Suburban sprawl has been widely criticized for its contribution to auto dependence. Numerous studies have found that suburban residents drive more and walk less than residents in traditional neighborhoods. Accordingly, smart growth programs have been advocated as a means to reduce auto travel. However, most studies have established only an association between the built environment and travel behavior, but not a causal relationship. Their connection may be more a matter of residential choice than of travel choice. For example, residents preferring walking may selectively live in walkable neighborhoods and thus walk more. If so, the effects of land use policies may be overstated. Using data collected from 1682 respondents living in four traditional and four suburban neighborhoods in Northern California in 2003, this dissertation explored this causal link by employing a quasi-longitudinal research design and controlling for residential self-selection (namely, residential preferences and travel attitudes). Specifically, we investigated the influence of the built environment on various measurements of personal travel choices including uses of different modes (driving, transit, walking, and biking), trip frequencies for different purposes (overall travel, nonwork travel, shopping travel, and strolling), auto ownership, and vehicle type choice. The results showed that residential preferences and travel attitudes have pervasive influences on all measurements of travel choices. The results also provide some encouragement that land-use policies designed to put residents closer to destinations and provide them with alternative transportation options will actually lead to less driving and more walking. Taking the evidence from all our analyses together, however, neighborhood design appears to have a stronger influence on walking than on driving. In other words, the residential environment promoted by smart growth programs may be an effective strategy to encourage walking but have less effect on driving, especially after attitudinal predispositions are accounted for. Given that walking is an inadequate substitute for driving, the smart growth movement seems to be more of a solution to public health problems than to transportation problems. Even so, it will give residents a choice to drive less and walk more and this choice is highly valued by a large proportion of respondents in our data as well as in other studies.

Suburban sprawl has been widely criticized for its contribution to auto dependence. Numerous studies have found that suburban residents drive more and walk less than residents in traditional neighborhoods. Accordingly, smart growth programs have been advocated as a means to reduce auto travel. However, most studies have established only an association between the built environment and travel behavior, but not a causal relationship. Their connection may be more a matter of residential choice than of travel choice. For example, residents preferring walking may selectively live in walkable neighborhoods and thus walk more. If so, the effects of land use policies may be overstated. Using data collected from 1682 respondents living in four traditional and four suburban neighborhoods in Northern California in 2003, this dissertation explored this causal link by employing a quasi-longitudinal research design and controlling for residential self-selection (namely, residential preferences and travel attitudes). Specifically, we investigated the influence of the built environment on various measurements of personal travel choices including uses of different modes (driving, transit, walking, and biking), trip frequencies for different purposes (overall travel, nonwork travel, shopping travel, and strolling), auto ownership, and vehicle type choice. The results showed that residential preferences and travel attitudes have pervasive influences on all measurements of travel choices. The results also provide some encouragement that land-use policies designed to put residents closer to destinations and provide them with alternative transportation options will actually lead to less driving and more walking. Taking the evidence from all our analyses together, however, neighborhood design appears to have a stronger influence on walking than on driving. In other words, the residential environment promoted by smart growth programs may be an effective strategy to encourage walking but have less effect on driving, especially after attitudinal predispositions are accounted for. Given that walking is an inadequate substitute for driving, the smart growth movement seems to be more of a solution to public health problems than to transportation problems. Even so, it will give residents a choice to drive less and walk more and this choice is highly valued by a large proportion of respondents in our data as well as in other studies.

Numerous studies have found that suburban residents drive more and walk less than residents in traditional neighborhoods. What is less well understood is the extent to which the observed patterns of travel behavior can be attributed to the residential built environment itself, as opposed to the prior self-selection of residents into a built environment that is consistent with their predispositions toward certain travel modes and land use configurations. To date, most studies addressing this attitudinal self-selection issue fall into seven categories: direct questioning, statistical control, instrumental variables models, sample selection models, joint discrete choice models, structural equations models, and longitudinal designs. This paper reviews and evaluates these alternative approaches with respect to this particular application (a companion paper focuses on the empirical findings of 28 studies using these approaches). We identify some advantages and disadvantages of each approach, and note the difficulties in actually quantifying the absolute and/or relative extent of the true influence of the built environment on travel behavior. Although time and resource limitations are recognized, we recommend usage of longitudinal structural equations modeling with control groups, a design which is strong with respect to all causality requisites. (C) 2007 Elsevier Ltd. All rights reserved.

The rapid growth of e-commerce is imposing profound impacts on modern society. On the supply side, the emergence of e-commerce is greatly changing the operation behavior of some retailers and is increasing product internationalization due to its geographically unlimited nature. On the demand side, the pervasiveness of e-commerce affects how, where, and when consumers shop, and indirectly influences the way in which we live our lives. However, the development of e-commerce is still in an early stage, and why consumers choose (or do not choose) online purchasing is far from being completely understood. To better evaluate and anticipate those profound impacts of e-commerce, therefore, it is important to further refine our understanding of consumers’ e-shopping behavior.

A number of studies have investigated e-shopping behavior, and reviewing them is valuable for further improving our understanding. This report aims to summarize previous e-shopping research in a systematic way. In this review, we are interested primarily in the potential benefits and costs that the internet offers for the business-to-consumer segment of e-commerce in the transaction (purchase) channel. An overview of the 65 empirical studies analyzed in this report is provided in the Appendix.

Most previous studies fall into one or more of several theoretical frameworks, including the theory of reasoned action, the theory of planned behavior, the technology acceptance model, transaction cost theory, innovation diffusion theory, and others. Among them, social psychological theories (the theory of reasoned action, the theory of planned behavior, the technology acceptance model) were widely applied. As shown in the applications of different theories, e-shopping behavior is not a simple decision process, and thus an integration of various theories is necessary to deal with its complexities. We suggest synthesizing these theories through the development of a comprehensive list of benefits and costs, using each of the key constructs of the pertinent theories as a guide to identifying the nature of those benefits and costs.

The dependent variables mainly include e-shopping intention and actual e-shopping behavior (a few studies used attitudes toward e-shopping). E-shopping intention was measured by various dimensions. Among them, the directly-stated intention to purchase online is the most frequently used measure. Although some studies used a unidimensional measure, most adopted a latent construct to assess consumers’ e-shopping intentions. Actual e-shopping behavior mainly includes three dimensions: adoption, spending, and frequency. Most studies examined one or more of these three dimensions directly, while a few studies constructed a latent variable to measure actual e-shopping behavior. When both behavioral intention and actual behavior are included in model development, attention should be paid to the time precedence between intention and behavior.

With respect to sampling, a choice-based sampling approach is probably preferable given that online shopping activity accounts for a minor proportion of all consumers, and a far smaller proportion of total retail sales. In previous studies, most chose internet/computer/email users or students as their subjects. Generally, a student sample is a natural choice for some particular products such as books. However, parameter estimates developed from a student sample lack generalizability to a larger population because of its homogeneity. By contrast, a more general sample of internet/computer/email users is more applicable for e-shopping behavior research.

The characteristics of products strongly influence the degree to which they are suitable for selling online. Mixing product categories in e-shopping behavior research tends to yield vague or inconsistent results. It is therefore necessary to explicitly consider product characteristics when exploring consumers’ e-shopping behavior. However, relatively little effort has been invested into product classification in the context of e-shopping. Although Nelson’s dichotomized system (search and experience goods) and Peterson et al.’s three-dimensional system (cost, tangibility, and differentiability) provide useful guides for product type classification, each has some shortcomings. Therefore, more research should focus on the construction of product classification systems.

Different methodologies have been applied in previous research. Generally, descriptive analysis is used to describe consumers’ e-shopping behavior; correlational analysis goes beyond descriptive analysis and attempts to analyze how two variables are related; and multivariate analysis is mainly used to explain consumers’ behavior using many variables considered together. Therefore, although descriptive and correlational analyses are important steps in helping to construct multivariate analyses, multivariate studies provide more information than these other two types of analyses. Multivariate analysis is ideal to study e-shopping behavior in depth. Among multivariate analysis techniques, multiple regression, structural equations modeling, and discrete choice models were most frequently used.

Previous studies have identified various determinants of consumers’ e-shopping behavior. These determinants mainly cover three essential elements: characteristics of e-shopping as a shopping channel, consumer characteristics, and vendor and product characteristics. Among these characteristics, the former two have been examined extensively in previous research, confirming their importance in understanding e-shopping behavior. Specifically, the dimensions of channel characteristics of e-shopping include e-shopping service quality, relative advantages, perceived risk of and confidence in e-shopping, and trust. Consumer characteristics include their shopping orientations, personality, social and psychological characteristics, computer/internet experience, in-home shopping experience, and socio-demographics.

Numerous studies have found that suburban residents drive more and walk less than residents in traditional neighborhoods. What is less well understood is the extent to which the observed patterns of travel behavior can be attributed to the residential built environment itself, as opposed to the prior self-selection of residents into a built environment that is consistent with their predispositions toward certain travel modes and land use configurations. To date, most studies addressing this attitudinal self-selection issue fall into seven categories: direct questioning, statistical control, instrumental variables models, sample selection models, joint discrete choice models, structural equations models, and longitudinal designs. This paper reviews and evaluates these alternative approaches with respect to this particular application (a companion paper focuses on the empirical findings of 28 studies using these approaches). We identify some advantages and disadvantages of each approach, and note the difficulties in actually quantifying the absolute and/or relative extent of the true influence of the built environment on travel behavior. Although time and resource limitations are recognized, we recommend usage of longitudinal structural equations modeling with control groups, a design which is strong with respect to all causality requisites.

Preparatory to an empirical analysis, this study conceptually discusses the influences of objective and subjective variables on the consideration of 16 travel-related strategies, reflecting a range of options individuals have to adapt to congestion. The variables considered here were measured by a 1998 survey conducted in the San Francisco Bay Area. The conceptual exploration shows that the consideration of travel-related strategies may be affected by the amounts of travel that individuals actually do, their subjective assessments, desires, affinities, and constraints with respect to travel. Individuals’ travel attitudes, personality, lifestyle and prior experience are also likely to affect their current consideration. Socio-economic and demographic characteristics may exhibit distributional effects with respect to the options individuals consider. These potential influences indicate that the individual adaptation process may be influenced by a wide range of qualitative and experiential variables, which are often ignored or omitted by policy makers and planners. A companion paper develops binary logit models of the consideration of each strategy.

This study operationalizes the conceptual analysis presented in a companion paper, to examine the effects of objective and subjective variables on the consideration of 16 travel-related strategies reflecting a range of individuals’ potential reactions to congestion. Using 1283 commuting respondents to a 1998 survey conducted in the San Francisco Bay Area, binary logit models were developed for the consideration of each individual strategy. The proportion of information explained by these models ranges from 0.18 to 0.63. It was found that the consideration of travel-related strategies is affected not only by the amounts of travel that individuals actually do, but also by their subjective assessments, desires and affinities with respect to travel, as well as their travel attitudes, personality and lifestyle. The previous adoption of these strategies greatly affects their current consideration, demonstrating an effect of past experience. Mobility constraints and socioeconomic and demographic characteristics exhibit distributional effects with respect to the options individuals consider. These findings imply that policies designed to alleviate congestion may be less effective than expected, because individuals’ responses to the travel-related strategies analyzed here—many of them directly tied to public policies intended to reduce vehicle travel—are influenced by a large variety of qualitative and experiential variables that are seldom measured and incorporated into demand models. Therefore, understanding the role of such variables will improve our ability to design effective policies and to accurately forecast the response to policy interventions as well as natural trends.

The rapid growth of e-commerce is imposing profound impacts on modern society. On the supply side, the emergence of e-commerce is greatly changing the operation behavior of some retailers and is increasing product internationalization due to its geographically unlimited nature. On the demand side, the pervasiveness of e-commerce affects how, where, and when consumers shop, and indirectly influences the way in which we live our lives. However, the development of e-commerce is still in an early stage, and why consumers choose (or do not choose) online purchasing is far from being completely understood. To better evaluate and anticipate those profound impacts of e-commerce, therefore, it is important to further refine our understanding of consumers' e-shopping behavior.

A number of studies have investigated e-shopping behavior, and reviewing them is valuable for further improving our understanding. This report aims to summarize previous e-shopping research in a systematic way. In this review, we are interested primarily in the potential benefits and costs that the internet offers for the business-to-consumer segment of e-commerce in the transaction (purchase) channel. An overview of the 65 empirical studies analyzed in this report is provided in the Appendix.

Most previous studies fall into one or more of several theoretical frameworks, including the theory of reasoned action, the theory of planned behavior, the technology acceptance model, transaction cost theory, innovation diffusion theory, and others. Among them, social psychological theories (the theory of reasoned action, the theory of planned behavior, the technology acceptance model) were widely applied. As shown in the applications of different theories, e-shopping behavior is not a simple decision process, and thus an integration of various theories is necessary to deal with its complexities. We suggest synthesizing these theories through the development of a comprehensive list of benefits and costs, using each of the key constructs of the pertinent theories as a guide to identifying the nature of those benefits and costs.

The dependent variables mainly include e-shopping intention and actual e-shopping behavior (a few studies used attitudes toward e-shopping). E-shopping intention was measured by various dimensions. Among them, the directly-stated intention to purchase online is the most frequently used measure. Although some studies used a unidimensional measure, most adopted a latent construct to assess consumers' e-shopping intentions. Actual e-shopping behavior mainly includes three dimensions: adoption, spending, and frequency. Most studies examined one or more of these three dimensions directly, while a few studies constructed a latent variable to measure actual e-shopping behavior. When both behavioral intention and actual behavior are included in model development, attention should be paid to the time precedence between intention and behavior.

With respect to sampling, a choice-based sampling approach is probably preferable given that online shopping activity accounts for a minor proportion of all consumers, and a far smaller proportion of total retail sales. In previous studies, most chose internet/computer/email users or students as their subjects. Generally, a student sample is a natural choice for some particular products such as books. However, parameter estimates developed from a student sample lack generalizability to a larger population because of its homogeneity. By contrast, a more general sample of internet/computer/email users is more applicable for e-shopping behavior research.

The suitability of the internet as a shopping medium depends to a large extent on the characteristics of the products. Mixing product categories in e-shopping behavior research tends to yield vague or inconsistent results. It is therefore necessary to explicitly consider product characteristics when exploring consumers' e-shopping behavior. However, relatively little effort has been invested into product classification in the context of e-shopping. Although Nelson's dichotomized system (search and experience goods) and Peterson et al.'s three-dimensional system (cost, tangibility, and differentiability) provide useful guides for product type classification, each has some shortcomings. Therefore, more research should focus on the construction of product classification systems.

Different methodologies have been applied in previous research. Generally, descriptive analysis is used to describe consumers' e-shopping behavior; correlational analysis goes beyond descriptive analysis and attempts to analyze how two variables are related; and multivariate analysis is mainly used to explain consumers' behavior using many variables considered together. Therefore, although descriptive and correlational analyses are important steps in helping to construct multivariate analyses, multivariate studies provide more information than these other two types of analyses. Multivariate analysis is ideal to study e-shopping behavior in depth. Among multivariate analysis techniques, multiple regression, structural equations modeling, and discrete choice models were most frequently used.

Previous studies have identified various determinants of consumers' e-shopping behavior. These determinants mainly cover three essential elements: characteristics of e-shopping as a shopping channel, consumer characteristics, and vendor and product characteristics. Among these characteristics, the former two have been examined extensively in previous research, confirming their importance in understanding e-shopping behavior. Specifically, the dimensions of channel characteristics of e-shopping include e-shopping service quality, relative advantages, perceived risk of and confidence in e-shopping, and trust. Consumer characteristics include their shopping orientations, personality, social and psychological characteristics, computer/internet experience, in-home shopping experience, and socio-demographics.

Numerous studies have found that suburban residents drive more and walk less than residents in traditional neighborhoods. What is less well understood is the extent to which the observed patterns of travel behavior can be attributed to the residential built environment itself, as opposed to the prior self-selection of residents into a built environment that is consistent with their predispositions toward certain travel modes and land use configurations. To date, most studies addressing this attitudinal self-selection issue fall into nine categories: direct questioning, statistical control, instrumental variables models, sample selection models, propensity score, joint discrete choice models, structural equations models, mutually-dependent discrete choice models, and longitudinal designs. This report reviews and evaluates these alternative approaches. Virtually all of the 38 empirical studies reviewed found a statistically significant influence of the built environment remaining after self-selection was accounted for. However, the practical importance of that influence was seldom assessed. Although time and resource limitations are recognized, we recommend usage of longitudinal structural equations modeling with control groups, a design which is strong with respect to all causality requisites.