英文翻译2(格式)

英文翻译B
《Mode choice of university students commuting to school and the role of active travel》
Author：Kate E. Whalen, Antonio Páez, Juan A. Carrasco School of Geography and Earth Sciences, McMaster University, 1280 Main Street West, Hamilton, Ontario L8S 4K1, Canada
Dept. of Civil Engineering, Universidad de Concepción, P.O. Box 160-C,
Concepción, Chile
Available online 9 July 2013
大学生去学校和旅游活动模式选择的作用
突出点
学生选择活动出行的比一般人群更普遍。

多项logit离散选择模型是用来识别影响模态选择的因素。

街道密集增加了电动模式的选择率。

更高的人行道密度减少了机动模式的概率。

我们找到一个由汽车和自行车旅游时间的积极效用。

摘要
近年来,有旅游兴趣的学生越来越多。

已经指出,学生倾向于使用各种运输方式,包括积极的旅行,比其他人群更频繁。

调查大学生的模态的选择提供了一个独特的机会来了解人口的大部分活跃的在一个主要trip-generating位置的通勤者。

反过来,这可以对影响活跃旅游的因素提供宝贵的见解。

在本文中,我们报告一个在大学生中模式选择分析的结果,使用在加拿大汉密尔顿的麦克马斯特大学为例。

这项研究的结果表明,成本的结合,个人态度,环境因素如街道和人行道密度是影响模态的选择的因素。

一个关键发现是,由汽车和自行车旅行时间积极影响这些模式的工具,尽管旅行时间以递减的速率增加。

而旅行时间的积极效用的汽车已经被堵在哪里,我们的分析证明骑自行车的人随处可去。

交通政策措施的例子需要进行考虑。

背景
为了有效地定位我们的研究在现有文献中,它是有用的考虑两个维度的分析:方法和变量。

第一个维度,先前的论文模态选择的学生使用描述性统计和多变量分析。

变量的分析而言,周(2012)确定了6类变量用于研究的模式选择:(1)细化因素(如社会经济和人口);(2)心理因素(例如态度);(3)使用模式特定因素(如旅行、舒适);(4)特征(特定的模式,如成本);(5)建筑环境和城市形态变量(如密度、十字路口);(6)TDM的措施(如停车成本)。

探索性数据分析使用描述性统计是一个宝贵的锻炼,帮助分析师识别值得注意的特征数据,编码错误或异常值(海宁et al .,1998)。

探索性数据分析本身在发展中工作假设可能导致重要的见解。

例如,Delmelle和Delmelle(2012,第4页)提供方式比例的样本使用七socio-demographic类(男/女,和五个学生类别从大一到毕业/法律系学生)。

因此,可以断言,“车”和“走”是最普遍的模式,由socio-demographic类变化。

这样,男性驾车出行的比例是0.332,而女性是0.376。

研究生或法律学生乘汽车旅行的比例是0.272。

虽然这表明性别差异,也
报告了模态分离特性的示例使用两个维度:状态(人员/学生)大学校园和基于距离的三个地区。

单独的一个限制使用描述性统计是统计学意义的因素被认为影响模式选择的不能确定,同时控制了潜在的混杂因素。

相比之下,其他研究已经使用多元技术分析模式的选择。

Klockner和Friedrichsmeier(2011)使用一个多层次结构方程模型,而周(2012)估计的多项物流模型有五个不同的模式。

Klockner Friedrichsmeier(2011)显示,多变量模型可以量化变量组合预测的变化模式,或计算边际影响使用多项logit模型的系数(无论Ben-Akiva和勒曼,1985)来评估一个变量的变化如何影响的概率选择模式。

在所有需要考虑的变量中,首先由一个结果变量。

例如，Klockner和Friedrichsmeier(2011)在倒塌的选择设置为一个二进制的情况,即汽车/其他模式(如自行车、散步,和交通)。

这时请注意,这简化了分析(见266页),这也让人无法评估影响因素单独的每个替代模式。

相比之下, 周(2012)估计五个不同的多项物流模型模式,包括,有趣的是,远程办公。

已知个人因素(社会经济、人口统计和心理)影响旅游行为。

一般文献表明,有男性和女性的旅游行为的差异。

金姆和Ulfarsson(2008)发现,女性比男性有更高比例的短汽车旅行。

运输方式差异也明显。

Gatersleben和阿普尔顿(2007)发现,自行车是常见的男性比女性多。

这一发现被证实男人更喜欢骑自行车,而女性更喜欢走路,在一项研究中评估了性别之间的联系当地基础设施和交通步行和骑自行车。

还表明,这可能是由于女性从家里由于家庭和家庭责任的情感感知安全和访问设施选择较短的距离。

复制了这些关注学生发现的一些研究。

Delmelle 和Delmelle(2012)也报告在他们的研究中对女性相对于男性比例较低的步行和骑自行车。

同样,周(2012)发现,男性更有可能步行或骑自行车相对于女性,但没有发现性别差异对任何其他的模式的影响。

家庭结构也是影响旅游行为。

曹 (2009)发现,数百个有一岁以下儿童家庭的数量会减少汽车旅行的频率,这可能是由于时间限制和孩子的不便,而金正日和Ulfarsson(2008)报告说,孩子的存在比没有孩子的家庭更大的延误走路。

接近这个主题,Delmelle和Delmelle(2012)报告说,儿童的数量与学生驾驶的概率呈正相关,但与步行和骑自行车的概率负相关。

在Klockner Friedrichsmeier或周的报告中，家庭变量不被认作是学生模态选择的影响因素。

心理因素被认为影响旅游模式一般(Van Acker et al .,2010和Van Acker et al .,2011)。

事实上, Klockner和Friedrichsmeier(2011)专注于汽车使用的心理因素,并包括一组丰富的态度,社会规范,有意的变量。

鉴于本文的重点,作者明确放弃任何考虑成本变量(p。

264)。

这是不幸的,因为成本变量已知影响使用的各种模式,并由于相关变量遗漏偏差有可能混淆的结果。

相反,周(2012)认为成本变量,但不是态度或其他心理因素。

Delmelle(2012)包括语句引起态度的反应,这对行为变化进行探讨,以确定潜在的激励因素。

各种模式的成本是模态选择的一个重要组成部分。

一般来说,文学提供了证据表明,人们希望限制他们花的时间旅行和访问他们的运输方式。

根据Mackett(2003)，时间约束和旅行时间高居榜首的原因是旅行者使用汽车。

通常距离是被认为是模式选择的因素。

例如,Shannon (2006) 在他们的研究中考虑三个距离区域(< 1公里,> 1公里,< 8公里,> 8公里),而Delmelle和Delmelle(2012) 根据距离校园的远近使用一个累积分布阴谋探索模态的选择。

这两项研究发现,驾车旅行者比例的增加迅速增加距离。

虽然直观,使用距离有点简单。

在一个探索性的设置,它留下了一个疑惑:为什么游客在相同的距离可能会
选择不同的模式(例如:见图 3 Delmelle Delmelle,2012)。

在建模环境中,使用距离(对于所有模式都是一样的)成为细化变量而不是一个模式——或者trip-specific变量,因此不能代表旅行准确的特点,即相对成本的模式。

在周(2012)的情况下,距离作为“成本”(在现实中,一个细化)多项logit模型变量,结果很难解释:骑自行车/步行距离系数并不显著,这意味着距离相对于距离没有影响(或者,换句话说,相当于独自驾车和拼车)。

在模态选择的分析,我们认为,这段时间或许货币成本概念和实践方面优于距离评估旅行的费用。

下一组变量与建筑环境和城市形态。

早期证据模式选择和城市环境之间的关系表明,建立行人活动的密度与更高的股票影响选择(Cervero Gorham,1995)。

某些形式的网络设计(即基于网格)也被证明增加行走模式的选择(Cervero Kockelman,1997)。

这些结果从最近其他交通研究中得到了证实。

例如,曹et al .(2009)报告说,居民在城市环境中更容易步行或骑自行车,一定程度上是由于混合土地使用支持步行和骑自行车旅行和阻止汽车旅行。

与此一致的是,金正日和Ulfarsson(2008)提供的证据表明,短的汽车旅行经常在城市化地区观察到,相比短巴士旅行和散步旅行更频繁的城市化地区。

马歇尔和灰吕(2010)发现,活动方式的分享旅行在加州的城市正受到街和交叉口密度的影响,但是如果涉及的街道主要道路会减少。

在学生中模态的选择,Delmelle和Delmelle(2012)发现,女性更有可能根据地形作为活跃的屏障模式,相对于男性。

唉,其他student-related研究综述(即Klockner Friedrichsmeier,2011年,Shannon et al .,2006,,2012)不考虑建筑环境变量。

最后,对TDM措施,Shannon et al(2006) 于停车,改变校园/淋浴设施,停车设施探讨应对态度声明关。

Delmelle和Delmelle(2012)报告说,学生在应对国家增加停车价格愿意转变模式;然而,从他们的探索性分析目前尚不清楚这是否响应将同时考虑其他潜在因素影响模式选择。

周(2012)调查停车许可证持有的影响,并发现选择人车辆以外的所有模式减少的概率。

Klockner和Friedrichsmeier(2011),另一方面,不考虑任何TDM措施分析。

在下面,我们调查学生使用的模态选择多项logit离散选择模型,考虑两个机动(汽车和交通)和两个活动模式(步行和骑自行车)。

解释变量被认为是个体社会经济、人口和态度。

我们还包括模式——和trip-specific因素、变量相关的建筑环境,和TDM的措施。

因此,我们分析了每一类周(2012)所做的报告。

原文B
Mode choice of university students commuting to school and the role of active travel
Kate E. Whalen, Antonio Páez, Juan A. Carrasco
Highlights
Active modes of travel are more prevalent among students than in the general population.
A multinomial logit discrete choice model is used to identify the factors that influence modal choices.
Higher street network density increases the probability of using motorized modes. Higher sidewalk density decreases the probability of using motorized modes.
We find a positive utility of travel time by car and bicycle.
Abstract
In recent years, interest in the travel behavior of students in institutions of higher education has grown. It has been noted that students tend to use a variety of transportation modes, including active travel, more frequently than other population segments. Investigating the modal choice of university students provides a unique opportunity to understand a population that has a large proportion of active commuters at a major trip-generating location. In turn, this can provide valuable insights into the factors that influence active travel. In this paper, we report the results of a mode choice analysis among university students, using as a case study McMaster University, in Hamilton, Canada. The results from this research indicate that modal choices are influenced by a combination of cost, individual attitudes, and environmental factors such as street and sidewalk density. A key finding is that travel time by car and bicycle positively affect the utilities of these modes, although at a decreasing rate as travel time increases. While the positive utility of time spent traveling by car has been documented in other settings, our analysis provides evidence of the intrinsic value that cyclists place on their trip experience. Examples of transportation policy measures suggested by the analysis are discussed.
Keywords
Mode choice; University students; Active travel; Positive utility of time; Discrete choice model
Background
In order to position our study effectively within the context of the existing literature, it is useful to consider two dimensions of the analysis: the methods used and the variables considered. Along the first dimension, previous papers on modal choices of students have used descriptive statistics and/or multivariate analysis. In terms of the
variables for the analysis,Zhou (2012) identifies six classes of variables used in the study of mode choices: (1) individual-specific factors (e.g. socio-economic and demographic); (2) psychological factors (e.g. attitudes); (3) mode-specific factors (e.g., comfort); (4) trip characteristics (specific to a mode, such as cost); (5) built environment and urban form variables (e.g., density, intersections); and (6) presence of TDM measures (e.g., parking cost).
Exploratory data analysis using descriptive statistics is a valuable exercise that helps analysts identify noteworthy characteristics of the data, coding errors, or outliers (Haining et al., 1998). By itself, exploratory data analysis can lead to important insights in developing working hypotheses. For example, Delmelle and Delmelle (2012, p.
4) provide mode-choice proportions for their sample using seven socio-demographic classes (male/female, and five student categories from freshman to graduate/law student). Thus, it is possible to assert that “car” and “walk” are the most prevalent modes, with variations by socio-demographic class. In this way, the proportion of males traveling by car is 0.332 whereas for females it is 0.376. The proportion of graduate or law students traveling by car is 0.272. While this suggests gender differences, it is not possible to assess the probability of walking for a female who is also a graduate student relative to a male who is a freshman. Shannon et al. (2006) also report the modal split characteristics of their sample using two dimensions: status at the university (staff/student) and three zones based on distance to campus. A limitation of the use of descriptive statistics alone is that the statistical significance of the factors thought to influence the choice of mode cannot be determined while controlling for potential confounders. Other studies, in contrast, have used multivariate techniques to analyze mode choices. Klockner and Friedrichsmeier (2011) used a multilevel structural equation model, whereas Zhou (2012) estimated a multinomial logistic model for five different modes. With multivariate models, as Klockner and Friedrichsmeier (2011) show, it is possible to quantify how variables combine to predict variations in mode use, or to calculate marginal effects using the coefficients of a multinomial logit model (q.v. Ben-Akiva and Lerman, 1985) to evaluate how changes in one variable influence the probability of selecting a mode.
In terms of the variables considered, there is first the outcome variable. Klockner and Friedrichsmeier (2011), for instance, collapsed the choice set to a binary situation, namely car/other modes (i.e. cycling, walking, and transit). While they note that this simplifies the analysis (see p. 266), it also makes it impossible to assess the factors that affect each of the alternative modes individually. Zhou (2012), in contrast, estimated a multinomial logistic model for five different modes, including, interestingly, telecommuting.
Individual factors (socio-economic, demographic, and psychological) are known to influence travel behavior. The literature in general shows that there are differences in the travel behavior of men and women. Kim and Ulfarsson (2008) found that females have a higher proportion of short automobile trips than males. Differences are also apparent with respect to active modes of transportation. Gatersleben and Appleton (2007) find that cycling is more common among men than women. This finding is corroborated by Stronegger et al. (2010), who find that men preferred cycling, while
women preferred walking, in a study that assessed gender-specific links between local infrastructure and amount of walking and cycling for transportation. Stronegger et al. (2010) also suggest that this is perhaps due to women’s feelings of perceived safety and choosing to access amenities at shorter distances from home due to household and family responsibilities. Some of these findings are replicated in studies that focused on students. Delmelle and Delmelle (2012) also report in their study lower walking and cycling proportions for females relative to males. Likewise,Zhou (2012) finds that males are more likely to walk or cycle relative to females, but finds no gender differences for any of the other modes.
Household structure is also known to impact travel behavior. Cao et al. (2009) find that the number of children under the age of five in a household tends to reduce the auto trip frequency, which may be due to time constraints and the inconveniences of taking children out, whereas Kim and Ulfarsson (2008) report that the presence of children is linked to a greater aversion to walking compared to families without children. Closer to the subject at hand, Delmelle and Delmelle (2012) report that the number of children is positively correlated with the probability of students driving, but negatively correlated with the probability of walking and cycling. Household variables are not considered by Shannon et al., 2006 and Klockner and Friedrichsmeier, 2011, or Zhou (2012) in their investigations of modal choices by students.
Psychological factors are increasingly recognized in the travel behavior literature in general (Van Acker et al., 2010 and V an Acker et al., 2011). Indeed, the study by Klockner and Friedrichsmeier (2011) concentrates on the psychological factors of car use, and includes a rich set of attitudinal, social norms, and intentional variables. Given the focus of the paper, the authors explicitly forego any consideration of cost variables (p. 264). This is unfortunate, because cost variables are known to influence the use of various modes, and potentially confound the results due to omitted relevant variable bias. Contrariwise, Zhou (2012)considers cost variables, but not attitudinal or other psychometric factors. Both Shannon et al. (2006) and Delmelle and Delmelle (2012) include statements that elicited attitudinal responses, which are explored to identify potential motivators for behavioral change. The effect of attitudinal factors has otherwise been observed in smaller scale, qualitative research on cycling, which shows that those who have adopted cycling as a mode choice have positive perceptions of cyclists and cycling in general, compared with those who have not adopted cycling as a form of transportation (Gatersleben and Appleton, 2007 and Gatersleben and Haddad, 2010). The effect has not, to our knowledge, been observed before in larger statistical samples.
The cost of the various modes is an important component of modal choice. In general, the literature provides evidence that people wish to limit the amount of time they spend traveling and accessing their mode of transportation. According to Mackett (2003), time constraints and travel time rank high on the list of reasons for travelers to use a car. Typically, time or monetary costs are used, although in some cases distance has been considered instead. For instance, Shannon et al. (2006) consider three distance zones in their study (<1 km, >1 km and <8 km, and >8 km), whereas Delmelle and Delmelle (2012) use a cumulative distribution plot to explore modal choices according
to distance to campus. Both studies find that the proportion of travelers moving by car increases rapidly with increasing distance. Although intuitive, the use of distance is somewhat simplistic. In an exploratory setting, it leaves open the question of why travelers at the same distance might choose different modes (e.g. see Fig. 3 in Delmelle and Delmelle, 2012). In a modeling setting, the use of distance (which is identical for all modes) becomes an individual-specific variable instead of a mode- or trip-specific variable, and thus fails to represent the characteristics of the trip accurately, namely the relative cost by mode. In the case of Zhou (2012), where distance is used as a “cost” (in reality, an individual-specific) variable in a multinomial logit model, the results are difficult to interpret: the coefficient for cycling/walking distance is not significant, which means that distance has no effect relative to (or, in other words, is equivalent to) distance by car (driving-alone and carpool). It is our view, in modal choice analysis, that time or monetary costs are conceptually and practically superior to distance in terms of evaluating the cost of travel.
The next set of variables relates to the built environment and urban form. Early evidence about the relationship between mode choice and the urban environment suggests that built density is associated with higher shares of pedestrian activity (Cervero and Gorham, 1995). Certain forms of network design (i.e. grid-based) have also been shown to increase walking (Cervero and Kockelman, 1997). These results find confirmation in other recent transportation research. For instance, Cao et al. (2009) report that residents in urban environments are more likely to walk or cycle, partly due to mixed land uses that support walking and biking trips and discourage auto trips. Consistent with this, Kim and Ulfarsson (2008) provide evidence that short automobile trips are observed more often in less urbanized areas, compared to short bus trips and walking trips that are more frequent in urbanized areas. Marshall and Garrick (2010) find that the share of active modes of travel in a selection of cities in California is positively affected by street and intersection density, but tends to decrease if the streets involved are major roads. In the case of student modal choice, Delmelle and Delmelle (2012) find that females are more likely to report topography as a barrier for active modes, relative to males. Alas, other student-related studies reviewed here (i.e. Klockner and Friedrichsmeier, 2011, Shannon et al., 2006 and Zhou, 2012) do not consider built environment variables.
Finally, with respect to TDM measures, Shannon et al. (2006) explore the responses to attitudinal statements regarding access to parking, changing/shower facilities on campus, and parking facilities. Delmelle and Delmelle (2012) report that students state a willingness to shift modes in response to increases in parking prices; however, it is not clear from their exploratory analysis whether this response would hold while considering other potential factors that influence mode choice. Zhou (2012) investigates the effect of parking permit possession, and finds that it decreases the probability of selecting all modes other than single-occupant vehicle. Klockner and Friedrichsmeier (2011), on the other hand, do not consider any TDM measures in their analysis.
In what follows, we investigate the modal choice of students using a multinomial logit discrete choice model, considering two motorized (car and transit) and two active modes (walking and cycling). The explanatory variables considered are individual
socio-economic, demographic, and attitudinal. We also include mode- and trip-specific factors, variables related to the built environment, and the presence of TDM measures. Thus, our analysis covers every category of those identified by Zhou (2012).。