Sage Journals: Discover world-class research

Abstract

Speeding on low-speed limit roads is a common traffic offense in China, which could be due to the mild traffic safety enforcement. The article aims to explicit the impact of traffic enforcement measures on the speeding behavior on low-speed limit roads. First, field data were collected to demonstrate the severity of speeding by investigating speed distribution; second, a virtual traffic enforcement was designed by considering three factors related to traffic enforcement, and a stated preference survey questionnaire including six scenarios was designed and implemented; finally, a series of generalized regret random minimization models were established to study the relationship of speeding behavior and traffic enforcement as well as drivers’ personal characteristics. From the stated preference survey analysis, the research figures out that other vehicles’ average speed is the most important reference to choose speed rather than traffic penalties, and the model estimation results show that speeding violation grows severe if traffic enforcements are lenient. Therefore, increasing the violation costs is a powerful means of lowering the probability of speeding for individual, thus proceeding the drop of vehicles’ average speed, and the fall of average speed will contribute to decrease speeding subsequently.

Keywords

Speeding traffic enforcement speed limit generalized regret random minimization model traffic safety

Introduction

Speeding is a common traffic violation resulting in crashes and casualties. According to the road traffic accident statistics report of the Ministry of Public Security,¹ traffic accidents caused by speeding accounted for about 2.88% of the total number of accidents in 2017, making for a death toll of 5.07%. Setting a maximum speed and installing traffic monitors on a specific road section are common ways to control speed and guarantee safety, and sanctions are also necessary to deter speeding offense.

In China, the Road Traffic Safety Law of the People’s Republic of China (PRC; TSL) and the 123th order of Public Security Ministry (123th order) stipulate the punishment mechanism for speeding. Traffic safety penalty catalogs distinguish speeding sanctions at a speed limit boundary of 60 km/h. The legislations claim that if the motor vehicle speed exceeds the prescribed speed per hour by 50% on a road with a speed limit over 60 km/h, drivers will be imposed on a fine from 200 Chinese yuan to 2000 Chinese yuan along with license revocation. While an administrative document released in 2013 announces that when the stipulated speed limit is under 60 km/h, even if speeding up to 50%, only a warning would be received. It is observed that the great contrast of sanctions emerges at a speed limit value of 60 km/h, so the article defines the roads with a speed limit under 60 km/h as low-speed limit roads.

Whereas in the United States, Virginia, the lower statutory speed limits are 25 mph (40 km/h, in a business or residential district) and 35 mph (56 km/h, on highways within a city or town excluding interstate or other limited access divided highways, on non-surface treated highways), reckless driving would be affirmed if driving 20 mph over the posted speed limit on a road, and it would be regarded as class 1 misdemeanor, carrying a fine of up to US$2500, 6-point reduction, imprisonment not more than 12 months, and license suspension for 6 months; all other speeding violations carry a fine of not more than US$200 and 3- or 4-point reduction. As for another state in the United States, New Jersey, the administration classifies speeding violation into three degrees: speeding, reckless driving and careless driving offenses; statutory law does not provide a clear quantitative criterion. But regardless of the speed limit, once a driver was judged to endanger unreasonably or to be likely to endanger unreasonably persons or property, he would be subjected to imprisonment at least 15 days and/or a fine ranging from US$50 to US$500 as well as point deduction and license suspension for 30 days. In Germany, the fine increases sharply and point reduction becomes mandatory when speeding over 20 km/h; more seriously, license suspension is enforced when speeding over 30 km/h in built-up areas and over 40 km/h on highways, regardless of what speed limit is posted. Generally, speeding sanctions are not concerned with the speed limit in many countries, only depending on the absolute value over the limit. License suspension is a common and effective means to resist speeding offense, which can be a short time like 30 days. The comparison of treatments among different countries proves the treatment in China is lenient.

However, evaluations of the effectiveness of sanctions are equivocal. Speed Check Services² demonstrated that severe punishment was effective in reducing speeding behavior; Ryeng³ proved stricter sanctions or increasing enforcement helped to reduce speed; Gehrsitz⁴ suggested that temporary license suspensions for traffic offenders were an effective tool in reducing major traffic violations, while Di Tella and Schargrodsky⁵ evaluated the effectiveness of electronic monitoring compared to incarceration and found that more lenient treatment reduced recidivism; similarly, Elvik and Christensen⁶ also found that rigor sanctions did not reduce speeding behavior; a report tried to prove speed limit and punishment were arbitrary by quoting the low fatality rate of the Autobahn highway system without the speed limit in Germany, and speed management should focus on identifying the speed limit value appropriately.⁷ On all accounts, the opposite views could be due to the diverse regulations (laws) in different countries; therefore, it comes to different conclusions.

Installation of speed cameras is an efficient way to enforce the speed limit and increase speed compliance. Many studies proved that installing monitor equipment helped to reduce mean speed and the variance^8,9 and had a positive effect on the reduction of crashes in intervention areas;¹⁰ but surveillance effects of controlling speed last on limited space,^11,12 enforcement avoidance behavior usually happened near the speeding surveillances, and the sudden change in speed increases the risk of crash occurrence and negates the benefits of speeding surveillances.¹³

In China, the sanctions on low-speed limit roads are lenient, and automated enforcement is often absent on these roads. Solomon’s¹⁴ curve illustrated that the possibility of accidents got higher along with the larger speed dispersion. Moreover, Parker et al.¹⁵ gave a conclusion by studying the relationship between vehicle’s speed and speed limit, found the rule that the higher the speed limit, the lower the speeding offenses. According to the two statements, can it be inferred that accident risks increase on low-speed limit roads? So the tasks for this research are to study the characteristics of speed choice behavior and to discuss the question whether the lenient (or rigor) treatment leads to more (or less) speeding on low-speed limit roads.

The article aims to demonstrate the factors influencing speeding behavior on low-speed limit roads and evaluates the effects of traffic enforcement especially. Two methods were used to study drivers’ speed choice: field data collection and stated preference (SP) survey. Field data were gathered for revealing the operating speed of low-speed limit roads through speed distribution, and an SP survey was undertaken to collect drivers’ speed preference under hypothetical scenarios by taking more influencing elements into consideration. Furthermore, a series of generalized regret random minimization model were established to elaborate the effect of the presumed factors on speeding behavior. Based on the discussion of model estimation results, countermeasures were put forward to deter speeding violation from standpoint of traffic enforcement.

Data sources and methodology

Field data of operating speed

In order to get a general knowledge of vehicles’ driving speed on the low-speed limit road, spot speed was measured for randomly selected cars in the free-flow condition using a speed radar. One hundred data were gathered in April 2017 at a section on one road in the suburb of Beijing.

The section has speed limit signs but no monitors, with four lanes in bi-direction with a central median, and the speed limit is set to 40 km/h. There are a few manufacture industries and logistic companies around, so vans and trucks hold an approximate proportion of 30% in mixed traffic flow.

The speed measuring sites were located at the downstream roadside of speed limit signs to make sure drivers had noticed the speed limit information, and it was set up at straight road sections without slope, at least 200 m away from the nearby crossing. It should be mentioned that the speed of each car was measured for three times at 5-s intervals, and the mean speed was adopted to form a dataset with 100 data describing speed distribution.

Descriptive analysis was made based on speed records, and the statistical characteristics were shown in Table 1 and Figure 1.

Table 1.

Speed data statistics.

	Statistics
Minimum	25 km/h
Maximum	82 km/h
Range	57
Standard deviation	10.8
Median	47.0 km/h
Mean	47.8 km/h
85% speed	57.8 km/h
50% speed	47.5 km/h
15% speed	36.5 km/h
Compliance rate	24%

Figure 1.

Operating speed data distribution.

It can be seen that the speed values ranged more widely, and the maximum speed was twice as the speed limit, resulting in a large variance. Only 24% of the measured cars complied with the speed limit, and speeding violation frequently occurred. Mean speed exceeded the speed limit by 20%. The 85th percentile speed is close to 60 km/h, 50% over the speed limit. The 15th, 50th and 85th percentile speed were used for the situation design of the following SP survey, and the approximate values of 35, 45, and 55 km/h were used as the average speed of other vehicles.

According to Solomon’s results, it can be concluded that accident risks rise at this status. Whether it is rational to attribute the widespread speeding violation to the absence of traffic monitor and moderate penalties, the inference needs to be testified.

SP survey

Survey design

An SP survey was carried out online to obtain speed choice data under hypothetical scenarios. Pictures and detailed information of the low-speed limit roads were presented at the beginning of the questionnaire, including road pavement, lane width, facilities, surroundings, and free-flow conditions, but the speed limit value was not informed to the participants so as not to interfere their perception of speed choice.

In the design of SP part, a hypothetical penalty was determined through three types of punishment: money fines, point deduction, and license revocation. The penalty can be classified into four hierarchies by four levels of speeding (see Table 2); if drivers were caught speeding, they would be imposed on the punishment as shown in Table 2, and license suspension was the most rigorous means. After introducing the potential penalties, respondents were asked about what speed they would choose under hypothetical scenarios.

Table 2.

Hypothetical penalty catalog for speeding in the SP part.

Speeding degree (%)	(0, 20]	(20, 50]	(50, 70]	(70, 100]	(100, ∞)
Approximate speed (km/h)	(40, 50]	(50, 60]	(60, 70]	(70, 80]	(80, ∞)
Money fine (Chinese yuan)	0/0	200/0	200/200	200/200	2000/200–2000
Point deduction	0/0	3/0	6/3	12/6	12/12
License revocation	0/0	0/0	0/0	0/0	1/1

SP: stated preference.

In Table 2, a hypothetical penalty was designed based on the current legislations in China. The numbers in front of oblique lines are the hypothetical penalties, and the numbers next to the oblique lines are current penalties. However, the percent value is not intuitional enough for drivers to judge their degree of speeding, especially when driving at a low speed. Therefore, the approximate speed value corresponding to the speeding degree was provided to the design of hypothetical penalty catalog. As shown in Table 2, penalties for money and points aggravate as speeding becomes severe. If speeding under 20%, drivers would not be punished because speeding was not deemed to be serious. Provided that speeding over 20% but under 50%, drivers shall be imposed on a fine of 200 yuan and 3-point deduction, instead of no punishment at present; in case of speeding over 50%, offenders would lose 6 points and suffer money punishment, which is half of the whole stock of points for 1 year; referring to the case that speeding over 70%, all the points would be deducted as well as financial penalties; the most severe behavior is supposed as speeding over 100%, and the penalty has been regulated by the Traffic Safety Law of the PRC, that is, drivers would be fined 200–2000 yuan, with all the points deducted, as well as license revoked.

The most important procedure for SP experiment is the design of hypothetical situations, that is, what factors should be imported and how factors should be combined. In this survey, three factors, ranging from money fine, point deduction, and license revocation, were exploited to define a series of virtual situations. To reduce the amount of questions constituted by the factors with different levels, a mixed uniform design was employed. In the survey, six hypothetical scenarios were offered to respondents and they were asked about their speed choice under each situation, as shown in Table 3. The numbers in the table can be explained as follows: (1) Speed limit sign: “0” means “no traffic sign” or “traffic sign was not observed” and “1” was on the contrary; (2) Speeding monitor: “0” means “no monitor” or “monitor was not observed” and “1” was on the contrary; (3) Other vehicles’ mean speed: it was set as 35, 45, and 55 km/h, for they are the approximate values to the 15th, 50th, and 85th percentile speed of field data (see Table 1), respectively. So “scenario I” can be explained as drivers did not notice the speed limit sign but a speeding monitor was installed at the road section; other vehicles’ mean speed was 35 km/h, and then, a question was presented: when you are in “scenario I,” if caught speeding and exposure to the potential penalties, what speed will you perceive to choose?

Table 3.

Mixed uniform design of three factors for hypothetical situations in SP part.

Hypothetical situation	Sign	Monitor	Mean speed of other vehicles (km/h)
I	0	1	35
II	1	0	35
III	1	1	45
IV	0	0	45
V	0	1	55
VI	1	0	55

SP: stated preference.

In addition, respondents’ individual characteristics, including gender, age, driving experience, vehicle type, and punished experience for speeding, were investigated. SP survey was conducted in April 2017, lasting for a week, and 268 valid samples containing 1608 speed choices data in hypothetical scenarios were collected.

Sample characteristics

Studying the 268 valid data, participants’ individual characteristics were summarized, as shown in Table 4. Because the survey was spread on the website, the sample was selected randomly.

Table 4.

Sample characteristics of SP survey.

Individual characteristics	Proportion (%)
Gender
Male	60
Female	40
Age (years)
≤25	37
26–35	24
36–45	19
46–55	15
≥56	5
Driving experience (years)
<2	32
2–5	29
6–10	21
>10	18
Vehicle type
Car	78
Coach	12
Truck	11
Other	9
Punished experience for speeding
Never	34
1–3 times per year	59
4–6 times per year	5
>6 times per year	2

SP: stated preference.

Statistical analysis for perceived speed choice

Driving speed choices under six hypothetical scenarios were studied. Figures 2 and 3 display the distributions of perceived speed. There are speed limit signs in scenarios II, III, and VI, so respondents were presented an extra option in these scenarios: driving at the speed limit. Some conclusions can be obtained as follows:

In any scenario, the highest proportion of perceived speed choice appeared in the range of other vehicles’ average speed, and nearly 90% of the drivers chose the speed less than or equal to other vehicles’ average speed. In scenarios I, III, and V, participants were fully aware that there were traffic monitors; however, most drivers did not choose to drive at the speed limit. Even in scenario III, both traffic signs and monitors existed, and 44.49% of drivers chose the speed near the average speed of other vehicles. This indicates that most of the drivers make their speed choice by judging the speed of other vehicles.

One-third of participants chose to drive at the speed limit in scenarios II, III, and VI, as shown in Figure 3, in which traffic signs were set up. While in scenarios I, IV, and V, as shown in Figure 2, no traffic signs existed, drivers made speed choice mainly depending on their perception, and the perception generally adapted to other drivers’ behavior. It reflects that traffic signs are effective in speed control but do not always work.

Figure 2.

Speed choice results in the scenarios without speed limit signs.

Figure 3.

Speed choice results in the scenarios with speed limit signs.

Results

Model for speed choice behavior

According to the above analysis, drivers keep the same speed with other vehicles because drivers are neither unwilling to afford time loss caused by low-speed driving nor ready to pay for the speeding penalties. In general, speeding choice behavior tends to be cautious and in conformity with majorities, which means drivers’ speed choice behavior is bounded rational, and this is similar to those of previous studies.^16,17 Whereas the most popular theory, expected utility maximum theory, is entirely rational assumptive, it is not suitable for describing drivers’ decision-making rule. Instead, many theories, such as prospect theory and regret theory, can be used to depict human’s bounded rational behavior. Comparing to the prospect theory, the regret theory has no difficulty on estimating reference point function, and a large number of studies have proved that the regret theory is more explanatory in solving multi-attribute scheme decision-making problems under risk conditions.¹⁸ The semi-compensatory nature of the regret theory makes it more explanatory for the behavior characteristics that decision-makers pay more attention to the loss attribute.¹⁹ However, due to the heterogeneity of drivers, it is difficult for the drivers to clearly determine whether the different attribute factors are utility pursuit or regret avoidance. More often, it is a mixed state of these two subjective psychological feelings, so a generalized random regret minimization (GRRM) is adopted to portray speed choice behavior on low-speed limit roads.

For the speed choice model, speed choices are classified into six alternatives by speeding levels: driving under the speed limit, speeding over 0% to 20%, 20% to 50%, 50% to 70%, 70% to 100%, and over 100%. The GRRM function can be expressed as

$\begin{array}{l} G R R_{n i} = G R_{n i} + ε_{n i} \\ = \sum_{j \neq i}^{6} \sum_{a = 1, \dots, n_{a}} \ln {r_{a} + \exp [β_{a} (x_{n j a} - x_{n i a})]} + ε_{n i} \end{array}$ (1)

where $GR R_{ni}$ is the generalized random regret value of the ith speed alternative for driver n; $G R_{ni}$ is the deterministic generalized regret value of the ith speed alternative for driver n; $ε_{ni}$ is the generalized random regret error of the ith alternative for driver n; $r_{a}$ is the regret weight coefficient for attribute a; $β_{a}$ is the parameter of attribute a; and $x_{nja} and x_{nia}$ are the attributes of the ith and jth alternative for driver n, respectively, which can be speed benefits, money fine, point deduction, and license revocation.

Compared to the classic regret random regret model, in which $r_{a}$ is equal to 1, the GRRM model presumes $r_{a}$ as a parameter to be estimated, in the range of 0 to 1 theoretically, and it is pragmatic to parameterize $r_{a}$ in terms of a binary logit function

$r_{a} = \frac{\exp (δ_{a})}{1 + \exp (δ_{a})}$ (2)

where $δ_{a}$ is the parameter treated as a dependent variable to account for the determinants of regret minimization versus utility maximization behavior toward different attributes. The larger the negative values of $δ_{a}$ , the more obvious the regret avoidance tendency toward the attribute. Otherwise, along with the larger positive values of $δ_{a}$ , the attribute is inclined to utility pursuit. When $δ_{a}$ equals to 0 and is estimated to be significant, regret minimization and utility maximization are equally important determinants of choice behavior for the attribute,¹⁹ and a hybrid model of random utility maximization (RUM) and random regret minimization (RRM) is proper to predict choice behavior.

Assuming $ε_{ni}$ as independent and generalized extreme value distributed (type I), the probability of drivers n to choose alternative i is expressed as

$\begin{matrix} P_{n} (i) = prob (G R_{ni} < G R_{nj}, \forall j \neq i) \\ = \frac{\exp (- G R_{ni})}{\sum_{j = 1}^{6} \exp (- G R_{nj})} \end{matrix}$ (3)

The maximum likelihood method can be adopted to estimate the parameters in the GRRM model, as expressed in equation (4)

$\begin{matrix} LL (β) = \ln (\underset{n}{Π} \underset{i}{Π} P_{n} {(i | β)}^{y_{n (i)}}) \\ = \sum_{n} \sum_{i} y_{n (i)} \ln (P_{n} (i | β)) \end{matrix}$ (4)

Three GRRM models were estimated by utilizing Biogeme package.²⁰ Model 1 fixed r at zero for all attributes, which can be seen as a linear-in-parameter RUM model; model 2 sets r at one, which can be seen as a linear-in-parameter RRM model; and model 3 formulates specific values for $δ_{a}$ of each attribute a. Table 5 presents the estimation results of models.

Table 5.

Estimation results of three speed choice models.

Attributes	Model 1 (r = 0)		Model 2 (r = 1)		Model 3 $(δ_{a})$
Attributes	B	t test	B	t test	B	t test
$β_{1}$ (speed benefits)	0.013	7.75	0.022	8.87	0.013	9.02
$β_{2}$ (money fine)	−0.151	−5.31	−0.368	−6.78	−0.450	−7.66
$β_{3}$ (point deduction)	−0.277	−8.62	−0.767	−7.45	−0.702	−8.04
$β_{4}$ (license revocation)	−1.880	−1.91*	−4.19	−4.02	−4.048	−4.58
$δ_{1}$ (speed benefits)	/	/	/	/	0.176	10.04
$δ_{2}$ (money fine)	/	/	/	/	−0.039	−30.98
$δ_{3}$ (point deduction)	/	/	/	/	−0.086	−28.11
$δ_{4}$ (license revocation)	/	/	/	/	−0.044	−32.51
Number of choices	1608		1608		1608
Null-loglikelihood	−2881.149		−2881.149		−2881.149
Final loglikelihood	−2325.562		−2323.357		−2322.439
ρ ²	0.193		0.194		0.194

Note: *means t-test value is significant in the 90% confidence interval.

First, in the three models, parameters of money fine, point deduction, and license revocation are all negative, indicating that the more severe the punishment was, the expected regret value of the driver would increase and the lower possibility of speeding would be. Second, the null and final loglikelihood can be applied to calculate the indicator ρ², which serves as evaluating the goodness of fit of the models; it can be learned that model 2 (RRM) provides a better t-test value and loglikelihood than model 1 (RUM), and model 3 achieves the maximum final loglikelihood, but the loglikelihood differences among them are a bit small; hence, there is little difference about the goodness of fit for these models, and the goodness of fit is not high. Third, referring to the estimation parameters, the attribute “point deduction” is of insignificance in the RUM model, and all attributes are significant in models containing regret definition.

The most stimulating results are the values of $δ_{a}$ ; all of them approximate to zero and are insignificantly different from zero, and this implies that utility maximization and regret minimization are equally important co-determinants of speed choice behavior for the benefits and loss attributes. The quantitative analysis results turn out to be consistent with drivers’ psychological characteristics and driving performance; drivers intend to keep a desired speed but try to escape sanction even if they are fully aware of speeding. The ambivalence can be depicted by GRRM with $δ_{a}$ .

Revised model for speed choice behavior

Note that the influence of drivers’ individual characteristics should be contained in the GRRM model, so equation (1) can be revised to be a hybrid model containing personal attributes and benefit–loss attributes, expressed as

$\begin{matrix} GR R_{ni} = G R_{ni} + ε_{ni} = \sum_{j \neq i}^{6} \sum_{p} β_{p} x_{np} \\ + \sum_{j \neq i}^{6} \sum_{a} \ln {r_{a} + \exp [β_{a} (x_{nja} - x_{nia})]} + ε_{ni} \end{matrix}$ (5)

where $β_{p}$ denotes the parameter of drivers’ attribute p. The attributes enter the model are shown in Table 4, and the estimation results can be seen in Table 6. This further estimation provides improvement in the loglikelihood value and better goodness of fit, dependent variables show similar tendency with model 3, and drivers’ characteristics are of significance, except the attribute “age”; thus, the model eliminated insignificant variables and re-estimated parameters.

Table 6.

Estimation results of speed choice model with drivers’ characteristics.

	B	t test		Δ	t test		B	t test
$β_{1}$	0.018	2.94	$β_{5}$ (gender)	−0.628	−12.48	$δ_{1}$	0.259	14.92
$β_{2}$	−0.382	−5.81	$β_{6}$ (driving experience)	0.225	5.67	$δ_{2}$	−0.021	−22.66
$β_{3}$	−0.553	−4.84	$β_{7}$ (vehicle type)	−0.364	−5.98	$δ_{3}$	−0.010	−32.96
$β_{4}$	−2.64	−2.35	$β_{8}$ (punishment experience)	−0.666	−6.13	$δ_{4}$	−0.013	−17.23
Number of choices			1608
Null-loglikelihood			−2881.149
Final loglikelihood			−2071.79
ρ ²			0.281

Among drivers’ attributes, the parameter of gender is negative, indicating that women have a higher perceived speed risk and are more willing to drive at a lower speed rather than speeding; the parameter of driving experience is positive, which means the longer the driving experience, the greater the probability of speeding; the parameter of vehicle type is negative, indicating that the larger the vehicle, the lower the driving speed; the parameter associated with speeding penalty experience is negative, showing that the driving speed is inversely proportional to the number of penalties received. Therefore, it can be concluded that the speeding penalty experience will reduce the probability of recidivism and have a positive effect on restraining speeding.

Discussion

The study was carried out to explain the influence of traffic enforcement on speed choice. For this sake, field data were collected to demonstrate the magnitude of speeding on low-speed limit roads, and SP survey as well as model analysis were conducted to explore the mechanism of speed choice. The research comes to a conclusion that speeding violation grows severe on low-speed limit roads if traffic enforcements are lenient.

The work of hypothetical speeding choice survey under the traffic penalties and surroundings reveals drivers’ psychological and behavior characteristics on low-speed limit roads. Statistical analysis of SP data found that the speed limit signs may constrain a third of vehicles from speeding, but whether setting traffic enforcement facilities or not, other vehicles’ average speed is the most important reference to choose speed. In other words, drivers intend to keep a desired speed but try to escape sanction even if they are fully aware of speeding. Therefore, it can be deduced that the positive effects of traffic monitors and signs are weakened by surroundings, and this is decided by drivers’ psychological risks. For drivers, mental activity toward speeding is complicated and incompatible, and the combination of different decision-making rules used to portray speed choice behavior is meaningful. Model analysis results show that utility maximization and regret minimization are equally important co-determinants of speed choice for the benefits and loss attributes, and drivers’ individual attributes affect speed choices.

To sum up, speeding behavior is not easy to eliminate. An appropriate speed limit should be posted under the traffic law, if not, it would be ineffective for eliminating speeding. Simply increasing traffic signs cannot solve the problem of speeding on low-speed limit highways, and the density of traffic violation monitors should be increased to put an end to drivers’ fluke psychology. At the same time, it should be re-examined that the rationality of the clause “speeding below 50%, causing no serious consequences,” shall be warned. According to the model outcome, increasing the violation costs, point deduction and license revocation are powerful means to lower the probability of speeding for individual and proceed the drop of vehicles’ average speed. Subsequently, the fall of the average speed helps to decrease speeding. Beyond that, various means related to speeding violation punishment should be adopted and applied, point accumulation and heavier punishment for recidivist are innovative means, and temporary license suspension for a few days is also a creditable attempt. However, due to the limitations of the data collection method, a few choices in the virtual situations may be exaggerated; in addition, more new ideas and methods other than sanctions to deter speeding offense should be discussed and measured in the further analysis.

Footnotes

Handling Editor: Hai Xiang Lin

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research,authorship,and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research,authorship,and/or publication of this article: The paper is supported by Public Security Theory and Soft Science Research Project of Public Security Ministry (2018LLYJGADX042),Fundamental Research Funds for Central Universities (2019JKF226),and the project “Data Support,Application Demonstration and Test Evaluation of Road Traffic Risk Warning and Intelligent Control Decision System.”

ORCID iD

Dan Zhao

References

Ministry of Public Security (MPS). Ministry of public security road traffic safety data statistical report. Beijing, China: Ministry of public Security, 2017.

Speed Check Services. Average speed enforcement solutions: safer, smoother, greener, fairer. London: Speed Check Services, 2010.

Ryeng

OE.

The effect of sanctions and police enforcement on drivers’ choice of speed. Accident Anal Prev 2012; 45(3): 446–454.

Gehrsitz

Speeding, punishment, and recidivism—evidence from a regression discontinuity design. Working paper, The City University of New York, New York, 2016.

Di Tella

Schargrodsky

. Criminal recidivism after prison and electronic monitoring. J Polit Econ 2013; 121(1): 28–73.

Elvik

Christensen

The deterrent effect of increasing fixed penalties for traffic offences: the Norwegian experience. J Safety Res 2007; 38(6): 689–695.

Shimizu

Srochers

Speed or greed? Automated traffic enforcement—safety? Or revenue? Report, Frontier Centre for Public Policy, Winnipeg, MB, Canada, 2015.

Goldenbeld

Schagen

The effect of speed enforcement with mobile radar on speeds and accidents. Accident Anal Prev 2005; 37(6): 1135–1144.

Walter

Broughton

Knowles

The effects of increased police enforcement along a route in London. Accident Anal Prev 2011; 43(3): 1219–1227.

10.

Martínez-Ruíz

Fandiño-Losada

Leon

, et al. Impact evaluation of camera enforcement for traffic violations in Cali, Colombia, 2008-2014. Accident Anal Prev 2019; 125(4): 267–274.

11.

Tay

The effectiveness of automated and manned traffic enforcement. Int J Sustain Transp 2009; 3(3): 178–186.

12.

Høye

Speed cameras, section control, and kangaroo jumps-a meta-analysis. Accident Anal Prev 2014; 73(12): 200–208.

13.

Shim

Park

Chung

, et al. Enforcement avoidance behavior near automated speed enforcement areas in Korean expressways. Accident Anal Prev 2015; 80(7): 57–66.

14.

Solomon

Crashes on main rural highways related to speed, driver and vehicle. Washington, DC: Bureau of Public Roads, U.S. Department of Commerce, 1964.

15.

Parker

Sung

Dereniewski

JL.

Review and analysis of posted speed limits and speed limit setting practices in British Columbia. Final report, British Columbia Ministry of Transportation, Victoria, BC, Canada, 2003.

16.

Sadia

Bekhor

Polus

Individual selection of driving speeds: analysis of a stated preference survey. J Transp Saf Secur 2015; 7: 291–306.

17.

Haglund

Aberg

Speed choice in relation to speed limit and influences from other drivers. Transport Res F: Traf 2000; 3(1): 39–51.

18.

Hess

Stathopoulos

Daly

Allowing for heterogeneous decision rules in discrete choice models: an approach and four case studies. Transportation 2012; 39(3): 565–591.

19.

Chorus

CG.

A generalized random regret minimization model. Transport Res B: Meth 2014; 68(10): 224–238.

20.

Bierlaire

. BIOGEME: a free package for the estimation of discrete choice models. In: The 3rd Swiss transportation research conference, Ascona, 19–21 March 2003. Swizerland: Springer.

Exploring the influence of traffic enforcement on speeding behavior on low-speed limit roads

Abstract

Keywords

Introduction

Data sources and methodology

Field data of operating speed

SP survey

Survey design

Sample characteristics

Statistical analysis for perceived speed choice

Results

Model for speed choice behavior

Revised model for speed choice behavior

Discussion

Footnotes

Declaration of conflicting interests

Funding

ORCID iD

References