Superdiversity

Diversification is one of the key social processes that defines our times. Over the past few decades, multiple causes and categories of migration – combined with migrants’ new and varying origins – have been transforming urban populations in complex ways, worldwide. The following graphics show us how.

Across the globe, most international migrants end up in cities. Urban populations today show fluctuating combinations of nationality, ethnicity, language, religion, age, gender, legal status, class and human capital. Such changing combinations of traits lead to conditions of ‘super-diversity’ – a concept highlighting the fact that current diversity patterns significantly supersede earlier ones.

An important feature of contemporary super-diversification is that new migrants tend to inhabit places where migrants from previous waves and generations still live. The characteristics of new migrants, therefore, are ‘layered’ on top of earlier kinds of diversity. Further, super-diversification also entails growing differences among people within ethnic, religious and national categories: although they share the same ethnic or religious label, their socio-economic and other characteristics are very different. In all these ways, urban populations are becoming ever more complex.

Super-diversification does not inherently present a threat to social cohesion. A range of research shows that people are creating and practicing new, crosscutting contacts and relationships in diversifying cities every day. However, super-diversification does present analysts, policymakers and the public with challenges to understand the complex ways that urban societies are changing. New methods for correlating complex data are needed, just as – if not more significantly – new ways of seeing emergent patterns in complex data. For these reasons and purposes, we have created an array of new, interactive data visualization tools for readily observing patterns of urban super-diversification.

Keynote Lecture, 2018 International Metropolis Conference, Sydney

Superdiversity in Metropolitan New York

Ethnoracial Transformations in The Quintessential Immigrant City

By Philip Kasinitz, Van Tran and Kasey Zapatka

New York has long been regarded as the quintessential immigrant city. Today close to a third of its residents and nearly half of its labor force were born outside of the United States and the majority of New Yorkers are immigrants or the children of immigrants. New York also stands out for its diversity. No single nation or sending region makes up the lion’s share of today’s immigrants. They arrive from dozens of countries speaking hundreds of languages. Nor do the immigrants easily map onto the usual North American racial categories. Most non-Hispanic Asian and Latinx New Yorkers are either immigrants or the children of immigrants, but so is about half of the non-Hispanic Black population, and a considerable minority of non-Hispanic Whites.

The newcomers are also diverse in terms of class and educational backgrounds with immigrants over-represented among both the most and least well-educated New Yorkers. They also vary greatly in terms of legal status and mode of entry. They include holders of temporary visas, legal permanent residents, naturalized US citizens as well as undocumented immigrants. They also include refugees and those hoping to be granted asylum, migrants with temporary protected status and “winners” of the “green card” lottery. Many live in mixed-status households which include legal and unauthorized immigrants as well as both naturalized and birthright citizens.

Other demographic factors make New York unique among American cities. Most obvious is its sheer size. The City of New York (i.e., the five boroughs) is the nation’s largest; considerably larger than the next two (Los Angeles and Chicago) combined. Its population density is by far the highest of any major American city and is exceeded only by several small suburban municipalities in the New York metropolitan area. Not surprisingly it is far more dependent on mass transit than other US or Canadian cities.

On this website, we concentrate on the NY-NJ-PA metropolitan region. This area powerfully demonstrates processes of urban super-diversification. The nature of this metropolitan area has been fundamentally reshaped by immigration since the mid 1960’s. The flood of new immigrants has continued to repopulate the metropolitan area, anchor its economy, and buoy its reputation as a diverse, global hub. Although the COVID-19 pandemic has halted immigration into the city since March 2020, the demographic patterns we describe here are still broadly representative of the city’s ethnoracial makeup.

On this data visualization tool, we begin by presenting an interactive chart that invites you to explore US migration patterns at the national level. Next, you can then examine the connections between ancestry and linguistic identity, and in another visualization, consider the relationship between ethnoracial diversity and socio-economic outcomes for the metropolitan area’s working-age (18-64) and working-age recent immigrant population. We investigate the neighbourhood scale of diversity by mapping social-demographic geographies and superdiversity “hot spots.” Finally, we use a dashboard tool to interrogate how key demographic characteristics intersect with socio-economic status.

For relevant methodological details, refer to the technical report here

Scale: National

Using data from the Yearbook of Immigration Statistics—published by the Department of Homeland Security—between 1998 and 2019, we explore immigration over time by region and sending country, for both temporary and permanent immigration categories. Overall, permanent migration has grown incrementally over the last two decades. By contrast, the size of temporary migration has exploded and tripled over the same period.

For permanent migrants, there are five categories based on their visa status at the time at entry: family-based, work-based, diversity visas, refugees or asylees, and a residual category (i.e., other). In the interactive graphic below, click on a particular year and one of the five types of visa categories to visualize a breakdown of the country of origin for that admission group.

Try to see these overall patterns and trends in the graphic: the different origins of people admitted into the US by visa categories, the changing nature of migration flows, and the diversification of sending countries with increasingly smaller number of immigrants coming from a larger number of countries (all the smaller rectangles).

In particular, note the dramatic decrease in all forms of permanent immigration in 2003 and the dramatic increase in Cuban immigration post-2016 after the end of “the wet-foot, dry-foot policy”, which allowed only Cubans who made it to the US soil to stay as refugees while turning away Cubans who were caught at sea.

Permanent migration to United States by category, 1998-2019

National origin of family immigrants in 2012

For family-based migration, Latin America is the largest sending region, followed by Asia. For work-based migration, the majority is from Asia, followed by Latin America and Europe. Among those who arrived on a diversity visa, we see the opposite pattern, with Africa and Europe making up the largest shares. Among refugees and asylees, Cuba is the single largest sending country of origin, but the share among non-Cuban arrivals has significantly shifted over time from predominantly European origins to mostly Asian origins.

The next visualization provides similar information, but this time for temporary non-immigrant admission to the US. This includes individuals granted temporary status because of a student visa, a temporary worker visa, or for diplomatic reasons. We decided not to include business and tourism visas because the sheer number—especially from specific countries—would overwhelm data visualization. Additionally, we created a residual category that included the children and spouses of foreign government officials, the minor children of fiancées, and individuals whose admission status was unknown. Unfortunately, data suppressed to protect confidentiality creates some minor discrepancies in country-level totals. Also note that the majority of short-term admissions from Canada and Mexico are excluded by the DHS and therefore from our calculations.

Temporary migration to United States by category, 1998-2019

National origin of students immigrants in 2012

Again, note the diversity of source countries of individuals arriving under these temporary categories. In particular, there was a huge growth in the share of temporary workers from American counties starting in 2010, primarily Canada and Mexico, alongside the shrinking shares from Asia and Europe. Among student visa entries, Asians comprise the growing majority over time, with Europeans accounting for a shrinking share and Africans being the least likely sending region across the two decades.

Sankey Diagram

This first metropolitan-scale interaction illustrates the growing demographic complexity in the relationship between ancestry identities and the language spoken in home. The lines connecting the two sides show the number of people with certain combinations of ancestral and linguistic identity. Hovering over the graphic will focus on a particular identity, which you can magnify with a click.

This pair of elements below illustrates this growing degree of complexity for NY-NJ-PA metropolitan area. The 2000 graphic shows the links between ethnicity and the language spoken at home for residents in the metropolitan area at that time. The left-hand labels indicate ethnic identity while the right-hand labels indicate linguistic affiliation. The connecting lines reveal which combinations exist in the population and indicate by line thickness the relative number of people in each ancestry-language combination. The same information is shown for 2018, with data from the 2014-2018 ACS 5-year estimates.

Toggling back and forth between years, you can see the growing complexity of these connections. Also note the mixture of thick and very thin lines – indicating that there are some common combinations of ethnic ancestry and language, as well as a very large number of these combinations that include only a small number of people. To gain a better sense of what this means, hover your mouse over the Spanish language category. You will see that there are many ethnic ancestry origins among those speaking Spanish, for example.

Overall, there is significant linguistic assimilation with many groups with diverse ancestry speaking the English language. Beyond English and Spanish, there is a relatively large group of small diverse other languages. Specifically, the dominance of English has declined slightly from 2000 to 2018, alongside the significant growth of Spanish as the second most common language spoken at home for individuals from many ethnic ancestries. This is tied to the changing sending regions of origin for immigrants. The use of Hindi and Chinese also grew over time, while the use of German, Irish and Italian shrunk.

This chart uses bubbles and relative proportions to explore the relationship between ethnoracial diversity and socio-economic outcomes. Specifically, each “bubble” reports the size of a particular ethnoracial group (the white circle), and the proportion of the group with that socio-economic attribute (the inner colored circle.) These data are visualized for two population groups: working-age population and working-age recent immigrants (i.e., those who arrived within the last five years). A technical note about definitions. To present a clearer picture of employment inequality across groups, we use total population as the denominator instead of total labor force population to calculate employment rates. We use “Working” instead of “Employed” to denote this methodological point. Also, we define low-income individuals as members of a family that earn collectively less than 150% of the poverty line, adjusted for family size.

For the NY-NJ-PA metropolitan area, we created ethnoracial groups and explored the relationship with specific outcomes. Analyzing race and ethnicity using the US census is complicated. The US census allows respondents to self-report and check more than one box when responding to questions about race. Further, ethnicity is asked as a completely separate question. One approach to simplify analyses is to combine both race and ethnicity into a single variable that is both mutually exclusive and exhaustive. This approach records all Hispanic responses as one category; reports those who identify only as White, Black, and Asian in three separate categories; and marks all other or multiple responses as a non-Hispanic Other category. This results in five mutually exclusive and exhaustive “racial” categories (non-Hispanic White, non-Hispanic Black, non-Hispanic Asian, non-Hispanic Other, and Hispanic). We also explore the outcomes for the top ten national origin groups in metropolitan New York in 2018. We classify individuals who responded to the ancestry questions by specifying one of these ten ethnic ancestries in either their first or second response. In descending order by group size, these ten ethnic groups are Dominican, Puerto Rican, Chinese, Mexican, Indian, Ecuadorian, Jamaican, Colombian, Haitian and Filipino.

The chart can be read vertically or horizontally. Looking up or down a column reports an outcome for a particular ethnoracial group. For example, the top-left-hand circle shows that the non-Hispanic White population in metropolitan New York is quite large, and that approximately a very high proportion of people in this group is working. The circles below this one indicate the proportion of the group that has a university degree, experiences low income, or lives in a home that they own. Generally speaking, we would expect higher educational attainment to be associated with higher employment and home ownership but with lower rates of low income. Scanning across a row compares outcomes across ethnoracial groups. So, for example, the likelihood of working is universally high across the ten ethnic groups. Chinese and Indians are most likely to be college-educated or to own a home whereas Dominicans, Puerto Ricans and Mexicans are more likely to be low-income.

Recent immigration status is an important variable to understanding where an individual falls along any given distribution. Overall, non-Hispanic Whites report the largest share of the working age population for all categories, whereas non-Hispanic Asians for the working age arrivals. Non-Hispanic Whites and non-Hispanic Asians report homeownership rates of over 50 percent, significantly higher than those among Hispanics and non-Hispanic Blacks.

As you toggle between the working-age and working-age recent immigrant population, note the different circle sizes. You can compare groups that have the highest rates of employment and see whether that translates to high rates of home ownership, for example. Why do you think this is true for some groups, but not for others?

Scale: Neighborhood

These two choropleth interactive maps reveal how different kinds of diversity are spatially distributed across the neighborhoods in the metropolitan area. The traditional maps identify immigrant groups, recent arrivals, ethnoracial groups, and high-income clusters across the metropolitan area. The superdiversity maps provide a sense of the lived experience of diversity across five dimensions: geographic mobility as well as ethnic, education, income, and immigrant generation diversity.

Note that for this visualization, we are using the 4,683 census tracts of the New York metropolitan area. These statistical subdivisions were designed by the US Census Bureau for the purpose of data gathering. They vary in both geographic size and population. Most contain between 4,000 and 5,000 residents but some, generally in industrial areas, have far fewer residents while others, mainly in densely populated areas dominated by large multiple unit residential buildings, have populations exceeding 8,000.

By most conventional measures, New York’s housing stock is among the nation’s most racially segregated, particularly between Whites and African Americans. Yet because the city is so compact, the physical distance between areas dominated by one or another racial group is often relatively small. Thus, despite living in highly segregated blocks or census tracts, people of different racial and ethnic backgrounds often share subway lines, commercial streets and other public places. Further, as the super diversity maps show, while Whites and African Americans generally live apart, many areas are home to a progressively complex and diverse mix of Asian and Latinx groups. Further, both “White” and “Black” neighborhoods are increasingly home to a combination of long-time natives and diverse groups of recent immigrant origin. In recent years a number of areas have become extremely diverse.

Our measure of ethnic diversity is based on the number of ethnic origin groups in each area. The mobility index is based on the percent of the population that moved into the census tract during the last year. The immigrant, income, and education indices all report the degree of variability along these dimensions, at the census tract level. For example, the income diversity index highlights the neighborhoods where people across the income spectrum meet, versus neighborhoods where households tend to have the same incomes. In each case, darker map colors indicate areas of more pronounced diversity or mobility as compared with lighter areas on the map that indicate less diversity.

You are also able to see the degree of correlation between two different types of diversity, by clicking once on each of two of the tags on the left side of the map. For example, if you select “ethnic diversity” and “income diversity,” you can see four major types of areas (the four corners of the legend at the top of the right side): areas of limited diversity of both types (the lightest blue colour), areas that have a high degree of one type or diversity and not the other (pink or green), and areas with the most diverse populations on both counts (dark blue).

Whereas the traditional maps plot the location of particular groups, the superdiversity maps enable users to visualize the spaces of intersecting diversity, where people encounter others who are different from themselves. This approach invites users to better see how people experience the complex interactions of multiple dimensions of diversity in their daily lives.

Immigrant Population

Recent Immigrant Population

High Income

non-Hispanic White

non-Hispanic Black

non-Hispanic Asian

non-Hispanic Other

Hispanic

Intersections

The final illustration is designed for users to consider the relationship between diversity and opportunity. The interactive buttons allow users to select key characteristics—be they sex, age, ethnoracial background, nativity and recency of immigration, and location within or beyond the five boroughs of New York City—and imagine socio-economic outcomes (likelihood to have a college degree, work, own a home, have low-income, speak English at home, or live in an unaffordable housing situation) of their imagined individual. The midpoint of each dial represents the average indicator value for the total working-age population. Note that intersections that have less than 100 observations will appear greyed out because there are too few observations to make any inferences.

Holding all but one button constant, users can explore the socioeconomic outcomes for specific sub-populations along a single dimension. This could be for different cohorts of immigrants, residence in different boroughs of New York City proper, or different ethnoracial background. For example, US-born Chinese between the ages of 18 to 64 are as likely as US-born Dominicans to speak English at home, but Chinese are significantly more likely to have a college degree, to be working, to have affordable housing, to not be low-income and to own a home. This illustrates the ethnoracial disparities in socioeconomic outcomes across the five racial groups and the ten largest ethnic groups.

I am of descent residing

I am likely as average to have a university degree.

I am likely as average to speak English at home.

I am likely as average to be working.

I am likely as average NOT to be low income*.

I am likely as average to have affordable housing.

I am likely as average to own a home.

*In the low income dial, the right side of the dial indicates greater likelihood of not being low income.

Conclusions

These interactive data visualizations serve several purposes. First, like most contemporary data visualizations, they help us to visualize patterns quickly and powerfully. In addition to innovative graphics, manipulating the data inspires users to seek information of particular interest. Second, the mix of tools offered here depict a variety of linkages between social and economic characteristics. This can help us understand more about how populations are shifting in myriad and overlapping ways. Third, when viewed with the graphics of the other superdiversity visualizations, they tell a common story of diversification. Together they show similar trends, but different patterns and processes. Probing similarities and differences across groups is a crucial step in comparative research—and undertaking such investigation both answers certain questions and raises others for further research. Fourth, these interactive data visualizations help foster a wider and more complex understanding of migration and diversity dynamics.

In this way, we can come to appreciate that the “diversification of diversity” does not entail chaos, but rather multifaceted and interconnected patterns that represent our changing urban fabric. In fact, the New York Metropolitan area has long been characterized by superdiversity, but this is the first time—to our knowledge—that these patterns are visualized. In building these interactive graphics, we hope to help people to see diversifying cities in new ways, to gain a better sense of the complexity of society, and to appreciate the evolving nature of everyday life in cities.

Acknowledgments

We wish to thank the following people and institutions for their help in assembling the reams of information behind this website, and its design and implementation. Steven Vertovec and the Max Planck Institute for the Study of Religious and Ethnic Diversity provided crucial funding support. Daniel Hiebert provided invaluable technical guidance on data and variables. Stamen Design provided crucial expertise on website design and visualization. CUNY Graduate Center and the Center for Urban Research also provided crucial funding and institutional support.

Max Planck Institute for the Study of Religious and Ethnic Diversity

University of British Columbia

City University of New York

Stamen Design

Visualization development and web design