Environment and Planning B: Planning and Design 2007, volume 34, pages 154 ^ 170 DOI:10.1068/b32124 The exposure of disadvantaged populations in freeway air-pollution sheds: a case study of the Seattle and Portland regions Chang-Hee Christine Bae, Gail Sandlin, Alon Bassok Department of Urban Design and Planning, College of Architecture and Urban Planning, University of Washington, Seattle, WA 98195-5740, United States; e-mail: cbae@u.washington.edu, gsandlin@u.washington.edu, abassok@u.washington.edu Sungyop Kim Department of Architecture, Urban Planing and Design, University of Missouri-Kansas, 213 Epperson, 511 Rockhill Road, Kansas City, MO 64110-2499, United States; e-mail: kims@umkc.edu Received 7 August 2005; in revised form 29 January 2006 Abstract. Freeway-related air pollution and its harmful health risks have been observed in recent research in the environmental-health sciences. In this study we investigate the impact of freeway and arterial-road air pollution on vulnerable populations ?for example, the poor, minorities, children, and the elderly?whose housing options are limited. Because many mobile-source emissions decay rapidly with distance, approaching background concentrations at 330 ft from the freeway, populations living near limited access roads are most at risk from exposure. Furthermore, microscale air monitoring systems are rarely in place at these locations in the United States. In this research we will define freeway air-pollution sheds with the aid of a geographic information system analysis and determine populations that may be at risk from exposure to mobile-source pollutants in two West Coast metropolitan areas (Seattle and Portland). We then use cluster analysis to identify key neighborhoods at risk in Seattle. Subsequently, we apply a hedonic pricing model to understand the extent to which house price values in Seattle are related to freeway proximity. Finally, we discuss policy options, planning implications, and mitigation measures, including an assessment of air-quality monitoring needs and land-use prescriptions. 1 Introduction Traditionally, environmental-justice studies have focused on the disparate impact of air pollution on low-income and/or minority populations residing near stationary or point sources from industrial facilities or locally undesirable land uses.(1) However, if roadways are viewed as pollution line sources then an examination of populations living near high traffic densities that release both criteria as well as hazardous air pollutants, also known as `air toxics') may also have environmental-justice implications. Six criteria pollutants are listed under the National Ambient Air Quality Standards required by the (1) A recent example is the results of a lawsuit filed by the Air Quality Management District (AQMD) in Southern California against the oil company BP/Arco in March 2005. The lawsuit sought $319 million because of violations at the company's Carson refinery between 1994 and 2002, the claim being based on the maximum penalty for each of thousands of violations. Within a week, BP/Arco settled for a total of $81 million ($6 million in past fees, $25 million in cash penalties, $30 million in community programs for asthma diagnosis and treatment, and $20 million in emissions-reduction measures). This was by far the largest settlement ever negotiated by the AQMD, the country's largest and most effective air-pollution agency. Although the Carson refinery is a stationary source and not located next to a freeway, it is adjacent to a major arterial (Sepulveda Boulevard). Moreover, although Carson is one of the most racially integrated cities in the United States, most of the residents near the refinery are minorities. Thus, although this paper is focused on mobile sources, we do not want to give the impression that stationary-source pollution is not a problem. Indeed, when stationary polluting sources are located near freeways or arterials, nearby residents are in double jeopardy (Bacerra, 2005). Disadvantaged populations in freeway air-pollution sheds 155 Table 1. Criteria pollutants?National Ambient Air Quality Standards. PM10 and PM2:5 refer to particulate matter of less than 10 mm and 2:5 mm, respectively. Clean Air Act? criteria pollutants 1 hour average Annual arithmetic mean 8 hour average 24 hour average concentration Carbon monoxide PM10 PM2:5 Ozone Sulfur dioxide Nitrogen dioxide Lead 35 ppm na na 0.12 ppm na na na na 50 mg m?3 15 mg m?3 na 0.03 ppm 0.053 ppm na 9 ppm na na 0.08 ppm na na na na 150 mg m?3 65 mg m?3 na 0.14 ppm na 1:5 mg m?3 Notes: ppm?parts per million; mg m?3 ?microgram per cubic meter; na?not applicable. Clean Air Acts (see table 1). Emissions from the transportation sector (in 2002, the latest year for which data are available) accounted for 77.3% of carbon monoxide (CO); 54.3% of nitrogen oxides (NOx ); 44.8% of volatile organic compounds (VOCs), 12% of lead (Pb) releases, 6.4% of particulate matter of less than 2.5 mm (PM2:5 ). 2.3% of particulate matter of less than 10 mm (PM10 ), and 4.5% of sulfur oxide emissions (SO2 ) (US Environmental Protection Agency, 2005). In addition, there are hundreds of elements and compounds emitted from the on-road and off-road vehicles. With support of recent findings from environmental-health sciences and epidemiology, the US Environmental Protection Agency recently identified a list of twenty-one mobile source air toxics. This listing is ``to capture the collection of emissions potentially responsible for the cancer and noncancer health effects related to diesel exhaust'' (US Environmental Protection Agency, 2001, page 8). Although the six criteria pollution levels have been reduced under the Clean Air Acts since the 1970s (Bae, 2004), researchers are finding more health risks related to, especially combustion-related, air toxics and ultrafine particles. For example, the Puget Sound Clean Air Agency reports that diesel soot is responsible for 70 ^ 80% of air-toxic-related cancer risks. The average air-toxic-related cancer risks are estimated to be in the 400 ^ 700 per million range in the Seattle region (Puget Sound Clean Air Agency, 2003, pages ES-4, ES-5). Moreover, traffic is also a source of noise and vibration with the greatest impact on those living within 500 ft of major roads (US Department of Transportation, 1997). Although criteria pollutants and urban air toxics may also have microscale impacts, unlike noise, which is commonly measured and mitigated, mobile-source microscale `hot spots'?for example, the roadside, freeways, rail roads, and bus depots ?are rarely monitored in the United States because the Clean Air Act only requires ambient airquality monitoring. These new findings are rarely reflected in local land-use planning decisionmaking processes (California Environmental Protection Agency and California Air Resource Board, 2005). As the urban-planning profession pays more attention to efforts to reduce urban sprawl via compact-city policies (for example, urban growth boundaries, smart growth), there is the threat of human-health costs to those living near freeways. In this paper we investigate land-use and population patterns near major roads in the Seattle and Portland metropolitan areas. Low-income and minority populations may be more at risk of exposure to mobile-source pollutants because of the search for affordable housing and because of land-use and transportation planning practices. Although the amount of pollution released from mobile sources such as trucks and automobiles is dependent on such factors as traffic volume, fleet composition, fuel type, control technology, and vehicle speed, the potential for population exposure to these pollutants is also dependent on area topography, meteorological conditions, and 156 C-H C Bae, G Sandlin, A Bassok, S Kim Table 2. Adverse health effects of mobile-source pollutants [modified from table 4.3 of Bates and Caton (2002)]. Pollutant Definitive effect Probable effect Fine particles aggravation of asthma (English et al, 1999) depressed lung function in school children (Wjst et al, 1993) increased risk of lung cancer increased prevalence of bronchitis aggravation of acute respiratory symptoms (van Vliet et al, 1997) increased risk of wheezy bronchitis in infants (Gehring et al, 2002) decreased rate of lung growth in children Ozone increased hospital admissions for acute respiratory diseases aggravation of asthma increased bronchial responsiveness reduced lung function increased school absences for respiratory illness effect on mortality Diesel/VOCs a (in addition to particle effects) increased response to allergens (Janssen et al, 2003) increased airway inflammation increased risk of lung cancer increased risk of childhood leukemia (Crosignani et al, 2004) Nitrogen dioxide increased respiratory morbidity (Nitta et al, 1993) aggravation of asthma in children (Delfino et al, 2003) reduced rate of lung growth Carbon monoxide increased cardiac ischaemia a VOCs?volatile increased hospital cardiac admissions organic compounds. pollutant dispersion patterns, in addition to demographic and human-activity factors. Despite the complexity of these variables, there is a growing body of environmentalhealth and epidemiological evidence that populations living near limited-access freeways or high-traffic-density arterials may be at increased risk of exposure to mobile-source pollutants, which result in adverse health effects, such a lung cancer, leukemia, and asthma (see table 2 for details).(2) 2 Literature review Proximity to high traffic densities can be considered a surrogate for individual exposure, although researchers have used a variety of methodologies to measure proximity and traffic volume and/or densities. Huang and Batterman (2000) examined forty-five epidemiological studies that used residential location as a measure of environmental exposure, with distance from the pollution source to the receptor (school, residence) as the most common methodology for determining exposure. Several studies have used circular buffer zones around the receptor, assuming equal pollution dispersion, and others have used a combination of receptor zones and proximity to major roads. A Tokyo study determined that, on the basis of the dispersion gradient of NOx, traffic emissions could have adverse health effects on populations residing within 150 m of major roadways (Nitta et al, 1993). The validity of distance from major roadways as a measure of exposure to air pollution from traffic was investigated by a Dutch team (2) Workers at sites close to freeways and major roads are also at risk, but these are not specifically included in our study. Disadvantaged populations in freeway air-pollution sheds 157 who measured outdoor levels of PM10, PM2:5, black smoke, benzene, and NO2 at different distances from the roadway (Roorda-Knape et al, 1998). This study concluded that traffic intensity, distance, and wind direction are all important variables when considering population exposure to mobile-source pollutants. Wind direction was also important in an Australian study (Hitchins et al, 2000). Another methodology adopted by researchers investigating an association of childhood asthma with proximity to roadway pollution was based on determining traffic density within residential buffer zones (Lin et al, 2002). In this study, traffic density was calculated by multiplying the length of road segment within the buffer by the annual average daily traffic count of the specific roadway. Traffic density was then categorized into low, medium, or high, with the latter specified as 5 4043 vehicle miles traveled (VMT) in a 200 m buffer or 5 18 765 VMT in a 500 m buffer. The study concluded that there was a correlation between high volumes of traffic or a high percentage of trucks within a 200 m residential buffer for children hospitalized with asthmatic episodes. Zhu et al monitored ultrafine particles (that is, less than 0.1 mm in diameter) at various distances from a nine-lane freeway in Southern California in order to determine the dispersion gradient of these particles (see figure 2 of http:// www.ph.ucla.edu/magazine/PHmgnov02research.pdf and Zhu et al, 2002 for details). The researchers concluded that people who live, work, or travel within 100 m downwind of major traffic sources may have a much higher ultrafine-particle exposure than those who live further from such sources. Of particular interest in epidemiological research is the effect of mobile-source pollutants on sensitive populations such as children. Wjst et al (1993) reported that with each increase of 25 000 cars on a main road near schools the lung function of children decreased by 0.71%. This work was soon followed by various research initiatives, especially led by Brunekreef and his colleagues in the Netherlands. Their findings indicated that children living or attending schools within 100 m (330 ft) of truck traffic had poorer lung function, leaving the researchers to hypothesize that this may be the result of long-term exposure to ultrafine particles (Brunekreef et al, 1997). The Brunekreef research team also conducted a detailed study of respiratory symptoms. After accounting for confounding factors of socioeconomic status and related lifestyle considerations (for example, smoking in the home, unvented gas-fired heaters) their results suggested that an association between traffic-related air pollution and respiratory health was mainly restricted to the children of intermediate and low socioeconomic status (van Vliet, 1997). Such findings imply that a demographic analysis of populations living near major arterials could reveal that minority or low-income populations are disproportionately impacted by mobile pollution sources. Several US-based studies were also concerned with the environmental-justice implications of exposure to mobile-source pollutants; that is, are low-income or minority populations more at risk? (Delfino et al 2003; English et al, 1999; Forkenbrock and Schweitzer, 1999; Green et al, 2004; Gunier et al, 2003; Kinney et al, 2000; Korenstein and Piazza, 2002; Lena et al, 2002; Loh and Sugerman-Brozan, 2002; Samet et al, 2001; Wilhelm and Ritz, 2003.) The Forkenbrock and Schweitzer study adopted a similar procedure to the one used in this research, a geographic information systems (GIS) approach at the census-block level, and developed a pollutant-dispersion model to measure the impact of highway projects on low-income and/or minority populations in the City of Waterloo, Iowa. However, their research is more of a demonstration of an air-pollution and noise-pollution modeling methodology to a very small microlocation than a metropolitan-level analysis. A more traditional epidemiological approach was conducted by English et al (1999), also using GIS, to determine whether living near busy roads may be associated with asthma among children of low-income populations 158 C-H C Bae, G Sandlin, A Bassok, S Kim in San Diego County, California. This study concluded, on the basis of traffic counts within a 550 ft ($ 165 m) buffer of children's homes, that living near high-volume traffic was a contributing rather than a causal factor in asthma development. The buffer size was selected on the basis of an examination of several air-emission dispersion models that indicated a 80 ^ 90% decay of pollutants between 492 ft and 656 ft. It is also important to note that the earlier Dutch study by Roorda-Knape et al (1998) suggests that traffic pollutants in or near schools are a more relevant measure of exposure because children spend most of the daytime at school during periods of high traffic flow. The California-based research of Korenstein and Piazza (2002), Green et al (2004), and Gunier et al (2003) further investigated the proximity of schools to major roads. Korenstein's team developed dispersion-modeling estimates of PM10 concentrations at four predominantly Hispanic urban schools, three of which were 150 m ($ 500 ft) from major roads with up to 250 000 vehicles per day (VPD). Although the results indicated that predictive concentrations were much lower than regulatory levels (either federal or California standards), the authors stated that below-regulation concentrations have still been shown to cause significant negative health effects. Green et al (2004) studied the demographics of California schools (173 schools with more than 105 000 students) located within a 150 m ($ 500 ft) high traffic buffer and found that, as the traffic-exposure category increased, then the percentages of Hispanic and of non-Hispanic black children attending schools in those categories increased and the percentage of non-Hispanic white children decreased. The study also found that poverty was related to traffic exposure, which supports the earlier findings of Gunier et al, in which the researchers concluded that Hispanic, African-American, and Asian children in the lowest income quartile were on average three to five times more likely than children in the highest income quartile to live in block groups with high traffic densities (Gunier et al, 2003). Community-based studies of the Bronx, New York (Kinney et al, 2000), and of Roxbury, Massachusetts (Loh and Sugerman-Brozan, 2002), found similar results. A survey of 1109 parents in East Bay, San Francisco, that inquired about their children, found a strong association between asthma symptoms and NOx and PM levels (Kim et al, 2004). Currently in the United States there are few policy initiatives beyond the regulatory requirements of the Clean Air Act to address the adverse health impacts from exposure to mobile sources. In California, Senate Bill 352 prohibits the siting of new schools within 500 ft of a busy road, defined as traffic in excess of 50 000 VPD in a rural area and 100 000 VPD in an urban area (California Department of Education, 2004). Two recent court cases also had the potential to influence public policy. The first involved a lawsuit filed by the Sierra Club and other litigants regarding the need for a supplemental environmental impact statement to address the potential impacts of a highway expansion in Las Vegas, Nevada, on the health of nearby residents. Considered a potentially precedent-setting case, the US District Court for the District of Nevada recently found that the Federal Highway Administration met its requirements in issuing an environmental-impact statement and denied all seven counts of the Sierra Club's summary judgment motion [Sierra Club v Mineta D.Nev., (No. CV-S-02-0578PMP-RJJ)]. In 2002 several environmental and labor groups filed suit against the US Department of Transportation with respect to the need for considering the localized adverse impacts of Mexican truck traffic. In 2003 the US 9th Circuit Court of Appeals ruled that the National Environmental Protection Act required the US Department of Transportation ``to consider the most likely localities to be affected by increased truck traffic and to perform more localized analyses for these areas'' and that simply placing the potential pollution increases in the context of US national emissions was inadequate [Public Citizen v Dept of Transportation (No. 02-70986 9d Cir. 2003)]. This ruling Disadvantaged populations in freeway air-pollution sheds 159 suggests that regional air-quality monitoring does not provide adequate data sources for microscale environments, such as populations living near major roads. 3 Data and GIS analysis In our research we examine land-use patterns near major roads to determine the number of residences impacted and to identify the populations at risk of exposure to mobile-source air pollution. We also examine the demographics of populations that live near limited-access freeways in the Seattle and Portland metropolitan areas, in order to determine if low-income, minority populations are at increased risk of exposure to mobile sources. In addition, we investigate the prevalence of schools and senior facilities located within the areas that we describe as a freeway air-pollution shed (FAPS), specifically a 100 m (330 ft) buffer from roadways with a minimum of 100 000 VPD, a definition that encapsulates the results of the research investigations. The 330 ft buffer is supported by several research studies, both in the United States and in other countries (especially Zhu et al, 2002 and Brunekreef et al, 1997), although, as pointed out above, California's statutory limit (as an example) is 500 ft. The rationale for the buffer of this extent is the distance decay rate for ultrafine particles from freeways (and arterial roads). Our study is an attempt to bridge the inderdisciplinary knowledge required for land-use, transportation, and air-quality planning to identify whether there is a disproportionate impact on minority and low-income populations. Our study areas include the urban-growth boundary regions of Seattle, WA, and Portland, OR. In Washington State this area was defined as approximately 310 000 acres of western King County. The Portland study area was defined by the three Oregon Counties of Clackamas, Washington, and Multnomah, totaling approximately 232 000 acres. From the Census 2000 Summary File 3, we compiled demographic, social, economic, and housing data at the census-block group level, the smallest geographic unit for the required socioeconomic data, for these areas. Parcel data were combined with tax-assessor records to develop a database of single and multifamily residences. The multifamily-residence data include the number of dwelling units in each building, which was used to assess the number of housing units in each block group. An exhaustive list of schools (K ^ 12) was collected from both King County and the Portland metropolitan area. The locational information was combined with specific data from each respective State's Department of Education database to show enrollment, demographics, and the number of students on free or reduced lunch programs. We obtained average annual daily traffic (AADT) volume data for 2000 for all limited-access arterials in western King, Clackamas, Multnomah, and Washington counties. High-traffic roads were identified as having more than 100 000 vehicles per day. Specifically, these were Interstates 5, 405, and 90; state road 520 in western King County; and Interstates 5, 84, 205, and 405 in the Portland urban growth area (UGA). Schools, parcels, census-black groups, and traffic files were compiled within a GIS (ArcView). Parallel lines were then created within 400 ft of the selected road network. This buffer accounted for both the 330 ft conservative dispersion estimate of mobile-source pollutants as well as an additional area of the roadway (ArcView measures distance from the middle of the roadways and not their edge). These areas, designated as FAPS, were analyzed for socioeconomic demographics, single versus multifamily residences, and schools. Although the census-block group is larger than the FAPS, we had little alternative than to ascribe to socioeconomic characteristics of the block group population to its FAPS subset. A proportion was created by calculating the number of residential units 160 C-H C Bae, G Sandlin, A Bassok, S Kim in the FAPS and those in the block group, and by using this as a ratio to estimate the demographic characteristics of the FAPS. 4 Descriptive results An interesting comparison emerges when we look at the Seattle and Portland FAPS. Although the Portland UGA is smaller than the King County's UGA discussed in this paper, in terms both of population and of area, it is slightly denser (5.7 and 4.8 persons per acre, respectively). Despite the higher density, the proportion of persons living within Portland's FAPS is considerably smaller than that living in King County's, 0.42% versus 1.81%. A striking finding is that single-family home development since 1990 was five times higher than in the 1980s in the Seattle FAPS. In both areas, poor and/or African-American residents are represented in disproportionately higher numbers in the FAPS. The number of poor living in the Seattle and Portland FAPS is more than 1.21 and 1.36 times higher, respectively, than that in the UGAs at large, and the concentration of African-Americans is higher than that of other minority groups (Asian-Americans and Hispanics), and 2 ^ 3 times higher than that of the general UGA population. In summary, there is a strong tendency for low-income and/or nonwhite populations to live in FAPS (see tables 3 and 4). Although the concentration of children in the core areas of Seattle and Portland is not particularly high, it is worthwhile to look at the number of schools and students in both FAPS. This illuminates the potential harm to young populations. Of the sixteen educational facilities within the Portland and Seattle FAPS, there are ten elementary schools, two middle schools, three high schools and an administration building. Total student enrollment is more than 6600. All these schools have a high proportion of minority students and students from low-income household (as reflected in the receipt of free or subsidized meal programs), except for the two private schools in suburban Bellevue.(3) Seven of the FAPS schools have a nonwhite majority (between 52% and 95%), and more than 50% of students in five schools received free or subsidized meals. 5 Cluster analysis Focusing on Seattle as our first and local study area, we applied a cluster analysis to assign all the 2000 Census-block groups in the Puget Sound region (2750 of them) to see whether there is a spatial residential association with income level and race. Though each block group has its own characteristics, it is useful to classify the census-block groups into several clusters that have similar characteristics; these may or may not be spatially contiguous. Cluster analysis partitions data into meaningful subgroups. It is particularly useful when the number of subgroups and other composite information are unknown a priori (Fraley and Raftery, 1998). The result of cluster analysis is a number of heterogeneous groups with a more or less homogenous content. A major task is identifying the optimum number of clusters. There is no generally accepted theory, so decisions are based on subjective interpretation. Fraley and Raftery (1998; 1999; 2002) developed a model-based clustering analysis technique using an expectation maximization (EM) algorithm. Traditional cluster analysis often has problems in determining the structure of clustered data. However, the EM approach significantly alleviates these problems. (3) Whereas FAPS schools in low-income communities tend to have minimal or no buffers next to the freeway, these two private schools are surrounded by dense trees. This hints that there may be some awareness among educators and parents in wealthy communities about the problems associated with proximity to freeways. On the other hand, it could be a signal of less-than-perfect information even among the wealthy and highly educated, because trees are a much better noise buffer than a pollution filter. Disadvantaged populations in freeway air-pollution sheds 161 Table 3. Socioeconomic characteristics of Seattle's urban growth area (UGA) and freeway air-pollution sheds (FAPS). Source: US Census of Population (2000). Variables UGA Percentage FAPS Percentage Area (acres) Population White Black Hispanic Asian Children Seniors Income below poverty Schools a 310 380 1 491 633 1 060 551 87 241 86 956 176 363 283 847 165 144 132 950 100 71.10 5.85 5.83 11.82 19.03 11.07 8.91 17 073 26 977 17 813 2 947 1 483 3 462 3 864 2 789 2 912 100 66.03 10.92 5.50 12.83 14.32 10.34 10.79 a Source: 427 8 Percentage of UGA FAPS/UGA share ratio 5.50 1.81 1.68 3.38 1.71 1.96 1.36 1.69 2.19 0.93 1.87 0.94 1.09 0.75 0.93 1.21 1.87 Washington State Department of Education. Table 4. Socioeconomic characteristics of Portland's urban growth boundary (UGB) and freeway air-pollution sheds (FAPS). Source: US Census of Population (2000). Variables UGB Percentage FAPS Percentage Area (acres) Population White Black Hispanic Asian Children Seniors Income below poverty Schools a 232 380 1 328 195 1 047 027 40 996 110 492 74 452 289 096 138 519 128 319 100 79 3 8 6 22 10 10 6 415 5 633 3 891 532 469 389 1 076 630 741 100 69 9 8 7 19 11 13 a Source: 366 8 Percentage of UGB FAPS/UGB share ratio 2.76 0.42 0.37 1.30 0.42 0.52 0.37 0.45 0.58 0.88 3.06 1.00 1.23 0.88 1.07 1.36 2.19 Oregon State Department of Education (2004). It assumes that the data are generated by a mix of underlying probability distributions in which each component represents a different cluster or group. The EM clustering algorithm computes probabilities of cluster memberships on the basis of one or more probability distributions instead of assigning cases or observations to clusters to maximize the differences. The clustering algorithm then chooses the number of clusters that maximize the overall probability or likelihood of the data using the Bayesian information criterion. In our study, this approach yielded seven clusters based on income and race. As suggested by the data in table 5, clusters 1 and 2 are significantly different from the others, especially in terms of racial mix and income. Cluster 1 is characterized by the only minority-group majority share (75%) and highest percentage of children and the elderly (33%). The general location of this cluster is south of downtown Seattle and west of Lake Washington and the Port of Tacoma. The next cluster (cluster 2) has a much lower income ($34 634) level than that of clusters 3 ^ 7 (which range approximately from $50 000 to $90 000) although its white-population share is not much lower than that of cluster 3. Another group near the freeway is characterized by high income 162 C-H C Bae, G Sandlin, A Bassok, S Kim Table 5. Cluster analysis of 2000 Census-block groups, Central Puget Sound, WA. Source: US Census of Population (2000). Cluster Percentage white population Median household income ($) Median housing value ($) Percentage children and the elderly of total population a 1 2 3 4 5 6 7 35 69 88 73 93 87 81 38 696 34 634 53 012 57 986 64 458 71 355 93 342 161 445 151 566 155 957 191 837 241 644 339 953 563 392 33 28 31 30 30 27 32 a Children defined as being less than 18 years of age, the elderly defined as those over 65 years of age. and is white dominant (cluster 6). The contrast with clusters 1 and 2 is striking, but the explanation is proximity to Lake Washington and mountain views, Microsoft, and somewhat lower traffic volumes. Figure 1 shows the spatial distribution of clusters in the Puget Sound region, and confirms our a priori expectation of lowincome and minority households living close to freeways (especially the Interstate 5 corridor) and major arterials. 6 Hedonic price analysis If there is residential market segmentation, how do residents value living near freeways? Do residents capitalize the negative externalities into lower house prices? To test this, we undertook a hedonic price analysis of house-price sales data (in 2000) in King County, WA, the core county of the Seattle metropolitan region. King County property sales records, parcel data, census-block-group-level information, and regional travel and employment data from the Puget Sound Regional Council (the regional Metropolitan Planning Organization) were used. Only single-family houses were included in the analysis. Housing sales transactions may not occur at random among the stock of housing units. Housing units that were sold in year 2000 may differ in measured and unmeasured characteristics from the houses not sold. However, we analyzed only the houses sold. Therefore, there is a potential selection bias problem. The unmeasured characteristics may lead to biased estimates of the parameters; to avoid this, we used a popular selection bias detection and correction method [the Heckman two-step estimation procedure (Heckman, 1979)]. The Heckman procedure consists of two stages. In the first stage a probit model is estimated to predict the houses sold in year 2000 with the data that contain all houses in the study area. The residuals of the probit model reveal information about the unmeasured characteristics that distinguish sold housing from unsold housing. The inverse Mill's ratio of selection-bias control factor called lambda is calculated in the process through the use of normally distributed residuals. In the second stage a regression analysis is performed with the selection-bias control factor lambda as an additional covariate. This factor reflects the effect of all the unmeasured characteristics related to the sales transaction choice, and the coefficient captures the partial effect of these unmeasured characteristics. The process produces unbiased estimates for the covariates in the hedonic pricing model. Additional steps to correct the standard errors Disadvantaged populations in freeway air-pollution sheds 163 Freeway and highway Water Cluster type White population percentage and median houeshold income ($) 35.3 and 38 696 69.4 and 34 634 72.6 and 57 986 81.4 and 93 342 87.7 and 53 012 87.7 and 71355 0 1.5 3 6 9 12 miles 93.2 and 64 458 Figure 1. The spatial distribution of residential clusters in Central Puget Sound, WA. were also performed (for detailed information for the application of the Heckman procedure, see Smits, 2003). Also, in the hedonic pricing model spatial autocorrelation is an important issue because it results in biased parameter estimates. We included census-block-grouplevel neighborhood information to capture the effects of neighborhood. However, spatial autocorrelation is largely attributed to omitted variables, and a degree of spatial autocorrelation may be unavoidable. Using CrimeStat, a spatial-analysis software program, we conducted a spatial-autocorrelation test. The results show that the Moran's 164 C-H C Bae, G Sandlin, A Bassok, S Kim I value and z-statistic value of normality for the residuals of the predicted dependent variable (log of sales price) were 0.0067 and 21.24, respectively. The Moran's I value indicates that there is positive and statistically significant spatial autocorrelation. However, the Moran's I value and z-statistic value for the observed dependent variable were 0.1253 and 383.02, respectively, suggesting a significant reduction in spatial autocorrelation. In table 6 the results of the hedonic regression model are adjusted for selection bias and the correction of standard errors. A significant negative parameter estimate of lambda implies that the houses sold in 2000 compared with the houses not sold have unmeasured characteristics negatively related to sales price. The dependent variable, housing sales price, was log-transformed because the price does not have negative values and is not bounded, and its distribution is skewed to the left. Only the statistically significant independent variables at the 0.05 level were retained in the model. Table 6 also includes a variance inflation factor (VIF) to examine multicollinearity among the independent variables. The VIF values indicate that the independent variables in the models are not collinear. Table 6. Results of the hedonic pricing model for King County, Seattle metropolitan area. Source: King County Department of Assessment (2000). Coefficient Housing characteristics Lot size (?1000 ft2 ) Total living area (?1000 ft2 ) Year built Construction quality Housing renovated Housing condition good Housing condition very good Parcel characteristics View Waterfront Tideland/shoreland Traffic noise moderate Traffic noise high Other nuisances Stream Within FAPS c Pr > jtj VIF b Elasticity 11.2008 Intercept SE a 0.5234 < 0.0001 ? ? 0.0017 0.1660 0.0006 0.1260 0.1177 0.0498 0.1512 0.0001 0.0090 0.0002 0.0063 0.0257 0.0112 0.0213 < 0.0001 < 0.0001 0.0070 < 0.0001 < 0.0001 < 0.0001 < 0.0001 1.1165 3.1739 2.9734 3.7865 1.1330 1.3385 1.2321 0.17 16.62 0.06 12.52 11.05 4.51 15.09 0.0677 0.2868 0.3330 ?0.0399 ?0.0457 ?0.0804 0.1818 ?0.0780 0.0205 0.0428 0.0663 0.0186 0.0159 0.0243 0.0414 0.0155 0.0010 < 0.0001 < 0.0001 0.0326 0.0042 0.0010 < 0.0001 < 0.0001 1.2632 1.7782 1.7496 1.0540 1.1557 1.0733 1.0911 1.1912 5.91 30.40 34.97 ?4.80 ?5.22 ?8.84 17.48 ?8.21 0.0004 0.0000 0.0136 0.0003 0.0007 0.0009 0.0077 < 0.0001 < 0.0001 < 0.0001 < 0.0001 < 0.0001 < 0.0001 < 0.0001 1.7844 2.1403 2.6643 2.8850 2.3376 1.4874 2.6194 0.26 0.13 ?10.56 0.26 ?0.38 ?0.53 8.23 0.0841 < 0.0001 2.6194 Neighborhood (census-block group) characteristics 0.0026 Population density (?100 000 ft2 ) Median housing value (?$1000) 0.0013 Average household size ?0.1044 Percent homeowners 0.0026 Percent commuting by automobile ?0.0038 Percent black population ?0.0053 Retail service accessibility 0.0823 Lambda 2 Adjusted R ? 0:7271 n ? 5237 a Standard error. inflation factor. c Freeway air-pollution shed. b Variance ?0.8205 ? Disadvantaged populations in freeway air-pollution sheds 165 The elasticity coefficient shows the proportional change in the dependent variable associated with a proportional change in independent variable. However, in this case we calculated the proportional change in sales price for a one unit change in the independent variable in order to permit an easy interpretation of the effects of explanatory variables on the response variable. For instance, a 1000 ft2 increase in total living space results in a 16.62% change in the predicted sales price of single-family housing, holding all other things constant. The elasticities of the binary variables were also calculated. The elasticity of a dummy independent variable indicates the proportional change in the dependent variable associated with a binary change in the independent variable. For example, the key to this analysis, the sales price of single-family housing inside FAPS is, ceteris paribus, 8.21%, 100fexp [ b ? (var b=2)] ? 1g, lower than housing outside FAPS. This is larger than the characteristic most closely associated with locations near freeways?that is, traffic noise (4.8 ^ 5.2% comparing moderate to high noise levels). The inference is that, even with imperfect information, pollution is more important than noise. 7 Qualifications There are three major qualifications to this research. First, our focus on the long-term health impacts of air pollution (especially PM) should be tempered by the increasing recognition that there may be health damages associated with even short-term exposure to air pollution. This has been noticed by several research studies in different locations with different air pollutants. Examples include the exposure of children to diesel-related pollutants during school-bus commutes in Los Angeles (Winer et al, 2005), the ozone exposure of schoolchildren, especially during school breaks (Peters et al, 1999), the exposure to fine PM of home dwellers (both indoor and outdoor) and workers in Helsinki (Jantunen et al, 2004), and the CO exposure of street sellers in Mexico City (Fernandez-Bremauntz et al, 1993). The second qualification is related to the fact that there is no widespread roadside monitoring in the United States, as opposed to many European countries, so that our research is based on inferences drawn from ambient air-quality levels and from the small number of sample surveys that have been undertaken. Furthermore, the officially sponsored studies in the United Kingdom reveal ambiguous conclusions about whether PM emissions close to roads exceed even long-term standard objectives. For example, data for 1999 or 2000 in fifteen cities exceeded the 2004 standard at only five of the thirty-four sites, whereas in London the standard was exceeded at only two (Camden and Marylebone Road) out of eight sites (Air Quality Consultants Ltd, 2002). Nevertheless, in a more recent official study (DEFRA, 2005), kerbside levels of nanoparticle emissions were much higher at the Marylebone Road site than background levels at other sites, both in London and elsewhere. In another study, in Winchester town centre, emissions failed to achieve the 24-hour mean standard of 50 mg m?3 (as measured by thirty-five failures per year or more) in five years 1997 ^ 2004, and last passed in 2001 (City of Winchester, 2005). A major problem with the UK monitoring hitherto is that it mainly tests for PM10 emissions; for example, there are only three PM2:5 monitoring sites in London compared with thirty-five PM10 sites (Air Quality Consultants Ltd, 2003). It is clear now that the dangers from roadside PM are much more serious for PM2:5 and ultrafine particles. A third and obvious qualification is that 90% of PM emissions are not transport related, so that a comprehensive assessment of their health impacts needs to look at other sources. This is a complex question, however, because there is not a one-to-one correspondence between emission sources and health impacts. For example, particulate emissions are particularly high in agricultural regions and desert areas, both of which are relatively sparsely populated. 166 C-H C Bae, G Sandlin, A Bassok, S Kim 8 Policy and planning implications As we have reported, there is an increasing mound of evidence that freeways, major arterials, and other traffic density zones inflict significant health damage on those who live and work nearby. Also, recent research shows that the most dangerous pollutants are ultrafine particles and diesel-related VOCs, whereas the jury is still out on the long-term health impacts of ground-level ozone, in part because of the possibility that, within certain ranges, NOx may have protective effects against ground level ozone exposure. A problem is that scientists disagree about the dollar value of these health costs, although all of them agree that these costs are very high. There is little effort to mitigate these freeway-related pollution impacts, although a start is being made in California (California Environmental Protection Agency and California Air Resource Board, 2005). However, without the advantage of a relatively precise cost ^ benefit analysis, it is difficult to prescribe the most appropriate policy and planning responses. Because the scientific community is only now beginning to appreciate the extent of the dangers, it is unreasonable to expect that people living or working near freeways would have the same degree of information. They are probably well aware of the noise and somewhat aware of the higher pollution levels. In making their location decision, residents trade-off lower rents and house prices against these disadvantages, but with limited information. So, a `first, do no harm' principle would be to give the information as it becomes available to residents to that they can decide (at lease end or some other stage) whether to move or stay. Another issue is that the conclusions about air-pollution emissions near freeways are still largely based on infrequent surveys or inferences from other locations. A key priority is to set up roadside air-pollution monitors at regular intervals along freeway routes. This is not a zero-cost option, but the costs are relatively low compared with the benefits from increased knowledge that can empower the will to act. Assuming that the knowledge of more damage will be built up over time, what else (if anything) should be done about it? One obvious option for planners is zoning remedies. These could range in scope from land-use restrictions on new housing (planners have allowed 4387 housing units to be built in the Seattle FAPS since 1990, one fifth of the housing built since 1900), schools, daycare centers, seniorcitizen centers, and other facilities, to the compulsory relocation of existing land uses (with or without compensation). An intermediate position is to allow commercial and industrial facilities, although worker exposure is also a problem (even if mitigated by reduced hours of exposure relative to residents, lower employment densities than residential densities and up-to-date filtration systems in modern office buildings). A different route is via the strengthening of the controls on the sources ? that is, trucks, cars, and other vehicles. However, even assuming the political will exists, these policy changes would take a long time (more than twenty years) to implement. Limitations on new-truck diesel emissions do not begin until 2007, and the truck fleet is very durable (with an average lifetime of up to thirty years). Restrictions on truck routes, and perhaps even times of travel, may merit inquiry, but many metropolitan areas (including Seattle and Portland) have a too-limited highway infrastructure to permit much route diversion. There are some modest technical fixes: ultra-sulfur and high-performance diesel fuel and financial incentives to retrofit particle traps, and retrofitting to use alternative fuels. New automotive technologies (from hybrids to fuel cells) are beginning to show promise, but rates of adoption and fleet turnovers are very slow. Disadvantaged populations in freeway air-pollution sheds 167 Yet another approach is via better health monitoring of and targetted healthcare for residents and workers within FAPS. This would help, but the sharply rising costs of healthcare are a major obstacle, given the evidence of the numbers with impaired health. Also, in any event, prevention is much better than a cure. Pragmatically, we may have to be content with more modest strategies: (1) more research and education on air quality (Stone, 2003), finding ways of making new knowledge accessible so that locators can make more informed decisions. (2) roadside monitoring of emissions; and (3) restrictions on new land uses within FAPS. The latter point brings up an important concern. Infill of all vacant sites and the redevelopment of obsolete land uses at higher residential densities are very consistent with the densification objectives embedded in the Growth Management Acts of Washington and Oregon. However, many of these sites are close to freeways, and developing or redeveloping them exposes people to significant pollution-related health damage. Land-use planning and regulations are currently ill prepared to mitigate these impacts. 9 Conclusions This research tested three related hypotheses: (1) minority and/or low-income households live disproportionately close to freeways compared with white and middle-income households; (2) households in each category cluster together in local subhousing markets; and (3) negative environmental externalities near freeways (especially air pollution) are capitalized in house prices and rents. First, the results support all three hypotheses and their corollaries: the clustering of low-income and minority population near freeways, and the higher concentration of minority and/or poor students in FAPS. Health consequences for these children can be more harmful, because of the effects of pollution on their lung development. Second, the cluster analysis suggests that the residential choices of the minority and/or low-income population are limited. Third, locations within a FAPS are negatively associated with housing prices when other negative environmental factors such as traffic noise are accounted for. Of course, for people living in such locations, trade-offs may have to be made: cheaper housing versus higher health risks. Of course, there are some qualifications to our research. Our hedonic price analysis is based on house prices, whereas many residents within FAPS are renters. Also, ideally, we would like a finer grain of detail for our socioeconomic data than the census-block groups. The most significant point of all is that, though Zhu et al (2002) have provided us with a useful grounding for deriving the FAPS boundaries, their study was based on one metropolitan area and hence cannot take account of location-specific variations (although the boundaries appear to receive some support from earlier European studies). At the practical level, in a perfect world we would need a series of air-pollution monitoring devices along each freeway in all metropolitan areas, given that this is a health issue of some importance. The United States could learn from the experiences of the United Kingdom and European Union with monitoring roadside air pollution. We also need a strategy to mitigate the costs of this problem (possibly, but not necessarily, to the extent of compulsory relocation of homes and workplaces within FAPS); whether this would justify relocation assistance is a question for policymakers. Only when problems of this kind are seriously addressed can we be assured that the parallel social problem of environmental justice is receiving the public policy attention that it demands and deserves. 168 C-H C Bae, G Sandlin, A Bassok, S Kim Acknowledgements. This research has been supported by the University of Washington Royalty Research Fund (65-5096). Preliminary results were presented at the 2004 Association of Colleges and Schools of Planning and the 2005 Western Regional Science Association Conferences. The authors appreciate valuable comments and encouragement from Professors Chris A Nelson (Virginia Tech University), Peter Flachsbart (University of Hawaii), Brian Stone (University of Wisconsin, and David Pitfield (Loughborough University, United Kingdom). Also, special thanks are due to Professors Jane Koening (Environmental Health) and Timothy Larson (Civil Engineering) at the University of Washington Particulate Matter Center. References 1000 Friends of Oregon, 1992 Making the Land Use, Transportation, Air Quality Connection (1000 Friends of Oregon, Portland, OR) Air Quality Consultants Ltd, 2002 Compilation of New Roadside Monitoring Data Obtained by Local Authorities as Part of the Review and Assessment Process http://www.defra.gov.uk/environment/ airquality/laqm/monitor/pdf/air roadside.monitor.pdf Air Quality Consultants Ltd, 2003 Review of Air Quality Monitoring in London Air Quality Consultants, 23 Coldharbour Road, Bristol BS6 7JT American Lung Association, 2003, ``Selected key studies on particulate matter and health: 1997 ^ 2001'', http://www.lungusa.org/air/air outdoor index.html Bacerra H, 2005, ``BP settles lawsuit for $81 million: in a record sum for AQMD, firm will pay for community health programs, upgrading Carson refinery'' Los Angeles Times 18 March, page B1 Bae C-H C, 1997, ``The equity impacts of Los Angeles' air quality policies'' Environment and Planning A 29 1563 ^ 1584 Bae C-H C, 2004,``Transportation and the environment'', in The Geography of Urban Transportation Eds S Hanson, G Giuliano (Guilford Press, New York) pp 356 ^ 379 Bates D, Caton R, 2002 A Citizen's Guide to Air Pollution David Suzuki Foundation, Suite 219, 221 West 4th Avenue, Vancouver V6K 4S2 Brainard J D, Jones A P, Bateman I, Lovett A, Fallon P, 2002, ``Modelling environmental equity: access to air quality in Birmingham, England'' Environment and Planning A 34 695 ^ 716 Brauer M, Hoek G, van Vliet P, Meliefste K, Fisher P H, Wijga A, Koopman L P, Niejens H J, Gerritsen J, Kerkhof M, Heinrich J, Bellander T, Brunekreef B, 2002, ``Air pollution from traffic and the development of respiratory infections and asthmatic and allergic symptoms in children'' American Journal of Respiratory and Critical Care Medicine 166 1092 ^ 1098 Brunekreef B, 1997, ``Air pollution and life expectancy: is there a relation? '' Occupational and Environmental Medicine 54 781 ^ 784 Brunekreef B, Janssen N A H, de Hartog J, Harssema H, Knape M, van Vliet P, 1997, ``Air pollution from truck traffic and lung function in children living near motorways'' Epidemiology 8 298 ^ 303 Bullard R D, Johnson G S, Eds, 1997 Just Transportation: Discriminating Race and Class Barriers to Mobility (New Society Publishers, Gabriola Island, BC) California Department of Education, 2004 Senate Bill 352 ? New Determinations and Findings Required for California Department of Education Site Approval Related to Air Emissions http://www.cde.ca.gov/ls/fa/sf/sb352.asp California Environmental Protection Agency, California Air Resource Board, 2005 Air Quality and Land Use Handbook: A Community Health Perspective http://www.arb.ca.gov.ch/handbook.pdf Chakravorty J, Schweitzer L, Forkenbrock D J, 1999, ``Using GIS to assess the environmental justice impacts of transportation system changes'' Transactions in GIS 3 329 ^ 358 Ciccone G, Forastiere F, Agabiti N, Biggeri A, Bisanti L, Chellini E, Corbo G, Dell'Orco V, Dalmasso P, Voltante T F, Galassi C, Piffer S, Renzoni E, Rusconi F, Sestini P, Viegi G, 1998, ``Road traffic and adverse respiratory effects in children'' Occupational and Environmental Medicine 55 771 ^ 778 City of Winchester, 2005 Air Quality 2004 http://www.winchester.gov.uk/Documents/Document/ air%20quality%202004%20PDF.pdf Crosignani P, Tittarelli A, Borgini A, Codazzi T, Rovelli A, Porro E, Contiero P, Bianchi N, Tagliabue G, Fissi R, Rossitto F, Berrino F, 2004, ``Childhood leukemia and road traffic: a population-based case-control study'' International Journal of Cancer 108 596 ^ 599 Davis L W, 2004, ``The effect of health risk on housing values: evidence from a cancer cluster'' The American Economic Review 94 1693 ^ 1704 DEFRA, 2005 Particulate Matter in the United Kingdom Department for Environment, Food and Rural Affairs, http://www.defra.gov.uk/environment/airquality/index.htm Disadvantaged populations in freeway air-pollution sheds 169 Delfino R J, Gong H Jr, Linn W S, Pellizzari E D, Hu Y, 2003, ``Asthma symptoms in Hispanic children and daily ambient exposures to toxic and criteria pollutants Environmental Health Perspectives 111 647 ^ 656 Dockery D W, Pope C A, Xu X, Spengler J D, Ware J H, Fay M E, Ferris B G, Speizer F E, 1993, ``An association between air pollution and mortality in six US cities'' New England Journal of Medicine 329 1753 ^ 1759 Duhme H, Weiland S K, Keil U, 1998, ``Epidemiological analyses of the relationship between environmental pollution and asthma'' Toxicology Letters 102 ^ 103 307 ^ 316 Edwards J, Walters S, Griffiths R K, 1994, ``Hospital admissions for asthma in preschool children: relationship to major roads in Birmingham, United Kingdom''Archives of Environmental Health 49 223 ^ 227 English P, Neutra R, Scalf R, Sullivan M, Waller L, Zhu L, 1999, ``Examining associations between childhood asthma and traffic flow using a geographic information system'' Environmental Health Perspectives 107 761 ^ 767 Fernandez-Bremauntz A A, Ashmore M R, Merritt J Q, 1993, ``A survey of street sellers'exposure to carbon monoxide in Mexico City'' Journal of Exposure Analysis and Environmental Epidemiology 3 Supplement 1, 23 ^ 35 Forkenbrock D J, Schweitzer L A, 1999, ``Environmental justice in transportation planning'' Journal of the American Planning Association 65 96 ^ 111 Fraley C, 1998, ``Algorithms for model-based Gaussian hierarchical clustering. SIAM '' Journal on Scientific Computing 20 270 ^ 281 Fraley C, Raftery A E, 1998, ``How many clusters? Which clustering method? Answers via model-based cluster analysis'' The Computer Journal 41 578 ^ 588 Fraley C, Raftery A E, 1999, ``MCLUST: software for model-based cluster analysis'' Journal of Classification 16 297 ^ 306 Fraley C, Raftery A E, 2002,``Model-based clustering, discriminant analysis and density estimation'' Journal of the American Statistical Association 97 611 ^ 631 Gehring U, Cyrys J, Sedlmeir G, Brunekreef B, Bellander T, Fisher P, Bauer C P, Reinhardt D, Wichmann H E, Heinrich J, 2002, ``Traffic-related pollution and respiratory health during the first two years of life'' European Journal of Respiratory Diseases 19 690 ^ 698 Gottlieb D J, Beiser A S, O'Connor G T, 1995, ``Poverty, race and medication use are correlates of asthma hospitalization rates: a small area analysis of Boston'' Chest 108 28 ^ 35 Green R, Smorodinsky S, Kim J J, McLaughin R, Ostro B, 2004, ``Proximity of California public schools to busy roads'' Environmental Health Perspectives 112 61 ^ 67 Gunier R B, Hertz A, von Behren J, Reynolds P, 2003, ``Traffic density in California: socioeconomic and ethnic differences among potentially exposed children'' Journal of Exposure Analysis and Environmental Epidemiology 13 240 ^ 246 Heckman J J, 1979, ``Sample selection bias as a specification error'' Econometrica 47 153 ^ 161 Hitchins J, Morawska L,Wolff R, Gilbert D, 2000,``Concentrations of submicrometre particles from vehicle emissions near a major road'' Atmospheric Environment 34 51 ^ 59 Huang Y, Batterman S, 2000,``Residence location as a measure of environmental exposure: a review of air pollution epidemiology studies'' Journal of Exposure Analysis and Environmental Epidemiology 10 66 ^ 85 Janssen N A H, Brunekreef B, van Vliet P, Aarts F, Meliefste K, Harssema H, Fischer P, 2003, ``The relationship between air pollution from heavy traffic and allergic sensitization, bronchial hyperresponsiveness, and respiratory symptoms in Dutch schoolchildren'' Environmental Health Perspectives 111 1512 ^ 1518 Jantunen M J, Koistinen K J, Edwards R D, Mathys P, Ruuskanen J, Kunzli N, 2004, ``Sources of fine particulate matter in personal exposures and residential indoor, residential outdoor and workplace microenvironments in the Helsinki phase of the EXPOLIS study'' Scandinavian Journal of Work, Environment and Health 30 Supplement 2, 36 ^ 46 Kim J J, Smorodinsky S, Lipsett M, Singer B C, Hodgson A T, Ostro B, 2004, ``Traffic-related air pollution near busy roads: the East Bay Children's respiratory health study'' American Journal of Respiratory and Critical Care Medicine 170 520 ^ 526 Kinney P L, Aggarwal M, Northridge M E, Janssen N A H, Shepard P, 2000, ``Airborne concentrations of PM2:5 and diesel exhaust particles on Harlem sidewalks: a community-based pilot study'' Environmental Health Perspectives 108 213 ^ 218 Korenstein S, Piazza B, 2002, ``An exposure assessment of PM10 from a major highway interchange: are children in nearby schools at risk? '' Journal of Environmental Health 65 9 ^ 19 170 C-H C Bae, G Sandlin, A Bassok, S Kim Lena T S, Ochieng V, Carter M, Holguin-Veras J, Kinney P L, 2002, ``Elemental carbon and PM2:5 levels in an urban community heavily impacted by truck traffic'' Environmental Health Perspectives 110 1009 ^ 1016 Lin S, Munsie J P, Hwang S A, Fitzgerald E, Cayo M R, 2002, ``Childhood asthma hospitalization and residential exposure to state route traffic'' Environmental Research Section A 88 73 ^ 81 Loh P, Sugerman-Brozan J, 2002, ``Environmental justice organizing for environmental health: case study on asthma and diesel exhaust in Roxbury, Massachusetts'' The Annals of the American Academy of Political and Social Science 584 110 ^ 124 Nitta H, Sato T, Satoshi N, Maeda K, Aoki S, Ono M, 1993, ``Respiratory health associated with exposure to automobile exhaust: I. Results of cross-sectional studies in 1979, 1982 and 1983'' Archives of Environmental Health 48 53 ^ 58 Oregon State Department of Education, 2004, ``Students eligible for free or reduced lunch'', http://www.ode.state.or.us/sfda/reportstudents.htm Peters J M, Avol E, Gauderman W J, Linn W S, Navidi W, 1999, ``A study of twelve Southern California communities with different levels and types of air pollution'' American Journal of Respiratory Critical Care Medicine 159 768 ^ 775 Puget Sound Clean Air Agency, 2003 Final Report: Puget Sound Air Toxics Puget Sound Clean Air Agency, 110 Union Street, Seattle, WA, http://www.pscleanair.org/news/other/psate final.pdf Roorda-Knape M C, Janssen N A H, de Hartog J J, van Vliet P H N, Harssema H, Brunekreef B, 1998, ``Air pollution from traffic in city districts near major motorways'' Atmospheric Environment 32 1921 ^ 1930 Samet J M, Dearny A, Eggleston P A, Ford J, Froines J, Gelobter M, 2001, ``Urban air pollution and health inequities: a workshop report'' Environment Health Perspectives 109 357 ^ 374 Smits J, 2003,``Estimating the Heckman two-step procedure to control for selection bias with SPSS'', Department of Economics of Nijmegen School of Management, Nijmegen University, http://home.planetnl/$smits.jeroen/selbias/Heckman-SPSS.doc Stone B, 2003, ``Air quality by design?harnessing the Clean Air Act to manage metropolitan growth'' Journal of Planning and Education Research 23 177 ^ 190 US Department of Transportation, 1997, ``Highway traffic noise in the United States: problem and response'', Federal Highway Administration, Washington, DC, http://www.fhwa.dot/gov/ environment/htnoise.html US Environmental Protection Agency, 2001 The Projection of Mobile Source Air Toxics from 1996 to 2007: Emissions and Concentrations EPA 420-R-01-038, http://www.epa.gov/otaq/regs/toxics/ r01038.pdf US Environmental Protection Agency, 2005 National Emission Trends, 1970 ^ 2002: Average Annual Emissions, All Criteria Pollutants http://www.epa.gov/ttn/chief/trends/index.html van Vliet P, Knape M, de Hartog J, Janssen N, Harssema H, Brunekreef B, 1997, ``Motor vehicle exhaust and chronic respiratory symptoms in children living near freeways'' Environmental Research 74 122 ^ 132 Wilhelm M, Ritz B, 2003, ``Residential proximity to traffic and adverse birth outcomes in Los Angeles County, California, 1994 ^ 1996'' Environmental Health Perspectives 111 207 ^ 217 Winer A M, Behrentz E, Fitz D R, Jeong S, Pankratz D V, Colome S D, 2005, ``Characterizing the range of children's pollutant exposure during school bus commutes'' Journal of Exposure Analysis and Environmental Epidemiology 15 377 ^ 387 Wjst M, Reitmeier P, Dold S,Wulff A, Nicolai T, von Loeffelholz-Colberg E F, von Mutius E, 1993, ``Road traffic and adverse effects on respiratory health in children'' British Medical Journal 307 596 ^ 600 Zhu Y, Hinds W, Kim S, Sioutas C, 2002, ``Concentration and size distribution of ultrafine particles near a major highway'' Journal of the Air and Waste Management Association 52 1032 ^ 1042 ss 2007 a Pion publication printed in Great Britain