MARRIAGE DESERTS

Mapping America’s Family Divide

 

Chris Bullivant, Ken Burchfiel 

April 2026

 

DRAFT 1

 

 

CONTENTS PAGE

 

 

 

 

ACKNOWLEDGEMENTS

We are indebted to several colleagues for their assistance in developing these maps. John Iceland of Penn State provided an external review of the methodology and results. Brad Wilcox offered valuable guidance on the scope of the report and the distinction between marriage deserts (share of adults married) and married-family deserts (share of households with children headed by married couples). Lyman Stone gave critical feedback on the use of the 2020 decennial Census, including the recommendation to group tract-level data by PUMA for more reliable calculations, and served as a key internal check throughout the project. Wendy Wang provided generous input during drafting to ensure the analysis met IFS research standards and supported the development of this map-based approach. Grant Bailey advised on suitable data sources and its presentation. Alysse ElHage provided invaluable editorial support throughout. We are also grateful to the team at Bevy Commerce, particularly Jafar Mahmood, for technical and design support in embedding the maps on the IFS website.

 

An author (Chris Bullivant) used AI-based tools in early drafts for proofreading and terminology consistency; all analysis and conclusions are the author’s own.

 

EXECUTIVE SUMMARY

Include Top 10 Marriage Deserts, Top 10 Married-Family Deserts

 

[Complete closer to final draft].

 

 

Introduction

This report accompanies two maps of the United States that are available online at the Institute for Family Studies website (ifstudies.org/marriagedeserts). These maps allow the user to view marriage and married-family shares within any part of the United States. The data informing these maps is made available at the same location in a series of tables ranking areas by their marriage rates and share of married families. The maps and the tables are both based on data from the 2020 US Census.

In this report we draw from these tables to present the Top 10 Marriage Deserts and Top 10 Married-Family Deserts in the United States. These areas are where marriage rates are low, and where the share of households with children headed by married couples is low.

We also list the Top 10 Marriage Gardens and Top 10 Married-Family Gardens. These are areas where the marriage rate is high, and where the share of households with children headed by married couples is high.

The areas we use to measure are Public Use Microdata Areas (PUMAs), roughly comparable units with an average size of 135,000 people. After presenting the Top 10 individual-PUMA deserts and gardens, we will then show top-10 lists of grouped PUMAs that share the same marriage or married-family category.

 

The assets described in this report are:

INDIVIDUAL PUMAs		GROUPED PUMAs
Marriage Maps and Tables		Marriage Maps and Tables
1: Map of PUMAs by marriage category 2: Table of PUMAs by marriage category 3: Table of top marriage deserts 4: Table of top marriage gardens		9: Map of grouped marriage deserts and gardens 10: Table of top grouped marriage deserts 11: Table of top grouped marriage gardens
Married-Family Maps and Tables		Married-Family Maps and Tables
5: Map of PUMAs by married-family category 6: Table of PUMAs by married-family category 7: Table of top married-family deserts 8: Table of top married-family gardens		12: Map of grouped married-family deserts and gardens 13: Table of top grouped married-family deserts 14: Table of top grouped married-family gardens

 

 

TERMINOLOGY

Marriage Desert: A PUMA where the share of householders aged 15-64 who are married is below 33%. (Here we look simply at the share of adults who are married versus not married—the latter which could include single, never-married, divorced, or widowed people. Those areas with very low levels of marriage are specifically termed marriage deserts.)

 

Marriage Garden: A PUMA where the share of householders aged 15-64 who are married is at or above 61%.

 

Married-Family Desert: A PUMA in which fewer than 52% of households with children are headed by a married couple. (In this report, 'children' refers to what the Census defines as 'own children' (rather than 'related children'); see methodology section for more details.

 

Married-Family Garden: A PUMA in which the share of households with children headed by a married couple is at or above 79%.

LEGEND

The categorical legend used for marriage and married-family maps are as follows.

Further details on how and why we established these categories are available in the methodology section.

 

Marriage categories

Married-family categories

 

 

 

Why this report?

 

Neighborhood family structures—whether experienced in tenements, apartments, public schools, or the local store—inform each other with a complex set of norms. As shown in research by Harvard economist Raj Chetty, “the most predictive factor of upward mobility in a community was the share of homes with two parents present in the household.”ⁱ

 

And yet, there is a sharp divide in the experience of family across the United States dependent upon income. As IFS Senior Fellow Brad Wilcox and report author Chris Bullivant wrote in a previous paper,ⁱⁱ higher levels of marriage and stability were prevalent among high-income families. Conversely, lower levels of marriage and stability were prevalent among low-income families.

 

This “social capital divide” is of concern because, across multiple metrics, children growing up in stable-married families have better life outcomes than their counterparts who grow up in less-stable cohabiting or single-parent households.

 

So that the predictive, though not deterministic, quality of the share of homes with two parents present in the household begs the question: where are these neighborhoods? And where aren’t they?ⁱⁱⁱ Because perhaps, if we can see where they are, it may be easier to bridge the social-capital divide to create equal access to stable, married family life and the relational security it confers.

 

Marriage deserts

This work was inspired by the Opportunity Atlas published by Opportunity Insights--in particular, their upward mobility maps, which include the share of people who are married by age 30.^iv Their maps reveal a patchwork quality to marriage rates at a census tract level. Bullivant described the low marriage rates neighborhoods in these maps as “marriage deserts” in an article with Brad Wilcox.^v Just as it is harder to buy healthier groceries in a “food desert,” it is likely harder to enter into stable married families in a “marriage desert.” This is simply where relationships styles are learned in the context of existing relationships.

 

Inspired by the Opportunity Insights maps, we wanted to create our own, humbler family-focused versions that homed in on two measures:

• The share of householders aged 15-64 who are married.

• A sense of the share of households with children that are stable, two-parent homes. (In this case, we use married couples as a proxy.)

 

What this report does not do

Beyond listing and describing the Top 10 marriage deserts/gardens, and married-family deserts/gardens, we offer minimal analysis as to causes.

 

It is also beyond the scope of our research to establish a causal relationship between low rates of marriage and low rates of married families, or their converse, high rates of marriage and high rates of married families. While the connection may seem obvious, there are areas where higher rates of marriage do not correlate with higher rates of married families, and there are areas where low marriage rates correlate with higher shares of married families.

 

However, these maps are presented with the knowledge that areas of low marriage and weak family stability are more prone to less upward social mobility. And we hope that these are a useful tool for inspiring further research and analysis.

 

INDIVIDUAL PUMAS

MARRIAGE DESERTS AND MARRIAGE GARDENS

The full table and map visualizing the share of householders aged 15-64 who are married are available here. The Top 10 marriage desert and top 10 marriage gardens are pulled from these assets.

 

Map:

 

Table:

Marriage Deserts Top 10

These are the top ten marriage deserts in the United States.

 

In this report, marriage deserts are defined as Public Use Microdata Areas (PUMAs) in which the share of householders aged 15 to 64 who are married falls below 33 percent (the rounded 10th percentile for this metric), representing the lowest-observed marriage rates nationally.

 

The table below ranks the ten PUMAs with the lowest shares of married householders ages 15–64. Across these areas, the married share of householders ranges from 13.0 percent to 16.9 percent. This implies that more than 80 percent of householders ages 15–64 in these PUMAs are not married.

 

These marriage deserts are located in ten major U.S. metropolitan areas across seven states—Georgia, Illinois, Massachusetts, Michigan, Missouri, Ohio, and Wisconsin—as well as the District of Columbia. Ohio appears three times in the top ten, making it the only state represented more than once. All ten PUMAs are located east of the Rocky Mountains. (Original PUMA names were simplified for easier readability.)

 

 

Marriage Gardens Top 10

These are the ten largest marriage gardens in the United States.

 

In this report, marriage gardens are defined as Public Use Microdata Areas (PUMAs) in which the share of householders aged 15 to 64 who are married is at least 61%, representing the highest-observed marriage shares nationally.

 

The table below ranks the ten PUMAs with the highest shares of married householders aged 15–64. Across these areas, the married share of householders ranges from 73.9 percent to 78.6 percent. This implies that around three in four householders aged 15–64 in these PUMAs are married, roughly the inverse of the pattern observed in the nation’s largest marriage deserts.

 

These marriage gardens are located in ten counties or county-equivalent areas across six states—Connecticut, Florida, New Jersey, Tennessee, Texas, and Utah. Texas appears three times in the top ten, while Utah and New Jersey each appear twice. Notably, none of the top-10 marriage gardens are located in major U.S. cities.

 

 

Observable differences between the Top 10 Marriage Deserts and the Top 10 Marriage Gardens

Family structure polarization

The differences between the Top 10 Marriage Deserts and the Top 10 Marriage Gardens are stark, with a 60+ percentage-point spread between these two extremes:

 

• Marriage deserts: 13.0%–16.9% married

• Marriage gardens: 73.9%–78.6% married

 

Even the marriage desert in tenth place, in Chicago, 16.9%, has less than one quarter the marriage share of the 10th-ranked marriage garden in Thompson Station Town/Fairview, 73.9%. This does not appear to describe small variations around shared national norms, but rather two distinct family-formation regimes, so stark it could even be described as family structure polarization.

 

No state appears on both top ten lists, suggesting an element of geographic polarization in family structure at the extremes. At the same time, this comparison should not be overinterpreted: a broader review of national maps shows that it is common for states to contain both marriage deserts and marriage gardens, often in close geographic proximity. The polarization observed here reflects the most extreme cases, not complete state-level segregation.

 

Ohio stands out as a negative outlier. Three Ohio PUMAs—Cleveland, Columbus, and Cincinnati—appear among the top ten marriage deserts, while none appear among the top ten marriage gardens. In the full national distribution, Ohio contains seven marriage gardens, with the largest—Delaware County (South), Ohio—ranking 41st nationally. Further research examining changes in marriage shares over time may help clarify whether Ohio’s pattern reflects long-term post-industrial family decline distinct from the experiences of coastal, urbanized areas. Ohio stands in contrast to Texas, which features three marriage gardens in the top 10 while not appearing in our list of the top 10 marriage deserts.

 

MARRIED-FAMILY DESERTS AND MARRIED -FAMILY GARDENS

What do these patterns imply for children? While the analysis above focuses on the share of householders aged 15–64 who are married, the underlying motivation is to understand how local family structure shapes the social capital environments in which children are raised. A substantial body of research indicates that children’s life outcomes are strongly influenced by relational stability within the family of origin, as well as by the prevalence of married-couple households in their surrounding community.

 

The top ten married-family deserts and top ten married-family gardens are therefore identified using the share of households with children of their own that are headed by married couples. This is with a view to considering a child’s relational security—the degree of stability, continuity, and reliability in the family and community relationships that surround them.

 

The Top 10 married-family deserts and top 10 married-family gardens are pulled from these assets.

Map:

 

Table:

 

Married-Family Deserts Top 10

In this analysis, married-family deserts are defined as Public Use Microdata Areas (PUMAs) in which the share of households with children that are headed by a married couple falls below 52 percent. In these areas, children are less likely to be raised in a household headed by married parents and less likely to be surrounded by peers growing up in married-couple families—conditions associated with lower levels of relational security at both the household and neighborhood level.

The table below ranks the ten PUMAs with the lowest shares of households with children headed by a married couple. Across these areas, the share of households with children headed by a married couple ranges from 18.0 percent to 25.7 percent. In all ten of these married-family deserts, fewer than one in four households with children are headed by a married couple.

Put differently, in these PUMAs, children are three to four times more likely to live in households with children that are not headed by married parents than in households that are.

These ten married-family deserts are located in eight major U.S. metropolitan areas across seven states—Georgia, Illinois, Kentucky, Michigan, Missouri, Ohio, and Wisconsin—and the District of Columbia. Chicago and Detroit each contain two married -family deserts. As with the marriage deserts, all ten married -family deserts are located east of the Rocky Mountains.

 

Married Family Gardens Top 10

In this analysis, married-family gardens are defined as Public Use Microdata Areas (PUMAs) in which the share of households with children that are headed by a married couple is at least 79 percent. In these areas, children are most likely to be raised in households headed by married parents and to be surrounded by peers growing up in married-couple families—conditions associated with higher levels of relational security at both the household and neighborhood level.

The table below ranks the ten PUMAs with the highest shares of households with children headed by a married couple, among PUMAs with substantial numbers of households with children. Across these areas, the share of households with children headed by a married couple ranges from 86.7 percent to 88.9 percent. In all ten married-family gardens, fewer than one in seven households with children are not headed by a married couple.

Put differently, in these PUMAs, children are six to eight times more likely to live in households with children headed by married parents than in households with children that are not headed by married parents.

These ten married-family gardens are located primarily in the suburban and exurban communities of major U.S. metropolitan areas across seven states—California, Connecticut, Illinois, Massachusetts, New Jersey, Texas, and Utah. California, New Jersey, and Utah each contain two married-family gardens in the top ten.

 

Observable differences between the Top 10 Married-Family Deserts and Gardens: a stark social capital divide

The differences between the Top 10 Married-Family Deserts and the Top 10 Married-Family Gardens are stark, with a difference of roughly 70 percentage points between the extremes.

 

• Lowest married-family desert: 18% of households with kids headed by a married couple

• Highest married-family garden: 89% of households with kids headed by a married couple

 

Married-couple households constitute a supermajority of families with children in the top 10 married-family gardens (>86%), but only a small minority in the top 10 married-family deserts (<26%). The household composition of the top married-family deserts is effectively the inverse of that found in the top married-family gardens.

 

As with the marriage deserts/gardens, this comparison between married-family deserts and gardens demonstrates not just a variance around shared national norms, but a structural divergence in family structures. Over the course of childhood, these differences imply substantially different levels of exposure to married-couple households in children’s everyday social environments.

• In the top 10 married-family deserts, children are three to four times more likely to live in households with children that are not headed by married parents.

• In the top 10 married-family gardens, children are six to eight times more likely to live in households with children that are headed by married parents.

This will effect a significant discrepancy in lived experience for children growing up in these distinct environments. Those in married-family gardens grow up with married parents, in neighborhoods, schools, and after-school activities where most peers experience the same. These children will be observing marriage as a stable, ordinary feature of adult life. This will create stronger social capital networks, with a greater alignment between school, families, and community, overlapping stable inter-family interactions over a childhood and adolescence, creating denser networks of adult supervision, stable peer networks, and deep relational literacy.

 

For children in married-family deserts, by contrast, it is likely that there will be much greater churn in adult relationship stability, and greater inconsistency in supervision, where marriage and stable relationship bonds are atypical or distant from everyday norms—both for themselves and their peers. Greater intergenerational kinship, such as grandparent care, and demand on public resources such as schools may seek to compensate for this gap; however, it is likely that children in these married-family deserts will still experience less social capital, and diminished relational literacy, than those in to married-family gardens.

 

The top marriage gardens have twice the number of children as the top marriage deserts

The preceding analysis focuses on the share of households with children that are headed by married couples. An additional question is whether places with different family structures also differ in the overall prevalence of households with children. In other words, do married-family deserts and married-family gardens differ not only in how children are raised, but in how common it is for households to include children at all?

A clear pattern emerges when comparing the top ten married-family deserts with the top ten married-family gardens. PUMAs with markedly different family structures also differ substantially in the share of households that include children.

Among the top ten married-family deserts, the share of households with children ranges from a low of 14.9 percent in Detroit, Michigan, to a high of 26.4 percent in Chicago (West), Illinois. In contrast, among the top ten married-family gardens, the share of households with children ranges from 34.0 percent in Northfield and New Trier Townships, Illinois, to 57.5 percent in Utah County (West), Utah.

On average, households with children account for 21.0 percent of all households in the top ten married-family deserts, compared with 44.0 percent in the top ten married-family gardens. In practical terms, the share of households with children is approximately twice as high in the top married-family gardens as in the top married -family deserts.

This pattern is consistent with prior research showing that fertility in the United States is increasingly concentrated among married households and those with greater economic stability. Research by IFS Senior Fellow Lyman Stone, for example, finds that recent declines in fertility have been driven disproportionately by lower birth rates among the unmarried^vi and among lower-income populations.^vii While the present analysis does not establish causality, the association between married-family structure and higher concentrations of households with children suggests that family norms and household formation are closely intertwined with contemporary fertility patterns.

Table: Top 10 Married-Family Gardens including percentage of households with children as a share of all households.

This comparison describes a considerable difference in local environments, especially as experienced by children.

Married-family deserts appear, on this evidence, to function increasingly as adult-oriented or post-family spaces, while married-family gardens appear to be environments that are more conducive to having and raising children. For children growing up in married-family deserts, this means not only being raised within a context of greater relational instability, but also living in environments where having children in the household is itself less common. Parenting, as well as marriage, is therefore less frequently observed as a dominant adult role within the community.

These differences have implications for schools, housing markets, tax bases, welfare systems, local politics, and the balance between services oriented toward children and those oriented toward older populations. Over time, environments with fewer children may also face challenges in sustaining the institutions and informal networks that support family life.

It is notable that within the top ten married-family deserts, even where the share of households with children headed by married couples is relatively higher, the overall share of households with children does not increase substantially. This pattern may point to deeper structural constraints on family formation in these neighborhoods. One possible interpretation is a process of social-capital erosion in which unstable relationship norms, fewer children, and limited opportunities for high-relational-literacy role-modeling reinforce one another over time.

By contrast, the married-family gardens suggest that, once a place crosses a certain threshold of married-family density, children remain visible and numerous. In these environments, family life appears more embedded in everyday community experience, with positive spillovers for schools, public services, and civil society norms. These conditions may be self-reinforcing, supporting stable family ecosystems over longer periods.

At the same time, caution is warranted. Marriage rates in the top ten married-family gardens are tightly clustered (approximately 86.7 to 88.9 percent), while the share of households with children varies much more widely (approximately 34.0% to 57.5%). While the association between higher marriage rates and higher child prevalence is clear, additional factors—particularly housing availability, affordability, and local institutional capacity—are likely to shape how strongly marriage translates into higher fertility in different places.

Taken together, the data suggest that married-family gardens are not only places with higher rates of married parenthood, but also places where children themselves are far more prevalent. Married-family deserts, by contrast, increasingly appear to be environments with relatively few households containing children at all. This divergence points to growing spatial separation not only in family structure, but in the presence of children, with important implications for the long-term social and civic life of communities.

 

 

GROUPED PUMAS

The preceding analysis listed the top marriage deserts and marriage gardens at the level of individual Public Use Microdata Areas (PUMAs). This is useful for identifying the most extreme examples, but it only provides a partial picture.

A cursory review of the maps shows that many marriage deserts and marriage gardens sit next to one another and are often co-joined or contiguous. In practice, this means that individual PUMAs frequently form part of larger clusters of areas with similar family-structure characteristics.

In the next set of analyses, we consider the same four features (marriage deserts, marriage gardens, married-family deserts, and married family gardens) but examine them using grouped PUMAs rather than individual units. These groupings are formed by adjacent PUMAs that meet the same criteria.

Grouped marriage gardens and deserts are ranked by the total number of householders ages 15-64 that they contain, and grouped married-family gardens and deserts are ranked by the total number of households with children that they contain.

The Top 10 grouped marriage deserts and top 10 grouped marriage gardens are pulled from these assets.

 

Maps:

All grouped marriage deserts and gardens

 

 

All grouped married-family deserts and gardens

 

 

GROUPED MARRIAGE DESERTS AND GROUPED MARRIAGE GARDENS

Marriage Deserts Top 10 Grouped

These are the top ten grouped marriage deserts in the United States.

 

The preceding analysis identified marriage deserts at the level of individual Public Use Microdata Areas (PUMAs). When these PUMAs are examined in contiguous groupings, a clearer picture emerges of the scale of low-marriage environments across major metropolitan areas.

The table below ranks the largest grouped marriage deserts by the total number of householders aged 15–64 (regardless of marital status) who live within contiguous PUMAs that meet the marriage desert criteria. In these groupings, the minimum and maximum marriage shares reflect variation across adjoining neighborhoods within the same metropolitan area, rather than isolated extremes.

This list pulls in new cities and new states, primarily because it draws in America’s largest cities, and with a higher maximum share than in the individual list. By contrast to the individual list, we now see we are researching a large phenomenon, with—in the case of the largest grouped marriage desert in New York—a total of 686,520 households.

The table below ranks the ten grouped PUMAs with the lowest shares of married householders aged 15–64. Across these areas, the married share of householders ranges from 14 percent to 33 percent. This means that over two thirds of householders aged 15–64 in each of these grouped PUMAs are not married.

 

 

 

These marriage deserts are located in major U.S. metropolitan areas across ten states (California, Georgia, Illinois, Massachusetts, Maryland, New Jersey, New York, Ohio, Pennsylvania, and Texas) and the District of Columbia.

 

Here we note,New York has two of the largest deserts, in Manhattan and Queens, sharing a total of over 1.2 million householders in PUMAs with low rates of marriage. California and Texas also join the Top 10, which means that grouped marriage deserts are present on both sides of the Rockies. California's grouped desert, which stretches from central Los Angeles to Santa Monica, is the fourth largest in the country.

 

Atlanta, Boston, Cleveland, Chicago and Washington D.C. are found in both the individual PUMA top 10 Marriage Deserts and the Grouped Top 10 Marriage Deserts. This indicates both scale and persistently low share in their marriage deserts.

 

 

 

Large sprawl normalization of non-marriage

What these top ten grouped marriage deserts show, and not least among the 1.2 million households in New York , is that low marriage rates are not necessarily restricted to extreme examples. For extensive geographies or commuter distances, low marriage rates are broadly normalized. Given New York’s unique importance in American cultural, media and corporate leadership, it is worth noting that the context in which significant numbers of participants in the region’s economy work does not appear to promote shared norms around stable, married family life. While it is likely that executives and leaders in these industries live in the adjacent marriage gardens and married-family gardens, there is a clear divide between the family-haves and the family-have-nots that likely mirrors other forms of inequality in the region.

 

Marriage Gardens Top 10 Grouped

In this report, marriage gardens are defined as Public Use Microdata Areas (PUMAs) in which the share of householders aged 15 to 64 who are married is at or above 61 percent, representing the highest observed marriage rates nationally.

 

The table below ranks the ten grouped PUMAs with the highest shares of married householders aged 15–64. Across these areas, the married share of householders at the PUMA level ranges from 61 percent to 78 percent. This implies that, at a minimum, three out of five householders aged 15–64 in these PUMAs are married, and in several parts of these grouped PUMAs, over three quarters are married.

 

These marriage gardens are located in major metropolitan areas across thirteen states—California, Connecticut, Illinois, Maryland, Massachusetts, Minnesota, New Hampshire, New Jersey, New York, Pennsylvania, Texas, Virginia, and Utah.

 

Here we note that the Rockwall–Greenville–Far Northeast Dallas, Texas marriage garden is the largest grouped marriage garden in the country. At 636,086 householders ages 15-64, it is around 7% smaller than the largest marriage desert (Manhattan, NY, 686,520).

 

Connecticut, Texas, Utah, and New Jersey are represented in both the top ten grouped marriage gardens and the top ten individual-PUMA marriage gardens. This suggests that marriage gardens in these states combine high marriage rates with breadth across multiple neighborhoods, indicating depth as well as scale in their marriage ecosystems. New states appearing in the grouped marriage gardens include California, Illinois, Maryland, Minnesota, New York, New Hampshire, and Pennsylvania.

 

Notably, there are seven states that appear not only in the top ten grouped marriage gardens but also feature in the top ten grouped marriage deserts: California, Illinois, New York, New Jersey, Maryland, Massachusetts, and Pennsylvania. While this reflects states with higher numbers of householders, it also describes a high degree of internal variation in family structure within these states and, as such, highly divergent, divided experiences of family life within state lines.

 

New York contains two of the largest grouped marriage deserts and grouped marriage gardens. While this partly reflects its large population size, it may also point to a sharp internal divide in marriage experience within the state.

 

Finally, the geography of marriage gardens is overwhelmingly suburban and exurban in character, standing in clear contrast to the urban-core concentration observed among marriage deserts.

 

 

 

Observable differences between the Top 10 Grouped Marriage Deserts and the Top 10 Grouped Marriage Gardens

Most grouped marriage gardens are larger than grouped marriage deserts

The top three grouped marriage deserts (by householders ages 15-64), Manhattan at 686,520, Chicago at 668,379, and Queens at 544,982, are between roughly 8% and 23% larger than their marriage-garden counterparts. However, the pattern reverses beyond the top three. From ranks four through ten, grouped marriage gardens are consistently larger than their similarly-ranked marriage deserts. In this range, marriage gardens are 10% to 47% larger than their desert counterparts.

 

In absolute terms, grouped marriage gardens ranked fourth through tenth contain between 484,576 and 301,877 householders ages 15-64, compared with 422,668 to 206,932 for similarly ranked marriage deserts.

 

This suggests that while the very largest low-marriage environments are concentrated in a small number of megacities, high-marriage environments are able to persist across broader suburban and exurban geographies at scale. Beyond the top tier of urban cores, marriage gardens tend to encompass larger populations than marriage deserts. This indicates that marriage gardens can persist with larger numbers of households, so at greater scale, than America’s largest group marriage deserts. GROUPED MARRIED-FAMILY DESERTS AND GROUPED MARRIED-FAMILY GARDENS

Married-Family Deserts Top 10 Grouped

In this analysis, married-family deserts are defined as Public Use Microdata Areas (PUMAs) in which the share of households with children that are headed by a married couple falls below 52%. In these areas, children are less likely to be raised in households headed by married parents and less likely to be surrounded by peers growing up in married-couple families—conditions associated with lower levels of relational security at both the household and neighborhood level.

 

The table below ranks the ten grouped PUMAs with the lowest shares of households with children headed by a married couple, where groupings consist of contiguous PUMAs that meet the married-family desert criteria. In these groupings, the minimum and maximum shares reflect variation across adjoining neighborhoods within the same metropolitan area, rather than isolated extremes.

 

Across these grouped married-family deserts, the PUMA-level share of households with children headed by a married couple ranges from 18% to 52%. Thus, married-family shares in these grouped deserts are generally higher than those observed in the individual-PUMA Top 10 married-family deserts, reflecting the inclusion of larger and more heterogeneous areas.

 

These ten grouped married-family deserts are located in eight major U.S. metropolitan areas across nine states—California, Georgia, Illinois, Maryland, Michigan, New York, New Jersey, Ohio, and Pennsylvania—and a significant married-family desert that includes parts of the Mississippi Delta in Louisiana, Mississippi, Arkansas, and Tennessee. Compared with the individual-PUMA Top 10 married-family deserts, Kentucky, Missouri, Wisconsin, and the District of Columbia no longer appear in the grouped Top 10.

 

America’s largest grouped married-family desert is located in the Bronx and Manhattan, with 178,144 households with children where the share headed by married parents ranges from 31% to 51%. New York is also home to the 7th-largest grouped married-family desert, which is located in Brooklyn. In this desert, which contains 76,805 households with children, only 28% to 48% of households with kids are headed by married parents. Combined, New York’s two married-family deserts hold over 250,000 households with children who are living in married-family deserts.

A significant member of this top-10 list is the Mississippi Delta, which ranks second. 155,492 households with children across a large, contiguous, mostly-rural geography fall within this married-family desert, where the share of households with children headed by a married couple ranges from 27 percent to 52 percent. Notably, this married-family desert does not coincide with one of the largest grouped marriage deserts. However, a review of the maps shows that it overlaps with contiguous PUMAs in which overall marriage rates are below the national average. This alignment suggests that married-family deserts may emerge in regions where marriage prevalence is already depressed, even if those areas have not yet reached the lowest national marriage shares.

 

Finally, four metropolitan areas—Atlanta, Cleveland, Chicago and Detroit—appear in both the individual and grouped Top 10 married-family desert tables. This suggests a prevalence and persistence in both depth and scale of lower shares of households with children headed by married couples in these cities.

 

 

Married-Family Gardens: Top 10 Grouped

In this analysis, married-family gardens are defined as Public Use Microdata Areas (PUMAs) in which the share of households with children that are headed by a married couple is at or above79% percent. In these areas, children are more likely to be raised in households headed by married parents and to be surrounded by peers growing up in married-couple families—conditions associated with higher levels of relational security at both the household and neighborhood level.

 

The table below ranks the ten grouped married-family gardens, where groupings consist of contiguous PUMAs that meet the married-family garden criteria. In these groupings, the reported minimum and maximum shares reflect variation across adjoining neighborhoods within the same metropolitan area, rather than isolated extremes.

 

Across these grouped married-family gardens, the share of households with children headed by a married couple ranges from 79% to 89%. Compared with the individual-PUMA Top 10 married-family gardens, the minimum shares in the grouped gardens are modestly lower, reflecting the inclusion of larger and more heterogeneous areas.

 

These grouped married-family gardens are located in ten mostly suburban and exurban regions across the District of Columbia and 12 states—California, Connecticut, Illinois, Maryland, Massachusetts, New Jersey, New York, Pennsylvania, Utah, Texas, Virginia, and Washington. This is the first part of the report in which Washington state appears within a top-10 ranking.

 

 

 

 

Observable differences between the Top 10 Grouped Married-Family Deserts and Gardens

Grouped married-family gardens are larger, and the experience more uniform, than grouped married-family deserts

The grouped married-family gardens encompass substantially larger numbers of households with children than the grouped married-family deserts. Among the Top 10 grouped married-family gardens, the number of households with children ranges from 108,357 to 385,561, compared with 59,715 to 178,144 in the Top 10 grouped married-family deserts. At every rank, the grouped married-family gardens include more households than their counterparts in the grouped married-family deserts, indicating that high married-family environments are able to persist across larger population scales. The grouped married-family gardens are between approximately 71% and 116% larger than their ranked married-family desert counterparts.

 

The degree of internal variation also differs sharply between the two groups. In the grouped married-family gardens, minimum shares of households with children headed by married couples cluster tightly between 79 and 80 percent, while maximum shares range from 85 to 89 percent, producing a spread of approximately 10 percentage points. By contrast, grouped married-family deserts exhibit much wider dispersion, with minimum shares ranging from 18 to 46 percent and maximum shares from 41 to 52 percent, a spread of around 34 percentage points. This contrast suggests that married-family prevalence is more internally consistent across grouped gardens, while grouped deserts encompass a wider range of family arrangements across adjoining neighborhoods.

 

Geographically, the grouped married-family gardens—like the individual Top 10 married-family gardens—are generally concentrated in suburban and exurban areas rather than central metropolitan cores. Many are located in high-income regions with limited housing availability, such as the San Francisco Bay Area and Northern Virginia. These characteristics are consistent with patterns of residential sorting and may help explain the relatively narrow range between minimum and maximum married-family shares observed within grouped gardens.

 

Finally, it is notable that several of the largest grouped married-family gardens are located in regions that are economically affluent and institutionally influential, including Northern Virginia, the Bay Area, and parts of Connecticut. In the New York area, two large married-family deserts (encompassing parts of the Bronx, Manhattan, and Brooklyn) are in close proximity to married-family gardens in New Jersey and Long Island. This juxtaposition highlights the degree to which sharply different family-structure environments can exist in close geographic proximity within the same broader labor and housing markets.

 

Six states (California, Illinois, Maryland, New York, New Jersey and Pennsylvania) are each home to top-10 grouped married-family gardens and grouped married-family deserts. This in part reflects that these are populous states, but it also means that, even within state lines, there is a huge discrepancy in childhood experience. It is also worth observing that these states all contain top-10 grouped marriage deserts and marriage gardens as well, making them the most divided states in America for family experience.

 

Four Features beyond the Top 10s

We have reviewed eight top-10 lists in our analysis so far. These lists showed the top 10 deserts and gardens for our % married and married-family metrics at both the individual-PUMA and grouped-PUMA level.

 

In this next section, we want to briefly outline some observable features of the entire map which point to interesting opportunities for future research.

1. Double Deserts

Marriage Deserts and Married-Family Deserts in same geography

Chicago

There are some areas that experience both marriage deserts and married-family deserts in a similar, overlapping neighborhoods. Here there is both a low share of married adults who are married, and a low share of households with children that are headed by married families. We see this, for examples, in parts of Atlanta, Baltimore, Chicago, Cleveland, Detroit, Los Angeles, Newark, New York, Philadelphia, and Washington D.C., among others.

 

Chicago: Marriage Desert PUMAs	Chicago: Married-Family Desert PUMAs

 

 

 

 

 

 

Figure: Comparing Chicago Marriage Desert PUMAs and Married-Family Desert PUMAs.

 

Here we can see there are some PUMAs that are both marriage deserts and married-family deserts. We have termed these areas double deserts.

 

Further research is needed to see if householders in these areas experience a persistent difficulty in entering into marriage and stable family life, and whether a lack of role models of stable, family life reproduces itself, limiting people’s lifestyle choices.

 

2. Delayed-Marriage Deserts

Marriage Deserts give way to Married-Family Gardens once children are in the picture

Manhattan and Brooklyn

There are some marriage deserts where there are significant, low rates of marriage. However, in the same geography, the share of householders with children who are married are in the higher share. In other words, marriage deserts do not equal married-family deserts. Such areas include college towns, for example. But these areas also appear in those economically-active zones that attract young people from across the country, such as certain neighborhoods within Washington D.C., Los Angeles, and New York. Once we compare these marriage deserts with the share of households with children headed by a married couple, we see the deserts quickly give way to regions with higher shares of married families.

 

Manhattan/Brooklyn: Marriage Desert PUMAs	Manhattan/Brooklyn: Married-Family Desert PUMAs

Figure: Comparing Manhattan and Brooklyn, NY, Marriage Desert PUMAs and Married-Family Desert PUMAs

 

In the above figure, we can see that while there are some PUMAs that are both marriage deserts and married-family deserts, there are other marriage deserts that become higher-share married-family PUMAs. We have termed these areas delayed -marriage deserts, though this hypothesis requires additional research.

 

It is possible that, in these areas, young people are remaining single longer but ultimately get married prior to having children, provided that they remain in the area. This may be in defiance of surrounding communities where there remain double deserts. If this is occurring, it could suggest that, when young people come to an area that has low marriage and low married-family rates as part of inward work migration, they bring high levels of social capital and relational literacy with them and replicate the norms they grew up with elsewhere. This interpretation is speculative and intended as a prompt for further research rather than a conclusion.

3. Desert Corridors

Lower-share marriage regions combined with with large married-family deserts

The Mississippi Delta

There are a number of areas where contiguous lower-share (not desert) regions of adults aged 15-64 who are married span large territories. Within these regions, contiguous married-family deserts appear in the same place. Low marriage rates appear to be associated with married-family deserts. This is unlike the delayed marriage deserts where the introduction of children eliminates the desert or low shares of marriage.

 

We see these desert corridors in parts of the Mississippi Delta and the New Mexico.

 

Mississippi Delta Marriage Desert PUMAs	Mississippi Delta Married-Family Desert PUMAs

Figure: Comparing the Mississippi Delta Marriage Desert PUMAs and Married-Family Desert PUMAs

 

The Mississippi Delta is home to America’s second-largest married-family desert. In this side-by-side comparison with its corresponding share of householders ages 15-64 who are married, we can see that these married-family deserts exist in the context of a large area where marriage rates are lower than normal.

 

Further research could investigate whether low marriage rates cause low married family rates and whether this share is growing across generations. If a given generation of children lacks exposure to married parents (except, perhaps, on TV), it is plausible that, upon entering adulthood, such children may reproduce the relationship patterns they experienced growing up. This would, in turn, reduce the share of householders aged 15-64 who are married. This may well put downward pressure on fertility rates and result in fewer children, but the share of children in married-family deserts would likely increase. These desert corridors, therefore, are vulnerable to becoming double deserts in the coming decades. In this desert growth, we perhaps witness in these corridors the growth of tomorrow’s double deserts.

 

New Mexico

The Southwest: Marriage Desert PUMAs	The Southwest: Married-Family Desert PUMAs

Figure: Comparing New Mexico Marriage Desert PUMAs and Married-Family Desert PUMAs.

 

In the Southwest, we see that New Mexico has lower-share marriage rates and several married-family deserts. The married-family deserts encompass both Native American reservations and more urban areas.

 

4. Inverse Desert Corridors - Rural America

High marriage prevalence but low married-family shares

 

An inverse pattern appears to be the case in rural areas of the United States and in particular central Appalachia.

 

In these PUMAs, we see above-average marriage rates among householders aged 15-64. However, when we look at the share of households with children that are headed by married couples, we see that these same areas have lower shares of married families, with the occasional desert.

 

This is the inverse of what we see in the Mississippi Delta, where low rates of marriage are correlated with low rates of married families.

 

Further research is required to understand this pattern conclusively and in what direction it is traveling. One possibility is that older adults in these regions largely chose to get married, but their adult children are choosing to have kids of their own without first getting married.

Central Appalachia

Central Appalachia: Marriage Desert PUMAs	Central Appalachia: Married-Family Desert PUMAs

Figure: Central Appalachia Marriage Desert PUMAs and Married-Family Desert PUMAs.

 

Central Appalachia, between West Virginia and Kentucky, contains large areas with higher-share marriage rates but lower-share married-family rates. This, for example, is a pattern visible in central California and rural Nevada also.

 

What this area does perhaps show is that, even with role- models of marriage (i.e. parents or grandparents) available to individuals, it does not necessarily follow that individuals will marry or have children in the context of marriage. Again, further research is required to understand the nature of this pattern. If, however, role models are important to establishing pathways to stable, married family life and the relational security and social capital that creates, then a reversal of trends may be easier to accomplish in central Appalachia—and regions like it—than in those areas where marriage is less of a norm.

Conclusion

This report set out to map marriage deserts in the United States. Using Public Use Microdata Areas (PUMAs) and 2020 Census data, the analysis centered on the Top 10 marriage deserts and gardens and the Top 10 deserts and gardens. We examined these areas on both an individual and grouped-PUMA level, which allowed us to explore not only the most severe deserts but the largest ones as well.

 

Marriage deserts are present across the United States. They cluster in specific neighborhoods and, in some cases, extend across large metropolitan regions. At the individual PUMA level, some of the lowest marriage shares appear in parts of cities such as Cleveland, Milwaukee, St. Louis, and Washington, D.C. When grouped, marriage deserts encompass far larger populations: the top three grouped deserts, which include parts of New York City and Chicago, each span over 500,000 households.

 

Taken together, the individual PUMA tables show the sharpest local extremes, while the grouped PUMA tables reveal how extensive and populous these family ecosystems are when contiguous areas are combined. The two approaches therefore measure intensity and scale.

 

Individual top-10 lists identify places with the lowest marriage shares, sometimes affecting relatively small populations; meanwhile, grouped Top 10-lists show where low marriage prevalence affects the greatest number of households.

 

The analysis also shows that marriage deserts and married-family deserts frequently overlap, though not uniformly. In cities such as Chicago, Philadelphia, Baltimore, Los Angeles, and Newark, some neighborhoods experience both low marriage among adults and low shares of households with children headed by married couples. In other areas—such as parts of New York City and Washington, D.C.—lower marriage shares give way to stronger married-family patterns once children are present, suggesting delayed or selective marriage rather than a complete absence of married family life.

 

By contrast, married-family gardens—both individual and grouped—are most often found in suburban and exurban areas, including parts of Northern Virginia, the Bay Area, Connecticut, Utah, and New Jersey. In these gardens, roughly 80 to nearly 90 percent of households with children are headed by married parents. We also found that the largest married-family gardens have higher numbers of households with children, on average, than do the largest deserts. In addition, the individual PUMAs with the highest married-family percentages have higher shares of households with children than do those PUMAs with the lowest percentages.

 

Mapping these patterns provides a foundation for further research into how marriage deserts and married-family deserts evolve over time and what they mean for families, children, and communities across the country.

Methodology

In order to create our analysis, we first needed to choose whether to base our calculations on 2020 Census data or the American Community Survey. We ultimately chose the former dataset, as it has a much larger sample size and should thus be less susceptible to sampling bias. (However, we did perform an alternative ACS-based analysis in order to corroborate our findings; see below for more details.)

 

Having selected the 2020 Census as our main data source, we needed to consider what level of geographic unit would be suitable for accurately representing the data.

Our choice of PUMAs

Our analysis features data at the PUMA (Public Use Microdata Area) level. We also considered using tracts as our reference region, as tract-level metrics would reveal much more variation within specific cities than would metro-area-, county-, or state-level data.

 

However, the 2020 Census dataset was modified in accordance with new differential privacy guidelines that sought to protect confidentiality at the local level by importing national data into the local level. This has introduced reliability concerns for census-tract-level data. To address this challenge, we grouped tract-level data into PUMAs, as doing so may help offset or diminish the effects of differential-privacy-driven alterations to those tracts' original data. (It would have been more straightforward to use pre-tabulated PUMA-level statistics, but this data was not available for the 2020 Census.)

 

Data retrieval

We used the 2020 Decennial Census API (Application Programming Interface) to retrieve Census-tract-level data that could then be grouped into PUMA-level data. These metrics included relevant Demographic and Housing Characteristics variables from groups H14 (Tenure by Household Type by Age of Householder*) and P20 (Households by Type and Presence of Own Children Under 18 Years).

 

The Census also provides a file that lists the 2020 PUMA that corresponds to each 2020 Census tract. We used this file to add PUMA information to each Census tract in our table, then created a new PUMA-level table that stored the sums of all values for all tracts within each PUMA.

 

We then used these PUMA-level sums to calculate relative percentages for our analysis. For instance, to calculate the percentage of householders aged 15-64 who were married, we first calculated the number of householders aged 15-64 who were married (i.e. H14_005N + H14_039N + H14_006N + H14_040N), then divided this sum by the total number of householders aged 15-64 (i.e. H14_005N + H14_006N + H14_014N + H14_015N + H14_010N + H14_011N + H14_029N + H14_030N + H14_033N + H14_034N + H14_020N + H14_021N + H14_024N + H14_025N + H14_039N + H14_040N + H14_048N + H14_049N + H14_044N + H14_045N + H14_063N + H14_064N + H14_067N + H14_068N + H14_054N + H14_055N + H14_058N + H14_059N).

 

Meanwhile, to calculate the percentage of households with their own children that were headed by married couples, we divided the number of married-couple households with children of their own under 18 (P20_003N) by the total number of households with children of their own under 18 (P20_003N + P20_006N + P20_011N + P20_017N). (All references to children in our analysis and this brief refer to householders' own children rather than related children; see the previous hyperlink for more details on the definitions of these terms.)

 

Analyzing PUMA-level data

In order to group PUMAs into 'deserts' and 'gardens,' we first needed to determine the percentage cutoffs for those groupings.

 

Age groups

Typically, in our research, we focus on prime-working-age adults (i.e. those aged 18-54 or 21-54) when analyzing marriage and family trends. However, the 2020 Census only made marital status data available for three householder age ranges within the H14 group: 15-34; 35-64; and 65+. Therefore, we chose an age range of 15-64 for our analyses of marriage prevalence. (In order to limit the effects of widowhood on our marriage prevalence calculations, we did not include the 65+ age group.)

 

Given that the average age at first marriage has been trending higher, we could also have set a higher minimum age for our marital-share analyses, such as 35. However, selecting a 35-64 age range rather than our 15-64 range would have obscured this delayed-marriage trend, which we believe is worth capturing as well. (It is worth noting that this wider age range likely resulted in lower marriage prevalence calculations in areas with large shares of young adults, such as college towns.)

 

Classifying PUMAS into marriage 'deserts' and 'gardens'

For our % married maps, we found the PUMA-level percent-married values that corresponded to the 10th, 50th, and 90th percentiles and rounded them to the nearest integer. (The resulting values were 33%, 49%, and 61%, respectively.)

 

PUMAs in which fewer than 33% of householders aged 15-64 were married were considered 'deserts'; those with shares greater than or equal to 33%, but less than 49%, were classified as 'lower share' regions; those with shares greater than or equal to 49%, but less than 61%, were classified as 'higher share' regions; and those in which 61% or more of householders aged 15-64 were married were classified as 'gardens.'

 

This approach means that, regardless of the actual prevalence of married households in the US, around one in every ten PUMAs will be classified as a desert, and another one in every ten PUMAs will be classified as a garden. An alternative methodology would have been to set a specific numerical value for 'deserts' and 'gardens', but the approach we took helps highlight areas with particularly high or low rates.

Classifying regions into married-family deserts and gardens

We used a similar approach for our married-family maps. The rounded 10^th, 50^th, and 90^th PUMA-level percentiles for our married-family metric (i.e. the % of households with children that are headed by a married couple) were 52%, 66%, and 79%, respectively.

 

As a result, PUMAs in which fewer than 52% of households with children were headed by a married couple were classified as deserts; those with a percentage greater or equal to 52%, but less than 66%, were classified as lower-share regions; those with a percentage greater to or equal to 66%, but less than 79%, were classified as higher-share regions; and those in which married-couple households represented at least 79% of households with children of their own were classified as gardens.

 

Here is a table that summarizes these cutoffs:

 

Metric	Region	Desert cutoff (rounded 10th percentile)	Median (rounded 50th percentile)	Garden cutoff (rounded 90th percentile)
% of householders aged 15-64 who are married	PUMA	33	49	61
% of households with kids headed by married couple	PUMA	52	66	79

 

Mapping

These maps were rendered using Plotly's choropleth_map() function and PUMA shapefiles from the Census Bureau. (In order to reduce map size and load time, these shapefiles were simplified using Geopandas' simplify_coverage() function.) Deserts are colored purple; lower-share regions have a very light orange color; higher-share regions are colored light orange; and gardens are colored orange.

 

Categorical versus gradient

We could also have used a gradient scale rather than a categorical one. However, the current approach and choice of colors helps highlight deserts while also making gardens easier to identify. 

 

Cartograms

We also created cartogram versions of our maps. The %-married cartogram is available here, and the married-family cartogram can be found at this link. These maps make each PUMA's area roughly equivalent to its underlying relevant population (i.e. householders aged 15-64 for our %-married maps and households with children for our % intact maps).

 

These cartograms make small, dense urban deserts and gardens easier to identify while also limiting the prominence of large, sparsely-populated regions. As a result, they better illustrate what percentage of the country's total population lives in a desert, garden, or other type of area.

 

We also tried out a 3D copy of our PUMA-level map in which the height of each region represented its density. (As a result, the volume of each region—i.e. its area times its height—would represent its actual population.) However, this map proved more difficult to interpret than the cartograms, as particularly tall PUMAs ended up blocking the view of those PUMAs behind them.

 

Grouping 'deserts' and 'gardens' together

To create our grouped-PUMA maps and tables, we used Geopandas, a Python library, to group contiguous desert and garden PUMAs together. (PUMAs that met only at one point were not considered contiguous.)

 

The following two screenshots of DC-area PUMAs illustrate this process. The first image shows ungrouped PUMAs:

 

 

 

Meanwhile, the following image shows the same marriage-desert PUMAs within the grouped-PUMA map. Note that the borders around the DC-area marriage deserts have been removed, thus creating one single grouped PUMA. Also note that desert and garden PUMAs with no adjacent PUMAs of the same category were still retained within this grouped map.

 

 

Determinations about which PUMAs to group together were based solely on the boundaries of our simplified PUMA shapefiles. For instance, the two marriage-garden PUMAs in the following screenshot, one in southeastern Pennsylvania and the other in northeastern Maryland, were grouped together because a tiny portion of their boundaries overlapped across the Susquehanna river. While it would have been justifiable to keep these two PUMAs separate, the approach we took made it easier to treat all PUMAs in a consistent manner.

 

 

Once our PUMAs were grouped together, we then ranked them by relevant household and householder counts. Grouped marriage-desert and marriage-garden PUMAs were ranked by the total number of householders aged 15-64 (regardless of marital status). Meanwhile, grouped married-family-desert and married-family garden PUMAs were ranked by the total number of households with children of their own (whether or not those households were headed by a married couple).

 

 

Checking our findings with American Community Survey microdata

 

The 2020 Decennial Census's massive sample size made it an ideal choice for our analysis. However, one major limitation of this data is that it only provided household-level data for our metrics of interest. Multiple factors (roommates, multigenerational households, etc.) can cause the % of married householders to differ from the % of married adults. Similarly, differences in family sizes can cause the % of households with children that are led by a married couple to differ from the % of children in married-couple households.

 

Therefore, both to check the validity of our decennial-Census analyses and to provide results at the individual level, we performed PUMA-level analyses of IPUMS-provided microdata from the 2023 American Community Survey 5-year dataset. (This dataset encompassed the years 2019 through 2023.) Because PUMA names were provided for each respondent, there was no need to map tracts to their corresponding PUMAs (as we did for our main analysis).

 

An alternative approach would have been to use pre-tabulated PUMA-level estimates; however, the microdata file allows for more flexible filters and simpler confidence interval calculations (which we computed within R).

 

To calculate the percentage of adults aged 15-64 who were married, we determined which adults had a MARST value of 1 (Married, spouse present) or 2 (Married, spouse absent), then divided the weighted sum of adults with one of these values by the sum of all adults within this age range. (Note that the 'Married, spouse absent' condition is distinct from the 'Separated' condition; we did not count separated adults as married for the purpose of this analysis. Visit this Census page for more details on the definitions on these terms.)

 

For our married-family calculations, we first determined whether each individual under age 18 had married parents. To do so, we linked unmarried children's records to their parents via IPUMS, then classified these children as being in a married-couple family if either their mother or their father had either a 'married, spouse present' status or a 'married, spouse absent' status.

 

Our methodology for this analysis was similar to that for our Decennial-Census analyses: we calculated rounded 10th, 50th, and 90^th percentiles, then used these percentiles to classify regions as deserts, lower-share areas, higher-share areas, or gardens. The following table shows the values of these percentiles for both metrics.

 

Metric	Region	Desert cutoff (rounded 10th percentile)	Median (rounded 50th percentile)	Garden cutoff (rounded 90th percentile)
% of adults aged 15-64 who are married	PUMA	35	47	56
% of children in married-parent families	PUMA	49	65	80

 

Overall, our individual-level results from this analysis of American Community Survey microdata lined up quite well with the household-level data that we retrieved from the 2020 Decennial Census. The Pearson correlation between our household and individual-level marriage metrics was 0.910, and the correlation between our household and individual-level married-family metrics was 0.939. In addition, 78.5% of PUMAs were classified in the same category category (garden, desert, etc.) for our marriage metrics, and 80.1% ended up in the same category for our married-family metrics.

 

 

 

*Here is how the Census Bureau defines 'Householder': 

 

The householder refers to the person (or one of the people) in whose name the housing unit is owned or rented (maintained) or, if there is no such person, any adult member, excluding roomers, boarders, or paid employees. If the house is owned or rented jointly by a married couple, the householder may be either the husband or the wife. The person designated as the householder is the "reference person" to whom the relationship of all other household members, if any, is recorded. The number of householders is equal to the number of households. (Source)

 

 

 

AUTHORS

 

Chris Bullivant

Chris Bullivant is Senior Fellow of the Social Capital Campaign. He was formerly Director of Communications at the Institute for Family Studies, and previously established the Social Capital Campaign, UK 2020 (a pro-Brexit think tank in London), and online commentary magazine UnHerd. He was a deputy director of the Centre for Social Justice in London with his early career based in nonprofits. His writing has been published in the USA Today, The Washington Examiner, The American Conservative, Deseret News, and more. He has a Masters in Political Sociology from the London School of Economics.

 

Ken Burchfiel

Ken Burchfiel is a Research Fellow at the Institute for Family Studies. Before joining the IFS, he served as a mental health counselor with Catholic Charities of Evansville and worked in a data analytics role at Seton Education Partners. He received a Master of Science in Social Work from the University of Texas at Austin and an MBA from Columbia Business School.