[雙語(yǔ)翻譯]酒后駕駛外文翻譯--影響酒后駕駛和非酒后駕駛車禍嚴(yán)重程度的因素研究（英文）

1、A study of factors influencing the severity of road crashes involving drunk drivers and non drunk driversS. Velmurugan a,*, S. Padma b, E. Madhu b, S. Anuradha b, S. Gangopadhyay ba TES Division, Central Road Research In

2、stitute, CRRI(P.O), Mathura Road, New Delhi 110025, India b TPE Division, Central Road Research Institute, CRRI(P.O), Mathura Road, New Delhi 110025, Indiaa r t i c l e i n f oArticle history:Available online 22 June 201

3、2Keywords:Road crash occurrenceDrunk and non drunk driversRoad crash severityMultinomial logit modelEconomic cost of road crashes and nationalhighwaysa b s t r a c tIn this study, an attempt has been made to develop Mult

4、inomial Logit (MNL) model by analysing thedrunken and non drunken drivers involved in road crashes on Indian highways. Multinomial Logit modelhas been deployed to assess the influence of various parameters like vehicular

5、, environment andgeometric factors on the set of drivers who were found to be drunk at the time of getting involved in theroad crash and those who were not under the influence of alcohol at the time of meeting with the r

6、oadcrash. The total economic cost of road crashes in the case of non-drunk driver road crash is Rs. 1046.27million whereas in the case of drunk driver road crashes it is estimated to be Rs. 204.50 million. Further,it can

7、 be observed that economic cost of drunk driver road crashes is varying from 13 to 19 % acrossdifferent types of road crashes.? 2012 Elsevier Ltd. All rights reserved.1. IntroductionDrunken driving is a menace throughout

8、 the world and the Indian scenario is not an exception. India has seen a steady increase in the drunken driving cases being involved in road crashes. The fast growing economy empowering individuals at a very young age wi

9、th higher spending capacities and with increasing avenues for enter- tainment there is no sight of these drunken driving crashes coming down in numbers. It is often argued that under the same circum- stances the probabil

10、ity of a drunken driver encountering a serious injury is more than that of a non drunk driver. The paper makes an attempt to study the road crashes involving drunk drivers and non drunk drivers in order to assess the imp

11、act of other variables on the severity of injury and thereby understand the intricacies of influ- ential variables on severity of injury. For this purpose, the road crash data available with the Road Safety Cell of Natio

12、nal Highway Authority of India (NHAI) has been employed. Multinomial Logit model has been used for the formulation of the relationship.2. Literature reviewNumerous studies have been carried out in different parts of the

13、world linking the severity of road crashes with the variousinfluencing factors. Eluru and Bhat (2007) studied the influence of seat belt usage on the crash related injury severity and highlighted the usage of a joint cor

14、related random coefficients binary-ordered response to underline the importance of moderating effects of unobserved individual crash related factors for the determination of injury severity and also the impact of seat be

15、lt use endogeneity. Similarly, Krull, Khattak, and Council (2000) deployed a logistic model to study how physical condition of the driver, vehicle type, roadway geometrics, Average Annual Daily Traffic, speed limit and r

16、ollover involvement affect the probability of fatal and incapa- citating injuries. They found that parameters responsible for increase in road crashes contributing to fatal and severe injury are rollover involvement, fai

17、lure to use a seat belt, alcohol consump- tion, rural roads (as opposed to urban) and higher speed limits. Similarly, O’Donnell and Connor, (1996) assessed the probabilities of four levels of injury severity as a functio

18、n of driver attributes. In this study, a comparison of the Ordered Logit (OL) and Ordered Probit (OP) specifications was made which demonstrated that the injury severity rose corresponding with increase in speed, vehicle

19、 age, occupant age, alcohol levels in the blood over .08 percent, non- use of a seat belt, manner of collision (e.g., head-on crashes) and travel in a light-duty truck. Further, according to their comparison of effects,

20、seating position of crash victims was of utmost signifi- cance (e.g., the left-rear seat of the vehicle was found to be most dangerous) and gender least relevant. Kockelman and Kweon (2002) used An OP model used to inves

21、tigate the risk of different injury levels sustained under all crash types. They used the U.S. General Estimates System (GES) data set in which severity was* Corresponding author.E-mail addresses: vms_04@yahoo.co.in (S.

22、Velmurugan), padmaseetharaman5@gmail.com (S. Padma), errampallim711@yahoo.com (E. Madhu), anuradha.crri@gmail.com (S. Anuradha), director.crri@nic.in (S. Gangopadhyay).Contents lists available at SciVerse ScienceDirectRe

23、search in Transportation Economicsjournal homepage: www.elsevier.com/locate/retrec0739-8859/$ e see front matter ? 2012 Elsevier Ltd. All rights reserved.doi:10.1016/j.retrec.2012.05.015Research in Transportation Economi

24、cs 38 (2013) 78e83The log of odds of various groups was calculated as follows:Zik ¼ bk0 þ bk1xi1 þ bk2xi2 þ . þ bkpxip (2)wherexij e is the jth predictor for ith case bkj e is the jth coefficient

25、 for the kth unobserved variable p e is the number of predictors.Its popularity is due to the fact that the formula for the choice probabilities takes a closed form and is readily interpretable. The basic limitation of t

26、he MNL model is that it has the property termed as the independence from irrelevant alternatives (IIA). The IIA property implies that the relative probability of choosing between any two alternatives is independent of al

27、l other alternatives. Correlation among unobserved factors across alternatives makes the MNL model ineffective under these conditions.3. MethodologyThe methodology employed in this study is presented in Fig. 1.4. Study a

28、rea and data descriptionThe study area has been chosen across the nation by considering selected National Highway (NH) sections falling under the Golden Quadrilateral (GQ) of India. In all a total of 620 road crashes inv

29、olving drunken drivers and 3316 road crashes involving non drunk drivers has been chosen for the analysis. The variables used for the study are described in Table 1.5. Development of MNL modelBefore carrying out the appr

30、opriate statistical analysis, it was felt prudent to assess the significance level of the variables by carrying out the significance testing of the variables considered for MNL modelling.5.1. Testing of significance of t

31、he model variablesThe variables used for model development have been primarily governed based on the data availability from the National Highway Authority of India crash records. In similarity with the studies of Al- Gha

32、mdi Ali (2002) and Wedagama and Dissanayake (2009) a hypothesis testing of proportions was performed in this study to decide whether the classification has to be reduced. Accordingly, Table 2 presents the hypothesis test

33、ing carried out on the variables/ parameters considered in this study which is aimed at elimination of the relevant variables/parameters in case of their inadequate representation in the overall data. The test used for t

34、he study was:H0: pi ¼ 0Ha: pis0where, pi is the proportion of the variable. Based on the hypothesis test it was found that the variables such as weather conditions like night, cloudy/misty/foggy and windy/ dust stor

35、m have been eliminated from the analysis in the case of non drunk drivers since these variables are insignificant at 5% level. These variables along with rain/sleet/snow have been eliminatedTable 2Hypothesis testing.Desc

36、ription Non drunk driver crashes Drunk driver crashesX N P-value 95% confidence limits X N P-value 95% confidence limitsLower Upper Lower UpperTime of road crashMid night 365 3316 .11 .10 .12 91 620 .15 .12 .18Early morn

37、ing 598 3316 .18 .17 .19 140 620 .23 .19 .26Morning 426 3316 .13 .12 .14 57 620 .09 .07 .12Late morning 539 3316 .16 .15 .18 84 620 .14 .11 .16Afternoon 417 3316 .13 .12 .14 74 620 .12 .09 .15Evening 451 3316 .14 .12 .15

38、 67 620 .11 .08 .13Late evening 325 3316 .10 .09 .11 74 620 .12 .1 .14Nighta,b 195 3316 .06 .05 .07 33 620 .05 .04 .07Nature of road crashUnknown (others) 450 3316 .14 .12 .15 205 620 .33 .29 .37Head on collision 506 331

39、6 .15 .14 .17 84 620 .14 .11 .16Rear end collision 1056 3316 .32 .3 .33 100 620 .16 .13 .19Collision due to sides wipe/right turn/left turn 555 3316 .17 .16 .18 114 620 .18 .16 .22Overturning/skidding 749 3316 .23 .21 .2

40、4 117 620 .19 .16 .22Classification of road crashFatal 371 3316 .11 .1 .12 63 620 .10 .08 .12Grievous injury 942 3316 .28 .27 .3 221 620 .36 .32 .39Minor injury 905 3316 .27 .26 .29 145 620 .23 .2 .27Non injury 1098 3316

41、 .33 .32 .35 191 620 .31 .27 .34Road conditionUnknown 255 3316 .08 .07 .09 38 620 .06 .05 .08Straight road/flat road 2551 3316 .77 .75 .78 507 620 .82 .79 .85Slight/sharp curve 510 3316 .15 .14 .17 75 620 .12 .1 .15Weath

42、er conditionUnknowna,b 21 3316 .01 N.A. N.A. 0 620 .00 N.A. N.A.Sunny day/fine 2493 3316 .75 .74 .77 472 620 .76 .73 .79Cloudy/mist/foga,b 99 3316 .03 .02 .04 20 620 .03 .02 .05Light/heavy rain/hail/sleet/snowb 234 3316

43、.07 .06 .08 22 620 .04 .02 .05Windy day/dust storma,b 7 3316 0 e e 1 620 0 e eVery cold/V. hot 462 3316 .14 .13 .15 105 620 .17 .14 .2a variables found insignificant at 5% level for non drunk drivers. b variables found i