University of Southern California Dissertations and Theses

The relationship between women's household responsibilities and commute lengths: a study on women in the US and Great Britain

(USC Thesis Other)

The relationship between women's household responsibilities and commute lengths: a study on women in the US and Great Britain

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content THE RELATIONSHIP BETWEEN WOMEN’S HOUSEHOLD RESPONSIBILITIES AND COMMUTE LENGTHS: A STUDY OF WOMEN IN THE UNITED STATES AND GREAT BRITAIN by Elif Karsi A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (PLANNING) May 2008 Copyright 2008 Elif Karsi ii ACKNOWLEDGEMENTS First of all, I would like to thank the USC Graduate School for the Dissertation Writing Fellowship. Secondly, I would like to acknowledge the people that have guided me, provided support or became company throughout the graduate school. I cannot thank Professor Gen Giuliano enough as my advisor who has generously shared her expertise and time with me. I have learned a lot from her, and this dissertation would not have been done without her help and support. Professor Harry Richardson’s support and guidance throughout the process has been very special also, with his kindness and dedication to his students. Professor Jim Moore was my first professor in Planning, and I am glad he was there in my dissertation committee as I graduated. I also would like to thank Professor Dowell Myers who is also among my first professors in the planning program, and served in my guidance committee, and whose encouragement has certainly helped me through the years. Thanks to my mom, Sabiha Karsi, and dad, Sadik Karsi, for all the unconditional love, devotion, and sacrifice. My brother, Murat has also been very supportive of me throughout the process. Part of the family now, Minnosh- the kitty deserves thanks, too, for helping me keep it in perspective with all his playfulness and calm. My friends HaeRan, and Mark, also from the planning program, have provided me with very valuable company, and a forum for exchange of iii ideas. Finally, Xudong An, and Hsi-Hwa Hu provided valuable insight into the NPTS and NTS data. iv IV.1.1 Age 63 TABLE OF CONTENTS ACKNOWLEDGEMENTS ii LIST OF TABLES vi LIST OF FIGURES viii ABSTRACT ix CHAPTER 1 INTRODUCTION 1 CHAPTER II LITERATURE REVIEW 6 II.1 Overview 6 II.2 Theoretical Explanations for Women’s Shorter Commute Lengths 8 II.2.1 Rational Choice Theory 9 II.2.1.1 Rational Choice (Utility) Theory for Households 9 II.2.2 Space-time Geography Approach 10 II.2.3 Sociological Explanation 13 II. 2.3.1 Household Responsibility Theory 16 II. 2.3.2 Labor Market Related Explanations 18 II.2.4 Summary of Theories and Their implications for the Future of the Gender Gap 20 II.3 Discussion of Empirical Evidence 22 II.3.2 Summary of Empirical Evidence 28 II.3.3 Cross-Cultural Differences on Gender Division in Household Responsibilities 30 II.3.4 International Research on Travel Behavior 32 II.4 Conclusion 34 CHAPTER III RESEARCH APPROACH AND METHODOLOGY 36 III.1 The Gender - Political Context and Labor Force Characteristics 37 III.2 Research Questions and Hypotheses 43 III.3 The Model 43 III.4 Data 46 III.5 Study Sample 47 III.6 Operationalizing Commute Length: Estimation of Commute Distances 49 III.7 Operationalizing Household Responsibility 51 III.7.1 The Measurement Variable 51 III.8 Specific Hypotheses to be Tested 54 III.9 Creating the Control Variable Categories 57 III.9.1 Variable Categories 57 CHAPTER IV RESULTS 62 IV.1 Descriptive Results 62 v IV.1.2 Effect of Children 64 IV.1.3 Job Status 68 IV.1.4 Household Income 69 IV.1.5 Car Access and Mode 71 IV.1.6 Mode by Life Stage 75 IV.1.7 Population Density 76 IV.1.8 MSA Size 78 IV.2 Method of Analysis 80 IV.3 Regression Results 83 IV.3.1 Results of Pooled Regression Estimation for Work Trip Distance 84 IV.3.2 Results of Pooled Regression Estimation for Work Trip Time 86 IV.4 Results According to Their Support for HRH 89 IV.5 Results According to Their Support for the “Difference” Hypothesis 92 CHAPTER V DISCUSSION 97 V.1 Assessment of the Results of the Test of the HRH 97 V.2 Assessment of the Results of the Test of the “Difference” Hypothesis 100 V.3 An Overall Assessment of the Results 105 V.4 Implications of the Results on Rationality Theory 107 V.5 Recommendations for Future Research and Practical Implications 108 BIBLIOGRAPHY 111 APPENDIX 120 vi LIST OF TABLES Table III.1 Percentage of Women 16 Years and Older, by Employment Status 41 Table III.2 A Comparison of Women’s Work Hours in US and GB 42 Table III.3 Women’s Continuity of Labor Force Participation, US and UK 42 Table III.4 Employment Rates, All Workers Aged 25-44 42 Table III.5 Family and Gender Gap in Mean Hourly Earnings 42 Table III.6 Wages According to Job Status by Sex, US and GB 42 Table III.7 Drawing the Study Sample 48 Table III.8 Life Stage Categories 51 Table III.9 Mode Categories 58 Table III.10 Household Income Categories in the Pooled Sample 59 Table III.11 List of Variable Categories Used in the Results 61 Table IV.1 Work Trip Distance and Time by Sex, US and GB 63 Table IV.2 Age Distribution by Sex, US and GB 64 Table IV.3 Work Trip Distance and Time by Age, US and GB 64 Table IV.4 Life Stage, US and GB 65 Table IV.5 Proportion of Full- and Part-Time Workers 68 Table IV.6 Distance and Time to Work for Full Time and Part-Time Workers 69 Table IV.7 Car Access of Commuters 72 Table IV.8 Car Access of All NPTS 1995 and NTS 1995/97 Participants 72 Table IV.9 Mode Share 73 Table IV.10 Work Trip Distance and Time by Mode, US and GB 73 Table IV.11 Percentage of Women Working Part Time by Life Stage 74 vii Table IV.12 Work Status of Women, NPTS and NTS participants 75 Table IV.13 Percentage of Men and Women Commuting by Car by Life Stage 76 Table IV.14 Population Density Distribution by Sex, US and GB 77 Table IV.15 Population Density Distribution, US and GB for LS4, LS5 and LS6 77 Table IV.16 Work Trip Distance and Time by Population Density 78 Table IV.17 MSA Size Distribution by Sex, US and GB 79 Table IV.18 Work Trip Distance and Time by City Size 80 Table IV.19 Testing HRH Among-Women 89 Table IV.20 Pooled US and GB Men and Women “HRH Between Men and Women” Analysis 89 Table IV.21 Separate Regressions for US and GB, “HRH Between Men and Women” US Men and Women for “the Difference” Hypothesis 92 Table IV.23 Women-Only, US and GB for “the Difference” Hypothesis 93 Table V.1 Percentage of Women Aged 15-64 Employed Part Time in US and GB According to 30 and 35 Hour Thresholds 103 Table 1 Estimation Results for Work Trip Distance, US and GB 121 Table 2 Estimation Results for Work Trip Time, US and GB 122 Table 3 Estimation Results for Work Trip Distance, US and GB Women 123 Table 4 Estimation Results for Work Trip Time, US and GB Women 124 Table 5 Estimation Results for Work Trip Distance, US Men and Women 125 Table 6 Estimation Results for Work Trip Distance, GB Men and Women 126 Table 7 Estimation Results for Work Trip Time, US Men and Women 127 Table 8 Estimation Results for Work Trip Time, GB Men and Women 128 viii LIST OF FIGURES Figure III.1 The Conceptual Model 45 Figure IV.2 Work Trip Distance by Life Stage, US and GB 67 Figure IV.3 Work Trip Time by Life Stage, US and GB 67 Figure IV.4 Work Trip Distance by Household Income 70 Figure IV.5 Work Trip Time by Household Income 71 ix ABSTRACT The research on the effects of household responsibilities on commute lengths so far is not conclusive. Further, little has been done to test the hypothesis cross-culturally. This study is an international comparative analysis of the relationship of household responsibilities and women’s commute lengths. I hypothesize that 1) Household responsibilities shorten women’s commute lengths both with respect to men and other women with lower amount of household responsibilities, and, 2) Women in US have a tighter household responsibility-commute length relationship than women in GB. Using 1995 US Nationwide Personal Transportation Survey (NPTS) data from the United States, and 1995/97 NTS data from Great Britain, I approximate household responsibilities with a composite life stage variable consisting of age, presence of children, and number of adults in a household. I present descriptive results and results on estimated models which test for both independent and interaction effects of gender and country on work trip distance and time, and presents results of estimated equations for pooled, women-only, and separate regressions for US and GB, controlling for land use indicators, number of workers and household income. I find little variation among-women, and more variation between men and women according to life stage. I find that household responsibilities affect women’s commute lengths differently than men. The household responsibility-commute length relationship is strongest for mothers of school-aged children in 2-adult households. I also find some differences between the countries which point x that, as expected, the relationship between household relationship and commute lengths are weaker in GB. I conclude that since Great Britain has a number of mediating factors such as mode (GB women use car in journey to work at a lower rate compared to GB men, and women with higher household responsibilities switch to car), and more attractive option for part time working for women with household responsibility which makes the direct relationship between household responsibilities and commute lengths weaker in Great Britain. CHAPTER 1 INTRODUCTION Travel behavior is related to larger issues in society. Differences in travel behavior between groups are usually examined to arrive at conclusions rooted much deeper. Women’s commuting behavior is such an issue. On average, women travel for different purposes, making more household-related trips such as shopping and chauffeuring children (unlike men who do more business-related trips), make more trip-chains, and shorter trips, especially shorter commute trips than men. This dissertation is about women’s commute trips. Women’s shorter distances to work than men has been documented by Madden, 1981; Hanson and Johnston, 1985; Gordon, Kumar and Richardson, 1989; Johnston-Anumonwo, 1992; Hanson and Pratt, 1995; Turner and Niemier, 1997; and Hjorthol, 2000. There is evidence that women commute for shorter commute times, as well (Fagnani, 1987; Gordon et al 1989; Hanson and Pratt, 1990; Singell and Lillydahl, 1986; Turner and Niemier, 1997; and Hanson and Johnston, 1988). Women’s commuting behavior is studied by many in relation to their changing roles at home and at work, since by definition commutes tie the two realms. Time use studies show that women have remained the main caretaker at home despite taking on greater economic roles in their 1 households (Bianchi, 2001). There has been an ongoing sociological debate about whether this inequality in the division household responsibilities is only temporary, and the gap will close naturally as more women continue to participate in the labor force, and gain more bargaining power with a “lagged adaptation” (Gershuny, 2000), or it just shows that women are working a “second shift” now and have to juggle both home and work responsibilities (Hochschild, 1989), the inequality being exacerbated. The debate is reflected in theories about women’s commuting trips. Women’s shorter commute distances and times than men’s are attributed by some to a rational choice; the idea is that differences are bound to diminish in the future. Others view it as an outcome of constraints the society poses to women at home and at work. According to constraint scholars, the differences will persist. Rational utility theorists argue that women’s lower attachment to labor force, and casuality in their job search are behind women’s shorter commute lengths; hence, the commute length differences between the sexes are bound to diminish in the future. Those that view women’s shorter commutes as an outcome of constraints are divided into two as those who attribute the source of the problem to gender discrimination in the labor markets, and those that attribute it to women’s household responsibilities, hypothesizing that disproportionate burden of household responsibility on women require short commute times and make it difficult for them to work away from home. This dissertation will test the latter, the household responsibility hypothesis. 2 There have been studies that have tackled the household responsibility hypothesis, but so far, the evidence is mixed. One reason for the lack of consistent evidence is the inadequate and inconsistent measurements of household responsibility. Studies have used a variety of proxies for household responsibility. Another reason for the evidence being mixed is the lack of cross-cultural studies. Since household responsibility is a socially constructed phenomenon, its effects can better be teased out across socially diverse settings. In addition, applicability to diverse settings and generalizability are furthered by such studies. With the dearth of comparative disaggregate research studies on travel behavior, most studies regard American studies as universal and the norm, and in doing so, disregard the importance of context in the determination of travel behavior. Without incorporating a cross-cultural perspective, one cannot appropriately determine the factors that lead to women’s shorter commute lengths, or measure the effect of household responsibility hypothesis. Thus, the lack of cultural context forms one of the key missing pieces in the current literature in household responsibility-commute length relationship. This is especially timely as gender role transformation has been and is taking place across the world. This dissertation seeks to fill the gaps listed above. It improves the method in measuring of household responsibility by using a composite life stage variable consisting of age, number of adults, presence of children, and the age of youngest child variables. It tests the household responsibility hypothesis in women’s commute lengths using data from national travel 3 surveys from the United States and Great Britain from mid-to-late 1990s (US 1995 NPTS and GB 1995/97 NTS). It argues that women’s shorter commutes are affected by the socially constructed notions of gender and culture. There are likely cultural differences between US and GB, which are likely to result in different support policies for working mothers, and different divisions of labor in households and consecutively different household responsibility- commute length relationships. My analysis shows that between US and GB, US women with children have a stronger place in the labor market, while GB women benefit from greater state-provisions in balancing work and home realms. One implication of these differences is that the different options in hours worked and handling childcare may have an effect on the ways that they adjust their home to work proximity. GB women seem to be handling the household responsibility and expensive child care cost with part time working and caring more for their children themselves, while US women are using relatively more affordable child care, part time working being a relatively less attractive working option for them due to lower levels of protection and benefits, US women may be driven to a greater extent than GB women to handle the conflicting roles at home and work by opting to adjust their home and work separation. As planning scholars, we are interested in this research topic, because we would like to understand the reasons behind the differences, and be able to better predict the future of the differences in commute lengths and inform policy makers accordingly. Urban models are usually based on the 4 journey-to-work choices of the man in the households. According to Crane (2007), understanding the effect of women’s roles on their commute lengths may help predict future housing and work place location preferences of women depending on their household responsibilities, lifestyles, the type of households they are in, and it can also help predict future location decisions of employers who want to employ women who may or may not be spatially restricted. A high level of restriction on women due to household responsibilities may cause firms to locate closer to suburban residences. In addition, if there are constraints for women to participate in the labor force, and physically access jobs and services, transportation and land use plans can be developed to overcome them. A gender-oriented insight to commuting may help planners better predict future travel and land use patterns, and to come up with gender-equitable planning policies. In Chapter II, I review the relevant literature in terms of theories and the empirical evidence. In Chapter III, I discuss the basic differences between US and GB that may affect the household responsibility – commute length relationship. In Chapter IV, I explain and discuss my model, dependent and independent variables, their relationships based on the literature review, describe the methodology I undertake, and finally in Chapter 5, I present my findings, and discuss the results. 5 CHAPTER II LITERATURE REVIEW Explanations behind women’s shorter commute lengths have been researched by geographers, economists, and planners. For longer than the last 4 decades, women – especially mothers- have increased their labor force participation rate. However, this has not translated into a proportional increase in gender equality. Division of labor at home and in the labor market remains gender-typed. This is reflected in gender differences in commuting patterns as well. Economists’ explanations view the gap to be on the way to be closing, while social constructionist sociologists focus on cultural and more permanent dynamics at home and at work for commute length differences. Empirical research has produced mixed results. There are methodological problems and a general lack of cross-cultural investigation. In this review, I first discuss the theoretical explanations given by economists’ rational choice theory, sociologists’ social constructionist theories at home and work, as well as geographers’ explanations for women’s shorter commute lengths. I then review the empirical evidence, and cross-cultural research in travel behavior. I conclude with a discussion of the research gap and my research questions. II.1 Overview Systematic involvement of travel behavior research with women’s issues started in the 1970’s. The focus of transportation research was to facilitate the development of U.S. highways and mitigate peak period traffic 6 congestion and thus, women who had not entered the work force in great numbers yet did not receive much attention (Rosenbloom, 2005). Accordingly, early studies in women’s travel behavior research examined constraints that women faced in accessing work (Hanson and Pratt, 1995). Studies on women’s journey to work have established that women travel considerably shorter distances to work than men, and this difference has persisted (White, 1977; Erickson, 1977; Madden, 1981; Hanson and Johnston, 1985; Fagnani 1987; Gordon, Kumar and Richardson, 1989; Johnston- Anumonwo 1992; Hanson and Pratt, 1995; Turner and Niemier, 1997; Hjorthol 2000). The explanations for women’s shorter commute lengths can be divided into two broad and distinct categories: those that explain it as an individual’s voluntary choice, and those that view it as the outcome of social constraints. The approach that regards a short commute as a voluntary choice for women employs utility theory from economics, and argues that individuals are rational utility maximizers; that is, they are capable of assessing which options provide them with the best outcome with regard to achieving a certain goal, and that their actions reveal such optimal choices. On the other hand, the constraint-based approach is from the field of sociology, and argues that individuals’ actions are the outcomes of societal constraints. In this literature review, I will look at first the choice, and then constraint-based explanations. 7 II.2. Theoretical Explanations for Women’s Shorter Commute Lengths II.2.1. Rational Choice Theory There are several explanations in which utility theory has been used to account for women’s shorter commutes. These regard commute as a disutility since commuting incurs costs 1 . These costs are both monetary (cost of car ownership, car use, parking fee, or transit fare) and temporal (commute time). It follows that rational individuals will want to weigh the cost of commuting against the reward they get from the access to their job. Women and men differ in the salaries they get. The main economic explanation for this difference is that women on average invest in ‘human capital’ at lower levels than men, i.e. lower educational investments, job training and thus skills, mainly due to their expectations that they will be in and out of labor force due to child birth and rearing. These different skills cause them to go into different type of occupations, and receive lower wages than men. The first explanation, then, simply attributes women’s shorter commutes to the lower monetary rewards they receive for commuting. The second explanation focuses on women’s occupations and their job search. Since wages from women’s jobs have lower variability, there is low variability in wages across space as well; commuting to jobs farther away offers little marginal benefit. In addition, individuals are likely to search for jobs and be more willing to travel further depending on their level 1 However, there are recent studies that found evidence for commute to be a pleasant time of transition from home to work and from work to home (Blumen 2000, Ory et al 2004, Mokhtarian and Salomon, 1999) 8 of educational attainment, training, or occupation status (see, for example, Hjorthol (2000), Fagnani (1987), and Hanson and Pratt (1995). A third explanation consistent with the first two is that women value working close to home. Due to their weaker labor force attachment and greater house and child care responsibilities, their value of time for working and commuting is high 2 . In addition, the fact that the usual rush hour times coincide with child pick-up and drop off, and meal preparation times also increases the value of the time allocated to long commutes for women (Van der Berg and Gorter, 1997). II.2.1.1 Rational Choice (Utility) Theory for Households Since most women do not live alone but within a household, their commute patterns need to be examined within the contexts of their households (Hofmeister, 2006). While household activities have been studied by economists, it is Gary S. Becker who is known to be the pioneer among those who suggested the notion of the household as a small factory (Lawson, 1998). According to Lawson (1998), Becker (1965) suggested that “It combines capital goods, raw materials and labor to clean, feed, procreate and otherwise produce useful commodities”. It followed that men’s and women’s different roles in households contribute to optimizing the household production function. In two-worker households, a short commute for women is considered a choice that maximizes household utility spatially. 2 This is despite their lower average wages, an indicator usually used to determine one’s value of time 9 Urban economists developed models in 1960s, which predict households’ residence location decisions based on household worker’s job location 3 . In these models, land price exponentially declines as one goes away from the central business district (Alonso, 1964; Mills, 1967; and Muth, 1969). Household’s location choice is a tradeoff between transportation cost of the household worker and land cost. The original theory assumed that there is only one worker in each household. White (1977) extended that to a two- worker household, proposing a model based on maximizing leisure time of the husband and wife, subject to household income constraint. She assumed that the work places of the husband and wife are fixed, and all married women work in the suburbs and all men work at the city center 4 . She demonstrated that, at equilibrium (household utility-maximizing housing location), two-worker households outbid single-worker households for residential locations in the inner suburbs, resulting in a shorter commute for women in two-worker households. II.2.2. Space-time Geography Approach Swedish geographer Torsten Hägerstrand (1970) is well known to have reconceptualized travel and argued that individuals’ travel patterns need to be understood within spatial and temporal constraints (Kwan 1999, 2000). 3 The neoclassical urban economics model assumes one worker per household. 4 White (1977) based her assumptions on 1970 Census of Population, Journey to Work Data, which showed that in six large U.S. cities, 54 percent of married women had suburban jobs, while only 49 percent of men workers and 48 percent of men and single women workers had suburban jobs. 10 According to his conceptualization, “time-space geography”, in an individual’s schedule, activities with a high degree of space-time fixity act as ‘pegs’ around which other activities are organized. Such activities serve to define the size of the individual’s “activity prisms”. Tivers (1985:12) outlines the three categories of constraints identified by Hägerstrand: “1) Capability constraints are those which limit the activities of the individual because of his biological construction and/or the tools he can command, ‘ and include principally the need for sleep and the ability to move around, these determining the size of daily ‘prisms’, 2) coupling constraints, operating within the capability constraints, ‘.. define where, when and for how long the individual has to join other individuals, tools and materials in order to produce, consume and transact, so controlling the individual’s path inside the daily ‘prism’ .. 3) authority constraints refer to limitations and control of access, occurring at different levels to produce hierarchies of ‘space-time domains’.” Hägerstrand’s work was applied to women’s travel by some researchers. Arguing that women have greater spatial and temporal rigidities in their schedules because of their greater responsibility for household activities, these researchers applied women’s greater responsibilities for many child care and household serving and consequent need to coordinate their schedules with their children to Hägerstrand’s coupling constraints. Similarly, women’s limited power in their households and in the society could qualify as authority constraints. Kwan (2000) used 11 Hägerstrand’s time-space prism to examine gender differences in commuting behavior. She constructed space-time “aquariums” for 41 working women and 31 men in Ohio, and found that full-time working women’s time-space patterns were much more restricted than similarly situated men’s. Analyzing 2001 NHTS data, McGuckin and Nakamoto (2005) found that working women in two-worker families were twice as likely as men to pick up and drop off school-age children at school during their commute. Comparing with 1995 results, they found that men are increasing trip chaining, too, but most to get a meal or coffee, while for women it’s to perform household-sustaining activities. Similarly, Nobis and Lenz (2005) using “Mobility in Germany” survey of 2002 of 61,729 persons found that the share of serve-passenger trips was higher for all female groups compared with any male group. The gap was particularly high for men and women living in multi-person households with children. For this household type, the share of serve-passenger trips for women was more than twice that of men (22.8 % versus 9.3 %). They also measured whether the gender differences according to household structure are modified if occupation is taken into account, and found that when working full time, women are more responsible for household duties and child care than men. Activity studies’ consistent results of women’s higher household- related trip making do not find direct relationship between household responsibility proxies and commute lengths, but as explained earlier, indirectly do so. With such rigidities, the activity prism’s size gets narrower, 12 and thus, long commutes are precluded from their time-space paths. Indeed, Kitamura (1983) observed that the spatial and temporal fixity of certain out-of-home activities (e.g. serve-passenger trips) is the most important determinant of a person’s activity-travel pattern. Although the time-space geography argues that it’s the physical and environmental constraints that determine one’s travel patterns, it uses the utilitarian framework like other activity studies, and puts forward the idea that individuals’ actions reveal their preferences within the given constraint framework. II.2.3. Sociological Explanation Both neoclassical utility and time-geography approaches have received criticism from sociologists, among others. The general criticism they received from sociologists was that individuals’ revealed actions may not always represent their preferences, and individuals are not always able to choose the most rational outcome for themselves, to trade-off different options in their decisions, or to bargain on them with others. Tivers (1985 : 11) cited Eyles’ (1971) work which said that “‘revealed preferences’ do not go far enough if an individual does not have the opportunity to behave in a certain way in which his preference can be revealed, and that due to societal constraints, individuals’ real choices may be repressed”. She went on to argue that the main constraint in the way of individuals’ “repressed 13 choices” is societal norms and “structure of society based on specific class and gender relations and cultural contents of choices and demands”. Regarding time-geography research, Tivers (1985) argued that the space-time studies have emphasized the economic, physical and institutional environment, and thus may be overlooking the existence of social constraints. She criticized Hagerstrand’s time-geography approach, and argued that a more sociological approach is needed: “Physical constraints and activity patterns are all the products of the way in which society is structured and organized. Thus, they cannot be treated in a simplistic manner as causal constraints”… ….The presence of young children would not seem to be adequately considered merely within the time-space notion of a coupling constraint. The depth of the gender-role constraint seems much greater than the simple idea of having to be at certain places at certain times in order to attend to the needs of children. Similarly, social class norms of behavior could be considered as ‘authority’ constraints, but again the subjective importance of such constraints is not adequately dealt with in the established terminology.” (Tivers, 1985: 15). In accordance with these sociologists’ suggestion, a social constructive look at women requires a need for looking into their socially and culturally defined roles, that is, the gender roles. According to Tivers (1985), “gender roles are based originally on the objective fact of biological sex roles in reproduction, but are expanded to include cultural roles, which “conditions the differentiation of roles of men and women in society.. and their positions” (Tivers, 1985:18-19). The motherhood role has become cultural, because child care responsibilities have been assigned to women and it has become their chief role. ‘In Western culture today, motherhood is 14 the chief occupation for which females are reared. It is a major component of the feminine gender role as taught to a female child. Further, ‘the father is usually not held accountable for what happens to the children…’ (Tivers, 1985: 20). She explains women’s role for housework. She writes: “while the specific form of gender differentiation varies from one society to another, the existence of such differentiation is in all times and places clearly visible, even in situations where considerable progress has been made towards equality” (Tivers, 1985: 19, citing Lane, 1983). Hence, since women’s chief responsibilities are motherhood and household care, when they join the work force, they still retain their chief roles. Time use studies have been used to illustrate the continuing existence of gender inequalities in the division of household and child care responsibilities. Rosenbloom reviews time use studies and she reports that although working women’s husbands or partners have been assuming more domestic and childcare responsibilities, women are responsible for the majority of all household and childcare chores (Marini and Shelton, 1993). She reviews Bianchi et al’s (2000) work, which analyzed two sources of data on unpaid work: repeated cross-sectional samples (i.e., 1965, 1975, 1985, and 1995) of time diaries and the most recent wave of the National Survey of Families and Households (NSFH2, 1992-94) and found that the type of tasks done by men and women remain different; women tend to do the “traditionally feminine” tasks such as cooking and cleaning while men do “episodic discretionary tasks” (Bianchi et al 2000). According to Rosenbloom 15 (2005), Bianchi et al (2000) report that having young children substantially increases the housework gender gap (Bianchi et al 2000), and in particular, having children under 11 increases the amount of time both spouses or partners put into household chores, but that amount is three times more for wives than husbands independent of employment status. Rosenbloom (2005) adds that, Bianchi et al (2000) report that in the past five years there has been a leveling off in men’s assumption of a greater share of household work, which could indicate merely that men will continue to increase their allocation to housework over the next decades, but at a slower rate than in the 1970s and 1980s. Rosenbloom (2005) also reviews a 2002 study which found that wives still do more than their husbands even in the most egalitarian countries (Batalova and Cohen. 2002, cited in Rosenbloom, 2005). II.2.3.1 Household Responsibility Theory Women’s chief role having been assigned to them by society as domestic responsibilities, trying to juggle the two roles results usually in choosing the employment around the household responsibilities – both temporally and spatially- looking for part time work, and/ or working close to home. Working close to home allows women to better juggle the roles; minimize commute time, be accessible to home in case of emergencies, be able to drop-off and pick-up of children from daycare or school on the way to or from work, etc. This explanation for women’s shorter commute lengths is called the household responsibility theory. 16 Two other household dynamics-related constraints other than household responsibilities that may also lead to women’s shorter commutes have been given in the literature. These issues are also linked to the power division within households (Hofmeister, 2006). First is an issue valid for women outside US: women’s lower level of mobility compared to men. Women have lower access to cars, especially in households with one car. This may lead women to rely on public transport or being a passenger in family cars in being transported to work, especially in countries other than US. Both situations are likely to result in shorter commutes for them in terms of distance. Second, the asymmetric power structure within households affects the sequence in which the housing and respective job locations choices are made. Several scholars have explained that usually, first, head of household’s -men’s- job location is determined, and then residential location, and finally, women’s job location (Kain, 1962; Sermons and Koppelman, 2001; Hanson and Pratt, 1995). This may limit women’s employment options, and thus they may have to settle with the jobs that are nearby. Hence, working close to home is one way women can blend the work and home-related roles. However, having to work close to home usually limits their employment options due to the phenomena called the separation of “home” and “work” in modern metropolitan areas, especially 17 in U.S (Wachs, 1996; Fainstein, 2003) -also referred to as the “separation of production and reproduction spheres”. II.2.3.2 Labor-Market-Related Explanations Dynamics in labor markets and the way labor markets respond to women’s spatial restrictions is another aspect that is put forward to explain women’s shorter commute lengths while complementing the household responsibility theory. According to Hanson and Pratt (1995), in the early 1980s, ”many feminists attempting to explain occupational segregation turned their attention away from the household towards the workplace, in line with a more structural reading of segmented labor markets” (Hanson and Pratt, 1995 : 5). In the current land use patterns in US, jobs that are located more centrally are male-dominated jobs; men commute to these jobs, but since they are far from residences, and women need to balance work with house responsibilities, this situation discourages women from commuting to these jobs. Women’s jobs, on the other hand, are located more peripherally, closer to suburban homes. This phenomenon is verified by scholars who found that commute lengths of women in female-dominated jobs are shorter than that of other women (Hanson and Pratt, 1995), although some scholars oppose the strength of the relationship between female- dominated jobs and shorter commute lengths. For instance, Weinberger (2007) in her study of 9 Philadelphia counties with 1990 and 2000 5 % Public Use Micro Sample (PUMS) found that whether employed in male dominated, 18 neutral, or female-dominated industries, women still have shorter travel times than men. She writes : “The gap narrowed by 2000, but this evidence suggests that sex membership trumps industry type as a determinant of travel time”. There are three reasons given by scholars for women’s jobs being close to homes: The first explanation views it as a natural option: some jobs are population-serving jobs, thus, they are naturally located closer to residences (jobs formerly done at home) such as schooling, cleaners, bakeries, etc. According to some, this reflects the gradual transfer of former- female-tasks to locations outside but near the home (Taylor, 1999). Rosenbloom writes that service industries, in which more women are employed, are substantially more dispersed across a metropolitan area than traditional manufacturing plants and do not create any kind of spatial agglomeration. Thus, women in the service industry can, and do, work anywhere in a region rather than at specific sites like factories, which tend to locate near one another (Rosenbloom 2005, citing TCRP 1998). The second explanation for jobs locating closer to women’s residences is that of employers locating near where women are to profit from women’s job segregation and lower pay [Rutherford and Wekerle (1988) citing a study by IBG Women and Geography Study Group (1984)]. Women usually cannot commute far from home due to household responsibilities and thus, they will accept lower pay for a short commute. This 19 will attract employers who are looking for cheap labor pools [usually educated, over-qualified suburban wives with little turnover rates] to suburbs (Hanson and Pratt 1995). It is argued that sometimes women’s unwillingness to travel as far as men creates localized labor markets (Rosenbloom, 2005) and Rosenbloom (2005) gives Worcester, Massachusetts in the study of Hanson and Pratt (1995) as an example for small labor catchment areas surrounding suburban firms that hire women. Not only do some jobs locate near closer to women’s residences, but new occupations also form to fill this new niche. This notion that women do not go into the occupations traditionally held by men located at central areas, but go into jobs that are close-by, female-dominated occupations, is explained by Hanson and Pratt (1995) as friction of distance causing occupational segregation. The third labor-market related reason given for women’s shorter commute lengths is the way in which employers and employees match in labor markets: Propensity of women to work closer to home is further perpetuated by individuals’ relying on social networks for jobs (individuals socialize within gender) and employers’ usually recruiting through word of mouth process. This notion that economic activities are embedded within social networks is known in the literature as Grannattover’s (1985) “the embeddedness” idea (Hanson and Pratt, 1995). II.II.4 Summary of Theories and Their implications for the Future of the Gender Gap Having discussed the two main constraints spheres of home and work, one may wonder which of the two constraint explanations is more influential 20 on women’s commute lengths: household dynamics or labor market dynamics. Because segregated labor market arguments seems to respond to women’s unwillingness to commute long distances, thus possibly strengthening the household responsibility effect further, it would be reasonable to assume that household responsibilities are the starting point. Overall, although their origins and focuses are different, research in the two arenas seems to support each other since the home and work constraints go together. For example, Hanson and Pratt (1995) conducted interviews with 162 women, and found that after a child-related break, women are more likely to go back to employment in a female dominated (a less prestigious, lower paying) occupation. In addition, among women with short commutes (<= 10 min), 32 % of those who were in female-dominated jobs listed the reason for their short commute as wanting to be close to home in case of an emergency, slightly higher than among all women with shorter commutes (29.5 %). Their findings show the interrelatedness of both household responsibilities and occupations. Summarizing, the choice and constraint theories have very different implications for the future of gender differences in commute lengths. The choice theory gives importance to economic factors and rationality. This means that when one controls for human capital, household responsibility should not affect on commute lengths (Madden 1981). Hence it argues that discrimination will diminish in the long run, and there will be a narrowing of 21 differences (Spain and Bianchi, 1996). The implication of the constraint theory is the opposite. It argues about the importance of cultural and social factors; not dismissing the importance of economic factors, but arguing that the economic factors exist within social contexts. Therefore, in the future, the differences will not diminish, unless some policy changes take place. Consequently, the future implications of the two theories on the gender gap in commute lengths are very different. II.3. Discussion of Empirical Evidence These divergent explanations have been investigated empirically by several scholars. It has proven that measuring the rational choice theory is easier, since data is relatively plentiful on economic factors such as education, occupation, and incomes. Measuring social constraints such as household responsibility, on the other hand, is more difficult. Large travel behavior surveys do not normally gather information on division of household responsibility, or attitudinal data. Therefore, most researchers have had to rely on proxies of household responsibility such as household composition and life cycle variables. One exception is Hanson and Pratt (1995) who directly asked survey participants with shorter commute lengths whether their household responsibilities were responsible for their short commute lengths. They got affirmative answers. Studies vary on which proxies are used to measure household responsibilities and how such variables as commute length are measured. Measuring the gender gap across different life cycle groups is one way. 22 Another way is measuring the extent to which life cycle variables explain the variability of commute lengths among women. Some studies furthered this by comparing the relationship for women with that for men. Overall, most studies utilized a combination of these methods. Finally, there were also differences among the studies in the ways they controlled for the economic factors. Most studies looked at marital status and the presence of children as proxies of household responsibility. Using 1977 and 1983 NPTS data and looking at effects of marital and parental status, Gordon, Kumar and Richardson (1989) found that all female groups made shorter commutes than men, and that there was no difference in trip lengths between single and married women or between women with children and women without children. However, they did not include significance levels in their analysis, and some differences, although small, may be statistically significant. In the second part of their study, the authors conducted a regression analysis of commute length and time on household income, household size, occupations and location (central city or suburb), gender, marital and parental status. They found that the gender variable was significant and negative for women, and that married people made longer commutes, while presence of children did not affect commute lengths. However, they did not account for the interactive effect of gender and presence of children and marital status. Turner and Niemier (1997) extended Gordon, Kumar and Richardson’s study with 1990 NPTS, doing a gender-interactive 23 regression analysis. Their regression analysis of commute times and distances on parental and marital status, controlling for education level, demographic factors, number of adults in the household, and level of urbanity of the residence concluded that both presence of children and being married were significant in explaining women’s commute distances, while they were not in explaining men’s commute distances. Turner and Niemier’s (1997) results are supportive of household responsibility theory. Villeneuve and Rose (1988) in a comparison of 1971 and 1981 data for Montreal, found that in 1971 both the presence of children and marital status was effective in explaining women’s commute lengths; in 1981, it was only marital status that was effective. A number of studies incorporated age of children into the investigation. Hanson and Johnston’s 1985 study of Baltimore MSA in 1977, utilizing ANOVA technique, found that marital or parental status did not explain differences among women, but they did explain the differences among similarly situated men and women (except for those with preschool children). They found that women’s shorter trips are mostly explained by their lower incomes, their concentration in female-dominated occupations, and their greater reliance on the bus and lower access to cars. Thus, their study has only ambiguous result on household responsibility. Hjorthol (2000) also included children’s age group in her examination of commute lengths of men and women in Oslo area. She found support for household responsibility theory in a 1990/91 study of married couples (both parties in 24 paid work) from a personal travel survey of 398 women and 303 men. Controlling for labor force and location variables, she conducted multiple regression analysis for both men and women, and found that women with children in the lower primary school (7-12 year old) were less likely to travel to work as long as women without children in this age group were. She attributed this to greater schedule constraints of mothers with children in this age group – explaining that first couple years of school being unlike kindergarten, has shorter hours, and includes fewer leisure activities. For married men, however, children had no effect on commute length. Since children shortened women’s commutes, but did not affect men, this study supports household responsibility theory. Pazy et al (1996) in a different way of measuring the dependent variable, used not commute length, but willingness to increase commute length in a hypothetical situation as their dependent variable, arguing that this measures individuals’ repressed preferences better than revealed preferences. They interviewed 162 Israeli professional women in information processing (the researchers chose survey participants from this occupation because it is not gender-typed high prestige) in the Tel-Aviv area. The majority of respondents expressed willingness to extend their journey to work for a career improvement. Willingness increased with career association. Education level, rank and weekly working hours did not have a significant influence. They also found that mothers of young children were less inclined to travel more, especially those who were dependent on public transport. Its 25 result points to importance of household responsibilities in limiting women’s willingness to extend their commutes. Some studies used number of children as an approximation of household responsibility. For instance, Fagnani (1987), using 1981 National Sample data for Paris, France of 1,827 working mothers from France, investigated the effect of number of children on commute times. She found that average commute time decreased as the number of children increased. However, although having children on average reduced commute times of women, it did not affect commute times of management and office workers and other professionals. Also, the highly educated (with high school diploma or more) were affected only when the number of children increased from two to three. White (1986) used data for New York City from the Annual Housing Survey. She estimated a reduced-form commute time equation for both male and female heads of households on the number of children under age 18 in the household, the presence of pre- school children, the presence of a secondary workers, and control variables such as household income, and the race of the household head. She found that female heads of households’ commute times were not significantly affected by the number of children, or by the presence of a secondary worker, but they were lengthened significantly by the presence of pre-school children: young children increased commute times of female heads of households by 8 minutes. It did not, however, significantly affect commute times of male heads of households. Turner and Niemier (1997) caution that 26 White’s results for the New York City overrepresents low-income and minority women, and thus can’t be generalized to larger populations. It should also be taken into account that New York is a very peculiar city in terms of its spatial structure and transportation system; thus commute times are affected by such different factors, and cannot represent other cities. One study that was very different from others in the operationalization of household responsibility in the explanations of commute lengths of women and men., and took a more cultural approach was Kawase’s (2004) study from Tokyo, Japan. Its particular results on unmarried young women have interesting implications. The author examined commuting behaviors of husbands, wife, sons, and daughters in 179 households, and found that both wives and daughters operated under constraints: while wives made shorter work trips, daughters had to lengthen their trips, because it is not culturally appropriate for a young unmarried woman to live alone and they lived with their parents. Kawase’s study shows how culture plays a role in commute lengths of unmarried young women and acts as a constraint. Some studies have focused on the number of workers in a household to explain differences in women’s and men’s commute lengths. Johnston- Anumonwo (1992), using 1977 Baltimore data, and controlling for income, travel mode, and occupation type, and utilizing one-tailed t-statistic, found that the presence of children did not affect gender differences in commute lengths significantly. However, number of workers in a household was significant in explaining gender commute time differentials- the differentials 27 were greater in households with two- workers than those with single worker. She argued that the dynamics in one worker families is very different than that in two-worker families. This makes sense, since women in female- headed families are under a different set of pressures and financial responsibility may generate longer commutes. In a large-scale study in the metropolitan area of Seoul, Korea, Lee and McDonald (2003) found, unlike Johnston-Anumonwo (1992), that the number of workers in the household did not affect women’s commute lengths. They also found that single workers had longer commutes than their married counterparts and married workers with a working spouse had shorter commutes than do married workers whose spouses did not work, and added that these differences are somewhat larger for women, and the presence or number of children shortened women’s commute lengths. One of their most interesting results that could also relate to cultural factors is that the presence of older parents has significantly positive impact on commuting time of both men and women, although the impact is smaller in the case of men. II.3.2 Summary of Empirical Evidence In sum, the literature reviewed above shows that there is some empirical support for the household responsibility hypothesis in the literature, although the support is very fragmented over time and place, mixed, and far from overwhelming, even contradictory. There is some ambiguity and inconsistency in results as in the findings of Hanson and Johnston, 1985; Gordon, Kumar and Richardson, 1989; Turner and Niemier, 1997; and 28 Fagnani, 1987; and White, 1986). One reason for this inconsistency could be that there is great variability in operationalization of household responsibility; studies have used very different measures of household responsibility. Most studies used proxies; examples being number of children, or marital status. The issue with proxies is that they only approximate the level of household responsibility in the totality of the household; they don’t say anything about the division of it. Hence, these studies are not really measuring the “disproportionate household responsibility” effect. This may be contributing to the inconsistency of the results. Another issue is that of generalizability. Studies done in small scales in one specific metropolitan area run the risk of being too specific to apply to other places. Time is also a factor; a large portion of the literature extends back to 1970s, thus their applicability to today may be limited. Also, the fact that some studies were limited in their sample sizes made it difficult for them to control for some variables (via for example, multiple regression which requires relatively large samples). One way to rectify this problem is improving the measures. Lifecycle encompasses more than just parental or marital status. It’s likely that a composite variable of life cycle would be a more appropriate measure of household responsibility. Another way is to enhance the contextual variability among research subjects. Although women’s taking on more household responsibilities is universal, division of household responsibilities is different in different societies and cultures. 29 As seen in this review, there is a good deal of work on the topic from different countries, and these enhance the knowledge on the topic greatly; however, it is difficult to directly compare their results against each other without doing a comparative study. According to social constructionist perspective, rational choice should explain behavior only to some extent, since gender differences according to household responsibility theory are socially constructed. Although women’s taking on more household responsibilities is universal, division of household responsibilities is different in different societies and cultures. II.3.3 Cross-Cultural Differences on Gender Division in Household Responsibilities There have been studies showing the importance of cultural and political context on the gender division of household responsibilities. For instance, Rosenbloom (2005) in her review of studies on the gender division of household labor, reviewed Fuwa (2004) which found that “macro-” level variables such as wage rates and the state of economic development were more effective in explaining the division of household responsibilities than “by any variables open to individual determination (such as personal attitudes about equality in household roles).” Hook (2006: 642) proposes: “ national context, conceptualized as women's employment practices and policies, influences men's unpaid work behaviors by shaping the benefits of specialization, the terms of bargaining, and the ease of adhering to gender ideologies and norms”. Using 44 time- 30 use surveys from 20 countries (spanning 1965 to 2003) combined with original national-level data, the author utilizes multilevel models to test hypotheses regarding the relationship between national context and men's unpaid work behaviors. She finds that “men's unpaid work time increases with national levels of women's employment. Furthermore, the effect of children on men's unpaid work time depends on women's national employment hours, the length of available parental leave, and men's eligibility to take parental leave, which indicates that particular public policies affect men in specific household situations. The analyses document the importance of national context for the unpaid work behaviors of all men, especially fathers, and shift the research focus from the attributes of individual men to the structures that hinder and facilitate men's unpaid work” (Hook, 2006:639). Such differences in cultural norms and political environments may have implications regarding (household responsibility’s affect on) women’s commute lengths. For example, the effect of family policies on the relationship between household responsibility and women's commute lengths can be two-fold: On one hand, provisions that decrease the work- home conflict can help women juggle work and home easily because women have more flexibility in their schedules. Total time they can allocate to commuting may not be as restricted as it could be for women in regimes without such provisions. Thus they may have more opportunities to participate in the labor force, and make longer commutes. On the other hand, the status of women with work-home balance provisions in the labor 31 market may not be as strong. For example, they may be penalized by lower wages for working part-time (part-time wage penalty). They may also have more reinforced traditional household roles since they have more time to be involved with housework. Therefore, such women may have a more traditional gender role in the household and thus, experience more temporal and spatial rigidities in their schedules to carry out household responsibilities, which may prohibit commuting longer distances. II.3.4 International Research on Travel Behavior A review of the literature shows that almost no study done on the topic had a cross-cultural outlook. Simma et al (2002: 2) write: “It is indeed surprising, how little work has been done to model national travel surveys, either individually or comparatively; only the recent availability of advanced modeling software is starting to change this situation”. Schafer (2000:1) writes: “Although travel demand characteristics have been analyzed at all aggregation levels (individual, urban, regional, national, world-regional, and global), surprisingly little research has been dedicated to quantifying and comparing travel characteristics across national boundaries. Such cross- country comparison is important, since it can reveal general trends and differences in the evolution of travel demand, possibly leading to a better understanding of underlying forces”. There have been quite a few studies that investigated travel behavior across countries at aggregate levels such as modeling studies that measured travel demand or motorization at aggregate levels (Schafer, 2000; 32 Dargay and Gately, 1999; Ingram and Liu, 1999). There have also been aggregate- level quantitative comparative studies; those comparing urban modal splits in European cities and relating them to respective land use and car taxation policies (Pucher, 1988); those measuring average transportation energy consumption and relating it to urban land-use (Newman and Kenworthy (1989 and 1999); and less quantitative, more descriptive analysis of European transportation systems (Salomon, Bovy and Orfeuil, 1993). However, studies that tested particularly the household responsibility hypothesis on gender differences in travel behavior –let alone commute lengths - in a comparative international way are much fewer. One exception is a Rosenbloom (1989) study, which employed a comparative approach in measuring trip chains in Lyon in France, Rotterdam in Netherlands, and Austin in U.S. She restricted her sample to those individuals working full time from 450 families. The study was based on attitude surveys carried out between 1982 and 1985. She found that women’s travel patterns varied significantly with the age of their youngest child in all three countries, and that women had more complicated activity patterns than comparably situated men. Further, men in all three countries were alike in that they tended to make fewer linked trips to work than either single or married mothers. She also found that US married women were more likely to link trips to work than their counterparts in Europe. Her results imply support for household responsibility hypothesis, and more so for US women when complex trip making is considered as an indicator of household responsibility. 33 As reviewed in the empirical review section, there are quite a few disaggregate travel behavior studies (e.g. Hjorthol, 2000; Lee and McDonald, 2003; Villeneuve and Rose, 1988), however, they are all within individual countries, that is, none of them are comparative. Studies that address travel patterns – let alone commute lengths- comparatively in a disaggregate fashion are rare. Exceptions are studies by Giuliano and Narayan (2003) and Giuliano and Dargay (2006). Giuliano and Narayan (2003) explored the relationships between urban form and travel behavior using individual travel diary data NPTS 1995 for US, and 1995/97 NTS for UK. Using the same two datasets, Giuliano and Dargay (2006) examined the relationship between car ownership, daily travel and urban form. II.4 Conclusion With the dearth of comparative disaggregate research studies on travel behavior, most studies regard American studies as universal and the norm, and in doing so, disregard the importance of context in the determination of travel behavior. One reason for this scarcity is the lack of large-scale comparable international data on travel behavior and household responsibility indicators. Another reason is the undermining and ignoring the value of a comparative approach in travel behavior. Without incorporating a cross-cultural perspective, one cannot appropriately determine the factors that lead to women’s shorter commute lengths, or measure the effect of household responsibility hypothesis. Thus, cultural context forms one of the key missing pieces in the current literature in 34 household responsibility-commute length relationship. This dissertation seeks to fill this gap. It tests the classical household responsibility hypothesis through an international comparison. Incorporating a cross-cultural approach will help expand the knowledge on the subject, and improve the generalizability of the findings to other countries as well. This is especially timely as gender role transformation has been and is taking place across the world. In chapter III, I explain how I test it, my model, and the methods employed in carrying out the study. In chapter IV, I present my results, and in chapter V, I discuss my findings, 35 CHAPTER III RESEARCH APPROACH AND METHODOLOGY The literature review chapter established that the household responsibility hypothesis should be tested in diverse populations in order to be generalizable to other contexts. This study will test HRH in the US and GB 5 . One difference between US and GB that is likely to have influence household responsibility-commute length relationship is in the relative incomes of their populations. In 1999, GB median household income was about 2 thirds of the corresponding US value ($ 21.8 K versus $ 33.9K, ppp- adjusted). No doubt related to that, car ownership in GB is much lower compared to US. Cost of driving is less favoring of private transport in GB. 1995 fuel price per liter was $ .90 for GB, and $ .30 for the US, a consequence of starkly different fuel tax policies (Giuliano and Narayan, 2003), thus, the price of travel by auto is relatively far higher in GB. This has an impact on women’s access to cars in GB, since if there’s one car in the household, it is usually men who get to drive it, and women to be the passenger. Another factor that shows significant difference between the countries is in land use and transportation infrastructure patterns. Since most GB cities were developed much before automobile became widespread, US cities are lower in density, while GB cities are denser, and GB has much 5 Great Britain refers to England, Scotland and Wales, but does not in include Northern Ireland. 36 greater transit accessibility. Additionally, GB has instituted greater control policies on driving in central cities (Giuliano and Narayan, 2003). Next, I discuss the place of women in US and GB, with respect to first, their place in the labor market, and second, state provisions in helping women’s dual roles through policies and provisions. I show that between US and GB, US women with children have a stronger place in the labor market, while GB women benefit from greater state-provisions in balancing work and home realms. US and GB have similarities and some important differences in support options and labor market roles for women, which make them comparable but still divergent contexts for women’s household roles and commuting. Both are Western countries with Anglo origins valuing individualism, market orientation of their economies, and both being at the post-industrial stage of economic development. Partly because of their individualism, self-reliance, and viewing of family as a private unit, they are among the most liberal countries, with minimal levels of state-provided social services to their populations among the Western countries (Saunders, 2001; Kamerman and Kahn, 1997). III. 1 The Gender - Political Context and Labor Force Characteristics Although both countries fall into liberal welfare regimes according to a classification system developed by Esping-Andersen (1990) 6 , several 6 Esping-Andersen (1990) classified welfare regimes into three groups as liberal, social democratic and conservative. 37 researchers classify US as or ‘pure’ liberal, and UK to be ‘mixed’ liberal (Tomlinson, 2007). Accordingly, there are greater provisions in policies that aim to balance women’s work and home responsibilities in GB. Greater maternity leave and part time worker protection are among such provisions. As of 1995, GB had had a paid maternity leave provision since 1978 of up to 18 weeks, while US merely in 1993 enacted a policy FMLA (Family Medical Leave Act) which provided a 12 weeks unpaid leave (Kamerman and Kahn, 1997). GB also has formally provided protection to part time workers since early 1995. Since then, part time employees have officially had the same rights as full-timers for purposes of the UK employment protection legislation 7 , including the provision of health care benefits 8 . In terms of another family-friendly policy, provision of child care, GB was ahead of US, more children were enrolled in publicly funded child care in UK compared to US. In US, 14 % of US children aged 3 to 5 were in publicly funded child care, while the proportion in UK was 38 %. Despite this, UK lags behind all of Europe (Kamerman and Kahn 1997) 9 . 7 Part-timers gained the right to the same rates of pay and conditions of employment as full timers through the Part-time Workers (Prevention of Less Favourable Treatment) Regulations 2000 SI 2000/1551 as from 1st July 2000. 8 Wegner (2001) reports that as many as 88 per cent of women working part-time in the US are not eligible for health insurance from their employers, while over 80 per cent of full-timers have healthcare insurance as a fringe benefit. 9 Tomlinson writes that childcare provisions and services in the UK have been flagged up time and again as poor given the services and provisions in other EU nation states: “Currently, childcare consists of uneven and patchy private services but, more often, women rely on informal care provided within the family.. the UK has historically been cited as having one of 38 Several work force indicators show that women in US are integrated into labor force at greater level than women in GB (see Tables III.1 through III.5). In general, US women work more hours (Jacob and Gerson, 2004), and have greater labor force continuity (Steier et al, 2001). In 1995, 25 % of US women worked on a part time basis, while the percentage was 37 % in GB (OECD, 1999). GB women’s part time work level increases with having young children. Table III.3 shows that 60 % of mothers of preschool children in GB worked part time in 1994, compared to 20 % in US (Steier et al, 2001). Percentage dropping out of labor force among mothers of preschool children was also substantially higher in GB. Harkness and Waldfogel (1999) did a comparative study that included UK and US, and found that gender gap in employment and earnings for parents were higher in GB than they were in US (Table III.4 and Table III.5). US also ranks higher than UK in GEM (Gender Empowerment Measure), a measure created by UNDP to measure gender disparities in participation in political decision-making, access to professional opportunities, and earning power" 10 (UNDP 1995: 72 cited by Fuwa, 2004). the worst provisions of childcare in the EU and, despite the recent expansion of childcare places under the National Childcare Strategy, as of 2002, there were still 6.6 children under eight years old for every available childcare place in the UK (Tomlinson, 2007: 410). 10 GEM is an index constructed from four dimensions (percentage of women in parliament, percentage of women among administrators and managers, percentage of women among professional and technical workers, and earned income gap between men and women). It ranges from 0 to 1, where 1 is the most egalitarian country. US index is 0.68 and GB index is 0.46. 39 Childcare is one of the main factors affecting women’s employment levels. British women’s greater involvement in part time work or discontinuity in employment with child birth is explained by several researchers by inadequate and expensive childcare in GB (Dex and Shaw, 1986; Wilman- Navarro, 2006; Tomlinson, 2007). In the same light, American women’s greater participation in labor force, greater level of full-time work, and higher levels of labor force continuity with respect to GB women is explained by Wilman-Navarro (2006) through market forces. She attributes American women’s greater employment levels to loose labor regulations and the availability of informal, mostly immigrant labor for childcare. Concurrent with Wilman-Navarro’s (2006) argument, a look at the results of a study done through Luxembourg Income Study (LIS, 2003) on the wages of care staff for preschool children shows that in UK, in 2000, equivalent (ppp-adjusted) full- year full-time wage of ECEC (Early Childhood Education and Care) staff ($ 15K to $ 20K) was substantially higher than that in US ($ 13K). Similarly, the share of care staff’s wages among all employed women’s annual wages was between 1.4 to 1.5 in UK, more than twice that of 0.66 in US. Having reviewed the family policies, women’s labor force characteristics, and government- and market-provided child care options in the two countries, one can conclude that US and GB provide two different contexts for women’s roles at home and work and for balancing home and work responsibilities. US seems to provide a context for greater “gender sameness” at work (Orloff, 2002), while there is more room for greater 40 balance in GB between roles at home and at work, albeit evidently at the expense of stronger roles in the labor market. Do the different contexts discussed above have any implications for relationship between women’s HR (Household Responsibility) and commute length relationships? One implication is that the different options in their hours worked and handling childcare may have an effect on the ways that they adjust their home to work proximity. GB women seem to be handling the household responsibility and expensive child care cost with part time working and caring more for their children themselves, while US women are using relatively more affordable child care, part time working being a relatively less attractive working option for them due to lower levels of protection and benefits, US women may be driven to a greater extent than GB women to handle the conflicting roles at home and work by opting to adjust their home and work separation. Table III.1. Percentage of Women 16 Years and Older, by Employment Status 1980 1990 2001 Ft pt N.E. 2 ft Pt N.E ft Pt N. E. US 34.4 13.3 52.3 39.9 14.4 45.7 43.1 13.9 43.0 UK 1 26.8 18.0 55.2 30.1 19.7 50.2 31.2 21.5 47.3 1 1983 instead of 1980 for the United Kingdom. 2 Unemployed or not in the labor force. SOURCE: In Martin and Kats, 2003 from US: BLS, April 2003 UK: OECD, Labor Force Statistics, 1980-2000; Labor Force Statistics, 1981-2001. Threshold for full time work: 30 hours 41 Table III.2. A Comparison of Women’s Work Hours in US and GB US GB Ratio of Mothers' to Fathers' Hours 0.78 0.6 Numbers of hours on an average work week With at least one spouse employed 72.3 57.4 Dual earner 81.2 74.3 Source: Jacob and Gerson (2004) Table III.3. Women’s Continuity of Labor Force Participation, US and UK US UK FT PT N.E. FT PT N.E. Employment before children 68 7 2571 10 19 Preschool children 35 20 45 15 57 28 School age children 55 20 25 23 51 26 N 1,447 984 Source: Steier et al (2001) from ISSP 1994 “Family and Changing Gender Roles” Survey Table III. 4. Employment Rates, all workers aged 25-44 UK 1995 US 1994 All women 0.64 0.66 Non-mothers 0.84 0.788 Mothers 0.552 0.598 Family Gap (mothers-non-mothers) -0.288 -0.19 Source: Harkness and Waldfogel (1999) Table III.5. Family and Gender Gap in Mean Hourly Earnings Women’s Wage/All Men’s Wage UK 1995 US 1994 All women 74.5 78.3 Non-mothers 82.2 82.9 Mothers 69.6 75.6 Gender Gap -25.4 -21.7 Family Gap (mothers-non-mothers) -12.6 -7.3 Source: Harkness and Waldfogel (1999) Table III.6. Wages According to Job Status by Sex, US and GB Full Time Part Time Male Female Gap Male Female Gap US 16 12.7 21% 9.3 9.9 38% GB 10.4 8.4 19% 7.9 7.1 32% Source: Hirsch(2004) for US, and Olsen and Walby (2004) for GB. 42 III.3 Research Questions and Hypotheses: Based on the literature review, and the section above that described the differences between the countries above, the following research questions are asked: • Do household responsibilities shorten women’s commute lengths? • Do household responsibilities shorten women’s commute lengths more than they do men’s? • Do household responsibilities shorten American women’s commute lengths more than they do British women’s? • Are the gender gaps in commute lengths larger in US than in GB? The following are the hypotheses that the study will test: • Household responsibilities shorten women’s commute lengths • Household responsibilities shorten women’s commute lengths more than they do men’s. • Household responsibilities shorten American women’s commute lengths more than they do British women’s. • Gender gaps in commute lengths are larger in than those in GB. III.4. The Model In order to test the hypotheses developed above, I develop a model in which I define dependent variables as work trip distance and work trip time. Both measures have been used as measures in studies on commute length, and they have values in measuring HRH in different ways: distance is 43 a better indicator of the limits of an individual’s activity space, while time is more directly related to behavior. As the literature reviewed illustrated, women are restricted in both time and space. Therefore, I include both as dependent variables in my model. Based on the literature review in Chapter II, commute length varies by individual demographic and economic characteristics; household characteristics such as family structure and household resources; and land use. The model can be represented, then, as follows: Y= f (X, L, H, S, R) Where, Y= Commute distance or time X= Demographic and economic attributes of the individual H= Household related factors (resources, household structure, roles) L= Land use characteristics of the individual’s residence S= Sex R= Country, representing contextual environment The relationships between these variables are assumed to resemble that inn Figure III.1. There were some considerations in the decision of which individual and household variables to include in the model. One issue was that car ownership and travel mode doubtlessly affect individuals’ commute lengths since faster modes make further destinations accessible. However, including them will cause inconsistencies in the model, because they are 44 endogenous to my model as they themselves are affected by the rest of the explanatory variables as described above. One way to avoid the endogeneity problem of car ownership and mode choice is using a reduced-form model - car ownership and mode choice will be omitted from the analysis. Excluding them will no doubt have repercussions on the results Figure III.1 The Conceptual Model: Political, and Cultural Environment Land Use and Transportation System Individual Characteristics Income Sex Age Employment Household Characteristics Roles Income Household Structure Commute Length Activities 45 since US and GB have differences in car ownership and travel mode in journey to work. I will rely on income variables to account for car ownership and mode. Another issue is that residential location and job location may have a bi-directional relationship with commute length, meaning that desire for a certain commute length itself can influence one’s job and residential locations (Mokhtarian and Salomon, 1999). In this model, I take a household’s residential location as a given, therefore assume that individual’s commute length is a short term decision relative to residential location decision. III.4 Data For the study, I used the US 1995 Nationwide Personal Travel Survey (NPTS) and the UK 1995/97 National Transport Survey (NTS). 1995/97 NTS survey was conducted from 1995 through early 1997. NTS data is collected for individuals and households in Great Britain (England, Wales, and Scotland). Both surveys are household-based surveys that provide detailed data on all trips of individuals during a period, social and demographic characteristics of the individuals, their households and some locational characteristics. Trip information in both surveys includes origin, destination, trip start and end times, trip distance, duration, trip mode and purpose. NPTS data is collected on an assigned travel day, while the NTS data is collected over 7 days. 1995 NPTS surveyed 42,033 households with 95,360 persons total. 46 1995/97 surveyed 9,688 households with 23,167 persons. NPTS Participating households were among those selected by a stratified random sampling method from a telephone number sample. The NTS is based on a stratified multistage random probability sample of households drawn from a Postcode Address file. NPTS 1995 was collected from all persons in household 5 years or older. NTS 1995/97 was a phone interview followed by a seven-day diary kept by each member of the household. Information was collected from all household members, and adults kept diaries for younger children and others unable to provide information on their own behalf. The NTS data are not weighted since the sample is presumed representative of the population based on the way the sample is chosen. In NPTS 1995 data, there is an option to weight the survey data to the total US 1995 population to account for sample design and selection probability, non-response bias and non-coverage bias. I elected not to use weighted NPTS data in order to have a compatible data set to unweighted NTS, and with the consideration that I controlled for potentially bias-causing factors such as household income, population density and metropolitan area size. III.5 Study Sample In US, of the 95,360 persons surveyed, 51,717 (54.2 %) were employed. Of them, 30,429 made a work trip on travel day (a trip originating or ending at work), and income data was available for 25,659 of them. Since income 47 is one of my key variables, I excluded people who made no commute trips or did not report income. These 25,659 people formed my sample. As discussed above, NTS travel diary was for 7 days. To be compatible with the US NPTS data, I needed to pick one day out of the 7. I chose day 1, considering that people would be less fatigued and more accurate in making and reporting their observations on day 1 11 . Day 1 is revolving (can be a different day of the week for each person), and could be a weekend day as well. In GB, of the 23,167 persons surveyed, 10,168 were employed. Of the employed individuals, 5,119 workers made a work trip on day 1. GB does not have the missing income data problem, thus, all 5,119 were in my sample. Table III.7 Drawing the Study Sample US GB Surveyed 95,360 23,167 Employed 51,717 10,168 Made a Work Trip on Travel Day 30,429 5,119 Income Data Available 25,659 5,119 Made a Direct Home-to- Work or Work-to-Home Trip 22,735 4,916 Estimated Work Trip was Less Than 75 miles and 120 minutes 22,112 4,874 11 Salathiel (2004) reports that in NTS, number of trips recorded declines through the week due to fatigue. 48 III.6 Operationalizing Commute Length: Estimation of Commute Distances Since I am interested in the geographic context of the separation of home and work for women, I am interested in the direct home to work or work to home trips as a definition of commute. About 4 % of individuals in GB, and 12 % of individuals in US had no direct home to work or work to home trips. Estimating the commute lengths of these individuals created a problem, since I only had information on their linked trips and no information on the direct distance between their home and work, or the time it takes for them to travel directly between the two points. Therefore, I eliminated from my sample all persons who did not make at least one home-to-work or work- to-home direct trip on the survey day. Since people may report varying number of work trips on a day, I took the first direct work trip of a person (either home to work or work to home) as representing that person’s work trip, and recorded that trip’s distance, duration and mode as the person’s typical work trip distance, duration, and mode. This estimate should represent the work trip distance well, since work trip distance should not change based on other conditions. In terms of work trip time, the time it takes to take a trip depends on several factors, including the time of day and congestion on the street, thus, the estimates on work trip time should be taken in that light. Also, I took the mode of the particular trip as work trip mode of the person, although in real life, it is possible that individuals use different modes to travel to work at different times. Finally, my estimation 49 assumes that a person works at only one job, and one job site, whereas in real life one may work at more than one job or job sites. In US, of the 25,659 in my sample, 22,735 made a (at least one) direct home-to-work or work-to-home trip. In GB, on the other hand, of the 5,119 in my sample, those who made at least one direct home-to-work or work-to- home trip numbered 4,916. I analyzed the work trips of those individuals who had commute distances of up to 75 miles, or commute time of up to 120 minutes, since the rest provided extreme cases, and affected the analysis disproportionately. These accounted for less than 1 % of all individuals in US and GB. In US, 22,212 survey participants had commute distances up to 75 miles and commute times up to 120 minutes. 4,874 of them had commute distance under 75 miles, and commute time under 120 minutes. About 1 % of work trip lengths were given as “under a block” coded as 9997. I coded them as 0.1 miles. The following section discusses my approach to measuring household responsibility. III. 7. Operationalizing Household Responsibility Understanding how a household’s is not an easy task since it requires a measure for a relatively unquantifiable concept such as household responsibility. In my inquiry about the role of household roles/responsibilities in determining commute distance and time, an ideal measure of household responsibility would be information on individuals’ actual share of household responsibility, possibly together with gender attitudes, and childcare arrangement. However, we lack this kind of data in travel surveys. Neither 50 do we have data on individuals’ time use or attitudes towards women’s employment, other gender attitude of the individual or his/her partner in NPTS and NTS (as in most large scale travel surveys), towards housework. Therefore, like many other studies, I used household structure as a proxy for household responsibility. III.7.1 The Measurement Variable I formed a composite life stage variable, aimed to represent the stages a typical traditional household goes through. It incorporates age of an individual, number of adults in a household, presence of children, and age of the youngest child in a household. Below are the categories I formed: Table III.8. Life Stage Categories LS1 1 adult, no child, age 16-34 LS2 1 adult, no child, age 35-64 LS3 2 adults, no child, age 16-34 LS4 2 adults, no child, age 35-64 LS5 2 adults, youngest child pre-school age LS6 2 adults, youngest child school age LS7 1 adult, youngest child 15 and below LS8 Other household The sequence LS1-LS3-LS4-LS5-LS6 represents the stages an individual traditionally goes through, from being young and single to being young and part of a couple, to addition of children, first preschool age and then to school age. LS2 represents an individual who did not follow the traditional life style, and LS7 is a single parent living with a child under 15. I considered those 16 and above as adults. 51 i) Number of Adults: The dynamics of household responsibility hypothesis best applies to 2 adult households. Based on the discussion above, married women take on more household responsibility than single women after they get married, with their husbands traditionally taking on the main breadwinner role. One can test the HRH by comparing the behavior of individuals in 1 and 2 adult households. I exclude those with greater than 2 adults because evidence on the effect of having more than 2 adults on women’s commute lengths are mixed . 12 ii) Age: The categories I formed are 16 to 34, and 35 to 64. These two age groups signify very different life stages. As individuals mature to their mid 30s, usually their tenure and earnings at work increase, so an increase in commute length is expected. However, according to household responsibility hypothesis, the effect of age may be the opposite, since women’s household responsibilities start reaching their peaks at early to mid 30s. iii) Presence of Children: Obviously, women with children have more household responsibilities than women without children, and the gender division of household responsibility becomes more disproportionate with presence of children based on time use studies. Therefore we should expect the household responsibility effect to be felt more strongly on the commute of women with children. As reviewed in the literature review section, 12 Some studies found that additional adults in a household may lessen the housework and child care load on the mother and relax constraints on commute length (Kamiya, 1999). In other cases, it may add to the existing household responsibility. Others found no effect of additional adults on women’s commute lengths (Turner and Niemier, 1997). 52 according to Rosenbloom (2005), Bianchi et al (2000) report that having young children substantially increases the housework gender gap, and in particular, having children under 11 increases the amount of time both spouses or partners put into household chores, but that amount is three times more for wives than husbands independent of employment status. iv) Age of Children: Many studies examining the HRH on women’s commutes do not consider the age of the children in their analysis. However, there can be a distinction in the household responsibilities’ effect depending on the age of children. The NPTS data provides categories for the age of the youngest child as: 0 to 5 and 6 to 15. The NTS categories are 0 to 4, 5 to 12, and 13 to 15. I combined the categories 5 to 12 and 13 to 15 in NTS to make it comparable to the NPTS categories. I categorized those with children older than 15 under ‘other’ (LS8) category. Women with children at preschool age are bound to home more than others, since children are more dependent on their mothers the younger they are. Women with preschool age children – especially with child birth- are more likely to drop out from the labor force than other women. Some go into part time work, while others may take a job closer to home as a strategy to balance home and work responsibilities (Tivers, 1985). Compared to mothers of school age children, situation of mothers with preschool children have more flexibility since school age children do have to be taken to school, but alternatives to day care can be arranged for some women, such as grandparents, relatives, or scheduling with 53 husband. On the other hand, women with school-age children are usually the ones in their households who have to transport children to and from school. This responsibility introduces fixities into their activity time and space, and ends up shaping up their travel behavior (Kitamura, 1993), and therefore may shorten their commutes. III.8 Specific Hypotheses to be Tested Based on the hypotheses listed in section III.3 and the measurement variable defined above, the specific hypotheses to be tested are as follows: If the HRH is true, then women in households with expectedly greater household responsibilities should make shorter commutes than women in households with expectedly lower household responsibilities. In addition, women in households with greater household responsibilities should make shorter commutes than corresponding men. 1) Expected Results for Household Responsibility Hypothesis i) Among Women: (a) Women 16-34 and living alone will make longer work trip lengths than other women. (b) Women 35-64 and living alone will make longer work trip lengths than other women. (c) Women 16-34 and living with another adult will make shorter work trip lengths than other women. 54 (d) Women 35-64 and living with another adult will make shorter work trip lengths than other women. (e) Women living with another adult and pre-school children will make shorter work trip lengths than other women. (f) Women living with another adult and school- aged children will make shorter work trip lengths than other women. ii) Between Men and Women (a) Women 16-34 and living alone will make longer work trip lengths than equivalent men. (b) Women 35-64 and living alone will make longer work trip lengths than equivalent men. (c) Women 16-34 and living with another adult will make shorter work trip lengths than equivalent men. (d) Women 35-64 and living with another adult will make shorter work trip lengths than equivalent men. (e) Women living with another adult and pre-school children will make shorter work trip lengths than equivalent men. (f) Women living with another adult and school- aged children will make shorter work trip lengths than equivalent men. 2) Expected Results for the “Difference” Hypothesis i) Between US and GB Women (a) Women 16-34 and living alone (LS1) in US will make longer work trip lengths than equivalent women in GB. 55 (b) Women 35-64 and living alone (LS2) will make longer work trip lengths than equivalent women in GB. (c) Women 16-34 and living with another adult in US (LS3) will make shorter work trip lengths than equivalent women in GB. (d) Women 35-64 and living with another adult (LS4) in US will make shorter work trip lengths than equivalent women in GB. (e) Women living with another adult and preschool children (LS5) in US will make shorter work trip lengths than equivalent women in GB. (f) Women living with another adult and school-aged children (LS6) in US will make shorter work trip lengths than equivalent women in GB. Additionally, in separate country regressions specific to US only, or GB only, I expect greater difference in household responsibility- commute length relationships between men and women in the US than in GB for individuals in the following categories: (a) Women 16-34 and living with another adult in US (b) Women 35-64 and living with another adult in US (c) Women living with another adult and preschool children (d) Women living with another adult and school-aged children 56 III.9 Creating the Control Variable Categories One issue was that not all desired variables were available. For example, some variables representing human capital such as wages – a variable that is rarely found in surveys- was not available in either survey. Another issue was that some variables of value to my research were available in only one of the two country data sets. Individual income category was such a variable. This information was not available in the NPTS data, although it was available in the British NTS data. Finally, categories for most variables were different and incompatible. Extensive cleaning and adjustments were done on the NPTS and NTS data sets. III.9.1 Variable Categories i) Mode Modes in both surveys were aggregated to create basic categories. Table III. 9 lists trip mode categories and how they were created for NPTS and NTS, as modified from Giuliano (2005). 57 Table III.9 Mode Categories Mode to work NPTS 1995 NTS 1995/97 POV Automobile Van SUV Pickup or other truck Recreation vehicle Motorcycle Other POV Car/van driver Car/van passenger Car/van either driver or pass. Car no details 2 wheel motor vehicle Bus Bus Bus Commuter Rail Commuter train Surface rail Rail Amtrak Streetcar/Trolley Elevated rail/subway LT underground Walk/bike Walk Bicycle Walk Bicycle Other School bus Other non-POV Taxi Airplane Other ii) Household Income: The reported annual household income for the household is identified for each person and in all cases, and the unadjusted income values are used. Household income was categorized into quartiles. Table III.10 gives unadjusted values for the quartiles for each survey. In order to make income categories compatible, British income in pounds was converted into US dollars using 1995 OECD PPP rates of 0.6539. 58 Table III.10 Household Income Categories in the Pooled Sample Quartile (US$) Lowest <25,000 Second 25,000-39,999 Third 39,000-59,999 Highest >=60,000 iii) Population Density: In NPTS, population density variable is based on 1990 census data at census tract level. NTS data has two estimates: one based on Primary Sampling Units (PSU) and the other is Local Area (LA). According to the NTS guide, the local area population density estimate is more reliable than the PSU density estimate, but it is less closely matched to the neighborhoods of the sampled addresses. Therefore I used the PSU density estimate. For the NPTS, intervals for the population density variable are: low (<1000/mi2), medium (1K-4K/mi2), high (4K-10K/mi2) and very high (>10K/mi2). In the case of NTS, the intervals are slightly different for the high and very high category which are (4K-9K) and > 9K respectively. The two measures do not compare exactly, but as Giuliano and Narayan (2003) wrote, they are the closest possible with what is available. iv) Metropolitan Area Size: Metropolitan size is measured in the US by Metropolitan Statistical Area (MSA) and by Consolidated Metropolitan Statistical Area (CMSA). For Great Britain, a comparable measure was developed by using 1991 British Census population figures and the household residence location data. The NPTS provides Metropolitan Statistical Area 59 (MSA) location for each household residence. MSA data are from the US census. In NTS, metropolitan size was obtained from the 1991 Great Britain census. Metropolitan size categories are urban areas < 250K population, 250 K to 500K, 500K to 1 million, 1 to 3 million, and not in a metropolitan area (designated rural area in NTS). Table III.11 gives the list of variable categories created. 60 Table III.11 List of Variable Categories Used in the Models Variable Category DESCRIPTION OF CATEGORIES FAMINCA1 Household Income is less than $25,000 (yes=1, no =0) FAMINCA2 Household Income is between $25,000 and $40,000 (yes=1, no =0) FAMINCA3 Household Income is between $40,000 and $60,000 (yes=1, no =0) Household Income FAMINCA4 Household Income is more than $60,000 (yes=1, no =0) JMODE1 Trip is made by private automobile (yes =1, no=0) JMODE2 Trip is made by bus (yes =1, no=0) JMODE3 Trip is made by rail other than commuter rail (yes =1, no=0) JMODE4 Trip is made by commuter rail (yes =1, no=0) JMODE5 Trip is made on bike/foot (yes =1, no=0) Mode JMODE6 Trip is made by other mode (yes =1, no=0) POPDN1 Population Density Low (<1000/mi2) (yes=1, no=0) POPDN2 Population Density Medium (1K-4Kmi2) (yes=1, no=0) POPDN3 Population Density High (4K-10K/mi2) (yes=1, no=0) Population Density POPDN4 Population Density Very High (>10K/mi2) (yes=1, no=0) MSA1 Location of household is not in MSA (yes=1, no=0) MSA2 Location of household is an MSA with population below 250,000 (yes=1, no=0) MSA3 Location of household is an MSA with population 250,000 to 500,000 (yes=1, no=0) MSA4 Location of household is an MSA with population 500,000 to 1,000,000 (yes=1, no=0) MSA5 Location of household is an MSA with population 1,000,000 to 3,000,000 (yes=1, no=0) MSA Size MSA6 Location of household is an MSA with population 3,000,000 and over (yes=1, no=0) LIFSTAG1 Individual is 16-34, lives alone, no children living at home (yes=1, no=0) LIFSTAG2 Individual is 35-64, lives alone, no children living at home (yes=1, no=0) LIFSTAG3 Individual is 16-34, lives with another adult, no children living at home (yes=1, no=0) LIFSTAG4 Individual is 35-64, lives with another adult, no children living at home (yes=1, no=0) LIFSTAG5 Individual is 16-64, lives with another adult, youngest child under 5/6 (yes=1, no=0). LIFSTAG6 Individual is 16-64, lives with another adult, youngest child 5/6 to 15 (yes=1, no=0). LIFSTAG7 Individual is 16-64, is the only adult, youngest child 15 and under (yes=1, no=0). Life Stage LIFSTAG8 Other household Job Status PT Individual is a full-time worker (yes=0), Individual is a part-time worker (yes=1) Additional Worker NWRKR Household has 1 worker (yes=0), more than 1 worker (yes=1) 61 CHAPTER IV RESULTS In this chapter, I discuss the findings of my analysis. First, I give a descriptive analysis of my sample, where I examine differences in some independent variables in US and GB that may explain the variations in commute lengths between the two countries. Secondly, I explore the relationships between the variables and work trip distances and times. Finally, I discuss my method and regression estimation results. IV.1 Descriptive Results Table IV.1 gives the average work trip distance and time statistics for men and women in US and GB. Men’s average work trip distances are larger than women’s in both countries, at 12 miles being 3 miles longer than women in US, and at 10.0, 3.6 miles longer than women in GB, corresponding to 24 % difference in US, and 36 % in GB. Women’s average work trip time is about 5 minutes shorter than men’s in both countries, corresponding to nearly nine-tenths of men’s. Men’s speeds are higher than women’s in both countries, although the gender difference in speeds is quite higher in GB. US women’s average work trip distance is larger than GB women, and their average work trip time is shorter. 62 Table IV.1 Work Trip Distance and Time by Sex, US and GB Men Mean Work Trip Distance (miles) Mean Work Trip Time (minutes) Mean Speed (miles/hour) N US 12.4 21.3 34.8 12,200 GB 10 25.9 23.4 2,866 Women Mean Work Trip Distance (miles) Mean Work Trip Time (minutes) Mean Speed (miles/hour) N US 9.4 18.2 31.2 9,912 GB 6.4 22.7 16.8 2,008 IV.1.1 Age Table IV.2 demonstrates that US individuals in the sample are slightly older than GB individuals. Commute distances and times seem to increase for men with age, but the opposite is true for women: among men aged 35- 64, commute distance and time seem to be slightly longer in both countries compared to those aged 16-34 while it is the opposite for women, especially those in GB. This finding is consistent with household responsibility hypothesis in that women since as women become middle aged and have children, give child-related breaks, and thus experience greater earnings differentials with men. 63 Table IV.2 Age Distribution by Sex, US and GB US GB Men Women Men Women 16-34 36.0 36.8 38.9 42.9 35-64 61.6 61.3 59.5 56.0 65 + 2.4 2.0 1.5 1.1 Total 13,782 11,877 2,976 2,143 Table IV. 3 Work Trip Distance and Time by Age, US and GB Work Trip Distance by Age (miles) US GB Men Women Men Women 16-34 11.7 9.8 9.5 7.2 35-64 12.9 9.2 10.4 5.9 65 + 8.5 6.1 8.2 3.8 Work Trip Time (minutes) US GB Men Women Men Women 16-34 20.0 18.4 25.7 24.6 35-64 22.3 18.2 26.0 21.2 65 + 17.2 14.7 25.0 25.0 N 12,200 9,912 2,866 2,008 IV.1.2 Effect of Children Consistent with the higher fertility rate in US of 2.0, compared to 1.7 in GB, individuals in US are 9 % more likely to be in households with children than those in GB. 43% of individuals in the U.S. sample are from households with children, compared to 34 % of those in the GB, Consistent with the household responsibility hypothesis, in both countries, as number of children in one’s household increases, men’s commute distance slightly increases. On the contrary, women’s commute 64 distance and times decrease as number of children increases in both countries. Table IV.4 Life Stage, US and GB Life Stage, Percent Share US GB men women men women 1 adult, 16-34 (LS1) 3.1 2.7 4.0 3.5 1 adult, 35-64 (LS2) 5.0 6.3 5.0 5.6 2 adults, no child, 16-34 (LS3) 8.4 9.9 9.5 13.6 2 adults, no child, 35-64 (LS4) 17.0 16.9 17.5 18.3 2 adults, youngest child preschool age (LS5) 19.9 13.1 17.1 9.6 2 adults, youngest child school age (LS6) 17.4 16.3 14.2 12.8 1 adult, child 15 and under (LS7) 0.9 4.9 0.2 3.0 Other(LS8) 28.3 29.9 32.4 33.6 N 13,782 11,877 2,976 2,143 Work Trip Distance by Life Stage US GB Men Women Men Women 1 adult, 16-34 (LS1) 10.1 8.7 7.8 6.9 1 adult, 35-64 (LS2) 10.8 8.8 9.0 6.4 2 adults, no child, 16-34 (LS3) 12.1 11.5 10.8 9.4 2 adults, no child, 35-64 (LS4) 13.0 9.7 9.6 5.8 2 adults, youngest child preschool age (LS5) 13.5 9.3 11.7 6.2 2 adults, youngest child school age (LS6) 13.0 9.0 11.0 5.5 1 adult, child 15 and under (LS7) 11.9 8.6 14.9 4.8 Other(LS8) 11.4 9.0 9.1 6.0 N 12,200 9,912 2,866 2,008 Work Trip Time by Life Stage US GB Men Women Men Women 1 adult, 16-34 (LS1) 18.5 19.4 23.2 25.3 1 adult, 35-64 (LS2) 20.7 18.5 27.0 24.7 2 adults, no child, 16-34 (LS3) 20.6 20.6 28.0 28.7 2 adults, no child, 35-64 (LS4) 22.7 19.0 24.9 21.8 2 adults, youngest child preschool age (LS5) 22.0 16.9 27.5 22.0 2 adults, youngest child school age (LS6) 21.9 16.7 26.9 19.2 1 adult, child 15 and under (LS7) 19.9 17.5 40.0 23.4 Other(LS8) 20.4 18.0 24.6 21.4 N 12,200 9,912 2,866 2,008 65 There are no stark differences between the countries in terms of the distribution of the life stage variable categories. However, there are still some distinctions: GB has a slightly higher proportion of women who are 16- 34 and in 2 adult households with no children (LS3) compared to US (14 % vs. 10 %). Percentage of both men and women who are in households with another adult and school age children (LS6) is higher in GB than US. In both countries, men coming from households with another adult and pre-school children (LS5) outnumbered women, most likely due to very busy life of mothers of preschool children. In both countries the percentage of individuals from single parent households is low among men – 0.9 % in US and 0.2 % in GB. As Figure 1 demonstrates, in both countries, average work trip distance and time are highest for young women (16-34) who live with another adult (LS3). The stages after LS3 are associated with a general pattern of decrease in work trip distances and times, particularly in GB. 66 Figure IV.1 Work Trip Distance by Life Stage, US and GB Work Trip Distance 0 2 4 6 8 10 12 14 US GB 1 adult, 17-34 (LS1) 1 adult, 35-64 (LS2) 2 adults, no child, 17-34 (LS3) 2 adults, no child, 35-64 (LS4) 2 adults, youngest child preschool age (LS5) 2 adults, youngest child school age (LS6) 1 adult, youngest child 15 and under (LS7) Figure IV.2 Work Trip Time by Life Stage, US and GB Work Trip Time 0 5 10 15 20 25 30 35 US GB 1 adult, 17-34 (LS1) 1 adult, 35-64 (LS2) 2 adults, no child, 17-34 (LS3) 2 adults, no child, 35-64 (LS4) 2 adults, youngest child preschool age (LS5) 2 adults, youngest child school age (LS6) 1 adult, youngest child 15 and under (LS7) 67 IV.1.3 Job Status Among women aged 16 and above, 60 % are employed in the US. In GB, 49 % of all women are employed, 11 percentage points lower than US. Part time work is defined as 10 to 30 hours in GB, while the US survey defines it as below 35 hours. According to these definitions, the share of part time among GB women was 22 %, 4 % higher than US. Table IV.2 gives the proportion of full and part time workers in each country. In the commuter sample, more than three quarters of employed US women and close to two thirds of employed GB women work full time. As expected, women in both countries work part time more than men. However, the proportion of employed women working part time in GB is more than half of the corresponding proportion in US: 36 % of women in GB work part time, while the corresponding percentage is 23 % in US. Table IV.5 Proportion of Full- and Part-Time Workers US GB Men Women Men Women Full-time 91 77 94 64 Part-time 9 23 6 36 N 13,782 11,877 2,976 2,143 68 Table IV.6 Distance and Time to Work for Full Time and Part-Time Workers Men Women US Mean Dist. Mean time Mean Dist. Mean time. Full-time 12.8 21.9 10.2 19.3 Part-time 7.3 15.5 6.7 14.4 N 12,200 9,912 2,866 2,008 GB Mean Dist. Mean time Mean Dist. Mean time. Full-time 10.3 26.2 7.5 25.3 Part-time 6.0 21.3 4.4 17.9 N 12,200 9,912 2,866 2,008 Table IV.6. shows that, as expected, full time workers commute substantially longer distances and times than part time workers. In both countries, the full time--part time gap is higher for men compared to women: overall 50-60 % for men, 70 to 80 % for women including both distance and time. IV.1.4 Household Income As expected, the proportion of individuals in higher income categories is substantially greater in the US sample compared to the GB sample. Percentage in the highest two income quartiles is 5 % higher among men compared to women in US, but the corresponding percentage in GB is the same for men and women, showing a slightly greater income disparity between the sexes in the US. As Figures 1 and 2 illustrate, in both countries, with higher household incomes, both men and women’s commute distances and times increase. However, increase with income is higher for GB, especially for GB men. GB women’s increase in commute distance with household income is at a 69 higher level compared to US women, most likely indicating an increase in mobility among GB women due to increased access to cars with higher income. Figure IV.3 Work Trip Distance by Household Income Figure IV. 3 Work Trip Distance by Household Income by Sex, US and GB 0 2 4 6 8 10 12 14 16 Lowest Lower middle Higher middle Highest US men GB men US women GB women 70 Figure IV.4 Work Trip Time by Household Income Figure IV. 4. Work Trip Time by Household Income and Sex, US and GB 0 5 10 15 20 25 30 35 Lowest Lower middle Higher middle Highest GB men GB women US men US women IV.1.5 Car Access and Mode As expected, commuters in US have much higher car access than those in GB (Table IV.7). In both countries, there is virtually no difference between the sexes in terms of car access, probably because the men and women are drawn from the same households. Commute distances are slightly longer with higher car access, but there is no change in commute times with higher car access. As Table IV.7 shows, car access among women commuters is higher among workers compared to all women 16 and older; showing that employment increases women’s car access in GB. The relationship of employment and car ownership is not as high for GB men as 71 much as it is for women, showing that their access to car is given, not conditioned on being employed. Table IV.7 Car Access of Commuters US GB male female male female 0 cars 2.9 4.2 9.8 11.0 Cars<persons 15.4 15.1 30.5 30.3 Cars=Persons 65.0 66.5 55.2 55.0 Cars > Persons 16.7 14.1 4.5 3.7 13,782 11,877 2,976 2,143 Table. IV.8 Car Access of All NPTS 1995 and NTS 1995/97 Participants US GB male female male female 0 cars 4.4 7.1 19.7 26.3 Cars<persons 17.4 16.4 27.3 24.2 Cars=Persons 62.3 62.7 49.4 46.5 Cars > Persons 16.0 13.9 3.6 2.9 N=95,360 N=23,167 Table IV.9. Mode Share US GB Men Women Men Women Auto 93.6 92.9 81.1 71.9 Bus 1.3 2.2 5.4 14.2 Regular rail 1.2 1.2 1.7 2.0 Commuter Rail 0.6 0.3 3.3 3.9 Walk or bike 2.5 2.7 8.0 6.8 Total 11,994 9,745 2,866 2,008 72 Table IV.10. Work Trip Distance, Time and Speed by Mode Work Trip Distance (miles) US GB Men Women Men Women Auto 12.7 9.7 10.7 6.7 Bus 10.5 8.7 6.1 4.4 Regular rail 10.5 10.9 7.8 7.7 Commuter Rail 32.1 23.5 21.2 16.7 Walk or bike 1.5 0.8 2.3 1.4 N 11,994 9,745 21,739 2,866 Work Trip Time (minutes) US GB Men Women Men Women Auto 20.9 17.6 24.0 18.7 Bus 35.4 35.2 35.1 30.9 Regular rail 42.5 40.4 44.3 48.2 Commuter Rail 61.2 62.5 64.3 60.5 Walk or bike 14.1 12.1 19.2 19.4 N 11,994 9,745 21,739 2,866 Speed by Mode (miles per hour) US GB Men Women Men Women Auto 36.5 33.1 26.8 21.5 Bus 17.8 14.8 10.4 8.5 Regular rail 14.8 16.2 10.6 9.6 Commuter Rail 31.5 22.6 19.8 16.6 Walk or bike 6.4 4.0 7.2 4.3 N 11,994 9,745 21,739 2,866 Table IV.10 gives work trip distance, time and speed in US and GB by mode and sex. It is seen that US commuters overwhelmingly travel to work by auto at 93 % compared to 77 % in GB. US women have similar rates of auto use as US men to travel to work, but GB women, on average, are 9 % behind GB men, and 17 % behind US women. Only 3 % of commuters in US 73 use public transit. 14 % of GB women use buses to travel to work, about three times as GB men (5 %). In both countries, auto users have much shorter commute trip times than public transit users. Contrary to expectations, in GB, average work trip distance of female auto users is only 0.3 miles longer than average work trip distance of all female British commuters. This finding shows that the difference between US and GB women’s commute distances are not explained by the greater propensity of American women to travel by car. One reason for this is the high transit access in GB, and relatively slow speed of all modes compared to US except for commuting by walking or bicycling. This explains why the times are longer and distances are shorter in GB than those in US. Table IV.11 Percentage of Women Working Part Time by Life Stage Percentage Working Part Time by Life Stage US GB LIFE STAGE M W M W 1 adult, 16-34 (LS1) 8.5 8.7 1.7 5.3 1 adult, 35-64 (LS2) 6.4 10.3 6.1 24.4 2 adults, no child, 16-34 (LS3) 7.4 13.1 4.6 9.9 2 adults, no child, 35-64 (LS4) 5.1 13.6 5.8 35.6 2 adults, youngest child 0-5/6 (LS5) 2.7 27.8 2.7 61.7 2 adults, youngest child 5/6-15 (LS6) 5.9 27.5 1.7 53.5 1 adult, child < 16 (LS7) 9.3 19.1 16.7 51.6 Other (LS8) 17.8 30.0 9.9 37.0 N 13,78211,8772,9762,143 In GB, women in high household responsibility stages (LS4 through LS7 in GB) work part time at substantially higher levels than women in lower 74 household responsibility stages. For example, in GB, the proportion of part time working women in households with preschool children (LS5) is 62% and with school children (LS6) is 53 %. Both stand relatively higher than the GB average of 36 %. However, the pattern in US is nowhere as striking: US women in LS5 work part time at 28 % compared to an average of 23 % among all US commuters. Table IV.12 Work Status of Women, NPTS and NTS participants US GB FT PT N.E. 1 FT PT N.E. LS1 74.7 13.0 12.3 69.9 7.4 22.7 LS2 69.4 12.1 18.5 40.6 14.5 44.9 LS3 65.4 18.2 16.4 77.7 10.1 12.1 LS4 59.9 16.2 23.9 29.5 25.8 44.7 LS5 39.6 22.8 37.6 18.4 33.0 48.7 LS6 50.5 25.8 23.6 28.6 36.8 34.6 LS7 53.7 16.1 30.3 12.3 21.4 66.3 LS8 24.6 14.5 60.8 16.7 12.9 70.4 Total 42.7 17.6 39.7 26.8 22.4 50.8 1. N.E: Not Employed That employment is very much conditioned by life stage in GB is shown by the data in the main sample as well. At all life stage categories, US women 16 and above are employed at a rate higher than corresponding British women, with the exception of those aged 16-34, living with another adult, and no children. At this stage, women in GB work at a higher full time level with a lower unemployment rate than US women. IV.1.6 Mode by Life Stages In GB, the proportion of women auto users commuting to work in LS6 (living with an adult and school age children, -a life stage known to include 75 chauffeuring of children for women) at 78 % is much closer to the level of men in the same life stage at 82 %. Hence it seems like GB women are indeed making some adjustments through part time work and mode to household responsibility. Such adjustments are very small in US compared to those in GB. Table IV.13. Percentage of Men and Women Commuting by Car by Life Stage % Commuted by Car by Lifecycle by Sex by Country US GB M W M W 1 adult, 16-34 (LS1) 87.1 87.1 70.2 76.7 1 adult, 35-64 (LS2) 88.9 88.9 76.8 65.5 2 adults, no child, 16-34 (LS3) 91.4 90.7 75.6 68.8 2 adults, no child, 35-64 (LS4) 95.0 94.4 83.8 70.8 2 adults, youngest child preschool age (LS5) 95.2 94.4 86.6 78.3 2 adults, youngest child school age (LS6) 94.8 96.1 81.8 77.8 1 adult, child 15 and under (LS7) 91.6 86.5 60.0 59.3 Other (LS8) 93.2 92.7 80.2 71.8 Total 93.692.981.171.9 N 11,2289,0502,3241,444 IV.1.7 Population Density As expected, commuters in the GB sample live in higher population densities than those in the US. There is not a big gender difference in terms of the population density. Lower density means longer commute distances in both countries. Effect of population density on work trip time is not as linear as it is for commute distance: Commute time mainly decreases with 76 increasing density, but it again increases at the highest density category, especially in GB. This should be due to the effect of congestion at high densities. Table IV.14. Population Density Distribution by Sex, US and GB US GB male female total male female total Density low (<1000/mi2) 43.7 41.3 42.1 26.9 26.5 26.7 Density Medium (1K-4Kmi2) 30.4 30.3 30.4 18.8 19.5 19.0 Density high (4K-10K/mi2) 18.2 18.7 18.5 33.3 30.8 32.3 Density very high (>10K/mi2) 8.6 9.7 9.1 21.0 23.2 21.9 total 13,700 11,793 25,493 2,976 2,143 5,119 Table IV.15. Population Density Distribution, US and GB for LS4, 5 and 6 US GB LS4 LS5 LS6 LS4 LS5 LS6 Density low (<1000/mi2) 46 45 50 31 26 26 Density Medium (1K-4Kmi2) 29 30 30 18 15 23 Density high (4K-10K/mi2) 17 17 15 31 37 31 Density very high (>10K/mi2) 8 7 6 20 22 19 N 4,331 4,261 4,313 914 716 699 Table IV.15 shows that in US persons in LS4, LS5 and LS6 (those with children and mature couples without children) live in slightly lower densities than other households. In GB, this effect is more subtle. This may be reflective of the greater suburbanization of families with children in the US. 77 Table IV.16 Work Trip Distance and Time by Population Density Work Trip Distance (miles) US GB Men Women Men Women Density low (<1000/mi2) 14.4 11.2 11.4 8.4 Density Medium (1K-4Kmi2) 11.3 8.6 11.3 6.1 Density high (4K-10K/mi2) 10.3 7.8 9.2 5.6 Density very high (>10K/mi2) 9.9 7.1 8.4 5.5 12,126 9,842 2,866 2,008 Work Trip Time (minutes) US GB Men Women Men Women Density low (<1000/mi2) 22.3 18.4 24.5 20.7 Density Medium (1K-4Kmi2) 19.7 16.9 26.5 19.7 Density high (4K-10K/mi2) 20.0 17.1 25.2 22.0 Density very high (>10K/mi2) 25.5 23.8 28.1 28.4 12,126 9,842 2,866 2,008 The above table shows that commuting to work by car in US is almost consistent regardless of life stages, although there is a 14 % gender gap for single parents (LS7). IV.1.8 MSA Size US has more people in large cities, and GB more in non-metropolitan areas and medium size cities. Two-fourths of the US sample live in very large MSAs, which is more than 3 times the rate in that of GB. Inversely, GB has more people in non-MSAs at 51 %, which is more than 3 times the rate in that of US. 78 Table IV.17 MSA Size Distribution by Sex, US and GB US GB male female total male female total Not in MSA 13.9 14.3 14.1 51.1 48.2 49.9 L.T. 250K 9.4 9.3 9.4 10.0 10.3 10.1 250K-500K 5.9 5.7 5.8 14.9 15.0 14.9 500K-1M 12.5 12.5 12.5 2.3 3.0 2.6 1M-3M 19.1 19.0 19 10.5 11.0 10.7 > 3M 39.2 39.2 39.2 11.2 12.6 11.7 N 13,782 11,877 25,659 2,976 2,143 5,119 Table IV.18 Work Trip Distance by City Size Work Trip Distance by City Size (miles) US GB Men Women Men Women Not in MSA 11.9 9.6 11.0 7.3 L.T. 250K 10.6 8.2 10.6 6.0 250K-500K 10.3 8.8 8.6 5.8 500K-1M 10.6 8.8 6.1 5.1 1M-3M 11.3 9.0 9.3 4.6 > 3M 14.3 10.1 8.5 6.1 N 12,200 9,912 2,866 2,008 Work Trip Time (minutes) US GB Men Women Men Women Not in MSA 18.4 15.7 24.3 19.8 L.T. 250K 17.7 14.9 25.6 21.7 250K-500K 17.3 15.8 24.0 21.5 500K-1M 18.2 16.3 22.4 22.5 1M-3M 19.1 17.0 26.8 22.5 > 3M 25.9 21.4 35.7 35.8 N 12,200 9,912 2,866 2,008 79 IV.2 Method of Analysis Since there are a number of factors contributing to the length of commute trips, a multiple regression analysis is appropriate to test the relationship between commute length and household responsibility. A series of estimations for work trip distance and time were carried out according to the following groups: all men and women, women only, men and women from the two respective countries, and finally, men and women from the respective countries with additional variables of education and individual income. In pooled regression, I estimate one-step equations for commute distance and time on a combined pool the individuals in both countries. The regression involves second-order interaction variables of country and sex with other variables. Female interaction terms are cross-products of the female dummy and the independent variables. They help us determine how the independent variables affect women differently than men. Similarly, GB interaction terms are cross-products of the country dummy and the independent variables, and they help us determine if and how variables affect GB individuals differently than US individuals in the equation below: Y = f(X, L, H) + f [female*(X, L, H)] + f[GB *(X, L, H)] where, 80 Y= Commute distance or time X= Demographic and economic attributes of the individual H= Household related factors (resources, household structure, roles) L= Land use characteristics of the individual’s residential location GB= Country, representing contextual environment female= Sex Commute Length= f ( life stage, job status, household Income, 2 or more workers in the household, Population Density, MSA Size) + f(female, female* life stage, female* job status, female* household income, female* 2 or more workers in the household, female*population density, female* MSA Size) + f(GB, GB* life stage, GB* job status, GB* household income, GB* 2 or more workers in the household, GB*population density, GB* MSA Size) + f(GB * female). The intercepts in this equation are: • The “female” intercept: Effects on women of factors not captured by any of the independent variables in the model • The “GB” country intercept: Effects on GB individuals of factors not captured by any of the independent variables in the model • The “GB*female” or “GBFEM” intercept: Effects on GB women of factors not captured by any of the independent variables in the model In this model, a reference person is an American male, full time worker, in a single-worker household with an annual income between $40 81 and $60 K, living in a MSA of 1 to 3 million population, with population density of 4K-10K/mi2 and in life stage LS8 (“other” household). The second model restricts the sample to women only. Commute Length = f(X, L, H) + f[GB *(X, L, H)] . Commute Length= f(life stage, job status, household Income, 2 or more workers in the household, Population Density, MSA Size), + f(GB, GB* life stage, GB* job status, GB* household income, GB* 2 or more workers in the household, GB*population density, GB* MSA Size). In this model, a reference person is American, full time worker, in life stage LS8 (“other”), in a household where there’s only 1 worker, income is between $40 and $60 K, located living in an MSA of 1 to 3 million population, with population density of 4K-10K/mi2. Finally, in order to compare the household responsibility-commute length relationship in the two countries, I ran separate regression equations in each country, in the following form: Commute Length = f(X, L, H) + f[female *(X, L, H)]. Commute Length= f(life stage, job status, household Income, 2 or more workers in the household, Population Density, MSA Size) + f(female, female* life stage, female* job status, female* household income, female* 2 or more workers in the household, female*population density, female* MSA Size). 82 In this model, a reference person is male, full time worker, in life stage LS8 (“other”), in a single-worker household with an annual income between $40 and $60 K, living in an MSA of 1 to 3 million population, with population density of 4K-10K/mi2. IV.3. Regression Results Table 1 through 6 gives coefficients and significance values. For the sake of simplicity, I use a p-value of 0.1. Tables 1 and 2 in the Appendix give the results of the estimation for work trip distance and work trip time from the pooled regression analysis of all individuals in the US and GB samples. The pooled sample allowed me to make comparisons between men and women, and between US and GB individuals in similar groups directly. In these models, in the first panel, general coefficients are given, the second panel gives the interactive coefficients for women relative to men. The effect for women is the sum of the first columns of the first two panels. The third panel gives the interactive coefficients for GB, that is, the effect for GB individuals relative to US individuals, and the effect for them are the sum of the first columns in the first and third panels. Tables 3 and 4 give the results of the estimation for work trip distance and work trip time from the pooled regression analysis of all women in the US and GB samples. This women-only sample regression allowed me to directly assess the effect of a variable on women, and to compare women in similar groups in US and GB. In Tables 3 and 4, in the first panel, coefficients for US 83 women are given; the second panel gives the interactive coefficients for GB women relative to US women. The effect for GB women is the sum of the first columns of the each of the two panels. Tables 5 through 8 give the results of separate country regressions. In these tables, in the first panel, general coefficients are given; the second panel gives the interactive coefficients for women relative to men. The effect for women is the sum of the first columns of the each of the two panels. Tables 5 and 6 give the results of the estimation for work trip distance and work trip time from the pooled regression analysis of all men and women in the US sample. Tables 7 and 8 give the results of the estimation for work trip distance and work trip time from the pooled regression analysis of all men and women in the GB sample. Below I discuss the results of the pooled regression estimation for work trip distance and time. I will not discuss in detail the results of women-only and separate regression unless they pertain directly to the hypothesis tested for the sake of simplicity (Their results are in alignment with the pooled regression). After that, I discuss the results according to their support for the “HRH” and the “Difference” Hypotheses. IV.3.1. Results of Pooled Regression Estimation for Work Trip Distance A general look at the coefficients shows that household income, job status, country-specific factors, land use variables, and household life stage 84 variables all contributed to the explanations of women’s commute distances. As expected, work trip distance with increasing household income in both countries. In GB, this relationship is stronger, especially at the lower income levels. The relationship of household income to distance is slightly weaker for women than for men at lowest income quartile. Also as expected, part time workers have shorter work trip distances than full timers in both countries. However, the difference between full and part time workers’ commute distances is greater in US. This is expected, given the greater pay and benefit gap between full and part time workers in US. The GB coefficient of 1.1 shows that the full time-part time gap is 1 mile narrower in GB (could be because of the penalty in US, and reward GB for part time workers). The positive coefficient for females shows that the gap between part- and full-timers is about a mile narrower for women than men. As expected, work trip distance decreases with increasing population density. In GB, the relationship is much weaker than it is in US (lowest density category was attenuated to about half, and highest was attenuated to 0). The effect of MSA size on females was no different. Commute distance increases with increasing MSA size. In GB, the signs are all opposite. For women, there is an attenuating effect of MSA0 (non-MSA) and MSA2 (size of 250-500K). The country intercept for GB for work trip distance is no different than the intercept for US (The GB interactive coefficient is not significant). 85 The “female” intercept for work trip distance shows that women have 2.2 miles shorter work trip distance than men due to factors that are not controlled by any of the variables used in the model. The GBFEM intercept for distance is negative and has a value of -1.4. This can be due to several possibilities: First, it may be due to lower human capital in GB that are not controlled in the equation, that is, lower wages of GB women. Secondly, it could be related to locational differences of jobs in the two countries. Thirdly, it could be related to a greater cultural role of women in GB to be supplementary earners or along the same line. IV.3.2 Results of Pooled Regression Estimation for Work Trip Time A look at the coefficients shows that household Income, job status, country-specific factors, land use variables, and household life stage variables all contributed to the explanations of women’s commute times. As expected, as household income increases, work trip time increases. In GB, this relationship is quite strong, especially at lower income levels. For women, like in the findings on work trip distance, the relationship is a little ore subtle (the female interactive effect of second income quartile is positive). Part timers’ work trip time is on average 5 minutes shorter than that of full time workers. Neither GB nor Female interactive variables are significant for part time workers’ commute time. Regarding land use variables, the Iowest and highest densities have longer work trip time than the reference category (lowest substantially more at 4.2 minutes). This effect is almost attenuated in GB. At highest densities, 86 females have larger trip time than men. Those in the largest MSAs had 6 minute longer work trips, others are shorter than the reference category. For women, the effect is attenuated to some extent at the MSA5 (> 3 million people); and GB interactive effect is stronger at MSA5 (GB*MSA5 is 3.9). This study used number of workers as a control variable, expecting that women’s roles would be very different in 1-worker households compared to two-- worker households. However, it was found that the effect of having more than one worker in the household was not significant for women’s work trip distance, and it was positive on work trip time. This finding was opposite of Johnston-Anowonwo’s (1992). The country intercept for work trip time in GB is 11 minutes (50%) longer than that in US. This is a substantial difference, due to the effects of differences not captured by the independent variables. The most likely factor is the slower modes of transportation in GB, wherein even the faster modes are relatively slower than the equivalent in US. An older transportation infrastructure, higher congestion levels and greater restrictions on car travel are likely to account for the slow speeds and longer work trip times in GB. The “female” intercept for work trip time is –2; showing that women have 2 minutes shorter work trip time due to factors not controlled by any of the variables used in the model. The Great Britain female intercept for time was negative and had a value of –1.1. 87 Tables IV.19 through IV.22 give a summary of expected signs for the supported hypotheses, and results for work trip distance and time. The results of three regression models: i) women-only regression, ii) pooled regression, and iii) separate country regression are tabulated and described below according to their support of HRH (Household Responsibility Hypothesis) and the “Difference” Hypotheses. The “Difference” hypothesis is that all else controlled, the effect of household responsibilities on women’s commute lengths is weaker in GB than US. For testing HRH, pooled and women-only regressions will be examined. For testing the “difference” hypothesis, women-only, and the separate country regressions results will be examined. 88 IV.4. Results According to Their Support for HRH: Table IV.19. Testing HRH Among-Women Model: US and GB Women Life stage Indicator Expected Sign for support of HRH Results: Work Trip Distance Results: Work Trip Time Being single (living alone) and age 16-34 (LS1) LS1 positive NS NS Being single (living alone) and age 35-64 (LS2) LS2 positive NS NS Living with another adult and being age 16-34 (LS3) LS3 negative positive positive Living with another adult and being age 35-64 (LS4) LS4 negative NS NS Living with another adult and preschool- aged children (LS5) LS5 negative NS NS Living with another adult and school- aged children (LS6) LS6 negative NS negative Being a single adult and having children aged 15 and under (LS7) LS7 inconclusive NS NS Table IV. 20. Pooled US and GB Men and Women “HRH Between Men and Women” Analysis Model: Pooled US and GB, Men and Women Life stage Indicator Expected Sign for support of HRH Results: Work Trip Distance Results: Work Trip Time Being single (living alone) and age 16-34 (LS1) FEM*LS1 Inconclusive positive positive Being single (living alone) and age 35-64 (LS2) FEM*LS2 Inconclusive NS NS FEM * LS3 negative positive positive FEM * LS4 negative negative NS FEM * LS5 negative negative negative Living with another adult and without children (LS3, LS4), or with children (LS5, LS6) for women versus men FEM * LS6 negative negative negative Being a single adult and having children aged 15 and under (LS7) FEM * LS7 inconclusive NS NS 89 1) Results that supported the HRH were the following: i) Among Women: (a) Women in living with another adult and school age children (LS6) had shorter work trip times than other women. ii) Between men and women: (a) Women 35-64 living with another adult and no children (LS4) had shorter work trip distances than similar men. (b) Women in couple families with preschool children (LS5) had shorter work trip distances and times than similar men. (c) Women in couple families with school children (LS6) had shorter work trip times than similar men. 2) Results that did not support the HRH were the following: i) Among Women: (a) Women 16-34 living with another adult and no children (LS3) did not have shorter work trip distances or times than other women. On the contrary, the relationships were both positive. (b) Women 35-64 living with another adult and no children (LS4) did not have shorter work trip distances or times than other women. The relationships were both not significant. 90 (c) Women in couple families with preschool children (LS5) did not have shorter work trip distances or times than other women. The relationships were both significant. (d) Women in couple families with school–age children (LS6) did not have shorter work trip distances than other women. The relationship was not significant. ii) Between Men and Women: (a) Women 16-34 living with another adult and no children (LS3) did not have shorter work trip distance and time than similar men. On the contrary, the relationships were positive. (b) Women 35-64 living with another adult and no children (LS4) did not have shorter work trip times (unlike distance) than similar men. The relationship was not significant. 91 IV.5. Results According to Their Support for the “Difference” Hypothesis Table IV.21 Separate Country Regressions Model: Separate Country, US Life stage Indicator Expected Sign for support of HRH Results: Work Trip Distance Results: Work Trip Time Being single (living alone) and age 16-34 (LS1) FEM*LS1 inconclusive NS positive Being single (living alone) and age 35-64 (LS2) FEM*LS2 inconclusive NS NS FEM * LS3 negative positive positive FEM * LS4 negative negative NS FEM * LS5 negative NS NS Living with another adult and without children (LS3, LS4), or with children (LS5, LS6) FEM * LS6 negative negative negative Being single and having children aged 15 and under (LS7) FEM * LS7 inconclusive NS NS Model: Separate Country, GB Life stage Indicator Expected Sign for support of HRH Results: Work Trip Distance Results: Work Trip Time Being single (living alone) and age 16-34 (LS1) FEM*LS1 Inconclusive NS NS Being single (living alone) and age 35-64 (LS2) FEM*LS2 Inconclusive NS NS FEM * LS3 negative NS NS FEM * LS4 negative negative negative FEM * LS5 negative negative negative Living with another adult and without children (LS3, LS4), or with children (LS5, LS6) FEM * LS6 negative negative negative Being single and having children aged 15 and under (LS7) FEM * LS7 inconclusive NS NS 92 Table. IV.23. Women-Only, US and GB for “the Difference” Hypothesis Model: US and GB Women Life stage Indicator Expected Sign for support of “The Difference” Results: Work Trip Distance Results: Work Trip Time Being single (living alone) and age 16-34 (LS1) GB*LS1 inconclusive NS NS Being single (living alone) and age 35-64 (LS2) GB*LS2 inconclusive NS NS Living with another adult and being age 16-34 (LS3) GB*LS3 positive NS positive Living with another adult and being age 35-64 (LS4) GB*LS4 positive NS NS Living with another adult and preschool- aged children (LS5) GB*LS5 positive NS positive Living with another adult and school- aged children (LS6) GB*LS6 positive NS NS Being single and having children aged 15 and under (LS7) GB*LS7 inconclusive NS NS 1) Results Supporting “the Difference” Hypothesis i) Separate Regressions: In separate country regressions, starker gender differences in US than in GB would support the “difference” hypothesis. Additionally, having the household responsibility effect at a greater number of household responsibility categories in US compared to GB would also support the “difference” hypothesis. In US, the negative and significant effect of high household responsibility on women is observed at a broader level; 93 (a) In GB, women 35-64 and living with another adult and no children (LS4) did not have shorter time than men. But in US they did, at both distance and time. (b) In GB, women living with another adult and preschool children (LS5) did not have shorter commute distance or times than men or other women, but in US they did. ii) Women-only Regression Longer commutes for high household responsibility stages for GB women, that is, positive and significant coefficients for GB interactive high household responsibility life stage variables would support the “difference” hypothesis. (a) Interactive GB* LS3 (living with another adult and no children), coefficient for work trip time was positive and significant, meaning that British women at this category made longer work trip time than American women. b) GB Interactive life stage coefficients were positive and significant for LS5, namely, British women living with another adult and preschool children made longer commute times (also distance at 0.12 significance) than corresponding American women. 94 2) Results Not Supporting “The Difference” Hypothesis: i) Separate Regressions: (a) In separate estimations in US and GB, gender difference was starker for women living with school age children (LS6) in GB than US for both work trip distance and time. The female interactive coefficients in GB for LS6 were -2 for distance and –3.1 for time, while they were more moderate in US at around 1.0. ii) Women-Only Regressions (a) Women 16-34 living with another adult and no children (LS3) in GB did not have longer work trip distances than similar women in US. The relationship was not significant. (b) Women 35-64 living with another adult and no children (LS4) in GB did not have longer work trip distances or times than similar women in US. The relationships were not significant. (c) Women living with another adult and with school –age children (LS6) in GB did not have longer work trip distances or times than similar women in US. The relationships were not significant. 95 This chapter first examined the relationship of work trip distance and time with explanatory variables. It then estimated the model regressions, and presented the results for HRH and the “difference” hypotheses. In the next chapter, I make conclusions on the results found in this chapter on HRH and the “Difference” Hypothesis. 96 CHAPTER V DISCUSSION In this chapter, I interpret the results presented in Chapter IV. First, I interpret the findings on the HRH. Second, I interpret the findings on the “Difference” Hypothesis. I conclude by recommendations for future research and practical implications. V.1 Assessment of the Results of the Test of the HRH The among-women analysis showed that Hypothesis 1 (that women at life stages with higher household responsibilities make shorter work trip distances and times than women at other life stages when all else is controlled) was supported for women living with another adult and school- aged children (LS6) for work trip time. Support for Hypothesis 1 in among- women analysis was not observed in any life stage categories other than LS6 13 . The between-men-and-women analysis showed that Hypothesis 2 (that women at life stages with higher household responsibilities make shorter work trip distances and times than men in similar life stages when all else is controlled) was supported for women living with another adult and preschool- aged children (LS5) and those with school-aged children (LS6) and work trip distance of women aged 35 to 64 and living with another adult 13 A look at the results of separate country regressions for women (the results not included here) shows that this relationship was found in US in particular. Since the US sample is much larger than the GB sample, the direction of the relationship between LS6 and work trip time in US influences the relationship in the combined sample. 97 and no children (LS4), and not supported for the work trip times of women at LS4. Hypothesis 2 was supported for only work trip distance of women aged 35-64, living with another adult and no children (LS4) (they had shorter work trip distances than corresponding men) while women living with another adult and children (LS5 and LS6) did have both shorter distances and times than men. Hypothesis 2 was not supported for childless women aged 16 to 34 and living with another adult (LS3). The women in this group made longer work trip distances and times than men, particularly in US. However, I will not draw strong conclusions from this finding because the reason for the positive relationship for commute length in this group could be a definitional issue. Although this category was originally conceived to represent a young married or cohabitating couple, and a life stage with relatively high amount of household responsibility, in GB 85 % of women in the LS3 category are married or cohabitating, while the remaining 15 % are single. The finding that women aged 16 to 34, single, and living alone (LS1) had longer work trip distances and times than men is not in conflict with the HRH, although why they had longer work trip distances and times than men is unknown. On the whole, the results showed that there was systematic support for the HRH in between-men-and-women analysis, and partially in among- women analysis (Women at LS 4, 5, and 6 have shorter work trip distances 98 than men; women at LS5 and LS6 had shorter work trip times than men, and women at ls6 had shorter work trip time than other women). Overall, this research found that the HRH was supported the most for women living with another adult and school-aged children (LS6) [as they made shorter work trip times than both other women and shorter work trip distances and times than similar men]. This finding confirms the results of the research by Nobis and Lenz (2005) who found that women with school-age children are usually the ones in their households who have to transport children to and from school, and that working women in two-worker families were twice as likely as men to pick up and drop off school-age children at school during their commute (McGuckin and Nakamoto, 2005). The results are also consistent with Kwan’s (1999, 2000) argument –that out–of-home trips shape women’s commutes, and Kitamura’s (1993) argument that child- serve trips in particular shaping travel behavior, and Hjorthol’s (2000) finding that women with children in the lower primary school (7-12 year old) were less likely to travel to work as long as women without children in this age group were. Hjorthol attributed this to greater schedule constraints of mothers with children in this age group – explaining that first couple years of school being unlike kindergarten, has shorter hours, and includes fewer leisure activities, thus imposing greater fixities in time and space, and ends up shaping up their travel behavior and therefore may shorten commutes (Kitamura, 1993). 99 The reason that the results were more supportive of HRH at LS5 compared to women at LS6 is likely that women at LS6 have less flexibility in their schedules than women at LS5: school-age children do have to be taken to school, and it is not as easy for their mothers as it is for mothers of preschool children to find alternatives to this task. Women with preschool children, on the other hand, can arrange for day care at home through nannies or relatives. Some time use studies also found that husbands are more likely to help mothers in child care at home than in chauffeuring children. Another finding is that differences among women were small compared to differences between men and women. This finding supports findings from other studies (Hanson and Johnston, 1988; Giuliano, 1988). One explanation is that societal upbringing influences a woman more than the particular situation in a household 14 and that gender is more important than “household responsibilities” as defined here. V.2 Assessment of the Results of the Test of the “Difference” Hypothesis Women-only regressions as presented on Table 3 and Table 4 show that Hypothesis 3 (that American women in households with higher household responsibilities have shorter commute lengths than British women in similar life stage categories when all else is controlled) is supported for work trip time of women aged 16 to 34, living with another adult and no children (LS3) and women living with another adult and preschool- aged 14 Gender socialization theory (Taylor, 1999) 100 children (LS5) as British women in these categories had longer work trip times than American women. Separate regressions run for US and GB to test Hypothesis 4 (that household responsibilities shorten American women’s commute lengths with respect to men in similar life stages more than they do British women’s when all else is controlled) is supported for LS4 and LS5. In US, there were significant gender differences in commute lengths in LS4, 5, and 6, and in GB, there were significant gender differences in commute lengths at LS6, but not LS4 and 5. In addition, the gender gap at LS6 was greater in GB. Therefore, there was not support for Hypothesis 4 at LS6. Hence, the results showed that there was some support for the “difference” hypothesis, which shows that context does matter. The results mainly confirm my proposition in Chapter III that US provides a context for greater “gender sameness” at work, while there is more room for greater balance in GB between roles at home and at work, albeit evidently at the expense of stronger roles in the labor market. GB women seem to be handling the household responsibility with part time working and caring more for their children themselves(due to the relatively more expensive child care cost), and with adjusting their journey-to-work mode to auto. And compared to GB women, US women are driven to a greater extent to handle the conflicting roles at home and work by opting to adjust their home and work separation. 101 However, the support for the “difference” hypothesis was partial. The following issues related to data and model specification could be making the observed difference lower between the two countries, and lead to a weaker support for the “Difference” Hypothesis: • A lower threshold for full time work in GB leads to an underestimation of part time workers in GB. • A greater range of levels of car mobility in GB makes switching to car when the household responsibilities are high a mediating factor for household responsibility. • Excluding trip chainers from the analysis may lead to an underestimation of the household responsibility effect in US. i) Lower threshold for full time work in GB One factor that could account for the lack of greater support for Hypotheses Three and Four is related to the definition of part time work in the two countries. First, thresholds between full time and part time employment in the two countries are different. The full time-part time threshold is 35 hours in US, but it is 30 hours in GB. About 6-7 % of employed women work between 30 and 35 hours in both countries (see Table V.1). This means that if the threshold for full time work had been 35 hours and above in GB, 6-7 % of women who were classified as full time in my GB sample would be considered part time. Since part time work is already one of the most important factors that differentiate the context for US and GB women, using the 35-hour threshold would mean that more women would be classified as 102 part time in GB, leading to a further differentiation between the two countries, therefore the household responsibility- commute length relationship in GB would be weaker compared to US. Table V.1. Percentage of Women Aged 15-64, Employed Part-Time Less than 30 hours Less than 35 hours US national data 18.7 25.2 UK national data 36.7 43.7 Source: Van Bastelaer et al (1997). ii) Differences in Trip Chaining A second factor that could have contributed to the somewhat weak support for the “difference” hypothesis could be that I have not included those persons who did not do a home-to-work or work-to-home trip on travel day in my analysis which constituted 12 % of the US and a smaller 4 % of the GB commuter samples. Greater share of persons who did not do a direct home to work or work to home trip in US is in alignment with Rosenbloom’s (1989) result that US women with children trip chain more than their European counterparts- most likely due to the difference of land use patterns between the two countries. Due to higher population densities, trip origins and destinations in GB are closer to each other than those in US, which leads to less need to link trips. This is particularly relevant to the measurement of household responsibilities since husband- and-wife 103 households with children in US are more likely to live in lower densities than other type households. Research shows that women with greater household responsibilities link trips more than others. Particularly women with more household responsibilities do more “out-of-home” trips on the way to or from work (Kwan, 2000). Therefore excluding women who did not do direct trips on the way to or from work as I have done in my analysis could mean excluding women who have higher household responsibilities, and quite possibly those with shorter commute lengths as a result of their household responsibilities. As discussed in the literature review section, researchers have linked women’s shorter commutes to their greater propensity to trip chain (McGuckin and Nakamoto, 2005). If they had been included, I might have gotten a stronger household responsibility- commute length relationship in US relative to GB. iii) Mode A third factor that could have contributed to the weak support for the “difference” hypothesis is the mode used in the work trip. The descriptive results section showed that there is a gender disparity in the use of auto in traveling to work in GB. Since US women already use automobiles at high levels, GB women have more room to make adjustments to their situations by changing mode of journey to work. For example, as shown earlier, British women living with another adult and school- aged children go to work by auto at 78 %, a rate 6 % higher than average GB women, and at only 3 % lower rate than GB men’s average rate, thus substantially closing the gender 104 gap in auto use in going to work. If mode were included in the models, it would act as a mediator between household responsibility and commute length, and the relationship between HR and commute length would be weaker in GB relative to US. V.3 An Overall Assessment of the Results Although the results showed that there was almost systematic support for the HRH through significance of LS variables, the overall effect of life stage variables were modest compared to some other variables such as household income, job status (full time-part time) and land use. The magnitudes of the coefficients for life stage variables, even if significant, hovered around one or two, while the coefficients of part time indicator, or density categories reached 4.0 or 5.0 in some cases. The relatively weak findings for the LS variables could be due to two main reasons: The first is the difficulty in measuring the household responsibility concept, and the limitation of the LS categories as a proxy to capture the household responsibility effect. This is a common problem in studies about this topic in the literature. Without information on attitudes towards housework and actual hours on household and market-related work, researchers will have to depend on household structures and life stage approximations to represent household responsibilities. The second reason for life stage variables being less than effective is the mediation of household responsibilities by other variables, the most important of which are part time working, and household income. As the study showed, there was a relationship between part time 105 working and household responsibility, as women in higher responsibility life stages had a significantly higher rate of part-time working. The support for HRH would have been higher if mediation through part time and other factors were considered. This study did not analyze how the interaction of part time work with household responsibility would affect commute lengths. In addition, the R-squared values in the regressions were quite low (all regression models explained about 1 % of the variation). The R-squared values being low implied that my model explains only a small amount of the variation in commute distance, and that there are other factors not included in the model that affect commute lengths. These factors may include residential location of individuals, individual income (in the case of US), wages, and car ownership. My assumption of fixed residential location may have been unrealistic. In real life, commute lengths could be changing (people keep on searching) since individuals change residential location, job location, and car ownership. Exclusion of residential location may have affected the R-sq values negatively. Moreover, the study did not control for individual income in the pooled form due to lack of availability of the variable in the US dataset 15 . Finally, excluding the car ownership variable may also have led to a low R-sq since car ownership may facilitate long commutes, helping explain the variation of commute lengths. 15 Including individual income in the GB separate regression improved the R-sq by 20 % (from 0.099 to 0.120) in the work trip time model. This showed that individual income is an important leverage factor in women’s commute lengths. Adding education variable, to the US model, on the other hand, provided only very modest increases in R-sq values. 106 V.4 Implications of the Results on Rationality Theory Some could argue that even the results supporting the HRH could be interpreted as choice according to rational utilitarian approach. For instance, the argument can be made that women with school-aged children prefer working close to home. However, these arguments are counteracted by several arguments: First, the attenuated effect of household income on women compared to men shows that the return to commutes is weaker for women, and thus imply that constraints apply to women as opposed to choices 16 . Second, the support for the “difference” hypothesis -even if partial- helps provide support for the HRH since both are based on a perspective of social and constraint-based construction of individuals’ actions as opposed to a utility-maximizing construction. If the data and method limitations explained above had been overcome, this support would have been even greater. Third, LS6 by definition is associated with higher household responsibility and LS4 with lower household responsibility. Third, women’s shorter commute times at high household responsibility life stage (LS6) without shorter commute distances and shorter commute distance without shorter commute time at a lower household responsibility stage (LS4) refurbishes the constraint interpretation as opposed to choice. Fourth, even if for one, the decision is a "choice”, the discussion becomes that of whether women’s shorter commute choice is the best outcome for them. First of all, individuals may not always choose the “best” 16 The argument is also made that women prefer shorter commutes over higher pay. 107 alternative for themselves since choices are based on perceptions and constraints. For example, a woman may not know that if she commutes to a job further away from her home, she might reap the positive outcomes from being in that particular job in the long term. Fourth, if there are too many women who are “choosing” to forego work for family (similar to choosing the job with a shorter commute), then this becomes a social problem. Namely, for girls and young women, the numbers of women role models who are not in traditional mommy track remain few, and therefore, perception of possibility of untraditional roles remains bleak. And finally, even if these women want to take on untraditional roles, they remain less likely to be part of social networks that lead to high status or male-dominated careers, and along the same line, they may not be as accepted into male-dominated occupations as easily as men, and even if they go into the untraditional route, they may remain subject to discrimination by men at work. Hence, one can also argue that theirs may be a rational choice, but not necessarily the best solution socially. V.5. Recommendations for Future Research and Practical Implications: Based on the discussion above, I can give several recommendations: First, it may be worthwhile to model to tease out the results in terms of determining to what extent part time workers’ shorter commutes come from household responsibility interactions, and to what extent from other factors such as lower financial return of part time for commuting longer or geographical explanations of women’s jobs. 108 Second, further cross-cultural tests of the HRH in countries with different child care options or other provisions that may affect women’s work-home balance must be done. In particular, it would be interesting to do an analysis in GB with new data since state-sponsored child care services and benefits to part time work have been increased at great scale in GB since mid to late 1990s, with the Labor Party’s coming to power in 1997. The changes should make the difference between US and GB starker. Hence, it would be invaluable to observe women’s commuting behavior and test the HRH in the new context. Third, in order to help US model household responsibility information better and to approximate household responsibility better, individuals’ childcare situation and choices, time use information, and gender attitudes must be incorporated into travel surveys. To approximate human capital factors, NPTS could include individual income, and NTS could include education. In addition, comparability of large-scale surveys between countries can be improved. Consistent definitions for what constitutes part time work can be chosen for part time work threshold, or information can be collected on the number of hours that individuals work. Implementing policies that could help women with school-aged children more easily combine work and home responsibilities. Increasing the provision of transportation to school children to schools could be one way. Another way could be provision of easier scheduling at work for women who have children at school age. In fact, GB has recently instituted “flexible 109 scheduling” policy, which is aimed for this purpose. In the longer run, equality inducing policies such as paternity leave policies may be helpful to reduce the gender inequality in household work division (Hanson and Hanson, 1981). 110 BIBLIOGRAPHY Alonso, W. (1964). Location and Land Use, Cambridge: Harvard University Press. Antecol, H. (2000). “An Examination of Cross-Country Differences in the Gender Gap in Labor Force Participation Rates.” Labour 7(4): 409-426. Antecol, H. (2003). Why Is There Cross-Country Variation in Female Labor Force Participation Rates? The Role of Male Attitudes Toward Family and Sex Roles. http://berger.claremontmckenna.edu/Summer/FacSupport/sexroles. Pdf. Bauder, H. (2001). “Culture in the Labor Market: Segmentation Theory and Perspectives of Place”. Progress in Human Geography, Volume 25, Number 1, March: 37-52(16). Batalova, J. A., and P. N. Cohen. (2002). “Premarital Cohabitation and Housework: Couples in Cross-National Perspective” Journal of Marriage and the Family, Vol. 64, pp. 743–755. Becker, G.S. (1965). "A Theory of the Allocation of Time." Economic Journal, 299 (75): 493-517. Bianchi, M., A. Milkie, L. C. Sayer, J. P. Robinson. (2000). “Is Anyone Doing the Housework? Trends in the Gender Division of Household Labor”. Social Forces, 79(1): 191-228. Blumen, O. (2000). “Dissonance in Women's Commuting? The Experience of Exurban Employed Mothers in Israel”, Urban Studies, Volume 37, Number 4, April 1: 731-748 (18). Cooke, L. P. (2005). Policy Pathways to Gender Power: State-Level Effects on the Division of Housework Running Head: Policy Effects on the Division of Housework. http://www.ccpr.ucla.edu/isarc28/Final%20Papers/PolicyPath way_Cooke.pdf Crane R. (2007). "Is There a Quiet Revolution in Women’s Travel? Revisiting the Gender Gap in Commuting,” Journal of the American Planning Association, 73, Summer. 111 Dex, S. and L.B. Shaw. (1986). British and American Women at Work. The Macmillan Press Ltd., London. Esping-Andersen, G. (2002). Towards a Post-Industrial Gender Contract. In: Peter Auer/Bernard Gazier (eds.): The Future of Work, Employment and Social Protection. Geneva: International Institute for Labour Studies, pp. 109-128. England, K. (1993). “Suburban Pink Collar Ghettos: The Spatial Entrapment of Women?” Annals of the Association of American Geographers 83: 225-242. Erickson, J. (1977): “An Analysis of the Journey-to-work for Women” Social Problems 24, 428–435. Eyles, J. (1971). “Pouring New Sentiments into Old Theories: How Else Can We Look at Behavioral Patterns?” Area, 3, 242-250. Fagnani J. (1987). “Daily Commuting Time: The Stakes for Working Mothers in France” Transportation Research Record, No. 1135: 26-30. Fainstein S. (2001). Gender and Planning: Theoretical Issues. Columbia University, New York. Fuwa, M. (2004). “Macro-Level Gender Equality and the Division of Household Labor in 22 Countries” American Sociological Review, Vol. 69. Fuwa, M. and P. N. Cohen. (2007). "Housework and Social Policy" Social Science Research, 36(2): 512-530. Gershuny J., (2000). Changing Times: Work and Leisure in Postindustrial Society. Oxford University Press, Oxford and New York. Giele J. and L. Stebbins. (2003). Women and Equality in the Workplace: a Reference Handbook (Contemporary World Issues). Figure 5.3 : compiled by the Institute for Women’s Policy Research from US Department of Labor, Bureau of Labor Statistics, 2001a, Table 17. Gottfried, H., Reese, L. (2004). Equity in the Workplace: Gendering Workplace Policy Analysis. New York: Lexington Books Giuliano G. (1999). "Land Use and Transportation: Why We Won't Get There From Here" Transportation Research Circular. Giuliano, G. and D. Narayan (2003). “Another Look at Travel Patterns and Urban Form: The US and Great Britain” Urban Studies, Vol. 40: 2295-2312. 112 Giuliano. G. and J. Dargay (2006).“Car Ownership, Travel and Land Use: The US and Great Britain” Transportation Research, A. 40: 106-124. Gordon P., Kumar, A. and H. W. Richardson (1989). "Gender Differences in Metropolitan Travel Behavior" Regional Studies, 23(6): 499-510. Granovetter M. (1985). “Economic Action and Social Structure: The Problem of Embeddedness,” American Journal of Sociology, 1: 481-510. Hägerstrand T. (1970). What About People in Regional Sciences? In Regional Science Association Papers. Ninth European Congress of the Regional Science Association. Hamilton K. (2002). Gender and Transport in Developed Countries. Background Paper for the Expert Workshop, Gender Perspectives for Earth Summit 2002: Energy, Transport, Information for Decision-Making. Hanson, S. and P. Hanson. (1981). "The Impact of Married Women’s Employment on Household Travel Patterns; A Swedish Example." Transportation. 10: 165-183. Hanson S. and G. Pratt. (1988). “Spatial Dimensions of the Gender Division of Labor in a Local Labor Market”. Hanson, S. and G. Pratt. (1991). “Job Search and the Occupational Segregation of Women” Annals of the Association of American Geographers 81: 229-53. Hanson, S. and G. Pratt (1995). Gender, Work, and Space. New York: Routledge. Harkness, S. and J. Waldfogel. (1999). The Family Gap in Pay: Evidence from Seven Industrialised Countries, Luxemburg Income Study, Working Paper No. 219. Haworth, A. , Manzi, T. and J. Kemeny. (2004). Social Constructionism and International Comparative Housing Research. In: Jacobs, Keith and Manzi, (eds.). Ashgate, Aldershot, UK, pp. 159-177. http://www.ibf.uu.se/PERSON/ jim/commers/compara.pdf Hirsch, B. (2000). “The Relative Compensation of Part-Time and Full-Time Workers: The Role of Worker and Job Skill”, July. 113 Hjorthol, R.J. (2000). “Same City – Different Options: An Analysis of the Work Trips of Married Couples in the Metropolitan Area of Oslo” Journal of Transportation Geography, 8: 213-220. Hochschild, A. R. (1989). The Second Shift. Working Parents and the Revolution at Home. New York: Viking. Hoffmeister, H. (2006). Literature on Job Mobility in Germany, State-of-the- Art of Mobility Research, Chapter Job Mobilities Working Paper No. 2006-01, Chapter 5, http://www.jobmob-and-famlives.eu/papers/JFW_06- 1_05_Ruppenthal _Limmer_Bonss.pdf Hook, J. L. (2006). “Care in Context: Men's Unpaid Work in 20 Countries, 1965- 2003”. American Sociological Review. Albany: Vol. 71, Iss. 4; 22 pgs. IBG Women and Geography Study Group (1984). Geography and Gender. London. Jacobs, J.A. and K. Gerson (2004). The Time Divide: Work, Family, and Gender Inequality. Harvard University Press. Johnston-Anumonwo, I. (1992). “The Influence of Household Type on Gender Differences in Work Trip Distance.” Professional Geographer 44: 161-69. Kain J F, 1962 "The journey to work as a determinant of residential location" Papers and Proceedings of the Regional Science Association, 9: 137-160 Kamerman, S. B. and A. J. Kahn. (1997). "Family Change and Family Policies in the United States", in Sheila B. Kamerman and Alfred J. Kahn, Family Change and Family Policies in Great Britain, Canada, New Zealand, and the United States, Oxford, Eng.: Oxford University Press. Kamiya H. (1999). “Day Care Services and Activity Patterns of Women in Japan”. Geojournal, Volume 48, Number 3, July. Kawase M. (2004). “Changing Gender Differences in Commuting in the Tokyo Metropolitan Suburbs”. Geojournal 61: 247-253. Kwan, Mei-Po (1999). “Gender and Individual Access to Urban Opportunities: A Study Using Space-Time Measures” The Professional Geographer, 51(2): 210-227. Kwan, Mei-Po. (2000). “Gender Differences in Space-Time Constraints” Area 32 (2): 145-156. 114 Lemaître A. G. and P. Marianna (1997). "The Definition of Part-Time Work for the Purpose of International Comparisons", OECD Labour Market and Social Policy Occasional Papers, No. 22, OECD Publishing. doi:10.1787/132721856632 Lawson C. T. (1998). “Household Travel/Activity Decisions: Who Wants to Travel?” Discussion Paper 98-4 Center for Urban Studies, College of Urban and Public Affairs Portland State University. Lee Bun S. and J. F. McDonald. (2003). “Determinants of Commuting Time and Distance for Seoul Residents: The Impact of Family Status on the Commuting of Women”, Urban Studies, Vol. 40, No. 7: 1283–1302. Lippe T. van and L. van Dijk (2001). Women’s Employment in a Comparative Perspective. New York. Luxembourg Income Study (2003). Family Policy Database, Version 2 (2003) Early Childhood Education and Care (ECEC). http://www.lisproject.org publications/fampol/fampol03.htm. Madden J. F. (1981). "Why Women Work Closer to Home," Urban Studies, 18: 181-194. Mallier A. and Rosser, M. (1987). Women and the Economy, MacMillan, London. Marini, M. M., and B. A. Shelton (1993). “Measuring Household Work: Recent Experiences in the U.S.” Social Science Research, Vol. 22: 361–382. McGuckin, N.and Y. Nakamoto (2005). “Differences in Trip Chaining by Men and Women”. Conference on Research on Women's Issues in Transportation. Chicago Illinois, United States. McLafferty, S. and V. Preston (1992). “Spatial Mismatch and Labor Market Segmentation for African-American and Latina Women.” Economic Geography 68: 406-31. Mokhtarian P. L. and I. Salomon (1999). How Derived is the Demand for Travel? Some Conceptual and Measurement Considerations by Institute of Transportation Studies, UC Davis in Working Paper Series with number UCD- ITS-REP-01-15. Newman, P. and J. Kenworthy (1989). Cities and Automobile Dependence: An International Sourcebook, Gower, England. 115 Newman, P. and J. Kenworthy (1999). Sustainability and Cities: Overcoming Automobile Dependence. Island Press. Washington, DC. Nobis, C., and Lenz, B. (2005).” Gender Differences in Travel Patterns: The Role of Employment Status and Household Structure” In Research on Women’s Issues in Transportation, Vol. 2: Technical Papers, Transportation Research Board Conference Proceedings 35 (pp. 114-123). Washington, DC: National Research Council. OECD (2001). Thematic Review on Early Childhood Education and Care. Background Paper on United States. Orloff, A.S. (2002) “Explaining US Welfare Reform: Power, Gender, Race and the US Policy Legacy”, Critical Social Policy, 22 (1): 96–118. Ory, D. T., P.L. Mokhtarian, L. Redmond, I. Salomon, G.O. Collantes and S. Choo. (2004). “When is Commuting Desirable to the individual?” Growth and Change, 5 (3): 334-359. Pazy, A., I. Salomon and T. Pintzov (1996). “The Impacts of Women’s Careers on Their Commuting Behavior: A Case Study of Israeli Computer Professionals” Transportation Research (A). 30 (4): 269-386. Pucher J. (1988). “Urban Travel Behavior as the Outcome of Public Policy” Journal of the American Planning Association, Summer: 509-519. Rosenbloom S. (1989). “Trip-Chain Behavior: A Comparative and Cross- Cultural Analysis of the Travel Patterns of Working Mothers” In Gender, Transport and Employment, edited by Margaret Grieco, Laurie Pickup, and Richard Whipp. Vermont: Gower Publishing Company. Rosenbloom S. (2005). “ Women’s Travel Issues: The Research and Policy Environment” in Gender and Planning : A Reader edited by S. Fainstein and L. J. Servon. New Brunswick, N.J. : Rutgers University Press. Rutherford, B., and G. Wekerle. (1988). “Captive Rider, Captive Labor: Spatial Constraints on Women’s Employment” Urban Geography, 9: 116-137. Salomon I., P. Bovy and J.P. Orfeuil (1993). A Billion Trips a Day: Tradition and Transition in European Travel Patterns, Kluwer Academic Publishers, Amsterdam. Saunders P. (2001). “Australia Is Not Sweden: National Cultures and the Welfare State, Policy”, Policy, 17, 3: 29-32. 116 Schafer, A. (2000). “Regularities in Travel Demand: An International Perspective”, invited paper in Journal of Transportation and Statistics, Vol. 3(3): 1-32. Schafer, A. and D. Victor. (2000). “The Future Mobility of the World Population” Transportation Research A 34, No. 3: 171–205. Shelton, B. A., and D. John. (1996). “The Division of Household Labor” Annual Reviews of Sociology, Vol. 22: 299–322. Sermons M. W. and F. S. Koppelman. (2001). “Representing the Differences Between Female and Male Commute Behavior in Residential Location Models” Journal of Transport Geography, V. 9, N. 2: 101-110. Simma A., D. Dubuluoz, A. Shafer, K. Ohta, and K. W. Axhausen (2002). “Data, Data Structures and Models”. Arbeitsberichte Verkehrs- und Raumplanung, 103, Institut für Verkehrsplanung, Transporttechnik, Strassen- und Eisenbahnbau, ETH, Zürich. http://e-collection.ethbib.ethz.ch/ecol- pool/incoll/incoll_699.pdf. Singell L. D. and J. H. Lillydahl (1986). “An Empirical Analysis of the Commute to Work Patterns of Males and Females in Two-Earner Households” Urban Studies, Vol. 2: 119-129. Spain D. and S. M. Bianchi (1996). The Balancing Act: Motherhood, Marriage, and Employment among American Women. Russell Sage Foundation, 112 East 64th Street, New York, NY 10021. Stier H., N. Lewin-Epstein and M. Braun (2001). “Welfare Regimes, Family- Supportive Policies, and Women's Employment along the Life-Course”. The American Journal of Sociology, Vol. 106, No. 6. (May): 1731-1760. Taylor, B. D. (1999). “Gender Socialization and Household Responsibility: An Analysis of Household-Serving Travel and Commuting” (Working Paper). Taylor, B. D. and M. Mauch (1998). Gender, Race, and Travel Behavior: An Analysis of Household-Serving Travel in the San Francisco Bay Area. Women’s Travel Issues: Proceedings from the Second National Conference, October. Tivers, J. (1978). “How the Other Half Lives: The Geographical Study of Women” Area, Vol. 10: 102–106. 117 Tivers, J. (1985). Women Attached. Croom Helm, London. Tomlinson, J. (2007). “Employment Regulation, Welfare and Gender Regimes: a Comparative Analysis of Women’s Working-Time Patterns and Work–Life Balance in the UK and the US” International Journal of Human Resource Management 18 : 3: 401–415. Turner T. and D. Niemier. (1997). “Travel to Work and Household Responsibility: New Evidence” Transportation, 24: 397-419. Van Bastelaer, A., G. Lemaître and P. Marianna (1997). "The Definition of Part- Time Work for the Purpose of International Comparisons", OECD Labour Market and Social Policy Occasional Papers, No. 22, OECD Publishing. doi:10.1787/132721856632. Van den Berg, G. J. and C. Gorter. (1997). “Job Search and Commuting Time” Journal of Business and Economic Statistics 15: 269-281. Van der Lippe T. and L. van Dijk (2002). “Comparative Research on Women's Employment” Annual Review of Sociology Vol. 28: 221-241. Villeneuve and Rose (1988). "Gender and Separation of Employment from Home in Metropolitan Montreal, 1971-1981." Urban Geography, 9: 155-179. Wachs, M. (1991). “Men, Women, and Urban Travel: The Persistence of Separate Spheres,” in Martin Wachs and Margaret Crawford, eds., The Car and the City: The Automobile, The Built Environment, and Daily Urban Life. Ann Arbor: University of Michigan Press, pp. 86-100. Walby, S. (1997). Gender Transformations. London: Routledge. Walby, S. (2004). “The European Union and Gender Equality: Emergent Varieties of Gender Regime”, Social Politics, 11(1): 4–29. Olsen W. and S. Walby. (2004). Modeling Gender Pay Gaps. Equal Opportunities Commission, First published Winter 2004, http://www.eoc. org.uk/PDF/modelling_gender_pay_gaps_wp_17.pdf Wegner, J. (2001). The Continuing Problems with Part-Time Jobs, Economic Policy Institute. Weinberger R. (2007). “Men, Women, Job Sprawl, and Journey To Work In The Philadelphia Region” Public Works Management & Policy, Vol. 11 No. 3, January: 177-193. 118 Willman-Navarro A. (2006). “All Things Unequal: Female Career Continuity and the Gender and Family Gaps” The New School International Affairs Working Paper 2006-01, http://www.gpia.info/files/u1/wp/2006-01.pdf White M. (1977). “A Model of Residential Location Choice and Commuting by Men and Women Workers" Journal of Regional Science, Vol. 17, No. 1, April. White, M. (1986). “Sex Differences in Urban Commuting Patterns”. The American Economic Review, 76, pp. 332–368. Zahavi Y. (1979). Travel Over Time, Report PL-79-004 (FHWA, U.S. Department of Transportation). 119 APPENDIX 120 Table. 1. Estimation Results for Work Trip Distance, US and GB R 2 (Adj) Total Sample Women GB B Sig. B Sig. B Sig. Constant 10.0 0.00 GB 1.0 0.24 FEM -2.2 0.00 GBFEM -1.4 0.00 Part Time Worker -4.4 0.00 1.2 0.00 1.1 0.03 2+ workers in household -0.5 0.03 0.5 0.22 -1.2 0.02 HOUSEHOLD INCOME: L.T. 25K -2.0 0.00 0.6 0.16 -2.3 0.00 HOUSEHOLD INCOME: 25K-40K -1.1 0.00 0.7 0.06 -1.7 0.00 HOUSEHOLD INCOME: G.E.60K 0.5 0.03 0.3 0.43 1.2 0.03 MSA Size: Not in MSA -1.6 0.00 0.2 0.62 2.6 0.00 MSA Size: L.T. 250K -1.9 0.00 -0.2 0.74 2.9 0.00 MSA SIZE: 250K-500K -1.9 0.00 1.2 0.04 1.3 0.08 MSA Size: 500K-1M -1.5 0.00 0.8 0.11 0.0 0.97 MSA Size: > 3M 3.1 0.00 -1.5 0.00 -3.4 0.00 DENSITY: low (<1000/mi2) 5.2 0.00 -0.6 0.14 -3.0 0.00 DENSITY: Medium (1K-4K/mi2) 1.4 0.00 -0.3 0.46 -0.1 0.92 DENSITY: very high (>10K/mi2) -1.9 0.00 0.4 0.44 1.8 0.00 1 adult, 16-34 (LS1) -0.7 0.24 1.4 0.11 0.5 0.63 1 adult, 35-64 (LS2) -0.6 0.23 0.8 0.23 1.2 0.20 2 adults, no child, 16-34 (LS3) 1.0 0.01 1.4 0.00 1.1 0.08 2 adults, no child, 35-64 (LS4) 0.8 0.01 -0.8 0.05 0.2 0.77 2 adults, youngest child preschool age (LS5) 1.3 0.00 -1.0 0.02 1.5 0.01 2 adults, youngest child school age (LS6) 0.8 0.01 -1.2 0.01 1.3 0.02 1 adult, child 15 and under (LS7) 1.0 0.37 -0.7 0.58 0.6 0.69 R 2 (Adj)= 0.096 S.E=11 F=46 F Prob=0.00 N=26,842 Bold=significant at p<= 0.1 121 Table. 2. Estimation Results for Work Trip Time, US and GB Total Sample Women GB B Sig. B Sig. B Sig. Constant 19.0 0.00 GB 11.1 0.00 FEM -2.4 0.01 GBFEM -1.1 0.08 Part Time Worker -5.2 0.00 0.7 0.26 0.4 0.62 2+ workers in household -1.4 0.00 1.1 0.07 -1.9 0.01 HOUSEHOLD INCOME: L.T. 25K -2.0 0.00 0.8 0.21 -3.1 0.00 HOUSEHOLD INCOME: 25K-40K -1.3 0.00 1.1 0.05 -2.8 0.00 HOUSEHOLD INCOME: G.E.60K 1.3 0.00 0.1 0.79 1.6 0.05 MSA Size: Not in MSA -2.4 0.00 -0.3 0.65 0.0 1.00 MSA Size: L.T. 250K -2.3 0.00 -0.4 0.61 1.3 0.24 MSA SIZE: 250K-500K -2.5 0.00 1.2 0.18 0.0 0.99 MSA Size: 500K-1M -1.7 0.00 0.8 0.29 -0.4 0.79 MSA Size: > 3M 6.0 0.00 -2.3 0.00 3.9 0.00 DENSITY: low (<1000/mi2) 4.2 0.00 -0.8 0.16 -3.2 0.00 DENSITY: Medium (1K-4K/mi2) 0.5 0.25 -0.2 0.69 0.2 0.76 DENSITY: very high (>10K/mi2) 2.6 0.00 2.5 0.00 -2.5 0.00 1 adult, 16-34 (LS1) -2.3 0.01 3.4 0.01 -1.0 0.53 1 adult, 35-64 (LS2) -0.8 0.26 1.1 0.30 1.9 0.17 2 adults, no child,16-34 (LS3) 0.2 0.73 2.1 0.00 3.1 0.00 2 adults, no child,35-64 (LS4) 1.4 0.00 -0.8 0.19 -0.9 0.24 2 adults, youngest child preschool age (LS5) 0.8 0.08 -1.3 0.04 1.9 0.03 2 adults, youngest child school age (LS6) 1.0 0.02 -2.1 0.00 1.0 0.22 1 adult, child 15 and under (LS7) 0.5 0.74 -0.4 0.85 2.5 0.28 R 2 (Adj)= 0.095 S.E=16 F= 46 F Prob = 0.00 N= 26,842 Bold=significant at p<= 0.1 122 Table. 3. Estimation Results for Work Trip Distance, US and GB Women Total Sample, US and GB Women GB Women B Sig. B Sig. Constant 8.01 0.00 GB -1.00 0.43 Part Time Worker -3.18 0.00 0.85 0.11 2+ workers in household 0.02 0.94 -1.59 0.06 HOUSEHOLD INCOME: L.T. 25K -1.66 0.00 -0.97 0.21 HOUSEHOLD INCOME: 25K-40K -0.49 0.06 -1.13 0.09 HOUSEHOLD INCOME: G.E.60K 0.88 0.00 0.84 0.26 MSA Size: Not in MSA -1.28 0.00 2.67 0.00 MSA Size: L.T. 250K -1.93 0.00 2.79 0.01 MSA SIZE: 250K-500K -0.67 0.13 1.61 0.09 MSA Size: 500K-1M -0.76 0.03 1.46 0.30 MSA Size: > 3M 1.42 0.00 -1.09 0.24 DENSITY: low (<1000/mi2) 4.34 0.00 -1.79 0.01 DENSITY: medium (1K-4K/mi2) 1.11 0.00 -0.51 0.45 DENSITY: very high (>10K/mi2) -1.56 0.00 1.63 0.03 1 adult, 16-34 (LS1) 0.63 0.33 0.29 0.85 1 adult, 35-64 (LS2) 0.32 0.50 0.17 0.90 2 adults, no child, 16-34 (LS3) 2.39 0.00 0.80 0.29 2 adults, no child, 35-64 (LS4) -0.10 0.72 0.17 0.80 2 adults, youngest child preschool age (LS5) 0.23 0.49 1.37 0.12 2 adults, youngest child school age (LS6) -0.38 0.20 0.83 0.28 1 adult, child 15 and under (LS7) 0.45 0.42 -0.74 0.63 R 2 (Adj)= 0.087 S.E= 9 F= 28 F Prob=0.00 N=11,850 Bold=significant at p<= 0.1 123 Table 4. Estimation Results for Work Trip Time, US and GB Women Total Sample, US and GB Women GB Women B Sig. B Sig. Constant 16.92 0.00 GB 7.56 0.00 Part Time Worker -4.39 0.00 -0.37 0.66 2+ workers in household -0.26 0.61 -1.63 0.23 HOUSEHOLD INCOME: L.T. 25K -1.82 0.00 0.22 0.86 HOUSEHOLD INCOME: 25K-40K -0.44 0.28 -1.49 0.16 HOUSEHOLD INCOME: G.E.60K 1.42 0.00 1.74 0.14 MSA Size: Not in MSA -2.55 0.00 0.59 0.67 MSA Size: L.T. 250K -2.62 0.00 2.11 0.19 MSA SIZE: 250K-500K -1.29 0.07 0.75 0.62 MSA Size: 500K-1M -1.03 0.06 1.67 0.46 MSA Size: > 3M 3.43 0.00 7.40 0.00 DENSITY: low (<1000/mi2) 3.26 0.00 -2.49 0.03 DENSITY: Medium (1K-4Kmi2) 0.41 0.35 -1.28 0.24 DENSITY: very high (>10K/mi2) 4.97 0.00 -2.52 0.03 1 adult, 16-34 (LS1) 1.10 0.28 -1.41 0.56 1 adult, 35-64 (LS2) 0.32 0.67 0.89 0.66 2 adults, no child, 16-34 (LS3) 2.17 0.00 3.40 0.01 2 adults, no child, 35-64 (LS4) 0.37 0.41 -0.08 0.94 2 adults, youngest child preschool age (LS5) -0.82 0.12 3.08 0.03 2 adults, youngest child school age (LS6) -1.09 0.02 0.48 0.69 1 adult, child 15 and under (LS7) 0.48 0.60 0.35 0.89 R 2 (Adj)= 0.091 S.E=15 F= 30 F Prob=0.00 N=11,850 Bold=significant at p<= 0.1 124 Table. 5. Estimation Results: Work Trip Distance, US Men and Women Total Sample, US Men and Women US Women B Sig. B Sig. Constant 9.92 0.00 Female -1.94 0.01 MSA Size: Not in MSA -1.72 0.00 0.44 0.44 MSA Size: L.T. 250K -2.03 0.00 0.10 0.88 MSA SIZE: 250K-500K -1.95 0.00 1.27 0.07 MSA Size: 500K-1M -1.44 0.00 0.67 0.22 MSA Size: > 3M 3.26 0.00 -1.83 0.00 DENSITY: low (<1000/mi2) 5.33 0.00 -0.98 0.03 DENSITY: medium (1K-4K/mi2) 1.36 0.00 -0.25 0.58 DENSITY: very high (>10K/mi2) -1.87 0.00 0.33 0.60 Job Status -4.53 0.00 1.36 0.00 1 adult, 16-34 (LS1) -0.64 0.31 1.29 0.20 1 adult, 35-64 (LS2) -0.65 0.24 0.99 0.21 2 adults, no child, 16-34 (LS3) 0.96 0.02 1.44 0.01 2 adults, no child, 35-64 (LS4) 0.82 0.01 -0.89 0.05 2 adults, youngest child preschool age (LS5) 1.26 0.00 -1.02 0.04 2 adults, youngest child school age (LS6) 0.72 0.02 -1.06 0.02 1 adult, child 15 and under (LS7) 0.78 0.51 -0.29 0.83 2+ workers in household -0.58 0.03 0.61 0.20 HOUSEHOLD INCOME: L.T. 25K -1.76 0.00 0.09 0.86 HOUSEHOLD INCOME: 25K-40K -1.05 0.00 0.55 0.18 HOUSEHOLD INCOME: G.E.60K 0.52 0.05 0.37 0.37 R 2 (Adj)= 0.09 S.E=11.1 F= 52 F Prob=0.00 N=21,906 Bold=significant at p<= 0.1 125 Table. 6. Estimation Results for Work Trip Distance, GB Men and Women Total Sample, GB Men and Women GB Women B Sig. B Sig. Constant 11.50 0.00 Female -4.48 0.00 MSA Size: Not in MSA 0.83 0.21 0.56 0.59 MSA Size: L.T. 250K 1.00 0.21 -0.14 0.91 MSA SIZE: 250K-500K -1.00 0.17 1.94 0.08 MSA Size: 500K-1M -2.82 0.03 3.52 0.06 MSA Size: > 3M -1.92 0.01 2.25 0.06 DENSITY: low (<1000/mi2) 1.45 0.01 1.11 0.18 DENSITY: medium (1K-4K/mi2) 1.75 0.00 -1.15 0.16 DENSITY: very high (>10K/mi2) 0.05 0.93 0.02 0.98 Job Status -2.63 0.00 0.30 0.74 1 adult, 16-34 (LS1) -0.09 0.93 1.01 0.56 1 adult, 35-64 (LS2) 1.04 0.28 -0.55 0.72 2 adults, no child, 16-34 (LS3) 2.31 0.00 0.88 0.36 2 adults, no child, 35-64 (LS4) 0.93 0.08 -0.86 0.29 2 adults, youngest child preschool age (LS5) 3.02 0.00 -1.42 0.16 2 adults, youngest child school age (LS6) 2.43 0.00 -1.98 0.03 1 adult, child 15 and under (LS7) 4.22 0.32 -4.50 0.32 2+ workers in household -1.68 0.00 0.11 0.90 HOUSEHOLD INCOME: L.T. 25K -5.07 0.00 2.45 0.01 HOUSEHOLD INCOME: 25K-40K -3.19 0.00 1.58 0.05 HOUSEHOLD INCOME: G.E.60K 2.00 0.00 0.56 0.59 R 2 (Adj)= 0.11 S.E=9.5 F= 15.5 F Prob=0.00 N=4,874 Bold=significant at p<= 0.1 126 Table. 7. Estimation Results for Work Trip Time, US Men and Women Total Sample, US Men and Women US Women B Sig. B Sig. Constant 18.79 0.00 Female -2.00 0.04 MSA Size: Not in MSA -2.59 0.00 0.06 0.94 MSA Size: L.T. 250K -2.33 0.00 -0.29 0.74 MSA SIZE: 250K-500K -2.56 0.00 1.28 0.20 MSA Size: 500K-1M -1.57 0.00 0.55 0.48 MSA Size: > 3M 6.22 0.00 -2.76 0.00 DENSITY: low (<1000/mi2) 4.35 0.00 -1.06 0.09 DENSITY: medium (1K-4K/mi2) 0.30 0.47 0.13 0.84 DENSITY: very high (>10K/mi2) 2.77 0.00 2.21 0.01 Job Status -5.50 0.00 1.14 0.08 1 adult, 16-34 (LS1) -2.19 0.01 3.40 0.02 1 adult, 35-64 (LS2) -0.85 0.27 1.27 0.25 2 adults, no child, 16-34 (LS3) 0.30 0.59 1.91 0.02 2 adults, no child, 35-64 (LS4) 1.62 0.00 -1.22 0.06 2 adults, youngest child preschool age (LS5) 0.86 0.05 -1.65 0.02 2 adults, youngest child school age (LS6) 0.99 0.02 -2.02 0.00 1 adult, child 15 and under (LS7) -0.08 0.96 0.65 0.73 2+ workers in household -1.40 0.00 1.17 0.07 HOUSEHOLD INCOME: L.T. 25K -1.54 0.00 -0.29 0.67 HOUSEHOLD INCOME: 25K-40K -1.23 0.00 0.79 0.17 HOUSEHOLD INCOME: G.E.60K 1.29 0.00 0.15 0.79 R 2 (Adj)= 0.08 S.E=15.6 F= 48.5 F Prob=0.00 N=21,906 Bold=significant at p<= 0.1 127 128 Table. 8. Estimation Results for Work Trip Time, GB Men and Women Total Sample, GB Men and Women GB Women B Sig. B Sig. Constant 31.56 0.00 Female -7.09 0.01 MSA Size: Not in MSA -2.98 0.02 1.03 0.59 MSA Size: L.T. 250K -1.51 0.31 0.99 0.66 MSA SIZE: 250K-500K -3.13 0.02 2.58 0.21 MSA Size: 500K-1M -3.86 0.10 4.51 0.20 MSA Size: > 3M 7.37 0.00 3.46 0.12 DENSITY: low (<1000/mi2) 0.63 0.51 0.14 0.93 DENSITY: medium (1K-4K/mi2) 1.82 0.06 -2.69 0.08 DENSITY: very high (>10K/mi2) 0.18 0.86 2.27 0.14 Job Status -3.14 0.03 -1.62 0.34 1 adult, 16-34 (LS1) -3.00 0.12 2.69 0.40 1 adult, 35-64 (LS2) 1.54 0.38 -0.33 0.91 2 adults, no child, 16-34 (LS3) 3.15 0.01 2.42 0.17 2 adults, no child, 35-64 (LS4) 0.09 0.93 0.21 0.89 2 adults, youngest child preschool age (LS5) 2.52 0.01 -0.26 0.89 2 adults, youngest child school age (LS6) 2.55 0.02 -3.16 0.06 1 adult, child 15 and under (LS7) 13.12 0.10 -12.29 0.14 2+ workers in household -3.48 0.00 1.60 0.35 HOUSEHOLD INCOME: L.T. 25K -6.95 0.00 5.35 0.00 HOUSEHOLD INCOME: 25K-40K -4.91 0.00 2.97 0.04 HOUSEHOLD INCOME: G.E.60K 2.87 0.01 0.29 0.87 R 2 (Adj)= 0.10 S.E=17.6 F= 14.0 F Prob=0.00 N=4,874 Bold=significant at p<= 0.1

Abstract (if available)

Abstract The research on the effects of household responsibilities on commute lengths so far is not conclusive. Further, little has been done to test the hypothesis cross-culturally. This study is an international comparative analysis of the relationship of household responsibilities and women's commute lengths. I hypothesize that 1) Household responsibilities shorten women's commute lengths both with respect to men and other women with lower amount of household responsibilities, and, 2) Women in US have a tighter household responsibility-commute length relationship than women in GB. Using 1995 US Nationwide Personal Transportation Survey (NPTS) data from the United States, and 1995/97 NTS data from Great Britain, I approximate household responsibilities with a composite life stage variable consisting of age, presence of children, and number of adults in a household. I present descriptive results and results on estimated models which test for both independent and interaction effects of gender and country on work trip distance and time, and presents results of estimated equations for pooled, women-only, and separate regressions for US and GB, controlling for land use indicators, number of workers and household income.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Essays on congestion, agglomeration, and urban spatial structure

PDF

Children’s travel behavior in journeys to school

PDF

The relationship between pubertal timing and delinquent behavior in maltreated male and female adolescents

PDF

The built environment, tour complexity, and active travel

PDF

Omentectomy in the canine: a unique model to investigate the physiologic relationship between visceral fat and insulin action

PDF

Understanding the dynamic relationships between physical activity and affective states using real-time data capture techniques

PDF

Healthy mobility: untangling the relationships between the built environment, travel behavior, and environmental health

PDF

A descriptive study of the “hardware” of education: the relationship between zoning and schoolsites in the Los Angeles Unified School District

PDF

Existence and response: living in quarantine during a pandemic

PDF

Role of purinergic P2X7 receptors in inflammatory responses in the brain and liver: a study using a mouse model of chronic ethanol and high-fat diet exposure

PDF

Breaking it down to make it stronger: examining the role of source credibility and reference group specificity in the influence of personalized normative feedback on perceived alcohol use norms a...

PDF

The influence of organizational ethics on job attitudes and government performance

PDF

Using Landsat and a Bayesian hard classifier to study forest change in the Salmon Creek Watershed area from 1972–2013

PDF

Age related macular degeneration in Latinos: risk factors and impact on quality of life

PDF

Protest of the oppressed: a case study on Black female administrators’ experiences in K-12 education from student to leadership

Asset Metadata

Creator Karsi, Elif (author)

Core Title The relationship between women's household responsibilities and commute lengths: a study on women in the US and Great Britain

School School of Policy, Planning, and Development

Degree Doctor of Philosophy

Degree Program Planning

Publication Date 05/06/2010

Defense Date 02/25/2008

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag commute,comparative,cross-cultural,gender,great britain,household responsibility,North America,OAI-PMH Harvest,Travel

Place Name islands: Great Britain (geographic subject), USA (countries)

Language English

Advisor Giuliano, Genevieve (committee chair), Moore, James Elliott, II (committee member), Richardson, Harry W. (committee member)

Creator Email ekarsi@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m1226

Unique identifier UC1101794

Identifier etd-Karsi-20080506 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-68100 (legacy record id),usctheses-m1226 (legacy record id)

Legacy Identifier etd-Karsi-20080506.pdf

Dmrecord 68100

Document Type Dissertation

Rights Karsi, Elif

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu