The research, published in the International Journal for Equity in Health, drew on the Brazilian National Health Survey (PNS), a large household survey conducted by the Brazilian Institute of Geography and Statistics that samples adults aged 23 years and older. The team, led by Luciana Alves de Medeiros and Catarina Machado Azeredo of the Federal University of Uberlândia, together with Renata Bertazzi Levy of the University of São Paulo and the Instituto de Investigación Biomédica de Salamanca, compared obesity prevalence across two survey waves, 2013 and 2019, using body mass index calculated from self-reported weight and height. Rather than simply comparing averages between groups, the researchers deployed sophisticated measures of health inequality that are increasingly standard in the global health literature: the Slope Index of Inequality (SII) and the Relative Index of Inequality (RII).

These two indices deserve explanation, because they are what allow the study to make claims about gradients rather than mere differences. The Slope Index of Inequality expresses, in percentage points, the absolute difference in obesity prevalence between the most educated and least educated individuals after ranking the population by educational attainment. A positive SII means obesity is more common among the more educated; a negative SII means the opposite. The Relative Index of Inequality expresses the same contrast as a ratio: an RII of 0.5, for instance, indicates that the most educated group has roughly half the obesity prevalence of the least educated, while an RII greater than one signals that obesity increases with education. By computing these measures separately for men and women, and within categories of race and skin color as self-identified in the Brazilian census tradition—White, Black, and Brown (pardo)—the team was able to probe what epidemiologists call intersectionality: the idea that social categories do not act independently but combine to produce unique patterns of health risk.

The headline finding concerns the direction of the educational gradient. In high-income countries, decades of research have established that obesity concentrates among the least educated, reflecting the greater availability of cheap, calorie-dense ultra-processed foods in disadvantaged neighborhoods and the protective effect of health literacy, leisure-time exercise, and access to fresh foods among the better off. As countries undergo what demographers call the nutritional transition, the burden of obesity is expected to migrate down the socioeconomic ladder. Brazil, the new analysis suggests, is following this script—but only for part of its population.

Among women, the expected inverse relationship between education and obesity was clearly visible for White and Brown women in 2019. White women showed a Slope Index of Inequality of −15.0 percentage points, meaning that obesity prevalence fell by fifteen percentage points as one moved from the least to the most educated; Brown women showed a negative SII of −9.7 percentage points. Their relative indices, 0.5 and 0.7 respectively, confirmed that higher education was associated with substantially lower obesity prevalence. But the same pattern was not statistically significant for Black women, a striking silence in the data that the authors interpret as evidence that education confers a weaker protective effect against obesity for this group.

For Brown women, the study also detected change over time. The relative index of inequality shifted from 0.5 in 2013 to 0.7 in 2019, a weakening of the protective gradient that the researchers attribute largely to rising obesity among women with secondary and tertiary education. In other words, the obesity epidemic in Brazil is no longer sparing educated women, particularly Brown women, and the social gradient that once clearly separated them from their less educated peers is flattening. This is consistent with the idea that as ultra-processed foods and sedentary lifestyles saturate an entire society, education loses some of its shielding power.

Among men, the picture is nearly inverted. Obesity in Brazilian men increases, rather than decreases, with education—a pattern frequently observed in early and middle stages of the nutritional transition, when sedentary white-collar work, restaurant meals, and higher disposable income accompany schooling. Crucially, the new study shows that this positive gradient is not uniform across racial groups. In 2019, both absolute and relative educational inequality in obesity were significantly higher for Black and Brown men than in 2013, with Slope Index values of 9.9 and 8.5 percentage points respectively, indicating that obesity prevalence among the more educated Black and Brown men exceeded that of their less educated counterparts by roughly nine to ten percentage points. Among White men, by contrast, no significant educational inequality in obesity was detected at all.

Perhaps the most revealing temporal finding concerns Black men: between 2013 and 2019, the burden of obesity shifted from the less educated to the more educated within this group. That reversal within a single six-year window offers a rare real-time glimpse of the nutritional transition at work, and it suggests that the pace of change can differ sharply across racial categories even within the same national population. For Black men, rising education appears to have become associated with rising obesity risk, a combination that complicates any assumption that socioeconomic advancement automatically translates into better cardiometabolic health.

The authors situate these findings within Brazil’s broader nutritional and epidemiological context. Over recent decades, Brazilian diets have shifted markedly toward ultra-processed foods, a trend documented extensively in national dietary surveys and linked to weight gain, and physical activity patterns have grown more sedentary. In such an environment, the classical inverse relationship between education and obesity among women is beginning to erode, while the positive gradient among men persists or intensifies. Overlaying race on this picture, the study argues, reveals structural disadvantages that would otherwise be invisible: Black and Brown individuals of both sexes fare worse, whether because education buys less protection for Black women or because the obesity risk of upward mobility weighs more heavily on Black and Brown men.

Methodologically, the study has both strengths and limits worth noting. The use of two waves of a nationally representative survey with large sample sizes lends statistical power, and the SII and RII indices are considered gold-standard measures for summarizing socioeconomic gradients in health because they take the full distribution of education into account rather than comparing only extreme categories. The intersectional stratification—crossing sex with race and education—represents a methodological commitment to the idea that health inequities are multiply determined. On the other hand, obesity was classified using self-reported weight and height, which is known to introduce measurement error, typically underestimating BMI; the authors also note that the cross-sectional design of each survey wave, even with two time points, cannot establish causation. Education is a marker of many things—income, occupation, neighborhood, food environments, stress exposure—and the survey data cannot disentangle which of these pathways drives the observed gradients.

The public health implications are considerable. Brazil’s national obesity rates have climbed steadily enough that the health ministry has repeatedly revised its surveillance projections upward, and noncommunicable diseases linked to excess weight now rank among the country’s leading causes of death and disability. If anti-obesity policy continues to be designed around the assumption that risk concentrates among the poor and less educated, it may miss growing segments of the educated population—Brown women with secondary and tertiary education, for instance—while simultaneously failing the Black and Brown men for whom education appears to magnify rather than mitigate risk. Targeted interventions, the authors imply, must be intersectional by design: food labeling and ultra-processed food regulation, areas where Brazil has been a global leader with its front-of-package warning labels, will need to be evaluated for their differential reach across racial and educational groups.

The study was funded by the Brazilian National Council for Scientific and Technological Development (CNPq) and approved by the Brazilian National Research Ethics Committee, with all participants providing written informed consent. As Brazil—and much of the middle-income world—continues its nutritional transition, this research serves as a reminder that averages conceal as much as they reveal. The obesity epidemic, the data insist, does not distribute itself randomly, nor does it follow a single socioeconomic gradient. It follows the fault lines of history, race, and gender, and only analyses sharp enough to see those intersections can guide policies that reach everyone the epidemic touches.

The distinction between being childfree and simply not yet having children is central to the study. Researchers define childfree adults as people who do not have children and do not want to have children in the future. This category therefore excludes adults who are childless because of infertility, circumstance, delayed family planning or uncertainty about whether they will eventually become parents. According to figures from the National Survey of Family Growth, the childfree share of non-parents aged 15 to 44 increased from 13.8 percent in 2002 to 29.4 percent in 2023. The new research examines how that expanding population is distributed geographically.

Zachary Neal and Jennifer Watling Neal of Michigan State University analyzed responses from 20,118 adults who participated in an online survey conducted between August 30 and October 8, 2024. The data came from the Civic Health and Institutions Project, a survey initiative designed to examine social attitudes, civic life and institutional conditions across the United States. By linking respondents’ reported reproductive intentions with their state of residence, the researchers estimated the prevalence of childfree adults in each state and tested whether state-level characteristics could help explain the differences.

The resulting map reveals substantial variation. California had the highest estimated proportion of childfree adults, at approximately 22 percent, followed by Michigan at 19.9 percent and New Hampshire at 19.4 percent. At the other end of the distribution, Louisiana had the lowest estimated share, at 9.4 percent, while North Dakota and South Dakota each registered about 10 percent. In practical terms, the estimates suggest that roughly one in five adults in some states may be childfree, compared with about one in ten in others.

Despite these differences, the pattern did not form the kind of clear regional clusters researchers might expect if childfree living were primarily driven by a shared political culture or geographic identity. The study found no strong evidence that childfree adults were concentrated in broad regions such as the Northeast, South, Midwest or Southwest. Instead, the highest proportions appeared in certain individual states, making the national distribution more fragmented than a simple cultural map of the country.

Some state characteristics were associated with higher childfree prevalence. Childfree adults were more common in highly populous states and in states with a higher cost of living. These relationships may reflect the pressures and opportunities associated with urbanized economies. Housing, childcare, education, healthcare and career advancement can all influence the perceived financial and logistical consequences of parenthood. Large states also tend to contain metropolitan areas where alternative lifestyles are more visible, social networks are more diverse and traditional expectations about marriage and family may be less restrictive.

However, state-level associations became less important when the researchers considered individual characteristics. Adults who were single, older or male were more likely to be childfree, according to the analysis. These factors do not necessarily operate independently. Relationship status can affect both the likelihood of having children and the expectations surrounding future parenthood, while age changes the probability that a person who currently has no children will eventually become a parent. Sex differences may reflect a combination of reproductive expectations, social pressure, relationship patterns and the way people report their intentions.

The researchers’ statistical approach is important because it separates contextual effects from individual effects. A state may have a high childfree rate not because living there directly causes people to reject parenthood, but because the state contains a larger share of people with characteristics already associated with being childfree. This distinction is known as compositional versus contextual explanation. A compositional explanation points to the makeup of the population; a contextual explanation suggests that the surrounding environment changes individual behavior. In this study, composition appeared to matter more than context.

The findings also complicate efforts to use state policy as a tool for reversing population decline. Governments concerned about falling birth rates sometimes consider policies intended to encourage parenthood, such as financial incentives, expanded childcare, housing assistance or restrictions affecting reproductive decisions. But if childfree intentions are driven predominantly by individual circumstances and durable preferences, state policies may have limited power to change them. The study does not show that state conditions are irrelevant, nor does it establish that people’s choices are completely fixed. It indicates instead that broad state characteristics alone are insufficient to explain why some adults do not want children.

The study has limitations. Because the survey data are observational, the researchers can identify associations but cannot prove that living in a particular state causes someone to become childfree. The analysis also operates at the state level, meaning it cannot reveal how childfree adults are distributed within cities, suburbs or rural communities. Online surveys may introduce sampling and reporting biases, and reproductive intentions can change over time for some respondents. Even so, the results offer one of the clearest national pictures yet of where childfree adults are estimated to live and why geography alone provides an incomplete explanation.

As the childfree population grows, the issue is becoming increasingly visible in debates about demographic change, housing, work and public policy. The new study suggests that the United States does not have a single “childfree region.” Instead, childfree adults are spread across the country, with their prevalence varying from state to state but their individual profiles proving more informative than any simple geographic divide. California and Michigan may have roughly one in five adults who are childfree, while Louisiana and the Dakotas may have closer to one in ten, but the deeper story is national: the decision not to become a parent is shaped primarily by people’s lives, identities and circumstances rather than by a single state-level formula.

Subject of Research: Not applicable

Article Title: The spatial distribution of the U.S. childfree population: Does state context matter?

News Publication Date: 12-Aug-2026

Web References: https://doi.org/10.1371/journal.pone.0352872

References: Neal ZP, Neal JW (2026) The spatial distribution of the U.S. childfree population: Does state context matter? PLOS One 21(8): e0352872. https://doi.org/10.1371/journal.pone.0352872

Image Credits: Neal, Neal., 2026, PLOS One, CC BY 4.0

Keywords: childfree adults, United States, reproductive intentions, population distribution, demography, family formation, social science, survey research, state context, birth rates