Year 2005 UCD— ITS— RR— 05— 24
The Travel Behavior of Immigrants and
Race/ Ethnicity Groups: An Analysis of the
2001 National Household Transportation Survey
Gil Tal
Susan Handy
Institute of Transportation Studies ◊ University of California, Davis
One Shields Avenue ◊ Davis, California 95616
PHONE: ( 530) 752- 6548 ◊ FAX: ( 530) 752- 6572
WEB: http:// its. ucdavis. edu/
The Travel Behavior of Immigrants and Race/ Ethnicity Groups:
An Analysis of the 2001 National Household Transportation Survey
Gil Tal
Susan Handy
Institute of Transportation Studies
University of California Davis
1 Shields Avenue
Davis, CA 95616
Prepared for the California Department of Transportation
as a part of Path Project Task Order 5111
Revised October 2005
The Travel Behavior of Immigrants and Race/ Ethnicity Groups:
An Analysis of the 2001 National Household Transportation Survey
ABSTRACT
The purpose of this paper is to examine the relationships between travel behavior and immigrant status. The National Household Travel Survey ( NHTS) allows us to explore the relationships between travel behavior and characteristics that are usually hard to discern in surveys with smaller samples. The place of birth and year of immigration to the US on travel behavior was tested for commute mode and for general travel variables such as yearly miles driven, number of weekly walk trips, and number of daily trips by all modes. Full models that include spatial and socio- demographic variables were estimated for each of the dependent variables. The effects of place of birth and year of arriving to the US were found to be significant in the full models that control for commute mode and yearly miles driven but not for weekly walk trips or number of daily trips. Understanding the differences in travel behavior and the possible explanations for these differences can help in modeling travel demand, finding policies best suited to meeting the travel needs of foreign born communities, and addressing environmental justice concerns.
ii
Table of Contents
1. Introduction................................................................................................................... .......................... 1
2. Prior Research on Immigrants and Travel............................................................................................ 2
2.1 Residential Location....................................................................................................................... .... 2
2.2 Travel patterns....................................................................................................................... ............. 4
3. Research Method......................................................................................................................... ............ 6
3.1 The Sample......................................................................................................................... ................ 6
3.2 Immigrants’ Characteristics................................................................................................................ 7
3.3 Race and Ethnicity...................................................................................................................... ...... 11
4. Travel Behavior by Race/ Ethnicity and Year of Immigration........................................................... 12
4.1 Travel Patterns by Year of Arriving To the US.................................................................................. 12
4.2 Travel Patterns by Race and Ethnicity.............................................................................................. 15
5. Perceptions of Transportation Services among immigrants and US Born........................................ 19
6. Multivariate Analysis of Travel Behavior by Year of Immigration and Race/ Ethnicity................. 24
6.1 Commute Mode Models For Foreign Born........................................................................................ 25
6.2 Foreign Born Miles Driven per Year................................................................................................. 27
6.3 Commute Mode for Full Sample........................................................................................................ 29
6.4 Miles Driven per Year for Full Sample.............................................................................................. 32
7. Conclusions.................................................................................................................... ........................ 33
8. Refrences...................................................................................................................... .......................... 34
iii
1. Introduction
Approximately 33.5 million foreign- born people live in the United States - 11.7 percent of the US population. ( US Census Bureau, 2005). Immigrants are expected to be a major source for population growth in many parts of the nation. In California, for example, the total population is expected to almost double between the years 1990 to 2040 mainly as a result of new immigrants ( California Department of Finance, 2004).
Previous research suggests that the travel behavior of immigrants is different from the travel behavior of US- born residents for the first five to ten years from arrival to the US and that new immigrants are more likely than others to use public transportation ( Myers; 1996, Deakin and Ferrell; 2001, Purvis; 2003, Casas et al; 2004). The differences in travel behavior are usually associated with the socio- demographic and location characteristics of immigrants. In many cases, the travel patterns of immigrants mirror those of racial and ethnic minorities in the US, although these similarities decline the longer immigrants have lived in the US However, immigration status and place of birth are seldom asked in travel and activity surveys, so that data on the travel behavior of immigrants are limited.
It is important to understand the travel behavior of minority groups in general and immigrant groups specifically for various reasons. First, understanding travel behavior and travel needs of specific groups in society enables the adoption of targeted policies and a more effective distribution of transportation resources; research on the travel behavior of these groups is thus important for addressing environmental justice concerns. Second, understanding the travel behavior of immigrants may help to improve travel demand forecasting, particularly for public transportation and ITS policies. Third, 1
immigrants who are not yet captives of American norms and attitudes may play an important role as agents of change, for example, by using new transit services.
The 2001 National Household Travel Survey ( NHTS) allows us to explore the relationships between travel behavior and characteristics that are usually hard to discern in surveys with smaller samples. We tested the effect of place of birth and year of immigration to the US on travel behavior for commute mode and for general travel variables such as yearly miles driven ( as reported by the respondent), number of weekly walk trips, and number of daily trips by all modes. Full models that include spatial and socio- demographic variables were estimated for each of the dependent variables. The effects of place of birth and year of arriving to the US were found to be significant in the full models that control for socio- demographic and location variables.
2. Prior Research on Immigrants and Travel
Travel patterns are the outcome of the needs and constraints of individuals and households and the location- specific set of opportunities provided by the transportation system and the distribution of activities. In this section we review previous research on immigrants’ residential location decisions and travel patterns. The residential location patterns of immigrants are different in many ways from those of long term residents of the US and may have a strong effect on travel behavior both by affecting the availability of activities and transportation options.
2.1 Residential Location
Most immigrants to the US live in one of the top 25 metropolitan areas within the US, compared to only 50% of the native born population ( Bartel, 1989). Two main theories are used to explain the location decisions of immigrants ( Pamuk 2004). The human 2
ecology approach suggests that immigrants are willing to live in congested conditions as a transition phase before improving their socio- economic conditions and moving to middle class neighborhoods. The second theory maintains that ethnic clusters provide immigrants with socioeconomic and cultural networks and therefore immigrants are not likely to move out when their socioeconomic conditions improve. Both theories suggest a clustering pattern of new immigrants that may affect their cultural and behavioral experience and therefore their travel behavior.
In San Francisco, Pamuk ( 2004) found three different types of immigrant clustering among Chinese, Mexican, and Filipino immigrants: ( 1) low income ethnic clusters, ( 2) more wide spread middle- income ethnic communities, and ( 3) a high income Chinese cluster. The outcome of this research suggests that both theories can be applied in the city and that immigrant groups will not necessarily blend with the general population when their economic situations improve.
In a national study, Bartel ( 1989) concluded that more immigrants that first locate outside of the major metropolitan areas eventually move inside these areas than immigrants that first locate inside these areas move out. Immigrants with higher education levels are more likely to live in the major metropolitan areas. Asians and Europeans that are more educated are more likely to choose a location outside of the major metropolitan areas. Hispanic immigrants that are more educated tend to move out of these areas as a second migration within the United States.
According to the theories outlined above, immigrant segregation depends on time of arrival and socio- demographic factors such as education, income, household size, and others. Rolf ( 2001) explored the connection between density, sprawl, and segregation by
3
race and income in US Metropolitan Areas and found that high density development does not reduce economic segregation and ethnic segregation. Economic segregation in this case is highly correlated with ethnic segregation, but new Hispanic immigrants tend to be more segregated regardless of their economic situation.
2.2 Travel patterns
The few studies that have examined the travel patterns of immigrants have focused on changes in behavior over time. Myers ( 1996) that shows that after ten years in the US, the travel behavior of immigrants becomes similar to that of the US born population. In their first years in the US, immigrants’ behave differently, for example by using more transit than the US born population. The extensive use of public transportation in the first years from arrival compared to the general population was also reported by Casas et al. ( 2004), who used 2001 NHTS data to evaluate travel behavior of “ newcomer Hispanic” versus “ settled” and native born residents.
Vehicle ownership is highly correlated with mode choice as households that cannot afford a car are more likely to use public transportation or other travel modes and as households that live in areas with high density and with good public transportation services have less motivation to purchase vehicles. McGuckin and Srinivasan ( 2003) found that 20.7% of the new immigrants live in households without vehicles versus only 8% of the immigrants who have lived in the US for ten years or more and 3.9% of the US born population. In areas where immigrants are highly concentrated, they may create an important portion of the demand for public transportation. Purvis ( 2003), for example, found that immigrants generate about one third of the public transportation commuting trips in San- Francisco.
4
The dependence of immigrants on public transportation together with unique activity patterns and cultural barriers that inhibit the use of regular public transportation can lead to ethnic- exclusive transportation services. For example, Camionetas are mini- vans privately operated as jitney services throughout cities in the US by Latino immigrants ( Valenzuela et al, 2005). Douma ( 2004) used focus groups of specific populations to analyze ways of better serving these populations using ITS. This study focused on “ non- traditional” populations such as immigrant, disabled, and retired groups. Focus groups were held with for homogenous groups of Latino, Somali and Hmong immigrants in both urban and rural areas. The focus groups showed that Latino immigrants are open to transit and more “ social” types of travel, while privacy was an important consideration for the Hmong. All groups were found to prefer driving themselves.
The evidence reviewed here suggest that the travel patterns of immigrants derive from both socio- demographic characteristics ( in ways similar to the US born population) and unique requirements and needs related to cultural and attitudinal differences. Immigrant travel behavior may be different from US born travel behavior with the same socio- demographic characteristics for a variety of reasons: ( 1) activity patterns: immigrants may have different needs such as shopping in special ethnic food shops, or they may have different social and recreational habits, etc., ( 2) cognitive maps: immigrants may have different level of knowledge about their area that may be reflected in their activity patterns and in their route choices, ( 3) attitudes and beliefs: immigrants may have a different set of attitudes and beliefs about transportation that influence the amount and mode of travel. In the work that follows, we examine the relationship
5
between immigrant status and travel behavior, recognizing that immigrant status itself does not have a causal effect but rather serves as an indicator of these underlying differences.
3. Research Method
The 2001 National Household Travel Survey ( NHTS) allows us to explore the relationships between travel behavior and characteristics that are usually hard to discern in surveys with smaller samples. The correlation of place of birth or alternatively race/ ethnicity and year of immigration to the US on travel behavior was tested for commute mode and for general travel variables such as yearly miles driven ( as reported by the respondent), number of weekly walk trips, and number of daily trips by all modes. Full models that include spatial and socio- demographic variables were estimated for each of the dependent variables. We tested separate models for the general population that includes the US born population and for immigrant only.
3.1 The Sample
The National Household Travel Survey ( NHTS) is a national level survey comprising a questionnaire and a travel diary survey that is conducted every five to six years. There are approximately 66,000 households in the final 2001 NHTS dataset, including about 160,658 people; we used a smaller sample of 97,694 people after taking out all cases where travel data where not complete. All the cases where travel was part of the individual’s job requirement and all cases were physical or other conditions limited the individual from traveling in any transportation mode were also excluded. The characteristics of the reduced sample are somewhat different from the regional sample;
6
for example, the reduced sample was 53.7% female in contrast to 51.9% female in the regional sample.
3.2 Immigrants’ Characteristics
The sample used in the analysis includes 5,396 cases of foreign- born individuals, about half of which arrived in the US in the ten years prior to the survey ( Table 1).
Table 1: Population by Year of Arriving to the US/ US born
Level
Count
Percent
Missing answer
250
4.63%
Move to the US Before 1981
2503
46.4%
Move to the US between 1981- 1991
1197
22.1%
Move to the US between 1991- 1996
722
13.3%
Move to the US between 1996- 2001
724
13.4%
Total
5396
100%
The gender break down for foreign born respondents is 42.5% male and 57.5% female, higher than for the overall sample. The new arrivals to the US are distributed across all ages, though the largest share is in their 30s ( Figure 1).
Figure 1: Current Age Distribution of Respondent Arrived to
the US between 1981 and 2001
7
Despite the fact that immigrants arrive in the US at all ages, there is a clear trend in life cycle: a few years after arriving in the US the number of people in the household increases and respondents have more children. As shown in Figure 2, around 45% of households that include people who arrived in the US in the last five years contain one or two adult with no children; for households with people who arrived five to ten years before the survey, this share is around 30%; and for people who arrived 10 to 20 years ago, less then 25% live in households without children.
10
adults, retired, no child
09
one adult, retired, no child
08
adults, youngest child 16- 21
07
one adult, youngest child 16- 21
06
2+ adults, youngest child 6- 15
05
one adult, youngest child 6- 15
04
2+ adults, youngest child 0- 5
03
one adult, youngest child 0- 5
02
2+ adults, no child
01
one adult, no children
Figure 2: Arrived in the US by Household Life Cycle
Before 1981 1981- 1991 1991- 1995 1995- 2001
The income of new arrivals ( defined here as immigrants who arrived in the US in the last five years) appears to increase over time, as shown in Figure 3: people who arrived in the US recently are generally poorer than people who arrived before them. It is 8
also interesting that the biggest differences are in the very low income and very high income categories. In the first five year from arrival, 30% of respondents make less than $ 30,000 per year; only about 20% of people who arrived 5 to 10 years before the survey and around 10% of the people who arrived more then 20 years before the survey are included in this income group. The share of high income households with income over $ 100,000 per year is twice that for people who arrived more then 10 years ago compared to people who arrived in the last five years.
income0.000.250.500.751.00
1981199119962001
YRTOUS_ 3
1
2
3
4
Before 1981 1981- 1991 1991- 1995 1995- 2001
Income groups
4
$ 75,000 +
3
$ 50,000 - $ 74,999
2
$ 25,000 - $ 49,999
1
< $ 5,000 - $ 24999
Figure 3: Arrived To the US by Household Income Group
Figure 4 shows the race and ethnicity of immigrants for four main groups ( White, African Americans/ Black, Asian, and Hispanic of any race). White immigrants account for more then half of immigrants who arrived more than twenty years before the survey but only about one- third for immigrants who arrived more recently. Asians and Hispanics
9
represent about quarter of immigrants for each group who arrived in the twenty years prior to the survey while Black immigrants are about 5% of the total in these time periods.
Hispanic
5
other
4
Asian
3
Black
2
White
1
Figure 4: Arrived To the US By Race
Before 1981 1981- 1991 1991- 1995 1995- 2001
Table 2 summarizes the socio- demographic variables of age, household size, and household yearly income by year of arriving to the US or US born status.
Table 2: Socio Demographic Variables by Year of arriving to the US
Respondent Age ( years)
HH Size ( people)
yearly HH income
Year of arriving to the US
Mean
Std Dev
t- test vs. US born*
Mean
Std Dev
t- test vs. US born*
Mean
Std Dev
t- test vs. US born*
US Born
37.17
23.13
-
3.23
1.50
-
54046
30467
-
Pre 1981
52.92
15.34
< 0.0001
2.92
1.50
< 0.0001
52791
32466
0.0567
1981- 1991
37.61
12.08
0.2208
3.85
1.73
< 0.0001
50868
32188
0.0007
1991- 1996
35.49
11.59
0.0001
3.64
1.69
< 0.0001
44625
30759
< 0.0001
1996- 2001
32.00
10.22
< 0.0001
3.32
1.55
0.1203
40109
30635
< 0.0001
* Probability that mean for this group is not different than the mean for the US born group.
10
3.3 Race and Ethnicity
Race and ethnicity constitute the second demographic variable of focus in this analysis. The NHTS data includes a race and ethnicity variable that specifies 17 categories of race or ethnicity or a combination of the two. We aggregated these categories into a new variable, as described in Table 3, with just five categories: ( 1) white only ( 2) African American/ Black and a combination of Black and White, Black and Asian, Black and Hispanic ( 3) Asian, including White and Asian ( 4) American Indian, Alaskan Native, Native Hawaiian, other Pacific Islander and other combinations of two races or more, and ( 5) Hispanic of any race.
Table 3: Population by Race and Ethnicity
Level
Count
Percentage
- 9 Missing
695
7.11%
1 White only
82151
84.09%
2 African American/ Black
4237
4.33%
3 Asian
3800
3.89%
4 Other
2692
2.75%
5 Hispanic of any race
4119
4.21%
Total
97694
100%
The socio demographic characteristics of these racial/ ethnic groups differ significantly ( Table 4). The average ages of White and Asian respondents are similar, at just under 41 years, and significantly older then Black and Hispanic respondents; the average age of Hispanic respondents is almost 10 years lower then the average age of White respondents. This difference can be partly explained by the number of children per household; the average household size for Hispanic respondents is significantly higher then the average household size for white respondents. Similarly, White and Asian
11
respondents have considerably higher incomes on average, around to $ 52,000 to 55,000$ per year, while households of black and Hispanic respondents have average annual incomes of around $ 41,000.
Table 4: Socio Demographic Variables by Race and Ethnicity
Respondent Age ( years)
HH Size ( people)
yearly HH income
Race
Mean
Std Dev
t- test vs. White*
Mean
Std Dev
t- test vs. White*
Mean
Std Dev
t- test vs. White*
White
40.93
21.51
-
3.16
1.45
-
55360
30378
-
Black
35.84
20.79
< 0.0001
3.40
1.67
< 0.0001
40308
28702
< 0.0001
Asian
40.99
21.96
0.8691
3.51
1.77
< 0.0001
52650
32497
< 0.0001
Other
35.70
20.45
< 0.0001
3.70
1.79
< 0.0001
48712
29797
< 0.0001
Hispanic
31.33
18.60
< 0.0001
3.97
1.59
< 0.0001
41338
28011
< 0.0001
* Probability that mean for this group is not different than the mean for the White group.
4. Travel Behavior by Race/ Ethnicity and Year of Immigration
In this section we examine differences in travel behavior in terms of commute mode, yearly miles driven, number of trips per day, and number of walk trips per week between racial/ ethnic groups and by year of immigration to the US.
4.1 Travel Patterns by Year of Arriving To the US
In this section we focus on the travel behavior of foreign born respondents differentiated by year of arrival in the US. Respondents who arrived in the US in the five years before the survey make somewhat fewer trips per day on average than other respondents ( Table 5). Given the large standard deviation, the differences are very small but still based on the student’s t test people that arrived in the last 10 years generate statistically significant less trips then US born population.
12
Table 5: Daily Trips per Person
Year of arriving to the US
Number
Mean
Std Dev
t- test vs. US born*
US Born
84900
4.36
2.7679
-
Pre 1981
2238
4.37
2.7535
0.8654
1981- 1991
1102
4.41
2.6416
0.5328
1991- 1996
642
4.09
2.4925
0.0065
1996- 2001
655
3.99
2.5062
0.0002
* Probability that mean for this group is not different than the mean for the US born group
The number of private vehicles per person in immigrant households differs by year of arriving in the US and compared to US born respondents ( Table 6). Recent arrivals ( within the five years before the survey) drive on average 7,230 miles per year, arrivals five to ten year before the survey drive about 9,500 miles per year, and immigrants that arrived more than 10 years before the survey drive about 10,500 miles per year. Recent arrivals may be driving less than others because of more limited availability of a car: the average number of vehicles per person is 0.45 for the households of recent immigrants, compared to 0.58 vehicles per person for respondents who arrived 10 to 15 years before the survey. The household of immigrants who arrived more then 20 years ago have a much higher level of automobile ownership, with almost 0.8 vehicles per person on average.
Table 6: HH Vehicle Ownership per Person
Year of arriving to the US
Number
Mean
Std Dev
t- test vs. US born*
US Born
92298
0.81
0.4702
-
Pre 1981
2503
0.79
0.4290
0.0218
1981- 1991
1197
0.58
0.3456
< 0.0001
1991- 1996
722
0.55
0.3514
< 0.0001
1996- 2001
724
0.45
0.3484
< 0.0001
* Probability that mean for this group is not different than the mean for the US born group 13
The average number of walk trips per week is highest for respondents who arrived in the last five years at 3.6 trips per week ( Table 7); respondents who arrived before 1981 make fewer than three walk trips per week by comparison. US born respondents make three trips per week on average, significantly lower than new arrivals. Similarly, recent immigrants make about 0.4 bicycle trips per week on average, versus 0.26 per week on average for immigrants who have lived in the US for five to ten years and 0.2 trips per week on average for the entire sample.
Table 7: Number of Walk trips per Week
* Probability that mean for this group is not different than the mean for the US born group
Year of arriving to the US
Number
Mean
Std Dev
t- test vs. US born*
US Born
67926
3.0153
5.2688
-
Pre 1981
2488
2.9501
4.8459
0.5112
1981- 1991
1189
2.9133
4.8714
0.4749
1991- 1996
715
3.200
6.2600
0.4321
1996- 2001
717
3.5815
6.2678
0.0162
For the entire sample, 93% of respondents commute by motor vehicle, 4% by public transportation, and 3% by walking. Among the 2866 foreign born commuters, the distribution across modes is significantly different depending on year of arrival in the US ( Figure 5). About 12% of the commuters who arrived in the US in the five years before the survey walk as their commute mode, though only about 3% of respondents who arrived 10 years before the survey walk. Differences for public transportation are also significant, with just fewer than 20% of recent arrivals commuting by public transportation but only 10% of immigrants who have in the US for more then 20 years using public transportation. However, the share using public transportation for the
14
immigrants who have been in the US for the longest is still higher than for US born respondents.
mode_ order0.000.250.500.751.001981199119962001YRTOUS_ 3123
Motor Vehicle
Public Transportation
Walk
Figure 5: Arrived To the US by Main Commuting Mode
Before 1981 1981- 1991 1991- 1995 1995- 2001
4.2 Travel Patterns by Race and Ethnicity
In this section we will explore the correlation between race and travel behavior. The total number of daily trips varies significantly: white respondents make more trips per day than the other race/ ethnicity groups, which average 4.41 for the White group and between 3.8 and 3.9 to black Asian and other race and ethnic groups. The Hispanic group travel similar number of trips as The White groups ( Table 8).
Table 8: Total Daily Trips per Person
Race
Number
Mean
Std Dev
t- test vs. White*
White
82151
4.41
2.7812
-
Black
4237
3.80
2.5142
< 0.0001
Asian
3800
3.87
2.6679
< 0.0001
Other
2692
3.92
2.7148
< 0.0001
Hispanic
4119
4.15
2.6587
< 0.0001
* Probability that mean for this group is not different than the mean for the White group 15
Differences in commute mode also differ for the race/ ethnicity groups. Although driving dominates as a commute for all groups, this dominance is greatest for white respondents and least for black respondents, whose share of commute trips by walking is twice as high as the share for white respondents and whose share of commute trips by public transportation is ten times as high as for white respondents and twice as high as for other race/ ethnicity groups ( Figure 6).
Walk
3
Public Transportation
2
Motor Vehicle
1
Figure 6: Commuter Mode Choice by Race
16
One of the explanations for these differences in commute mode may be differences in vehicle ownership per person, which varied from 0.84 on average for white respondents to 0.51 on average for black respondents ( Table 9).
Table 9: HH Vehicle Ownership per Parson
Race
Number
Mean
Std Dev
t- test vs. White*
White
82151
0.84
0.4682
-
Black
4237
0.51
0.4362
< 0.0001
Asian
3800
0.68
0.3755
< 0.0001
Other
2692
0.70
0.4397
< 0.0001
Hispanic
4119
0.56
0.3810
< 0.0001
* Probability that mean for this group is not different than the mean for the White group
The yearly mileage driven by drivers in each group paints a different picture from the previous variables ( Table 10). White respondents drive about 12,000 miles per year on average while black respondents drive just over 10,000 miles per year on average. Asian respondents drive the least on average, at 8600 miles per year.
Table 10: Yearly Miles Driven ( Drivers only)
Race
Number
Mean
Std Dev
t- test vs. White*
White
47862
12091
10082
-
Black
1524
10275
12860
< 0.0001
Asian
1789
8619
7267
< 0.0001
Other
1197
11497
10859
0.0614
Hispanic
1527
11750
13542
0.3295
* Probability that mean for this group is not different than the mean for the White group
Walk trips may be the outcome of necessity or of choice. White respondents make the most walking trips per week ( 3.13), followed by black respondents ( 3.00) and Hispanic respondents ( 2.82) ( Table 11). All of these groups make over twice as many walking trips as Asian respondents. White respondents also make more bicycle trips per
17
week then the other groups with average of about 0.19 trips per week. Asian respondents make the fewest bicycle trips per week, just under 0.01 trips per week. Note that bicycle trips are significantly less frequent than walking trips.
Table 11: Walk Trip per Week by Race/ Ethnicity
Race
Number
Mean
Std Dev
t- test vs. White*
White
62302
3.1277
5.3040
-
Black
2946
2.9966
5.7502
0.2251
Asian
2910
1.3879
3.7862
< 0.0001
Other
1880
2.3473
5.0466
< 0.0001
Hispanic
2675
2.8231
5.1928
0.0030
* Probability that mean for this group is not different than the mean for the White group
18
5. Perceptions of Transportation Services among immigrants and US Born
As part of the NHTS survey the respondents were asked about their perceptions regarding different traffic situations. The questions focused on infrastructure conditions and traffic conditions on roads and did not address public transportation. The response rate varies considerably from question to question as some questions were asked randomly and not to all the respondents.
When asked about “ worrying about a traffic accident,” immigrants express more concern then US born respondents ( Table 12). For those born in the US, less than 40% of respondents consider traffic accidents as a somewhat of a problem, very much of a problem, or a severe problem, in contrast to around 50% in the immigrant population. More than 18% of immigrants who arrived since 1996 see traffic accidents as a severe problem, compared to less than10% of the US born population.
Table 12: Perceptions of “ Worrying about a traffic accident” as a Problem
by Year of Arriving to the US/ Born in the US ( n= 18,882)
Not a problem
A little problem
Somewhat of a problem
Very much of a problem
A severe problem
US Born
34.1
27.44
19.73
8.78
9.96
Pre 1981
30.9
22.92
20.1
9.63
16.45
1981- 1991
25.47
25.47
20.97
13.11
14.98
1991- 1996
25.33
21.33
24
11.33
18
1996- 2001
31.61
20.65
20
9.03
18.71
Overall
33.79
27.16
19.79
8.89
10.37
* Chi square = 70.3, p- value< 0.001
About three times as many respondents answered the question on highway congestion ( Table 13). There are no differences between US Born respondents and foreign born respondents who arrived in the US prior to 1996 ( i. e, have lived in the US
19
more then five years). About 50% of these respondents do not consider highway congestion a problem or consider it a little problem. New immigrants ( with less then five years in the US) tend to consider congestion less of a problem than the other groups, probably as a result of lower use of private vehicles. The results for traffic or road conditions in general ( Table 14) are similar to those for highway congestion ( Table 13)
Table 13: Perceptions of “ Highway congestion” as a Problem by Year of
Arriving to the US/ Born in the US ( n= 53,518)
Not a problem
A little problem
Somewhat of a problem
Very much of a problem
A severe problem
US Born
32.72
23.12
21.67
11.29
11.20
Pre 1981
29.95
18.61
20.62
13.31
17.51
1981- 1991
28.27
18.59
23.30
12.43
17.41
1991- 1996
32.31
17.69
25.55
12.88
11.57
1996- 2001
32.24
22.00
26.36
12.20
7.19
Overall
32.56
22.85
21.74
11.39
11.47
* Chi square = 133.26; p- value< 0.001
Table 14: Perceptions of “ Traffic or road congestion” as a Problem by Year
of Arriving to the US/ Born in the US ( n= 13,632)
Not a problem
A little problem
Somewhat of a problem
Very much of a problem
A severe problem
US Born
31.97
24.67
19.93
12.35
11.09
Pre 1981
26.22
24.59
19.95
17.63
11.60
1981- 1991
23.00
25.50
21.50
11.00
19.00
1991- 1996
30.36
21.43
15.18
16.96
16.07
1996- 2001
29.67
31.87
13.19
10.99
14.29
Overall
31.62
24.70
19.87
12.52
11.28
* Chi square = 27.3; p- value< 0.0375
Respondents were also asked about their perceptions of driver behavior as a problem – distracted drivers ( Table 15), drunk drivers ( Table 16), aggressive drivers ( Table 17), and speeding ( Table 18). On almost all questions, the differences between immigrant groups and the US born population are significant, though the patterns of differences are not consistent across questions. For example, a higher share of those who
20
arrived between 1981 and 1991 than other groups think that distracted drivers are a problem, while a higher share of those who arrived between 1996 and 2001 than other groups think that drivers speeding are not a problem. The reasons for these differences are not readily apparent, though they could be tied to age differences for the different immigrant groups.
Table 15: Perceptions of “ Distracted drivers” as a Problem by Year of Arriving
to the US/ Born In The US ( n= 13,102)
Not a problem
A little problem
Somewhat of a problem
Very much of a problem
A severe problem
US Born
16.70
23.97
27.63
17.13
14.57
Pre 1981
18.94
19.86
24.94
17.55
18.71
1981- 1991
14.97
22.75
23.35
18.56
20.36
1991- 1996
15.09
26.42
21.70
16.98
19.81
1996- 2001
24.79
27.35
21.37
11.11
15.38
Overall
16.81
23.87
27.39
17.10
14.83
** Chi square = 15.7; p- value< 0.047
Table 16: Perceptions of “ Drunk drivers” as a Problem by Year of Arriving to the US/ Born in the US ( n= 13,541)
Not a problem
A little problem
Somewhat of a problem
Very much of a problem
A severe problem
US Born
42.87
18.46
11.71
6.52
20.45
Pre 1981
44.52
18.88
11.89
6.76
17.95
1981- 1991
36.87
15.66
13.64
7.07
26.77
1991- 1996
43.36
14.16
8.85
7.96
25.66
1996- 2001
40.22
17.39
8.70
10.87
22.83
Overall
42.82
18.39
11.70
6.57
20.52
* Chi square = 140.0; p - value< 0.001
21
Table 17: Perceptions of “ Aggressive drivers on the road” as a Problem by Year of Arriving to the US/ Born in the US ( n= 13,570)
Not a problem
A little problem
Somewhat of a problem
Very much of a problem
A severe problem
US Born
17.28
23.02
25.14
17.76
16.80
Pre 1981
18.60
18.60
21.88
17.94
22.98
1981- 1991
18.50
24.00
20.00
20.50
17.00
1991- 1996
16.26
21.14
29.27
17.07
16.26
1996- 2001
29.13
22.05
16.54
16.54
15.75
Overall
17.44
22.86
24.92
17.79
16.99
* Chi square = 77.9; p- value< 0.001
Table 18: Perceptions of “ Drivers speeding” as a Problem by Year of Arriving to the US/ Born in the US ( n= 13,224)
Not a problem
A little problem
Somewhat of a problem
Very much of a problem
A severe problem
US Born
20.20
23.31
23.64
16.81
16.05
Pre 1981
18.87
19.61
27.45
16.42
17.65
1981- 1991
14.29
23.98
23.98
20.92
16.84
1991- 1996
18.10
24.14
21.55
18.10
18.10
1996- 2001
27.05
23.77
19.67
14.75
14.75
Overall
20.11
23.22
23.71
16.85
16.11
* Chi square = 67.1; p- value< 0.001
The results suggest that recent immigrants do not consider gas price a problem to the same degree as less recent immigrants and US born respondents: only about 11% of recent immigrants consider gas price a severe problem compared to about 15% of immigrants who have been in the US for 5 to 10 years and over 20% for US born respondents. This difference in perception can be attributed to the lower use of private vehicles for recent immigrants and by the recent memory of higher gas prices in the country of origin for the recent immigrants.
22
Table 19: Perceptions of “ Price of gasoline” as a Problem by Year of Arriving to the US/ Born in the US ( n= 47,773)
Not a problem
A little problem
Somewhat of a problem
Very much of a problem
A severe problem
US Born
18.20
18.90
26.03
15.43
21.44
Pre 1981
21.51
15.84
24.38
17.67
20.60
1981- 1991
22.04
18.25
22.63
17.81
19.27
1991- 1996
29.83
21.00
22.20
12.41
14.56
1996- 2001
31.90
22.14
22.38
12.14
11.43
Overall
18.58
18.84
25.86
15.48
21.23
* Chi square = 140.3; p- value< 0.001
Only one question focused on non- motorized modes in asking about “ lack of sidewalks and walkways” as a problem ( Table 20). Compared to the previous questions, few respondents in any group considered lack of sidewalks a problem. However, recent immigrants perceive lack of sidewalk as a problem to a greater degree than the other groups. The concern on the part of recent immigrants may reflect their lower use of private vehicles and their residential location within urban areas. Immigrants that arrived to the US prior to 1981 also see a lack of sidewalks as a problem to a greater degree than other groups; this difference may be attributable to the fact that this group is on average a little older than the general population.
Table 20: Perceptions of “ Lack of walkways or sidewalks” as a Problem by Year of Arriving to the US/ Born in the US ( n= 13,547)
Not a problem
A little problem
Somewhat of a problem
Very much of a problem
A severe problem
US Born
57.99
17.23
10.83
6.30
7.65
Pre 1981
46.72
19.00
14.85
8.73
10.70
1981- 1991
39.80
21.39
18.41
12.44
7.96
1991- 1996
49.19
19.35
14.52
9.68
7.26
1996- 2001
42.52
19.69
16.54
5.51
15.75
Overall
57.11
17.39
11.17
6.50
7.83
* Chi square = 77.3; p- value< 0.001
23
6. Multivariate Analysis of Travel Behavior by Year of Immigration and Race/ Ethnicity
In order understand the relationship of travel behavior with immigration status and race/ ethnicity group while controlling for socio- demographic characteristics such as income as well as location, we estimated a series of multivariate models for two measures of travel behavior: commute mode and yearly miles driven. First we present models for foreign born respondents only, focusing on the year of immigration and either race/ ethnicity group or place of birth as explanatory variables. Second, we present models for the entire sample, to compare travel behavior for foreign born versus US born respondents, focusing on foreign born and race/ ethnicity group as explanatory variables. Table 21 includes the definitions of all variables used in this section.
Table 21: Variables Names and Definitions
Dependent variables
Variable name
Definition
Miles Driven per Year
Miles respondent reported on driving in the last 12 months
Foreign Born Commute Mode
Transportation mode to work last week by two categories: ( 1) motor vehicle ( car, Van, SUV, Pickup truck, other truck, motorcycle ) ; ( 2) public transportation ( local public transit bus, commuter bus, city to city bus, AMTRACK, commuter train, subway, street car/ trolley )
Commute Mode for all Sample
Transportation mode to work last week by three categories: ( 1) motor vehicle ( car, Van, SUV, Pickup truck, other truck, motorcycle ) ; ( 2) public transportation ( local public transit bus, commuter bus, city to city bus, AMTRACK, commuter train, subway, street car/ trolley ) ( 3) non motorized trip ( walk, bicycle)
Independent Variables
Variable name
Definition
YRTOUS_ N
Number of years in the US between 0 for arriving in 2001 to 45 for people entered the US prior to 1958
YTOUS_ L
US born
YTOUS_ 5
In the US for 0 to 5 years
YTOUS_ 10
In the US for 5 to 10 years
YTOUS_ 15
In the US for 10 to 15 years
24
Income_ m
House Hold income per year in $
race_ W1
Race = White Non Hispanic
race_ B1
Race= Black Non Hispanic
race_ A1
Race= Asian
race_ H1
Race= Hispanic
veh per driver_ to1
Vehicle per Driver up to a ratio of one
R_ SEX
Sex 1= male
R_ AGE
Respondent age in years
HHSIZE
Number of people in the household
MSACAT_ 1t
Metropolitan statistical area of 1 million or more, with heavy transit
MSACAT_ 2n
Metropolitan statistical area of million or more, not in 1
MSACAT_ 3
Metropolitan statistical area less than 1 million
Bornin01
Place of birth: Canada, US Territories
Bornin02
Place of birth: Central and South America
Bornin03
Place of birth: Europe & Scandinavia/ Polar Regions
Bornin04
Place of birth: Eastern Europe & Russia/ USSR
Bornin05
Place of birth: East Asia
Bornin06
Place of birth: Indian Subcontinent
Bornin07
Place of birth: Caribbean/ Atlantic Islands
6.1 Commute Mode Models For Foreign Born
A sample of 2450 foreign born commuters was used to estimate a model to test the association between commute mode and number of years in the US, race/ ethnicity group, other socio- demographic variables, and location variables. We limited the modes in the model to motor vehicle or public transportation because of the small number of respondents in this sample that walked ( 14) or used other modes. The initial model with all variables, estimated using logistic regression, had a pseudo R- square of 0.34. The initial model was used to identify variables that have no effect on commute mode for foreign born respondents. Insignificant variables included income, all race/ ethnicity groups, and medium MSA size.
The final model presented in Table 22 has a pseudo R- square of 0.337. Not surprisingly, the average number of vehicles per household is highly significant, with a higher number of vehicles per driver in the household associated with a lower likelihood
25
of taking transit. Size of metropolitan area is also significant: residents of smaller metropolitan areas are less likely to use transit, while residents of large metropolitan areas with heavy transit are more likely to use transit. Gender is also significant ( men are less likely to use transit), as is household size ( larger household size means less likely to take transit). Although no race/ ethnicity groups were significant in the initial model, the variable for the black group was significant in the final model: black respondents were more likely to take transit. Once these variables have been accounted for, the effect of years in the US is significant: more recent immigrants are more likely to take transit. Although household income is often shown to be associated with mode, it was not significant in this model; the number of vehicles per household may partially account for the effect of income.
Table 22: Model for Commute Mode – Foreign Born and Race/ Ethnicity
Public transportation over private vehicle
Term
Estimate
Std Error
ChiSquare
Prob> ChiSq
Intercept
1.1270
0.3789
8.85
0.0029
YRTOUS_ N
- 0.0192
0.006
8.55
0.0035
race_ B1[ 0]
- 0.3977
0.1165
11.65
0.0006
veh per driver_ to1
- 3.5144
0.2212
252.33
<. 0001
R_ SEX[ 1]
- 0.1374
0.0752
3.33
0.0680
R_ AGE
0.0135
0.0072
3.46
0.0629
HHSIZE
- 0.1585
0.0496
10.18
0.0014
MSACAT_ 1t[ 0]
- 0.6196
0.0822
56.69
<. 0001
MSACAT_ 3[ 0]
0.4421
0.1446
9.34
0.0022
A second model with a sample of 2862 foreign born commuters was used to estimate a model to test the association between commute mode and number of years in the US using a region- of- birth variable instead of race/ ethnicity and the same variables as in the previous model ( Table 21). The final model had a pseudo R- square of 0.33. This model shows some similarities with the model presented in Table 23, which included year of immigration: vehicles per driven and household size were significant in both
26
models, although age dropped out in the second model. Most interesting, two places of birth were significant: immigrants that were born in the former USSR and immigrants from the Caribbean are more likely to use public transportation for their commuting trips. In this model, the effect of years in the US was insignificant, perhaps because it is related to place of birth.
Table 23: Model for Commute Mode – Foreign Born and Place of Birth
Public transportation over private vehicle
Term
Estimate
Std Error
ChiSquare
Prob> ChiSq
Intercept
- 1.52368640
0.2884688
27.90
<. 0001
HHSIZE
0.15968678
0.0464779
11.80
0.0006
veh per driver_ to1
3.62745344
0.2031541
318.83
<. 0001
MSACAT_ 1t[ 0]
0.68642611
0.0767311
80.03
<. 0001
MSACAT_ 3[ 0]
- 0.40679020
0.1310651
9.63
0.0019
born04[ 0]
0.26425604
0.1280314
4.26
0.0390
born07[ 0]
0.21936416
0.1182788
3.44
0.0636
6.2 Foreign Born Miles Driven per Year
A multivariate linear model was estimated for the sample of foreign born respondents for the dependent variable miles driven per year. An important limitation of this analysis is the lack of data on the role of the respondent as a member of the household; household responsibilities influence the activity and travel patterns of individual household members. In addition, the annual mileage as reported by the respondent may be inaccurate. Nevertheless, the multivariate model can help to identify variables associated with driving levels and provide a basis for further research.
The initial model for miles driven in the last 12 months with race/ ethnicity variable, estimated using ordinary least squares regression is based on 2865 observations and has an adjusted R- square of about 0.08. The low R- square suggests that relatively little of the variation in miles driven per year is explained by the variables in the model. The variables found insignificant in the initial model were dropped one by one and the
27
model re- estimated with the reduced set of variables. The initial model shows that race/ ethnicity, location, and household size are not significant in predicting miles driven per year.
The final model, presented in Table 24, includes just seven variables but achieves a similar adjusted R- square of 0.076. Scaled estimates, the standardized effect of each variable on miles per year, shows more clearly the relative importance of each variable in predicting miles driven. These estimates show that the number of vehicles per driver ( capped at one to account for the limited effect that having more vehicles than drivers is likely to have on travel behavior) is the most significant variable, followed by age, Hispanic race/ ethnicity, and gender. An increase of 0.5 in the ratio of vehicles per driver is associated with an increase of 2425 miles per year. Women drive 2212 fewer miles per year than men. Respondents of Hispanic origin drive 1357 more miles per year on average than non- Hispanic respondents; white respondents drive 945 more than non- white respondents. After accounting for these effects, recent immigrants drive 968 fewer miles per year than less recent immigrants and US born respondents.
Table 24: Model for Miles Driven Last 12 Months –
Foreign Born and Race/ Ethnicity
Term
Scaled Estimate*
Estimate
Std Error
t Ratio
Prob>| t|
Intercept
10736.22
6356.2342
1185.167
5.36
<. 0001
veh per driver_ to1
2425.50
6468.0195
1090.769
5.93
<. 0001
income_ m
1389.32
0.0292489
0.007151
4.09
<. 0001
race_ W1[ 0]
- 945.62
- 945.6208
227.8208
- 4.15
<. 0001
race_ H1[ 0]
- 1357.40
- 1357.403
315.0955
- 4.31
<. 0001
R_ AGE
- 3510.76
- 73.1409
13.53064
- 5.41
<. 0001
R_ SEX[ 1]
2212.27
2212.2764
196.7093
11.25
<. 0001
YTOUS_ 5[ 0]
968.03
968.03405
342.4276
2.83
0.0047
* Nominal factors expanded to all levels
* Continuous factors centered by mean, scaled by range/ 2 i. e. the scaled estimate shows the change in the dependent variable for an increase of ½ of the range of the independent variable.
28
A similar model was estimated for the sample of 3024 foreign born respondents using the region of birth variable instead of the race/ ethnicity variable. The model, presented in Table 25, has an adjusted R- square of 0.133 - higher than the model that used race/ ethnicity to define immigrant groups. Scaled estimates show that this model is similar to the previous one, in that the number of vehicles per driver is the most important variable. The effect of the year of arriving to the US is stronger in this model, with both five and ten years in the US significant. Respondents born in Canada, American territories, and Western Europe drive about 600 to 700 more miles per year than other immigrants.
Table 25: Model for Miles Driven Last 12 Months –
Foreign Born and Place of Birth
Term
Scaled Estimate*
Estimate
Std Error
t Ratio
Prob>| t|
Intercept
9636
6721.4362
729.2614
9.22
<. 0001
R_ AGE
- 2114
- 66.06994
11.06299
- 5.97
<. 0001
R_ SEX[ 1]
2016
2016.5933
148.2866
13.60
<. 0001
income_ m
1305
0.027491
0.005223
5.26
<. 0001
vec per driver
10584
4233.7262
373.028
11.35
<. 0001
YTOUS_ 5[ 0]
1312
1312.9011
256.6887
5.11
<. 0001
YTOUS10[ 0]
516
516.28949
234.5614
2.20
0.0278
bornin01[ 0]
- 660
- 660.0625
270.4028
- 2.44
0.0147
bornin03[ 0]
- 715
- 715.0786
182.1884
- 3.92
<. 0001
* Nominal factors expanded to all levels
* Continuous factors centered by mean, scaled by range/ 2 i. e. the scaled estimate shows the change in the dependent variable for an increase of ½ of the range of the independent variable.
6.3 Commute Mode for Full Sample
In this section we identify factors associated with commute mode for the full sample that includes foreign- and US- born respondents. This larger sample of 37,565 respondents enables an analysis of three modes: private vehicle, public transportation, and non- motorized modes ( such as walk and bicycle). The model, estimated using
29
logistic regression, thus comprises two submodels – one for the likelihood of public transportation relative to private vehicle, and one for the likelihood of non- motorized relative to private vehicle. The initial model included all variables and had a relatively high explanatory power with a pseudo R- square of 0.26. The final submodel for public transportation includes size of metropolitan area as well as socio demographic variables such as age, income, number of vehicles, and household size ( Table 26). US- born was not significant in this model, and white race/ ethnicity was only marginally significant. In contrast, US- born and white, black, and Asian race/ ethnicity were significant: white respondents walk more than others, while Asian and black respondents walk less, and respondents born in the US walk less than foreign- born respondents. These results thus suggest that race/ ethnicity and place of birth have more of an effect on the use of non- motorized modes than they do on transit.
30
Table 26: Commute Mode Model* - Full Sample
Term
Estimate
Std Error
ChiSquare
Prob> ChiSq
Odds Ratio
Intercept
- 3.5539
3.5864
0.98
0.3217
.
YTOUS_ L[ 0]
- 4.2436
3.5788
1.41
0.2357
0.0002
race_ W1[ 0]
- 0.1364
0.0779
3.07
0.0799
0.7613
race_ B1[ 0]
0.0790
0.1068
0.55
0.4596
1.1712
race_ A1[ 0]
- 0.0963
0.1400
0.47
0.4913
0.8248
income_ m
0.0000
0.0000
55.44
<. 0001
0.3581
veh per driver_ to1
- 3.4230
0.1156
877.08
<. 0001
0.0326
R_ AGE
- 0.0101
0.0027
14.46
0.0001
0.4780
HHSIZE
- 0.0609
0.0271
5.04
0.0247
0.4528
MSACAT_ 1t[ 0]
- 0.3483
0.0542
41.29
<. 0001
0.4982
MSACAT_ 2n[ 0]
0.1523
0.0582
6.84
0.0089
1.3562
public transportation/ private vehicle
MSACAT_ 3[ 0]
0.1062
0.0467
5.16
0.0231
1.2366
Intercept
1.7927
0.1993
80.88
<. 0001
6.0057
YTOUS_ L[ 0]
0.1800
0.0495
13.24
0.0003
1.4334
race_ W1[ 0]
0.2102
0.0587
12.81
0.0003
1.5227
race_ B1[ 0]
- 0.2462
0.0685
12.93
0.0003
0.6112
race_ A1[ 0]
- 0.2862
0.0782
13.39
0.0003
0.5641
income_ m
0.0000
0.0000
13.01
0.0003
1.5385
veh per driver
- 4.0302
0.1006
1605.3
0.0000
0.0178
R_ AGE
- 0.0038
0.0025
2.35
0.1253
0.7564
HHSIZE
- 0.1676
0.0245
46.76
<. 0001
0.1132
MSACAT_ 1t[ 0]
- 1.1465
0.0578
392.92
<. 0001
0.1010
MSACAT_ 2n[ 0]
- 0.4384
0.0621
49.8
<. 0001
0.4161
walk/ private vehicle
MSACAT_ 3[ 0]
0.1124
0.0671
2.8
0.0940
1.2520
For log odds of ( public transportation/ private vehicle), ( walk/ private vehicle)
31
6.4 Miles Driven per Year for Full Sample
The final model examines the association of miles driven per year with race/ ethnicity and immigration status, after accounting for socio- demographic variables and location variables ( Table 27). The initial model was based on 54260 observations and had an adjusted R- square of 0.092. The final model, presented in Table 30, includes six variables and has a similar explanatory power with an adjusted R- square of 0.090.
The scaled estimates for the variables help to show the relative influence of each variable on miles driven per year. As expected, the number of vehicles per driver has the largest effect, but gender, age, and household income also have substantial effects. US- born respondents drive 339 miles more per year then foreign- born respondents, and recent immigrants drive 580 miles less per year than others. Thus, immigrants in general drive less but the influence is more significant for recent arrivals. Among race/ ethnicity groups, black respondents drove 1700 fewer miles per year than others.
Table 27: Linear Model of Yearly Mile Driven
Term
Scaled Estimate*
Estimate
Std Error
t Ratio
Prob>| t|
Intercept
8954
1781.88
391.34
4.55
<. 0001
YTOUS_ L[ 0]
- 338.77
- 338.77
95.91
- 3.53
0.0004
YTOUS_ 5[ 0]
580.42
580.42
274.54
2.11
0.0345
MSACAT_ 1t[ 0]
593.38
593.38
73.39
8.08
<. 0001
MSACAT_ 2n[ 0]
223.00
223.00
64.67
3.45
0.0006
MSACAT_ 3[ 0]
346.25
346.25
56.19
6.16
<. 0001
race_ A1[ 0]
1710.88
1710.88
124.12
13.78
<. 0001
income_ m
1908.05
0.03
0.01
26.38
<. 0001
R_ AGE
- 2514.90
- 52.39
2.58
- 20.27
<. 0001
R_ SEX[ 1]
2081.95
2081.95
42.15
49.39
0.0000
HHSIZE
533.71
82.11
35.41
2.32
0.0204
veh per driver_ to1
5801.35
7735.14
245.71
31.48
<. 0001
* Nominal factors expanded to all levels
* Continuous factors centered by mean, scaled by range/ 2
32
7. Conclusions
The analysis presented here shows that recent immigrants have different patterns of travel than people born in the US and than immigrants who have lived in the US for longer periods of time. Travel patterns also differ for immigrants by race/ ethnicity and by place of birth. The descriptive analysis revealed significant differences in income level, household lifecycle stage, and age for immigrant groups living in the US for different periods of time and for different race/ ethnicity groups. These socio- demographic variables may in part explain differences in travel behavior. However, multivariate analyses show that immigrant status, race/ ethnicity, and place of birth are associated with certain aspects of travel behavior even after accounting for these socio- demographic factors.
Although the evidence for associations between travel behavior and immigrant status as well as both race/ ethnicity and place of birth is strong, the evidence for a causal relationship is not. It is hard to come up with a plausible explanation for how or why race/ ethnicity or immigrant status itself would influence travel behavior. Rather, these variables are likely associated with factors such as needs, limitations, preferences, attitudes, culture, and prior experiences that have some influence on travel behavior. Understanding the factors that explain the observed differences in travel behavior requires further research, and both qualitative and quantitative methods may be helpful. This understanding can help in modeling travel demand, finding policies best suited to meeting the travel needs of foreign born communities, and addressing environmental justice concerns.
33
8. References
Bartel, A. P. ( 1989). Where Do the New United- States Immigrants Live. Journal of Labor Economics 7 ( 4): 371- 391.
California Department of Finance ( 2004). Census 2000 California Profile. Sacramento, CA: State of California.
Deakin, E., C. Ferrell, J. Mason, and J. Thomas. ( 2002). Policies and practices for cost- effective transit investments - Recent experiences in the United States. In Transit: Planning and Development, Management and Performance, Marketing Fare Policy. Casas, J., C. Arce, and C. Frye ( 2004). Latino Immigration and Its Impact on Future Travel Behavior. Available: http:// trb. org/ conferences/ nhts/ Casas. pdf
McGuckin, N. and N. Srinivasan ( 2003). “ National Summary,” Journey to Work Trends in the United States and its Major Metropolitan Areas 1960 – 2000. Publication No. FHWA - EP- 03- 058.
Myers D. ( 1996). “ Changes over Time in Transportation Mode for Journey to Work: Effects of Aging and Immigration,” Decennial Census Data for Transportation Planning: Case Studies and Strategies for 2000, Volume 2: Case Studies. Washington, DC: Transportation Research Board.
Myers, Dowell. ( 1997). " Changes Over Time in Transportation Mode for Journey to Work: Effects of Aging and Immigration." Decennial Census Data for Transportation Planning: Case Studies and Strategies for 2000 Volume 2: Case Studies ( Irvine, California. Washington, D. C.: National Academy Press).
Pamuk, A. ( 2004), “ Geography of immigrant Clusters in Global Cities: A case study of San Francisco, 2000” International Journal of Urban and Regional Research, 28( 2): 287- 307.
Park, R. E., E. W. Burgess and R. D. McKenzie ( eds.) ( 1925) the city. The University of Chicago Press, Chicago.
Pendall Rolf ( 2001) “ Exploring Connections between Density, Sprawl, and segregation by Race and income in US Metropolitan Areas, 1980- 1990” presented at a Lincoln Institute course titled “ International Seminar on Segregation in the City” held from July 26- 28, 2001.
US Census Bureau, 2005 Population Size and Composition 2003 Available: http:// www. census. gov/ population/ www/ socdemo/ foreign/ slideshow/ fb2003/ TextOnly/ Slide2. html
Portes, A. ( ed.) ( 1995) The economic sociology of immigration: essays on networks, ethnicity, and entrepreneurship.
34
Purvis C. ( 2003). Commuting Patterns of Immigrants. Washington, DC: US Department of Transportation, Federal Highway Administration, Bureau of Transportation Statistics, Federal Transit Administration, CTPP 2000 Status Report.
Valenzuela Jr. A., L. Schweitzer and A. Robles ( 2005) Camionetas: Informal Travel Among Immigrants. Transportation Research A, In Press.
35