The U.S. Census Bureau’s new report on the living arrangements of U.S. children has seriously misinformed the public. The report’s highly problematic definition and use of the term “traditional nuclear family” has led most news accounts of the study to report erroneously that the proportion of U.S. children living with their two married, biological parents rose from 51 percent in 1991 to 56 percent in 1996. In fact, data from the report itself reveal that about 62 percent of all U.S. children in 1996 lived with their married, biological (or adoptive) parents, and no evidence presented in the report justifies the assertion that the proportion of children living with both biological parents rose in the 1990s.
These errors demand immediate correction. More broadly, they underscore the fact that, to date, the Census Bureau has failed to collect and competently report trend line data on one of the key issues facing our society: the proportion of children growing up with their two married parents.
The Public is Misinformed
On Friday, April 13, 2001, the U.S. Census Bureau released a new report, Living Arrangements of Children: Fall 1996, which contained an apparently startling piece of news: The intact family is making a comeback. The press release accompanying the report was titled: “The ‘Nuclear Family’ Rebounds, Census Bureau Reports.” The first sentence of the press release presented the main finding: “The proportion of children living in a traditional nuclear family with their biological mother and father increased from 51 percent in 1991 to 56 percent in 1996.”
In light of seemingly endless reports in recent decades showing that increases in divorce and unwed child bearing mean that steadily smaller proportions of U.S. children are growing up with their own two married parents, news organizations across the country understandably treated this new finding as major news. The April 13 page one story in USA Today began this way: “The traditional nuclear family – a married mom and dad living with their biological children – is making a comeback, according to a Census report released today.” And then a reporting of the specific finding: “The proportion of the nation’s children living with both biological parents jumped from 51 percent in 1991 to 56 percent of 71.5 million children in 1996.” That evening on television, on ABC’s World News Tonight, Peter Jennings reported: “The Census Bureau said today that the number of children who live with both their parents increased during the 1990s. The percentage of children in these families grew from 51 to 56 percent between 1991 and 1996.”
The Christian Science Monitor began its story this way: “The nuclear family is staging a comeback. Between 1991 and 1996, the percentage of American children living with both biological parents jumped from 51 percent to 56 percent.” The Washington Post began its story this way: “The share of children living with their biological parents in a married-couple family rose during the 1990s.”
The story from the Associated Press, which was used in scores of newspapers around the country, began this way: “Ozzie and Harriet are not dead yet. The prototypical nuclear family of black and white TV – where mom, dad and their biological children all live together – may not be as endangered as it sometimes seems. The percentage of children living in these traditional families rose during the early 1990s, from 51 percent in 1991 to 56 percent in 1996.”
An increase of this magnitude, in only five years, in the proportion of U.S. children living with their own two married parents would be very important news. And presumably, for most people – certainly for us – very good news. The only problem is, it ain’t true.
The Untrue Assertion
Consider again the (absolutely typical) evidence summation sentence in the story from USA Today: “The proportion of the nation’s children living with both biological parents jumped from 51 percent in 1991 to 56 percent of 71.5 million children in 1996.” No part of that sentence is true. In 1996, the proportion of U.S. children living with both biological parents was significantly higher than 56 percent. There is no evidence in the Census Bureau report that the proportion of children living with both biological parents was any higher in 1996 than it was in 1991. And finally, it is not true that 56 percent of 71.5 million children is the number of U.S. children in 1996 living with their biological parents.
Confirmation from the Census Bureau
Interviewed on April 15, Jason Fields of the Census Bureau, the author of Living Arrangements of Children: Fall 1996, confirmed that the data presented in this report do not show that the proportion of children living with both biological parents is increasing. Compared to 1991, “Children [in 1996] are neither more or less likely to live with their own biological parents,” Fields told us. Instead, he stated, the main reason why more children in 1996 were classified as living in “traditional nuclear families” seems to be an increase since 1991 in the likelihood that married couples with children were living on their own, without other relatives or unrelated individuals living in the household.
The Flawed Methodology
What went wrong? Basically, the Census Bureau has arbitrarily conflated two separate issues. The first issue is the proportion of children living with their two biological parents (and within that category, the proportion of these households in which the parents are married to one another). The second issue is who else – grandparents, other relatives, other individuals – is or is not residing in these particular family households. For some reason, the Census Bureau chose to merge these two issues. Specifically, the Census Bureau created a category of family household that it provocatively calls the “traditional nuclear family.” For this report, the Census Bureau defines such a family as a household containing two married parents, their biological offspring under age 18, and no one else.
As a result of this definition, any household characteristic that tilts a family in the direction of a three-generation household, an extended-family household, or simply a large household – toward what in many respects is the very definition of “traditional” family life – immediately knocks them out of the Census Bureau’s “traditional” box. Similarly, in the scheme of things designed by the Census Bureau, within the universe of two-biological-parent families, any decrease in the proportion of children living three-generation, extended-family, and large or complex households almost necessarily means an increase in the proportion of children living in “traditional” households. And that appears to be exactly what happened in the U.S. in the early and mid 1990s. Fewer married couples with children taking in a relative or a boarder, thus resulting (in Census Bureau parlance) in more “traditional” families.
Consider some of the implications of this curious way of measuring marriage and family life. Take the hypothetical case of Fran and Jim, married to one another and living together with their two teenage children in River City, Iowa. So far, they are what the Census Bureau calls “traditional.” But if Fran’s aging mother, Betty, gets sick, can’t take care of herself any more, and moves in with Fran and Jim, Fran and Jim’s household is no longer “traditional.” If Fran’s sister and brother-in-law die tragically in a car accident, and their orphaned child, Fran and Jim’s niece, comes to live with her aunt and uncle in River City, Fran and Jim’s household is no longer “traditional.” If Jim loses his job, and the couple rents out one room in their home to a boarder in order to help pay the mortgage, they are no longer “traditional.” If their grown daughter, Alice, who lives in Connecticut, gets a divorce, returns to River City with her young child, and moves back in with her parents, Fran and Jim’s teenage children no longer live in a “traditional” household.
On the other hand, if Fran and Jim permit none of these changes – if they put Grandma Betty in a nursing home instead of the upstairs bedroom; if they send their orphaned niece off to foster care; if they make sure to take in no boarders, even a family friend who is willing to teach swimming to their two children; and if they tell their daughter Alice to stay in Connecticut and learn to stand on her own two feet as a single mother – they are still a “traditional” family household.
Demographically, what appears to have occurred in the U.S. in the early and mid 1990s is a decrease in the proportion of married-couple families with children living in extended-family and large or complex households. The likely reasons for this trend would include a booming economy, which makes it easier for couples to get by without sharing living space, and an increase in couples who delay having children until they are in their 30s, when they are more likely to be able to afford their own households. This decrease had no effect either way on the proportion of U.S. children living with their two biological parents. Its only effect was to increase the size of an artificial category that the Census Bureau calls “traditional nuclear families.”
It is this statistical artifact of a problematic research model that apparently led to the Census Bureau press release suggesting that the married-couple home is “rebounding,” when in fact, based on the evidence presented, nothing of the sort is taking place. This flawed model also led directly to the clearly false “finding” that was reported in good faith by news organizations across the country. Recall, for example, the Christian Science Monitor: “Between 1991 and 1996, the percentage of American children living with both biological parents jumped from 51 percent to 56 percent.”
The U.S. Census Bureau is trusted and respected by scholars, journalists, policy makers, and the general public. But in this case, it has constructed a category for measurement that has obviously frustrated rather than met the public’s desire to know what is actually going on in U.S. families. Why would the Census Bureau do such a thing? We don’t know. But we are convinced, and we are confident that many leading marriage and family scholars will concur, that the Census Bureau in this case has created much unnecessary confusion, and in doing so, has clearly done a disservice to the public.
In the view of numerous leading family scholars, a key trend to measure – a primary indicator of child and societal well-being – is the proportion of children living with their own two married parents. The body of social science evidence to support this point is large and growing. The ways in which people use these words in everyday life are similarly revealing. When people speak of an “intact” or “traditional” family, they are typically thinking of children living in a two-biological-parent, married-couple home. When experts or ordinary citizens opine that we need to “strengthen” the family, they typically mean that we need less divorce and unwed child bearing, and more two-parent, married-couple homes.
On the other hand, there is no social science evidence suggesting that renting a room to a friend, or letting Grandpa or a cousin live with you, in any way undermines family bonds, harms children, or weakens the family as a social institution. Nor would popular opinion support the idea that such things constitute worrisome departures from what is “traditional.” Indeed, in most cases, quite the opposite.
The result is that the Census Bureau has defaulted on its attempt to measure a trend that would be almost certainly beneficial (more two-parent homes) by entangling it conceptually with a trend that is arguably harmful (fewer three-generation and extended homes). Moreover, given the ways in which they intertwine their measurements of these two trends, their definition of “traditional nuclear families” quickly leads most readers, and in fact has now led almost all journalists, to the empirically false conclusion that, within the universe of two-parents-with-biological-children homes, a decline in the proportion of three-generation and extended households somehow mean an increase in the proportion of children living with (take your pick) “their biological parents” or in “traditional” families. The main result of this flawed procedure and confusing presentation is a seriously misinformed public.
What’s the Truth?
Based on the data presented in the detailed tables in the report, including Internet tables, here is what we believe to be the statistical truth of the matter. In 1996, of all U.S. children under age 18, about 62 percent were living with both of their biological (or adoptive) married-to-each-other parents. If we remove the criterion of marriage, the proportion of U.S. children living with their two biological or adoptive parents (including parents who are cohabiting but not married) is about 64 percent. The figure of 56 percent, so widely reported in the media, is clearly inaccurate.
As the Census Bureau’s Jason Fields, the report’s author, has now confirmed, no evidence presented in the report makes it possible to determine whether the proportion of U.S. children living with their two biological parents has gone up, gone down, or stayed the same since 1991. Consequently, the notion of the “rebound” of the two-biological-parent home, so widely reported in the media, is clearly unsupported by the data in this report. Our own best estimate at this time is that the proportion of U.S. children living with their own two parents remained roughly constant during the early and mid 1990s. The proportion of U.S. children living with two parents (including stepparents) continued its decline, from about 73 percent in 1990 to about 71 percent in 1996.
The Reporting or the Report?
If the accuracy of this report were being tested in the scholarly equivalent of a court of law, the Census Bureau could cling, probably successfully, to a formal defense. Critics may not agree with the report’s intellectual assumptions and procedures, or the provocative way in which the Census Bureau chose publicly to announce its findings, but the Census Bureau does, right there in the report, spell out its definitions (however confusing) and its methodology (however flawed). So the Census Bureau did not, technically speaking, make demonstrably false empirical assertions. From this perspective, journalists and others interested in this report should simply have done more or better homework before reaching their conclusions. Well, maybe.
But consider this except from the report: “However, the data for the 1990s indicate that the rapid increase in the proportion of children growing up in a single-parent household may have leveled off. Data from the SIPP show an increase in the proportion of children living in traditional two-parent nuclear families. . .” Reading these two sentences at face value, it is hard to escape the conclusion that the Census Bureau is telling us that fewer children are growing up in one-parent homes and that more children are growing up in two-parent homes. But of course, based on the data in the report, and as confirmed in our interview with the report’s author, no such trends are occurring at all. Only the most persistent examination of the report’s definitions, and the most diligent digging into its various statistical tables and assertions, would reveal to the reader that these two sentences, taken together, are so misleading that they border on public education malpractice.
The U.S. Census Bureau is an important and usually highly competent federal agency. Why, then, has it proven to be so difficult for scholars, policy makers and citizens to get reliable trend line information from the Census Bureau regarding what many family scholars believe to be the single most urgent family question facing our society: What proportion of our children are growing up with their own two married parents, and what is the direction of the trend? It is distressing that, to date, the Census Bureau has been unable or unwilling to discover and competently report the answer to this fundamental question.
There is a small but significant and growing “marriage movement” in our society – scholars, policy makers, educators, counselors, religious and civic leaders and others who are working together to strengthen marriage. But they ought not to be forced to work in the dark. They need – we all need – to know the answers to basic questions about the state of marriage and child well-being in our society. This problem with the Census Bureau demands immediate correction.
- Jason Fields, Living Arrangements of U.S. Children: Fall 1996, Current Population Reports, P70-74 (Washington, D.C.: U.S. Census Bureau, April 2001).
- Haya El Nasser, “More children live in traditional families. Women marrying later in life may be behind trend reported by Census,” USA Today, April 13, 2001.
- The reporting on this issue on the April 13 edition of the CBS Evening News, with Dan Rather, was almost identical to that which occurred on ABC. The main point of the CBS report, which aired excerpts from an interview with Jason Fields, the Census Bureau family demographer who wrote the new Census report, was that a smaller proportion of U.S. children are living in one-parent homes, while a greater proportion are living in two-biological parent homes.
- Laurent Belsie, “Surprise: more nuclear families,” Christian Science Monitor, April 13, 2001; “More Children Living in `Traditional Families,’” Washington Post, April 13, 2001; and Laura Meckler, “American Families: More Traditional,” Associated Press, April 13, 2001. Here is the relevant evidence summation sentence from the Philadelphia Inquirer: “The study found that between 1991 and 1996, the share of children living with both of their biological, married parents and any full siblings – the traditional Ozzie-and-Harriet nuclear family – rose from 51 percent to 56 percent of all children.” Thomas Ginsberg, “Study sees increase in traditional families, cohabitation,” Philadelphia Inquirer, April 18, 2001.
- When we posed this question to Jason Fields of the Census Bureau, he replied: “That is the classification used in the 1991 report, and we wanted to be able to compare across two points in time.” The desire to compare two points in time is laudable, but it does not justify the confusion generated by this recent report. While the report for 1991 does, as Fields states, deploy the category of “traditional nuclear family,” it does so in a way that, in our view, is much clearer and therefore considerably less problematic. See Stacy Furukawa, The Diverse Living Arrangements of Children: Summer 1991, Current Population Reports, P70-38 (Washington, D.C.: U.S. Census Bureau, September 1994).
- For example, see Linda J. Waite and Maggie Gallagher, The Case for Marriage: Why Married People Are Happier, Healthier, and Better Off Financially (New York: Doubleday, 2000); Paul R. Amato and Alan Booth, A Generation at Risk: Growing Up in an Era of Family Upheaval (Cambridge, MA: Harvard University Press, 1997); Steven L. Nock, Marriage and Men’s Lives (New York: Oxford University Press, 1998); and Sara McLanahan and Gary Sandefur, Growing Up With a Single Parent: What Hurts, What Helps (Cambridge, MA: Harvard University Press, 1994).
- One exception is Cindy Rodriquez of the Boston Globe. In her story, she points to the decline (from 73 percent in 1991 to 71 percent in 1996) in the proportion of U.S. children living in a home with two parents (including cohabitants and stepparents). She also reports that the Census Bureau study shows “an increase in the percentage of children living in a traditional nuclear family.” She then writes: “The discrepancy comes in the way the Census Bureau defines families. A two-parent household can include non-married parents, stepparents, grandparents, or any other member of an extended family. The traditional nuclear family includes only the married biological parents.” While we believe that Rodriquez’s summary of Census Bureau definitions contains some inaccuracies, and also that the root of this report’s failure is not directly illuminated by the reported decline in two-parent homes, among the news accounts to date that we have seen, Rodriguez’s story comes closest to presenting accurately the actual meaning of the data contained in this report. See Cindy Rodriguez, “Line-up shifts in U.S. family portrait. Survey sees fewer two-parent homes,” Boston Globe, April 13, 2001.
Another exception is Elizabeth Simpson of the Norfolk, Virginia Virginian-Pilot. In her story, Simpson first accurately reports that “the percentage of children in single-parent families grew” during the early and mid 1990s. Then, after reporting a “slight rebound” in the proportion of children in “traditional nuclear families,” Simpson clearly defines this term for readers. Unfortunately, however, the headline for this story announces (wrongly) that: “Survey Shows More Children in Two-Parent Homes.” See Elizabeth Simpson, “Traditional Families Make Comeback in 90s. Survey Shows More Children in Two-Parent Homes,” Virginian-Pilot, April 13, 2001.
- About 70 percent of non-Hispanic white children under age 18 in 1996 lived with their two married biological (or adoptive) parents. For African American children, the figure is about 30 percent. For Hispanic children, the figure is about 59 percent.
- Data from this report show that, in 1996, about 2 percent of all children under age 18 were adopted. Of all children under age 18 living in two-parent homes in 1996, about 2.5 percent were adopted.
- See a jointly authored public statement, The Marriage Movement: A Statement of Principles (New York: Institute for American Values, 2000).