inspected, and the expected numbers of "correct positives", "false positives" and "false negatives" serve as measures of performance. This theory is converted into a procedure for a special but common case. This procedure is applied to a population of Ontario highway ramps and to a population of California drivers. In both cases the object of the screening process is to identify deviant units and the efficiency of the sieve is described by the aforementioned measures of performance. (TRRL)
In this paper, the existence in before-and-after studies of a consistent bias resulting from a ph... more In this paper, the existence in before-and-after studies of a consistent bias resulting from a phenomenon known as regression-to-the-mean (bias-by-selection) is explained. Many convincing examples drawn from real world data demonstrate the reality and magnitude of the phenomenon, the importance of which is two-fold. Firstly, it tends to make results of specific projects look better than they are. Secondly, the cumulative effect of several biased before-and-after comparisons is to establish overly optimistic "accident reduction factors" for various treatments which, in turn, can lead to misallocation of resources in designing countermeasures. In this paper, a simple, yet efficient, procedure for the elimination of bias-by-selection is described. Revised estimates of treatment effectiveness for three studies reported in the literature showed that the effectiveness could be significantly overestimated due to bias-by-selection. In this paper, the procedure described is used to imply that good judgement was applied in the definition of the control group - the treatment, indeed, had no effect on the control group. It was concluded that the regression-to-mean phenomenon is both real and important and that the elimination of bias from the results of before-and-after studies should therefore be undertaken by the best method available. Examples in this paper show that, in the absence of control groups, the simple procedure described achieves this end. The limitations of the method were noted. For the covering abstract of the conference see TRIS 378581. (Author/TRRL)
... Title: HOW TO ESTIMATE THE SAFETY OF RAIL-HIGHWAY GRADE CROSSINGS AND THE SAFETY EFFECTS OF W... more ... Title: HOW TO ESTIMATE THE SAFETY OF RAIL-HIGHWAY GRADE CROSSINGS AND THE SAFETY EFFECTS OF WARNING DEVICES. ... Current estimates of the safety effect of warning devices used at crossings are incorrect. This claim is supported in this paper. ...
In earlier work, accident prediction models for the general population of Ontario drivers were de... more In earlier work, accident prediction models for the general population of Ontario drivers were developed. This paper presents an extension of that work to the specific population of older drivers. Two sets of 16 models were calibrated. Each set had different combinations of variables depending on the weightings and groupings assigned to convictions and accidents, and whether or not age and gender were included as variables. The set of models for which all demerit point offenses were placed in one conviction group performed much better than the other set for which the grouping was identical to that used for the general driver population. The results indicate that models using age, gender, accidents and assigning equal weights to all demerit point convictions are somewhat better in identifying older drivers likely to be involved in accidents than the variable weight demerit point system. While it seems possible that one could develop conviction groupings that would be even more appropriate for older drivers, the indication is that the gain in prediction accuracy would be modest and probably not worth the effort required for a separate demerit point system for older drivers. (A)
This project had two main objectives. The first was to assemble existing information on the relat... more This project had two main objectives. The first was to assemble existing information on the relationship between accident frequency and traffic flow (AADT) for various parts of the roadway system. This information could then be used to update the MicroBENCOST software for the evaluation of highway investments when used in Canada. The second objective was to highlight possible areas of future research for improving safety information. This would benefit not only the users of MicroBENCOST, but practitioners of road safety management in general. Part 1 of the report provides a description of MicroBENCOST and the safety-related information used in it. Part 2 reviews the relationship between accident frequency and traffic flow. In addition, a list of Accident Modification Factors (multipliers to estimate the effect of certain changes or interventions) is provided to supplement existing MicroBENCOST data. Lastly, Part 3 discusses how to improve on currently available data. Recommendations are made as to data needed, how to prioritize needs, and implementation of the work.
The reasons for doubting the causal interpretation of single equation regression models based on ... more The reasons for doubting the causal interpretation of single equation regression models based on cross-section data are well known. But hope springs eternal. The hope is founded on the belief that if the function linking the response variable to the predictor variables was known and its parameters estimated from plentiful data then one could predict what change in the response variable is caused by a change in a predictor variable. But what if this foundational belief was incorrect? As this paper shows even perfect models can lead to incorrect conclusions. The problem is that to say what change in the response variable is caused by a change in a predictor variable one must assume that all the other predictor variables remain unchanged. This may not be possible or may require changes to reality that are entirely outside of the model, changes that almost certainly will not exist. Predictions of the effect of change for conditions that do not exist are not trustworthy.
Papers presented at this session include: preliminary location of passing lanes using a simulatio... more Papers presented at this session include: preliminary location of passing lanes using a simulation model (morrall,j); safety effect of resurfacing rural highways (sabo,pa and hauer,e). For the covering abstract of the conference see IRRD 291074. (TRRL)
A sample of 827,955 records of Ontario drivers containing information about age, gender, convicti... more A sample of 827,955 records of Ontario drivers containing information about age, gender, convictions, accidents, demerit points, and suspensions for 1981-1984 has been examined. On this basis 16 alternate models to estimate a driver's accident potential have been formulated. It appears that the currently used demerit point system, wherein the number of points associated with an offense reflects the perceived seriousness of the offense, is not a good predictor of accident potential. One can predict better by relying on the driver's record of accidents and convictions and still better by making use of a model for which the "regression weights" have been rigorously estimated. The performance of alternative models for the estimation of drivers' accident potential is described in terms of "hits" and "false alarms." It is shown, for example, that if the top 10,000 drivers are selected by the best model, 3,757 of these are expected to have an accid...
Most agencies with responsibility for extensive road systems use some variant of the rate-and-num... more Most agencies with responsibility for extensive road systems use some variant of the rate-and-number method to identify hazardous locations or blackspots. Sites so identified are later examined in detail to diagnose deficiencies and to suggest remedial measures. In this paper the degree to which the rate-and-number method is successful in identifying the unidentified, and what proportion of the sites that are subjected to detailed examination are not deviant at all, is examined. The first part of the paper is devoted to the analysis and development of the mathematical machinery. In the second part the use of the analytical results is illustrated by application to two data sets--one dealing with highway ramps in Ontario and the other with California drivers. The main result of this research is the facility to examine the performance of various identification procedures on the basis of measures of performance that are easy to understand. Such an examination should lead to a realistic ...
In earlier work, the authors developed accident prediction models for the general population of O... more In earlier work, the authors developed accident prediction models for the general population of Ontario drivers based on age, gender, conviction record and accident history. This paper presents the results of an extension of that work to the separate populations of truck and car drivers. For each sub-population, a set of linear regression models was calibrated using different combinations of variables depending on the weightings and groupings assigned to convictions and accidents, and whether or not age and gender were included as variables. The models were used to predict which drivers were likely to have accidents in a subsequent two-year period based on their driving record in an initial two-year period.
The current demerit point system in use in Ontario allocates points to offences on the basis of t... more The current demerit point system in use in Ontario allocates points to offences on the basis of the perceived seriousness of the offence. The goal of the work reported here was to allocate points to offences with a different purpose in mind. The purpose here was to use a driver's record of convictions and accidents in order to predict, as well as possible, which drivers, based on their past record, are most likely to have an accident in the near future. A sample of 827,955 records of Ontario drivers containing information about age, gender, convictions, accidents, demerit points and suspensions, for 1981-1984 was examined. On this basis 16 new models to estimate a driver's accident potential were formulated. Some models used age and gender information, others did not. Some models assigned different weights to each of the conviction categories; in other models all convictions were weighted equally (equivalent to 1 point per conviction). In some models accident data were used;...
Data on accidents, road characteristics, and traffic for rural, two-lane roads in seven states ha... more Data on accidents, road characteristics, and traffic for rural, two-lane roads in seven states have been assembled. It was found that, for the same amount of traffic, different states record widely discrepant numbers of accidents. The discrepancy does not disappear even when roads with the same lane width, shoulder type, and terrain are examined. It is concluded that (a) data from different states should not be pooled, (b) warrants and standards based on accidents should be tailored to each state, and (c) the cause of the noted differences should be investigated.
Decisions that highway and traffic engineers make significantly affect the safety of road users. ... more Decisions that highway and traffic engineers make significantly affect the safety of road users. The documents that guide highway and traffic engineering practice suggest that many of these decisions be made by 'engineering judgment'. One would like this judgment to be informed by evidence-based anticipation of their likely safety consequences and by a professional ability to balance safety against mobility and other dimensions of 'utility'. I show that these desiderata are largely unfulfilled. The many implications of this finding are discussed.
... Suppose then that there exists a cause-effect relationship between flow and accident frequenc... more ... Suppose then that there exists a cause-effect relationship between flow and accident frequency µ=µ(q) in which both µ and q pertain to the same short time period. ... Estimation of the Relationship Between Accidents and Traffic Flow ABRAHAM MENSAH AND EZRA HAUER ...
The Secretary of Transportation rightly says that the road safety status quo in the US is unaccep... more The Secretary of Transportation rightly says that the road safety status quo in the US is unacceptable. The new National Strategy is supposed "to make a meaningful difference over the next few years" by embracing the Safe System Approach. The approach has been impressively successful in Europe and Australia. The NRSS version, however, has been so watered down that it is not likely to do the job.
Pedestrians cross the road between intersections and this is where most are killed. Where there i... more Pedestrians cross the road between intersections and this is where most are killed. Where there is no pedestrian signal, no median and no refuge island they must contend with fast traffic coming from both left and right, a risky maneuver. This article shows that the amount of time during which the crossing by a pedestrian is not obstructed by traffic varies along the road and how it depends on the chosen signal cycle time, the two 'greens', and the 'offset' between them. This knowledge can be useful in the design of livable streets and help to incorporate the interest of pedestrians in signal timing plans.
Decisions that highway and traffic engineers make significantly affect the safety of road users. ... more Decisions that highway and traffic engineers make significantly affect the safety of road users. The documents that guide highway and traffic engineering practice suggest that many of these decisions be made by 'engineering judgment'. One would like this judgment to be informed by evidence-based anticipation of their likely safety consequences and by a professional ability to balance safety against mobility and other dimensions of 'utility'. I show that these desiderata are largely unfulfilled. The many implications of this finding are discussed.
Abstract
Knowledge of crash causes is important because it directs the mind to the consideration ... more Abstract Knowledge of crash causes is important because it directs the mind to the consideration of potential prevention actions and because knowledge of the frequency with which various causes arise in crashes is necessary for determining the promise of potential prevention actions. Clinical crash causation studies consistently found that in the majority of crashes the road user was the sole cause and that in almost all crashes the road user was one of the causes. This is a ‘quasi-finding’ which provides false respectability to a style of road safety management that makes the road-user the primary target of prevention actions. For the knowledge obtained by clinical crash causation studies to be useful ‘cause’ has appropriately defined.
Abstract. What is the relationship between road safety research and the practice by which the roa... more Abstract. What is the relationship between road safety research and the practice by which the road infrastructure is built and operated? The question is seldom asked. I discuss the complexities of the research-practice symbiosis in the light of two historical anecdotes. These allow me to point out twelve issues of concern. My general conclusion is that the relationship as it evolved over time is unpremeditated and occasionally dysfunctional. Findings can be easily disregarded and questionable results given a ring of consensual truth. Practice based on such knowledge cannot be evidence-based. In the interest of road-user safety it is time to endow the research-practice relationship with a clear and purposeful structure.
The essence of CMFs is that they are to capture the safety effect of causes. The commonly used 'b... more The essence of CMFs is that they are to capture the safety effect of causes. The commonly used 'before-and-after' and 'regression modeling' approaches to CMF estimation often run into problems and consensus about trustworthy CMF estimates is slow to emerge. The nature of these problems and some promising directions to alleviate them are examined in the companion paper . In view of the identified directions, what alternative approaches to the determination of the safety effect of causes hold promise? This is the question to be examined here.
This paper was developed within a project initiated and funded by the National Cooperative Highwa... more This paper was developed within a project initiated and funded by the National Cooperative Highway Research Program (NCHRP 17-63) entitled "Guidance for the Development and Application of Crash Modification Factors". Crash Modification Factors (CMFs) allow one to anticipate the safety effect of manipulations (interventions, design changes etc.). The main aim of this project is to suggest ways to make CMF estimates more trustworthy and more widely applicable.
With the design of an existing bike-lane in mind I discuss several general issues: accident causa... more With the design of an existing bike-lane in mind I discuss several general issues: accident causation and its linkage to the formulation of prevention strategies; the myopia afflicting major studies of causation and their misleading 'the-driver-did-it' message; the question of who is responsible for what in the management of road safety; and the difficult position in which the professionals find themselves when the 'State' does not embrace its responsibility to road safety. I think that were the public aware of this state of affairs in North America it might insist on change.
Uploads
Papers by Ezra Hauer
As this paper shows even perfect models can lead to incorrect conclusions. The problem is that to say what change in the response variable is caused by a change in a predictor variable one must assume that all the other predictor variables remain unchanged. This may not be possible or may require changes to reality that are entirely outside of the model, changes that almost certainly will not exist. Predictions of the effect of change for conditions that do not exist are not trustworthy.
As this paper shows even perfect models can lead to incorrect conclusions. The problem is that to say what change in the response variable is caused by a change in a predictor variable one must assume that all the other predictor variables remain unchanged. This may not be possible or may require changes to reality that are entirely outside of the model, changes that almost certainly will not exist. Predictions of the effect of change for conditions that do not exist are not trustworthy.
Knowledge of crash causes is important because it directs the mind to the consideration of potential prevention actions and because knowledge of the frequency with which various causes arise in crashes is necessary for determining the promise of potential prevention actions. Clinical crash causation studies consistently found that in the majority of crashes the road user was the sole cause and that in almost all crashes the road user was one of the causes. This is a ‘quasi-finding’ which provides false respectability to a style of road safety management that makes the road-user the primary target of prevention actions. For the knowledge obtained by clinical crash causation studies to be useful ‘cause’ has appropriately defined.