Given the objectives of this study, the research framework is illustrated in Figure 1.1. This dissertation contains seven chapters that are organized as follows. Chapter 1 introduces our research motivations, objectives, and framework. Chapter 2 presents the results of a literature review on sustainable transportation and policy beliefs.
“Sustainability” is a complicated concept with multiple and arguable meanings.
Transportation issues are increasingly discussed in the context of sustainable development, most commonly under the rubric of sustainable transport (Banister &
Button, 1993; Greene & Wegener, 1997; Nijkamp, 1994; Whitelegg, 1993). However, discussions of sustainable transportation are usually limited to the environmental impact of transportation and possible measures to address these effects (Feitelson, 2002). Here we applied policy beliefs as empirical perceptions and normative opinions about relevant sustainable transportation policy questions and/or policy behaviors. Chapter 3 illustrates our methodology to explore the stakeholders’ policy beliefs. Chapter 4 is an empirical study that demonstrates the policy beliefs of senior officials regarding sustainable transportation policies. Chapter 5 is another empirical study that explores the policy beliefs of the general public related to sustainable
transportation policies. Chapter 6 is a comparison of the cognitive processes related to sustainable transportation policy beliefs between senior officials and the public. At the end of the dissertation, in Chapter 7, we provide a discussion of the findings from the policy beliefs of the senior officials and the general public, and propose suggestions for future study.
Figure 1.1 Research framework
C HAPTER 2 L ITERATURE R EVIEW 2.1 Sustainable Transportation Concepts and Policy
The concept of sustainable development has long been implicitly or explicitly accepted as an important component in formulating long-term strategies, although discussions often remained in the qualitative scope. The concept of sustainability emerged in the 1970s as the result of the polarization between advocates of environmental preservation and backers of economic development. At the time, environmentalists claimed that the continued exponential growth in a finite environment would soon meet natural limitations. Gradually, this ‘‘limits to growth’’
argument lost steam and credibility, primarily because it seemed to somehow ignore claims that economic growth was vital in alleviating starvation, disease, and poverty (Torgerson, 1995).
The World Commission on Environment and Development (1987) issued a report that defined sustainable development as ‘‘development that meets the needs of the present without compromising the ability of future generations to meet their own needs.” This would be “a type of development that integrates production with resource conservation and enhancement, and that links both to the provision for all of an adequate livelihood base and equitable access to resources”. In the Commission’s view, sustainability would require action at the global, national, and local levels.
Five years later, at the United Nations Conference on Environment and Development in Rio de Janeiro, representatives of several heads of states embedded the idea of sustainable development into a package of agreements, including a biodiversity convention; a climate change convention; a statement on forest principles; an agreement to work towards a desertification convention, the Rio Declaration on
environment and development; and Agenda 21, an 800-page plan for implementing the Rio Declaration (Ryan & Throgmorton, 2003). Given all of this, considerations of external effects in the environment, stakeholders’ equity in society, and efficient use of natural resources in the economy were simultaneously required of all policies of sustainability.
With the exceptions of climate change and atmospheric pollution concerns in urban areas, the emergence of a great number of additional sustainability concerns in recent years (bio-diversity, transport congestion, social exclusion, regional imbalances with their attendant political risks, etc.) have posed particular challenges to analysts with respect to integration and quantification of these problems (Zachariadis, 2005).
One of the major issues in this agenda is transportation, which is accepted worldwide as a priority area in sustainability discussions (EEA, 2002; European Commission, 2001; IEA, 2002; WBCSD, 2001; World Bank, 1996). Work on sustainable transport is progressing well both in the research arena and in policy-oriented studies concentrating primarily on emissions of air pollutants (causing health problems) and greenhouse gases (affecting climate change), and is expanding to other sustainability concerns, such as congestion, noise, and accidents. Because of the inherent complexity of this sector in comparison to most other branches of economic activity, and due to the millions of travelers affected, policy measures often have to be viewed at the local level and take into consideration local particularities. In such cases, instead of concrete, quantified proposals, it is necessary to provide policy guidelines only, pointing to successful pilot projects around the world (OECD, 2002).
Additionally, The Organization for Economic Cooperation and Development (OECD, 2002) identified sustainable indicators along a causal generator, namely the
“Driving force–State–Response Model,” which is adapted to take into account the
specificities of the public sector. The OECD indicators are established according to the tendencies for economic and environmental impact in the various sectors.
2.2 Policy Beliefs
In the process of policy-making, stakeholders bring different types of social values into the partnership process. Not all individuals are good candidates for participating in a collaborative process. Collaborative processes embody a particular set of social values, especially a belief in inclusive public participation, reciprocity, and the belief that environmental and economic values are not mutually exclusive.
Another congruent social value is a general belief in consensus-based processes as an appropriate decision-making technique. To the extent stakeholders have social values congruent with the structure and purpose of the relevant policies, they are more likely to cooperate, and less likely use alternative venues to question the effectiveness and legitimacy of any outputs (Lubell & Leach, 2005). To discuss the social values of environmentalism and conservatism, and the possible conflicts between these values, the Advocacy Coalition Framework has been broadly applied by many authors. Case in point: Hovardas and Poirazidis, (2007) examined environmental policy beliefs of stakeholder groups engaged in protected area management and found environmental policy beliefs can be used to effectively divide stakeholders into well-defined segments that override the product-oriented definition of stakeholders. The use of K-means clustering revealed innovation-introduction and implementation-charged sample segments. The instrument utilized in that research proved quite reliable and valid in measuring stakeholders’ environmental policy beliefs. Furthermore, the methodology implied that stakeholder groups differ in a significant number of belief-system elements.
In several studies, pro-environmental behavior typically involves a tradeoff between individual and collective benefit, and it has often been conceptualized within models of altruism. In point of fact, Schwartz’s (1977) theoretical framework of normative influences on altruism was extended to the environmental domain (Black, Stern, & Elsworth, 1985; Hopper & Nielsen, 1991; Widegren, 1998). In the value-belief-norm (VBN) theory of environmentalism (Stern, 2000; Stern, Dietz, Abel, Guagnano, & Kalof, 1999), pro-environmental behavior is explained by a hierarchical sequence of variables. According to the theory, values, general environmental beliefs (e.g., general problem awareness, awareness of the adverse environmental effects of human actions (awareness of consequences)), and belief that one’s own actions could prevent those effects (ascription of responsibility), activate a personal norm. In turn, that personal norm, experienced as a feeling of moral obligation to act, is stipulated to create a willingness to act pro-environmentally (Eriksson, Garvill, &
Norlund, 2006).
Different parts of this theoretical framework have been applied to environmentally significant intentions and behaviors. In the New Environmental Paradigm (NEP) (Dunlap & Van Liere, 1978), egocentric beliefs and problem awareness have been found to be positively related to pro-environmental behavior (Nordlund & Garvill, 2002; Stern, Dietz, & Guagnano, 1995; Thompson & Barton, 1994) and the acceptability of different transportation demand management (TDM) measures (Eriksson et al., 2006; Poortinga, Steg, & Vleck, 2002, 2004; Steg & Vlek, 1997). More comprehensively, Nordlund and Garvill (2003) demonstrated the importance of collective values, egocentric values, and problem awareness for a personal norm, which in turn is positively related to willingness to reduce car use. In addition, the full VBN theory has been used to explain acceptability of various energy
policies influencing households (Steg, Dreijerink, & Abrahamse, 2005).
As described by Collantes (2008), policy beliefs here are viewed as empirical perceptions and normative opinions about relevant policy questions and/or policy behaviors. Essentially, empirical perceptions are subjective assessments of cause–
effect relationships. Normative opinions are subjective value assessments of policy questions and/or behaviors—they relate to the question of what policy-related behavior should be. Normative opinions are affected by empirical perceptions and by the expectations of relevant sectors of social pressure weighted by the stakeholder’s motivation to comply with social pressure.
However, a policy preference is a behavioral intention and it can be defined as the level of support that a stakeholder is ready to give to a specific policy course of action. Reliable measures of true policy preferences are often difficult to obtain.
Public statements on policy preferences can be more reliably considered a mix of true policy preferences and strategic behavior. In general, what a stakeholder expresses in a public setting (public hearing, media, conferences, etc.) will be the result of his/her true policy preferences, the coordination with policy allies, and the expectations of the audience (peers, policy-makers, the general public, etc.). Such dissonance between what is true and what is stated may, to some extent, apply to policy beliefs as well.
Following, Collantes (2008), “policy belief” is defined as an individual’s level of confidence that a policy is practicable or effective. Presumably, each individual has a unique value representing his/her policy belief regarding sustainable transportation.
Such a latent trait can be revealed by the person’s answers to items in a questionnaire.
That is, people who have stronger beliefs regarding sustainable transportation will
respond with higher scores on a greater number of items than those who have weaker beliefs about the same issues. In addition, some policy strategies might be regarded as better than others in promoting sustainable transportation. Therefore, it can also be presumed that each item has a unique value of inherent resistance against the individual’s belief in sustainable transportation.
C HAPTER 3 M ETHODS FOR M EASURING A L ATENT T RAIT 3.1 Review of Item Response Theory
In order to provide objective and valid rating scales for addressing a situation like that outlined above, the item response model was developed and, subsequently, improved. Item response theory (IRT), which is a model-based measurement in which trait level estimates depend on both persons’ responses and on the properties of the items that were administered, has become the mainstream of psychological measurement (Hambleton, Swaminathan, & Rogers, 1991). Among the various models of IRT, the Rasch model is one that is widely applied for exploring psychological constructs. A review of IRT and the Rasch model are provided in the following parts of this chapter.
Psychological constructs are usually conceptualized as latent variables that underlie behavior. Latent variables are viewed as unobservable entities that influence manifest variables (e.g., test scores or item responses). Thus, the observation of these manifest variables can only serve as indicators of a person’s standing on the latent variables. As a result, measurement of psychological constructs is usually indirect;
that is, latent variables are measured by observing behavior on relevant tasks or items.
A measurement theory in psychology must provide a rationale that both persons and items on a psychological dimension should be inferred from behavior. Based on such a rationale, IRT was elaborated to serve as a methodology in developing or executing a psychological test.
Item response models are designed to estimate the values of latent variables on an interval scale from item scores that form an ordinal scale. Items scores, or linear combinations of item scores, are called “raw scores”. If the raw scores form a
uni-dimensional ordinal scale, then when the data is displayed with the items ordered according to item raw scores (the sum of each subject’s responses to a given item) and with the subjects ordered according to individual raw scores (the sum of each subject’s responses across all items), the data matrix will conform to a Guttman scale (Guttman, 1950).
A Guttman scale suggests that item raw scores are monotonic with item difficulty, and test scores are monotonic with the subject’s ability. The sum of scores across items for each person is the person’s raw score and the sum of scores across people for each item is item’s raw score. If the raw scores form a Guttman scale, then when people are rank-ordered by person raw score and items are rank-ordered by item raw score, the person rankings are the same for each item and item rankings are the same for each person. There are likely to be inconsistencies with this rigid rule, but the overall statistical pattern of responses should agree with these expectations. The more closely the data agree with the Guttman scale, the more likely it is that the raw scores represent at least an ordinal scale.
3.2 Brief Introduction of the Rasch Model 3.2.1 Formulation of the Rasch Model
The Rasch model has been intensively used in psychometric studies to estimate values on an interval scale based on ordinal responses (Fisher, Harvey, Taylor, Kilgore, & Kelly, 1995; Massof & Fletcher, 2001). To simplify, we initially consider only dichotomous responses; “Do you feel this strategy is practicable for implementing to achieve sustainable transportation?” A score of 1 is assigned to the response “yes”, while a score 0 is assigned to the response “no”. The probability that a respondent senior official n will respond “yes” for Item is expressed as i
; (1)
and the probability that the response is “no” is expressed as:
(2)
Therefore, the odds ratio that a respondent senior official will say “yes” to Item is
(3)
giving the logit specification;
(4)
that isolates the parameters of interest.
The person and item parameters in the case of dichotomous responses can be estimated from response odds ratios in the data set using the formulation in Equation (4). In addition to dichotomous responses, the Rasch model can be modified to be applicable to polytomous rating-scale instruments, such as a five-point Likert scale (Andrich, 1978; Masters, 1982). The modified Rasch model decomposes a polytomous response into several dichotomous responses, and formulates one multinomial-choice problem into several binary-choice problems. That is, it assigns bix as the value of the item parameter (i.e., the inherent resistance against belief in this study) for rating category to Item , and assumes that Equation (1) refers to the probability of subject n responding with rating category x rather than x-1 to Item i.
Thus, we can model the log odds of the probability that a person responds in category for Item i, compared with category x-1, as a linear function of the person parameter (i.e., the person’s policy belief in this study) θn and the relative parameter of category x, namely, for Item i:
(5)
Following Andrich’s (1978) modification of the Rasch model for a polytomous response, two types of formulations are widely applied in assessing the values of item and person parameters, namely the “rating-scales model” and the “partial-credit model”. The rating-scales model is used for instruments in which the definition of the rating scale is identical for all items, whilst the partial-credit model is employed when the definition of the rating scale differs from one item to another. The partial-credit model differs from the rating-scales model in the possession of its own threshold parameters Fix, for each category k (Wright, 1977). This is achieved by a re-parameterization of Equation (5):
(6)
the partial-credit model can be demonstrated as:
. (7)
The partial-credit model (Masters, 1982) is used for items where (1) credit is given for partially correct answers, (2) there is a hierarchy of cognitive demands on the respondents for each item, (3) each item requires a sequence of tasks to be completed, or (4) there is a batch of ordered response items with individual thresholds
x
for each item. In assessing the policy beliefs of decision makers (DMs), it is not necessary to assume the rating scales of the items are the same; thus, we adopted the partial-credit model for our empirical study.
The Rasch model is regarded as a prescriptive approach rather than a descriptive approach (Bond & Fox, 2001). In other words, the data must fit the model, or the assumptions of the model must be rejected for a particular data set. As a result, some assumptions must be made when we try to apply the Rasch model to measure policy beliefs: (1) people differ in their policy beliefs, (2) people’s responses to items depend only on their policy beliefs, (3) responses are probabilistic and conditional on their policy beliefs, and (4) the odds of achieving an item increases monotonically with the difference between the people’s policy belief parameters and the inherent-resistance parameter of the item.
Indices of reliability and validity for assessing a latent construct are also provided by the Rasch model via the person and item aspects, respectively (Wright and Masters, 1982). Reliability indices help us examine whether the model is convincing and the material is replicable, and validity indices help us examine whether the properties of our material are consistent with the assumption of the Rasch model.
3.2.2 Parameter Estimation of the Rasch Model
Based on different statistical assumptions, there are several approaches for estimating the parameters of the Rasch model. Among them, joint maximum likelihood (JML) estimation is a relatively simple and effective approach, which is also the core technique of the related computer programs: WINSTEPS and FACETS
θn
bi
(Linacre and Wright, 1997). A simple introduction of JML estimation is given as follows.
In JML estimation, unknown construct levels are handled by using provisional trait level estimates as known values. The provisional trait level estimates themselves are improved by using subsequently estimated item parameters, which are successively improved. In other words, JML estimation is an iterative procedure that typically involves sequential estimates of person and item parameters. In the initial stage, person parameters are estimated.
The first iteration of the two-stage procedure involves specifying starting values for the item parameters so the maximum likelihood estimates of person parameters can be obtained. Then the item parameters are estimated using the person-parameter estimates. In the following iterations, person and item parameters are iteratively estimated using the improved person or item parameters change very little between the successive iterations (convergence status).
JML has been extensively applied in the estimation of many IRT models and it has several advantages in applications. First, this algorithm is easily programmable.
Second, JML is applicable to many IRT models. Both 1PL IRT (e.g. the Rasch model) and 2PL IRT (e.g. the Multi-Facet Rasch Model) can be estimated with JML. Third, JML is efficient on computation. One thing that has to be noted in applying the JML estimation is that there is a strong limitation in applying the JML algorithm. In JML estimation, items or persons with perfect scores (all passed or all failed) provide no information about the parameters because there are no constraints placed on the solution.
Therefore, estimates of such items or persons with perfect scores are not
available in the JML estimation. In fact, measures of items or persons with perfect scores mostly occur in the data of educational tests but rarely in psychological exploration. In psychological exploration, items with perfect scores are regarded as inappropriate because they provide no information about evaluating construct levels of the respondents; a person with perfect scores can be also considered as an ineffective observation because their construct levels are not comparable. It is generally suggested that these items or persons be excluded from the original data or withdraw the data and redesign the whole investigation program.
3.2.3 Reliability and Validity Statistics in the Rasch Model
In latent construct measurement, reliability indices help us to examine whether the model is convincing and the material is replicable, and validity indices help us to examine whether the properties of our material are consistent with the assumptions of the measurement. In the Rasch model, indices of reliability and validity are calibrated respectively via person and item factors (Wright & Masters, 1982) to provide the
In latent construct measurement, reliability indices help us to examine whether the model is convincing and the material is replicable, and validity indices help us to examine whether the properties of our material are consistent with the assumptions of the measurement. In the Rasch model, indices of reliability and validity are calibrated respectively via person and item factors (Wright & Masters, 1982) to provide the