Documentos de Académico
Documentos de Profesional
Documentos de Cultura
To cite this Article Mollenkopf, John and Champeny, Ana(2009) 'The Neighbourhood Context for Second-Generation
Education and Labour Market Outcomes in New York', Journal of Ethnic and Migration Studies, 35: 7, 1181 — 1199
To link to this Article: DOI: 10.1080/13691830903006283
URL: http://dx.doi.org/10.1080/13691830903006283
This article may be used for research, teaching and private study purposes. Any substantial or
systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or
distribution in any form to anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses
should be independently verified with primary sources. The publisher shall not be liable for any loss,
actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly
or indirectly in connection with or arising out of the use of this material.
Journal of Ethnic and Migration Studies
Vol. 35, No. 7, August 2009, pp. 11811199
Over the last five decades, immigration has profoundly transformed the population of
metropolitan New York, long divided by race and class. The almost-forgotten ‘underclass’
debate established that New York was the nation’s capital of concentrated poverty, which
grew significantly worse during the 1970s and 1980s. Though more recent data show that
New York has achieved a remarkable turnaround since 1990, most probably associated
with immigration, it remains a city of economic extremes and stubbornly high poverty.
Concern about where new immigrants*and their children*might fit into this matrix of
urban inequality led several leading social scientists to hypothesise that some members of
the second generation would be downwardly mobile. To investigate this possibility, in
1999 and 2000, the Immigrant Second Generation in Metropolitan New York (ISGMNY)
surveyed 3,415 young people aged 18 to 32 years, from five immigrant and three native-
born racial and ethnic backgrounds, about their life trajectories. This paper conducts an
analysis of the contextual effects of the neighbourhoods in which respondents grew up on
their later experiences in terms of educational attainment and labour market success.
Using OLS and HLM modelling, we find small but consistent and theoretically
interesting effects. In particular, growing up in a poor neighbourhood has a negative
effect on later outcomes, while growing up in a black neighbourhood does not, once
poverty is taken into account.
John Mollenkopf is Distinguished Professor of Political Science and Sociology at The Graduate Center, City
University of New York, and director of its Center for Urban Research. Ana Champeny is the supervising analyst
for the Housing, Environment and Infrastructure team of the Independent Budget Office of the City of New
York and was previously a Research Associate at the Center for Urban Research. Correspondence to: Prof. J.
Mollenkopf, Center for Urban Research, CUNY Graduate Center, 365 Fifth Avenue, New York NY 1001, USA.
E-mail: jmollenkopf@gc.cuny.edu.
American citizens, Puerto Ricans, make up 9 per cent of the city’s population and 36
per cent of its Hispanic population. Foreign-born people thus make up less of the
city’s Hispanic population than in LA or other immigrant gateways. Nevertheless,
New York has attracted many people from the Dominican Republic, Ecuador, Mexico,
Colombia and other parts of Latin America. Moreover, while its large Puerto Rican
population holds US citizenship, it has also undergone a migratory process. New York
and the surrounding area contain large numbers of Chinese, Korean and Indian
immigrants as well.
New York stands out because it has also attracted many white and black
immigrants. Indeed, New York currently has almost as many white immigrants as
Hispanic ones. It received many Europeans displaced by the Holocaust and World
War II, some of whom are still living, and also attracted many emigrants from the
former Soviet Union, mostly classified as Jewish, before and after the fall of the old
regime. Finally, New York receives small but continuing flows from Italy, Ireland,
Poland and now the former Yugoslavia and Albania. Los Angeles, by contrast,
attracted large populations of Iranians and Armenians.
New York is also noteworthy for attracting black migrants from the Caribbean, a
distinction it shares with Miami and Washington. Many of its Hispanic Caribbean
migrants have significant African ancestry. The largest number, however, are
immigrants from the Anglophone West Indies and from Haiti. They have a long
history in New York City and a significant part of the city’s black population has
Caribbean roots. In earlier periods, the West Indian population tended to blend into
the larger African American one, but it has become a culturally and politically distinct
community in recent decades (Kasinitz 1992). The city’s African immigrant
population, while small relative to other immigrant groups, consistently exceeds
those of other immigrant-receiving areas. The distinction between natives and
immigrants thus cuts across each basic racial category in New York City, while
differentiating native whites and blacks from immigrant Hispanics and Asians
elsewhere in the country.
Journal of Ethnic and Migration Studies 1183
New York has a persistently high level of blackwhite segregation, surpassed only
by Detroit and a few other old industrial cities (Logan et al. 2004). The now-distant
‘underclass’ debate identified New York as the nation’s capital of concentrated
poverty, a phenomenon that worsened during the 1970s and 1980s. While more
recent data show that the city achieved a remarkable turnaround, most likely
associated with immigration (Jargowsky 1997; Jargowsky and Yang 2005), it remains
a city of great economic extremes and a stubbornly high poverty rate (New York City
Commission on Economic Opportunity 2006). While the city has rebounded
tremendously in both demographic and economic terms from its low point in the
mid-1970s, and the crime rate has fallen sharply since the early 1990s, it is still a city
in which neighbourhood contexts can pose sharp risks for young people. This was
particularly true in the late 1980s and early 1990s, when those who are now young
adults were growing up.
It is reasonable to expect that exposure to the high levels of crime and social
Downloaded By: [HEAL-Link Consortium] At: 22:15 9 November 2010
and ethnic groups in metropolitan New York. Because we are particularly interested
in the possibility that local social context and neighbourhood conditions might affect
their trajectories, we also asked people to identify where they had grown up, obtained
the names of the high schools and colleges they attended and identified their current
address. For those who grew up in New York City, we were able to geocode the
neighbourhoods in which they lived the longest between the ages of 6 and 18 by Zip
Code (and Public Use Microdata Areas or PUMAs). We also geocoded the 1990 and
2000 Census Tracts and PUMAs for respondents’ current places of residence.
their native-born counterparts. Among those who grew up in and around New York,
the average educational attainment among the Russian and Chinese respondents
approached that of native whites, while the West Indian average exceeded that of
African Americans and the South American and Dominican averages exceed those of
Puerto Ricans (as well as African Americans). The native-born minority groups*
African Americans and Puerto Ricans*seem to suffer the most from continuing
disadvantage. Dominicans are faring the least well of the second-generation groups,
though they are doing marginally better than Puerto Ricans and African Americans.
This pattern grows stronger after controlling for age, sex, family form growing up and
parental levels of education (Kasinitz et al. 2008). For example, many Chinese
youngsters have poorly educated parents, but they obtained more education than
counterpart native whites from poorly educated families. Such outcomes seem to be
somewhat contrary to what might have been predicted by the segmented assimilation
model and have led us to develop a perspective we call ‘second-generation advantage.’
Labour market outcomes follow similar patterns, although our relatively young
respondents are just beginning their careers. Labour force participation is fairly high
across all the groups, generally a bit higher among men than women, and is lowest
among African Americans and Puerto Ricans. Working is less prevalent than might be
expected among the Chinese, Russians and West Indians because many are still
pursuing full-time studies at age 21 or above. Of the second-generation groups,
Dominicans once more fare the least well and resemble the Puerto Ricans in many
respects, although they are marginally less likely to be unemployed and more likely to
be in school. Although less likely to work, the median woman in every group is
typically earning at least as much as the median male. Mean wages are consistently
higher than median wages, however, and men tend to have higher mean earnings
than the women in their group. Males tend to hold more of the higher-paying jobs,
skewing their averages upward. Finally, the native whites, Russians and Chinese earn
significantly more than the West Indians, South Americans and African Americans,
who in turn earn more than the Dominicans, while Puerto Ricans earn the least.
1186 J. Mollenkopf & A. Champeny
Neighbourhood Contexts
To test whether the neighbourhood in which they grew up has shaped the later life
chances of our respondents, we needed to know where they lived the longest as a
child so that we could match that location with neighbourhood-level indicators from
roughly ten years prior to when we interviewed them in 2000. Since we could
establish a relatively exact 1990 geographic location only for those who grew up in
New York City, and some 1990 contextual data was only available for the city as well,
we restrict our analysis to the 2,790 respondents who grew up in the city for whom
we could identify a specific Zip Code of residence for the neighbourhood in which
they lived the longest between the ages of 6 and 18 (81.7 per cent of the total sample).
Since most of those who came from the suburbs of or from outside New York grew
up in middle-class areas, this sample retains most of those who may be at risk from
exposure to negative neighbourhood effects. At the same time, neighbourhood
Downloaded By: [HEAL-Link Consortium] At: 22:15 9 November 2010
circumstances within New York City vary sufficiently to provide a good basis for
analysis. Knowing the Zip Code of residence allowed us to draw neighbourhood
context indicators from the 1990 SF3 Zip Code file of the US Census, from the 1990
Public Use Microdata File (for PUMAs) and from administrative records on crime
and health patterns. In the dense neighbourhoods of New York, the Zip Code is a
more valid geographic proxy for neighbourhood than it might be in less-densely
populated cities and suburbs.
As would be expected in a highly segregated city, members of our eight study groups
tended to grow up in different kinds of places. The Puerto Ricans grew up the most
frequently in the tough, declining neighbourhoods of the South Bronx and Bushwick
and East New York in Brooklyn, but also in better areas like Sunset Park in Brooklyn.
The Dominicans mostly grew up in the denser, poorer neighbourhoods of Washington
Heights in Northern Manhattan and Corona in Queens, but some also grew up in
Bushwick and Sunset Park in Brooklyn. Washington Heights and Bushwick had a great
deal of trouble with drugs and crime in the 1980s. By comparison, the South
Americans were the most likely to grow up in middle-class Queens neighbourhoods
like Jackson Heights, Flushing, Corona and Elmhurst*a third of them grew up in
these four neighbourhoods alone. These neighbourhoods are characterised by single
family dwellings and small apartment buildings and have many Asian as well as
Hispanic residents.
African Americans were the most likely to grow up in Bedford Stuyvesant and
Brownsville in Brooklyn, West Harlem, Morissania in the South Bronx and South-
East Queens, areas which are overwhelmingly black and, with the exception of South-
East Queens, relatively poor. The West Indian young people grew up in the heavily
immigrant working-class areas of Flatbush, East Flatbush and Crown Heights in
Central Brooklyn, which had formerly been Jewish neighbourhoods. Some also grew
up in African American neighbourhoods like Bedford Stuyvesant. These neighbour-
hoods have some tough sections, but also have middle-class housing. There was more
Journal of Ethnic and Migration Studies 1187
blending of West Indians and African Americans across neighbourhoods than was
true among the Hispanic groups or between them and the black groups.
The Chinese grew up in Manhattan’s Chinatown and the Lower East Side,
Flushing, Queens and Sunset Park, Brooklyn. Flushing and Sunset Park tend to be
better off than the dense first-generation immigrant settlement in Chinatown. The
Russian Jewish respondents mainly grew up in the middle-class neighbourhoods
of South Brooklyn, including Borough Park, Bensonhurst, Brighton Beach and
Sheepshead Bay. The Chinese and Russians had less proximity to blacks than did the
Hispanic groups, but many Chinese settled near Hispanic immigrants. Finally, the
native-white New Yorkers were reared in middle-class neighbourhoods in Flushing,
Queens, Bensonhurst and Midwood in Brooklyn, and Manhattan’s Upper West Side.
Despite these selective settlement patterns, however, members of each group grew up
in a variety of neighbourhood conditions.
Table 1 uses 1990 Census Zip Code data to summarise the demographic
Downloaded By: [HEAL-Link Consortium] At: 22:15 9 November 2010
characteristics for the median respondent in each study group. Several patterns are
noteworthy from the perspective of segmented assimilation. First, the median child of
Russian, Chinese and South American immigrants grew up in neighbourhoods where
whites formed the majority and blacks had only a small presence; Hispanics were
generally the most prevalent group after native whites. Second, the median child of
West Indians had the highest contact with blacks (though a lower exposure to
poverty), while those of Dominicans and Puerto Ricans had the highest exposure to
poverty (and less exposure to blacks, though still more than other groups). Finally,
since the median child of African American parents grew up within segregated black
neighbourhoods characterised by high rates of poverty, poorly educated adults and
many high-school dropouts, the segmented assimilation hypothesis would predict
that they should suffer the most from negative neighbourhood contexts.
1188
J. Mollenkopf & A. Champeny
Table 1. Demographics (1990) of neighbourhood growing up by group (percentage in zip code of median respondent except for household
income)
Pop. 25 Pop. 25 1619 High- Median
Non-Hispanic Non-Hispanic Non-Hispanic Post-1965 no H-S with BA or School household
Group white black Asian Hispanic foreign-born diploma more dropouts income Poverty
CEP 33.5 7.9 10.9 32.3 32.5 32.7 18.9 12.2 $29,490 13.5
DR 14.7 24.1 2.8 50.7 30.2 45.0 11.9 16.4 $21,991 28.7
PR 13.1 27.5 2.4 43.5 21.2 45.0 11.2 16.6 $21,270 28.7
WI 5.9 63.1 1.6 13.5 24.0 32.5 13.2 11.3 $26,712 18.6
NB 4.6 55.7 1.1 17.1 20.1 40.2 11.2 16.1 $20,609 28.7
CHI 45.4 5.4 18.0 20.4 35.8 37.5 19.8 10.6 $28,836 16.2
RJ 77.0 2.3 9.3 8.5 25.1 28.1 20.6 11.2 $27,234 16.9
NW 77.6 5.0 6.4 12.4 17.2 28.3 18.9 10.4 $33,913 11.9
Notes: CEPColombia, Ecuador, Peru; DRDominican Republic; PRPuerto Rico; WI Anglophone West Indies; NB Native Blacks; CHIChinese; RJRussian Jews;
NWNative Whites.
Source: 1990 Census SF3 by zip code.
Journal of Ethnic and Migration Studies 1189
homework done), whether he or she had been arrested or had become a parent,
and age and gender, to explain 28 per cent of the overall variation in educational
outcomes (see the left-hand panel of Table 2.) It is comforting to know that
individual choices count, although it is much more important not to get arrested
than to do more homework. Age is obviously a strong positive factor, matched by the
aggregate of an individual’s parents’ education (which also indicates family form,
since the education of an absent parent is coded as a zero contribution to their
combined education level). Women do consistently better than men. Once we control
for these individual and family characteristics, only Puerto Ricans and African
Americans seem to fare worse*though not statistically significantly so*than the
excluded group, the native whites, while the members of the second-generation
groups all seem to be doing statistically significantly better, especially the Chinese.
Our initial effort to include the racial composition and poverty level of the
neighbourhood in which the respondents grew up did not produce significant
Downloaded By: [HEAL-Link Consortium] At: 22:15 9 November 2010
coefficients for these variables (results not shown). By themselves, the per cent black
and per cent Hispanic of the Zip Code both have negative, but non-significant
relationships with educational attainment. Neighbourhood poverty has a larger
negative relationship and adding it turns the coefficients for black and Hispanic share
into positive coefficients, but the results are still not significant (not shown). The
situation did not improve when we add the Zip Code-share of post-1965 immigrants,
which also has a negative, but insignificant, relationship to educational attainment.
It is substantively worth noting, however, that the most obvious measures of
Table 2. Multivariate models (OLS) of educational attainment
Beta Sig Beta Sig.
While these results are highly preliminary, they have some interesting implications
for the segmented assimilation hypothesis. The negative impact of the prevalence of
adult male joblessness in the neighbourhood where a respondent grew up certainly
supports this particular extension of Wilson’s argument to segmented assimilation
theory. However, the fact that growing up in poverty and in neighbourhoods with
larger populations of minority groups does not lead to significantly lower levels of
educational attainment raises doubts about segmented assimilation theory’s conten-
tion that growing up in poor, native-minority neighbourhoods will push second-
generation youngsters toward oppositional and dysfunctional behaviour. Moreover,
growing up among recent immigrants does not seem to offer much protection, while
a milieu in which children are shifting to English more rapidly than their parents
seems to help them rather than hurt them, as predicted by the segmented assimilation
theory.
Each variance component’s share of the total variance represents how much variation
may be due to individual or to neighbourhood effects. Most of the variance (92 per
cent) is between individuals within neighbourhoods, with the remaining 8 per cent
being between neighbourhoods. This result supports the findings of the OLS
regression that the contextual effects are less significant than the individual and
family characteristics, but it nonetheless confirms that context does, in fact, count.
The model in Table 4 is a random-intercept regression model with Level 1 variables
to predict individual educational outcomes within neighbourhoods. The model is
thus-named because only the Level 2 equation for the intercept has an error term.
This model assumes that the Level 1 (individual) effects are the same across all
neighbourhoods. For example, if we find that being female has a positive relationship
with getting more education, the model assumes that this positive relationship is the
same in all the neighbourhoods. Additionally, this model does not yet test specific
neighbourhood effects because we have not added any neighbourhood characteristics.
In this model, the predicted educational outcome for a native-white, male, single
respondent of average age (within the neighbourhood), from a two-parent family,
who has not moved during his childhood, has not had a child and has not been
Table 3. Random effects (no predictors) Level 1 dependent variable: educational
outcome
Level 1 model: Yij B0jrij
Intercept B0
Intercept G0 0 6.428043 0.206427 31.14 49 0.000
Dominican Republic B1
Intercept G1 0 0.045559 0.178647 0.255 2737 0.799
Puerto Rican B2
Intercept G2 0 0.619745 0.161556 3.836 2737 0.000
West Indian B3
Intercept G3 0 0.048983 0.220666 0.222 2737 0.825
Native Black B4
Intercept G4 0 0.456037 0.158262 2.882 2737 0.004
Chinese B5
Intercept G5 0 1.462759 0.191052 7.656 2737 0.000
Russian B6
Intercept G6 0 0.599222 0.207258 2.891 2737 0.004
Downloaded By: [HEAL-Link Consortium] At: 22:15 9 November 2010
Age (group-centred) B7
Intercept G7 0 0.196065 0.014566 13.461 2737 0.000
Female B8
Intercept G8 0 0.678371 0.116476 5.824 2737 0.000
Moves between 618 years old B9
Intercept G9 0 0.080871 0.028156 2.872 2737 0.005
No father figure B10
Intercept G10 0 0.636719 0.17297 3.681 2737 0.000
Single-parent household B11
Intercept G11 0 0.438999 0.123908 3.543 2737 0.001
Mother’s education (group-centred) B12
Intercept G12 0 0.450794 0.052011 8.667 2737 0.000
Father’s education (group-centred) B13
Intercept G13 0 0.289102 0.065807 4.393 2737 0.000
Is a parent B14
Intercept G14 0 1.166943 0.16051 7.27 2737 0.000
Is/was a teenage parent B15
Intercept G15 0 0.600898 0.169658 3.542 2737 0.001
Has been arrested B16
Intercept G16 0 0.752398 0.181126 4.154 2737 0.000
Number of siblings grew up with B17
Intercept G17 0 0.085833 0.030014 2.86 2737 0.005
Random effects
Level 1 R 4.61771 2.149
Level 2 U0 0.30142 0.549 182.32 49 0.000
slightly less than half a point. On the other hand, being Chinese lifts educational
attainment by almost 1.5 points. Russians are also doing significantly better than the
native-white baseline, with the coefficient showing an increase of 0.6 points.
It was important to control for age, as many of our survey respondents were still of
school age. Age was group-centred, with each year of age above the average of
respondents in the neighbourhood being associated with a .19 increase in educational
attainment. Being female has a fairly strong positive effect on educational attainment,
with a coefficient of 0.68. Mobility during childhood, measured as the number of
moves between the ages of 6 and 18, is negatively associated with educational
outcome. Growing up in a single-parent household is also negatively associated with
it, but not having a father figure paradoxically has a positive coefficient of 0.64. The
number of siblings a respondent grew up with has a coefficient of 0.09.
Centred parental educational attainment also helps us to capture the effect of
having more- or less-educated parents than the average mother or father. Higher
Downloaded By: [HEAL-Link Consortium] At: 22:15 9 November 2010
educational attainment among either the father or the mother is positively associated
with the respondent’s educational attainment. The effect is stronger for increases in
maternal education (0.45 compared to 0.29). Since father’s educational attainment
was centred and not having a father was given the lowest value, it appears that the
coefficient for no father figure might be positive partly to offset the negative effect of
this variable. Having had a child, however, is negatively associated with educational
attainment, with a coefficient of 1.17. Having become a parent as a teenager has a
further negative effect on attainment of 0.60. Ever having been arrested is also
negatively associated with educational outcome.
These coefficients all closely parallel those of the individual-level model presented
in the previous section. It is noteworthy that being West Indian or Dominican does
not carry a statistically significant penalty relative to being a white native, while being
African American or Puerto Rican does. Being Chinese or Russian brings a benefit.
The interaction between single-parent households, lacking father figures and paternal
education also appears complex. We can compare the Level 1 variance in this model
with that in the random-effects model to estimate the proportion of the variance we
have explained. The Level 1 variance decreases from 6.20043 to 4.61771, a decline of
26 per cent from that in the model with no predictors (see bottom of Table 4). Again,
this finding is similar to the R2 value for the OLS regression in the previous section.
We then use the random-intercept model described above as a base from which to
test whether specific Level 2 neighbourhood characteristics exert a common influence
on all individuals, separate from the effect of the previously modelled individual
characteristics (the next three models discussed are not presented in tables). In this
step, we keep the Level 1 equation and add neighbourhood variables to more fully
specify the model for the Level 1 intercept. It is still a random-intercept model
because the variation is limited to modelling the Level 1 intercept, not the individual
slope coefficients in the Level 1 model. The Level 1 intercept equation contains its
own intercept and additional terms for the Level 2 variables. In other words, the
intercept for Level 1 (the base predicted education outcome) now varies across
1194 J. Mollenkopf & A. Champeny
neighbourhoods as a function of neighbourhood characteristics, while the individual
effects remain the same in all the neighbourhoods,
We test the impact of three different neighbourhood features (details of the analysis
not shown in the tables). First, we look at the relationship between the variation in
the neighbourhood’s black population and the variation in educational outcome. In
addition to looking at the non-Hispanic black share of the PUMA, we considered the
level of black/white and Hispanic/black segregation.1 In its most straightforward
form, the segmented assimilation hypothesis views African Americans as having been
subjected to the worst forms of discrimination and social exclusion, and therefore as
having developed the most negative reactions to this restricted opportunity structure.
The hypothesis would seem to predict that growing up among them would therefore
have a negative impact on individual educational outcomes. In fact, however, just as
in the OLS model above, this model shows that people growing up in neighbour-
hoods with higher shares of non-Hispanic black residents did not have different
Downloaded By: [HEAL-Link Consortium] At: 22:15 9 November 2010
assault rate declines by about 1 standard deviation (rounded up to 500 assaults). The
models suggest that prevalence of neighbourhood violence, as measured by felony
assault rate, is negatively associated with educational outcomes of youth.
However, testing these factors individually can miss effects that are brought out by
the multivariate model presented in Table 5. We investigated a large set of Level 2
variables chosen to reflect different theories about neighbourhood effects. First, we
used the (uncentred) share of non-Hispanic black residents to investigate the effect of
living in a black neighbourhood. To look at segregation, we tested indices of black to
white and black to Hispanic segregation for each neighbourhood. Since immigration
is a central focus of our larger study, we used the neighbourhood’s post-1965 foreign-
born population in the Level 2 models as well as an index of native- to foreign-born
segregation. The shares of the neighbourhood’s population who were of the
predominant study groups (Dominican, Puerto Rican, West Indian and Chinese)
were also tested. All these indicators were left uncentred.
Downloaded By: [HEAL-Link Consortium] At: 22:15 9 November 2010
Intercept B0
Base G0 0 7.061996 0.406915 17.355 43 0
Share non-Hispanic Black G0 1 0.008346 0.002602 3.208 43 0.003
Share post-1965 foreign-born G0 2 0.018887 0.006214 3.039 43 0.004
Share of 618 years enrolled in school G0 3 0.089729 0.039929 2.247 43 0.03
Felony assault rate (per 100,000) G0 4 0.000767 0.000146 5.267 43 0
Drug-overdose hospital admissions, 1994 G0 5 0.013225 0.004185 3.16 43 0.003
Native-born to foreign-born
segregation index (1100 scale) G0 6 0.022441 0.010384 2.161 43 0.036
Dominican Republic B1
Intercept G1 0 0.112103 0.182927 0.613 2731 0.54
Puerto Rican B2
Intercept G2 0 0.584219 0.153718 3.801 2731 0
West Indian B3
Downloaded By: [HEAL-Link Consortium] At: 22:15 9 November 2010
Conclusion
While this analysis is no more than an initial foray into understanding how growing
up in different neighbourhood contexts might affect the later trajectories of the
young adults we have interviewed in the Immigrant Second Generation in
Metropolitan New York Study, it does support some preliminary conclusions. First
of all, we believe that context does count. Both the OLS regressions bringing in Zip-
Code data and the hierarchical models of the PUMA data indicate that contextual
1198 J. Mollenkopf & A. Champeny
factors can have a clear and significant impact on individual educational outcomes.
These impacts are far weaker than those of individual and family characteristics, just
as we would expect them to be. At the same time, they are meaningful.
Second, living amidst poverty and associated conditions, such as male non-
participation in the labour force, has a more negative impact than simply living in a
black or Hispanic minority neighbourhood. Moreover, growing up in a first-
generation zone of immigrant concentration also seems to have a negative impact on
individual educational outcomes, net of many individual and family characteristics.
Growing up surrounded by youngsters who speak English even when their parents do
not is positively associated with individual educational attainment. This seems to run
counter to the idea posed by Portes and Rumbaut that ‘dissonant acculturation’ will
lead to bad educational outcomes. On the other hand, growing up in an area with
high crime rates clearly has a negative impact*although perhaps not as large as we
might have expected*while the incidence of drug-overdose admissions (admittedly
Downloaded By: [HEAL-Link Consortium] At: 22:15 9 November 2010
Acknowledgements
This paper was originally given at the conference Local Contexts and the Prospects for
the Second Generation held at the West Coast Poverty Research Center, University of
Washington, 1920 October 2006. We are deeply grateful for financial support from
the Russell Sage Foundation, the Rockefeller Foundation, the Ford Foundation, the
Mellon Foundation, the MacArthur Foundation, the United Jewish Appeal-Federa-
tion and the National Institutes of Child Health and Human Development. We also
thank Henry Brady, Gunnar Almgren and Mark Ellis for constructive criticism of a
previous version of this paper.
Journal of Ethnic and Migration Studies 1199
Note
[1] The segregation indices were calculated based on the formula for an index of dissimilarity
(see http://www.geo.hunter.cuny.edu/imiyares/Segregation.htm). The value for each
neighbourhood (PUMA) was based on the census tracts contained within that PUMA.
References
Bourgois, P. (1995) In Search of Respect: Selling Crack in El Barrio. New York: Cambridge University
Press.
Bryk, A. and Raudenbush, S. (1992) Hierarchical Linear Models: Applications and Data Analysis
Methods. Newbury Park: Sage Publications.
Gans, H. (1992) ‘Second generation decline: scenarios for the economic and ethnic futures of the
post-1965 American immigrants’, Ethnic and Racial Studies, 15(2):17393.
Industrial Areas FoundationMetro New York and Public Education Association (1997) Futures
Denied: Concentrated Failure in the New York City Public School System. New York: IAFMNY
Downloaded By: [HEAL-Link Consortium] At: 22:15 9 November 2010
and PEA.
Jargowsky, P. (1997) Poverty and Place: Ghettos, Barrios, and the American City. New York: Russell
Sage Foundation.
Jargowsky, P. and Yang, R. (2005) ‘The ‘‘underclass’’ revisited: a social problem in decline’.
Washington, DC: Brookings Institution Working Paper, May.
Kasinitz, P. (1992) Caribbean New York: Black Immigrants and the Politics of Race. Ithaca, NY:
Cornell University Press.
Kasinitz, P., Mollenkopf, J. and Waters, M. (eds) (2004) Becoming New Yorkers: Ethnographies of the
New Second Generation. New York: Russell Sage Foundation.
Kasinitz, P., Mollenkopf, J., Waters, M. and Holdaway, J. (2008) Inheriting the City: The Children
of Immigrants Come of Age. New York: Harvard University Press and the Russell Sage
Foundation.
Logan, J., Stults, J. and Farley, R. (2004) ‘Segregation of minorities in the metropolis: two decades of
change’, Demography, 41(1): 122.
New York City Commission on Economic Opportunity (2006) Increasing Opportunity and Reducing
Poverty in New York City. New York: Office of the Mayor.
Portes, A. and Rumbaut, R. (2001) Legacies: The Story of the Immigrant Second Generation. Berkeley
and Los Angeles: University of California Press; New York: Russell Sage Foundation.
Portes, A. and Zhou, M. (1993) ‘The new second generation: segmented assimilation and its
variants among post-1965 immigrant youth’, Annals of the American Academy of Political and
Social Science, 530: 7496.
Raudenbush, S., Bryk, A, and Congdon, R. (2000) HLM 6: Hierarchical Linear and Nonlinear
Modeling (Version 6.02a). Lincolnwood, IL: Scientific Software International, Inc.
Sampson, R., Raudenbush, S. and Earls, F. (1997) ‘Neighbourhoods and violent crime: a multilevel
study of collective efficacy’, Science, 277: 918924.
Sampson, R., Morenoff, J. D. and Gannon-Rowley, T. (2002) ‘Assessing ‘‘neighbourhood effects’’:
social processes and new directions in research’, Annual Review of Sociology, 28: 44378.
Wallace, D. and Wallace, R. (1998) A Plague on Your Houses: How New York was Burned Down and
National Public Health Crumbled. New York: Verso.
Williams, T. (1989) The Cocaine Kids. New York: Addison-Wesley.
Wilson, W. (1996) When Work Disappears: The World of the New Urban Poor. New York: Random
House.