To a kid with a hammer, everything looks like a nail. So I used the same kind of figure for two different datasets. Materials at the end.
Marriage
Regardless of how you think about the causal relationship between marriage and men’s economic wellbeing, it’s an important fact that marriage in the US has become more economically polarized, with the social class gap in marriage prevalence widening.
Recently, Scott Galloway wrote a bad blog post about marriage and men, which included this truly terrible and misleading figure, which pours bad data analysis of the General Social Survey (see here) into a manipulated-axis clustermuck, which doesn’t even manage to show much of a correlation:
Anyway, Galloway also recycled a figure from bad 2012 blog post from the Hamilton Project. Bad work, but the trend is real, so I updated it and made a different kind of figure, using a heatmap with geom_tile in R, inspired by Kieran Healy’s Baby Boom heatmap. And I added women, separately.
Using the Current Population Survey (CPS) Annual Social and Economic Supplement (downloaded from IPUMS.org), I broke men and women down into 10 income deciles in each year from 1980 to 2021, and calculated the percentage of each cell that was married (and not separated) at the time of the survey. This is men:
This shows that rich men are much more likely to be married than poor men, and the gap has grown even as marriage rates have fallen across the board. The figure for women is more complicated, and is a good way to remind yourself that the causal story here is not as simple as some people make it sound.
In 1980, women with higher incomes (their own incomes) were the least likely to be married (not get married, be married). The most likely to be married were women with just a little income. Now, women with the highest incomes are more likely to be married than all but the bottom 20 percent. The biggest drop has been among women with low incomes. (Remember, these are cross-sections, so it’s not necessarily reflecting change over time in these women’s lives.) This is an inequality story, as high income women are more likely to be married (with spouses who have incomes as well), and low income women are more likely to be single (without spouses). Cohabitation, which is not included here takes some of the edge off this, but not that much.
Working from home
Starting in May 2020, some forward-thinking people at the Bureau of Labor Statistics added a question to the monthly CPS:
At any time in the LAST 4 WEEKS, did (you/name) telework or work at home for pay BECAUSE OF THE CORONAVIRUS PANDEMIC? (Enter No if person worked entirely from home before the Coronavirus pandemic)
At the time, the great majority of workers in some occupations — especially teaching — were working from home, as their workplaces were shut down by epidemic mitigation policies. Others, such as cooks and waiters, were either unemployed or working in dangerous conditions. Since that first survey in May (through August), the pattern has changed a lot, and there is much less teleworking. But some occupations are still staying home at pretty high rates, including college teachers, programmers, lawyers, and management analysts.
There is a sharp distinction between high- and low-telework occupations. It’s not quite a map of status and income, but it’s not not that, either. As in all things, apparently, the pandemic has been a seismic inequality event. Everything has changed, but very differently for different groups of people. More and different inequalities.
I can’t share the CPS data I got from IPUMS, but you can get it yourself with a free account. I shared the Stata code I used to manipulate the data, and the R code I used to make the figures, on the Open Science Framework, here: https://osf.io/2k86a/. My R skills are very limited so I just use it to make the figures, but if you are at a functioning beginning level the code might help.
Your work is important. Don’t hide it. (PNC video)
This post is about the practice of putting your dissertation under an embargo, which means your university library, and probably its agent, ProQuest, don’t let people read it for a certain amount of time, sometimes only a few months, sometimes many years. At my school, the University of Maryland, the graduate school is implementing a new policy that allows two-year embargoes without special permission (down from six years), and longer embargoes only with permission of the advisor and the dean.
Are you in this to advance knowledge? If so, don’t embargo your dissertation. By definition, a dissertation is a contribution to knowledge. By definition, keeping people from reading it stops that from occurring.
Many PhD graduates embargo their dissertations because it feels like the safer thing to do, because they’re vaguely worried about sharing their work, either because it’s so good someone will steal it, or it’s so bad it will embarrass them — or, weirdly, both. Many people don’t seriously think about it, don’t read up on the question, don’t discuss it with knowledgeable mentors (which your PhD advisor is very likely not, at least when it comes to this question). Lots of good people make this mistake, and that’s a shame. I’m writing this post so that, if you see it before you face this choice, there’s a chance my nagging voice will get stuck in your head.
Some graduate students think they’re being exploited and someone is going to make money off their work. Probably not. (You may have been exploited as a graduate student, and you might have good reasons for disliking your university, but this isn’t about making your university happy.) Maybe your dissertation will lead to an important book that lots of people will read — that is wonderful, and I hope it does. Of course, that’s a very small minority of dissertations, even among really good ones that make important contributions to knowledge. That’s just not in the cards in the vast majority of cases. But unless you already have a contract and a publisher telling you that without an embargo the deal is off — a situation that is vanishingly rare if it occurs at all, at least in sociology — making your dissertation publicly available will not hurt (and will probably help) your chances of accomplishing that goal. And if you’re going to publish articles based on your dissertation, no reputable journal will turn them away because they have overlapping content with your dissertation.
Some graduate students are afraid they will get “scooped” or their ideas will be “stolen.” This is profoundly misguided. You are doing the work so that people will read it. People are going to do what they do. You might be taking a small risk to your personal interest by making your work public, but consider it against the benefit of people reading it (which is, after all, the reason you should have written it). This is your finished work. It’s done. By definition it can’t be scooped. It can be plagiarized, like anything else. Would it be awkward or disappointing if someone published something similar that made similar contributions? Maybe. Will that substantially harm your career or personal interests? Very unlikely.* If you had a good idea, it will probably lead to more. Your ideas and your efforts in the dissertation are on the record now. Be proud of them, take credit for them, encourage people to engage with them, and hope that they will be inspired to do work that follows your lead. If your dissertation is good, it’s worth the risk — because you want people to read it. If your dissertation is bad, there is no risk anyway.
Will making your dissertation public hurt your chances of publishing a book? Almost certainly not. As an editor at Harvard University Press wrote:
“Generally speaking, when we at HUP take on a young scholar’s first book, whether in history or other disciplines, we expect that the final product will be so broadened, deepened, reconsidered, and restructured that the availability of the dissertation is irrelevant.”
And they quoted an assistant editor who went further: making your dissertation available improves your chances of getting a book contract:
“I’m always looking out for exciting new scholarship that might make for a good book, whether in formally published journal articles and conference programs, or in the conversation on Twitter and in the history blogosphere, or in conversations with scholars I meet. And so, to whatever extent open access to a dissertation increases the odds of its ideas being read and discussed more widely, I tend to think it increases the odds of my hearing about them.”
Or, as the editorial director at Columbia University Press, Eric Schwartz wrote in a tweet about sharing dissertations: “No problem. Book and dissertation are for different audiences.”
Of course there may be exceptions. If you have an editor on the hook who insists on an embargo, consider the pros and cons. If you have only a vague hope of publishing it down the road, don’t bother.
Do you want to win awards so everyone is talking about your dissertation? Don’t embargo it. Thanks to a 2015 change in policy at the American Sociological Association:
“To be eligible for the ASA Dissertation Award, nominees’ dissertations must be publicly available in Dissertation Abstracts International or a comparable outlet. Dissertations that are not available in this fashion will not be considered for the award.”
There are real, important principles at stake. Hate on your universities all you want, but some of their lofty rhetoric is true and good — and we should be holding them to it, not scoffing at it. Many universities, like the University of California system, have policies based on such high-minded statements as this:
“The University of California is committed to disseminating research and scholarship conducted at the University as widely as possible…. The University affirms the long-standing tradition that theses and dissertations, which represent significant contributions to the advancement of knowledge and the scholarly record, should be shared with scholars in all disciplines and the general public.”
Embargoing the work for years absolutely violates the spirit of such a principled policy, even if they do allow an embargo. Making your work accessible years later is clearly depriving the public of “significant contributions to the advancement of knowledge and the scholarly record” for the most important period in the life of the work — the years right after it’s done.
Here’s the statement from the University of Chicago:
“The public sharing of original dissertation research is a principle to which the University is deeply committed, and dissertations should be made available to the scholarly community at the University of Chicago and elsewhere in a timely manner. If dissertation authors are concerned that making their research publicly available might endanger research subjects or themselves, jeopardize a pending patent, complicate publication of a revised dissertation, or otherwise be unadvisable, they may, in consultation with faculty in their field (and as appropriate, research collaborators), restrict access to their dissertation for a limited period of time.”
Some people might skim through this policy and say, “Oh, cool, they allow an embargo,” and just check the box requesting it. But that’s making a powerful statement against the important principle articulated in this policy. If you don’t have a really good reason to embargo your dissertation — and you almost certainly don’t — the public interest demands that you make it public. Take the value of your work seriously. Not it’s commercial value, it’s actual value — which is to people who want to read it.
There is also an important accountability principle at stake. Should PhDs be awarded in secret, with no accountability beyond the committee room walls, until years later? For those of us on the faculty, how are we to evaluate programs and their candidates if we can’t scrutinize their most important works? How can we claim to be reputable programs if we shroud our work behind embargoes. Without at least this bottom-line transparency, there can be little accountability.
I write this post out of a certain sense of shame. I’m the director of graduate studies in our department, and I haven’t made it a priority to talk to students about this, because I didn’t know it was happening. When I looked at the dissertations from our department, which are archived in the Digital Repository at the University of Maryland (or, if they are embargoed, merely listed), I saw that among the last 19 dissertations, 12 were currently embargoed. The seven that were made public have been downloaded 1,200 times.
If you want to embargo your dissertation, or if someone is telling you that you should, the burden is on you (or them) to prove that the real benefits of the embargo — not just for you, but for the contribution to knowledge that your work represents — are greater than the harm of denying readers access to your research. The default must be to share our dissertations, with rare exceptions only when real (not imagined or rumored) circumstances demand that the public interest in access to knowledge be sacrificed.
* My dissertation, completed in 1999, although excellent, was not especially original. My major contributions were updating research on a longstanding theory to (a) use more recent data, (b) include women, and (c) use hierarchical linear models. My dissertation was titled, “Black Population Size and the Structure of United States Labor Market Inequality.” In 1997, as I was hard at work, and had a chapter under review at Social Forces (which I had already presented at two conferences), an article appeared (in Social Forces!) titled, “Black Population Concentration and Black-White Inequality: Expanding the Consideration of Place and Space Effects.” The authors used (a) the new data I was using, they (b) included women, and their (c) models were fancier than mine. I was crushed. And then, with my advisor’s help, I got over it. My article (with a citation to theirs added) got published the next year anyway, titled, “Black Concentration Effects on Black-White and Gender Inequality: Multilevel Analysis for U.S. Metropolitan Areas.” People read both articles. And then I went on to do a bunch more work in that area, with great collaborators, building up a body of research that drew from my dissertation but went much further in terms of theory, methods, and data. My article got cited plenty, partly because it was part of a group of articles that traveled together. I was “scooped,” but they didn’t get their ideas from sneaking a look at my brilliant work in progress, they were logical next steps in a 40-year trajectory of research on an established set of questions. Their publication strengthened the field in which I was working. (In fact, if they had stolen my ideas their paper would have been worse for them, and less damaging to me.)
After reviewing a paper for JAMA Network Open I was invited to write a comment about it. The paper is here, reporting a large drop in the percentage of mothers who are planning or thinking about having another child in a sample from New York City in mid-2020. After summarizing the results, I wrote this:
Before the COVID-19 pandemic, the US was in a period of declining fertility following the 2008 financial crisis and subsequent recession—a decline that was linked to economic precarity and hardship [2]. Then, in 2020, the total number of US births decreased 3.8%, which was the largest annual decline on a percentage basis since the early 1970s. The decreases were steeper at the end of the year, −6% in November and −8% in December, compared with 2019 [3]. In some large states with public monthly reports (California, Florida, and Ohio), it appears that January and February 2021 had fewer births still, with some recovery in the months that followed [4]. This timing suggests a direct association with the onset of the pandemic and closures that began in the spring of 2020. The evidence presented by Kahn and colleagues [1] supports this interpretation and suggests that when people faced the uncertainty and hardships associated with the pandemic, one common response was to pull back from plans to add children to their families. Future research will examine whether family decision-making in more advantaged families was similarly affected.
The current evidence concerns shifts in pregnancy planning. However, in the US, a substantial portion of births results from unintended or mistimed pregnancies, and these are concentrated among disadvantaged women [5]. The inability to predict, much less control, the trajectory of their lives leads many women to postpone the lifelong commitments implied by intentional births, but also makes unintentional pregnancy more likely. How the pandemic may have affected such births is not yet known. If mobility restrictions, unemployment, illness, care work burdens, and social distancing all reduced social interaction, coupled with increased motivation to prevent pregnancy, we may suspect unintended births will have declined as well.
The impacts of the pandemic within and between families points to the complex interrelationships among family structure, health disparities, and social inequality in the US [6]. The COVID-19 pandemic has been an inequality-exacerbating event on a large scale, widening existing health disparities, especially along the lines of socioeconomic status, race, and ethnicity. Excess mortality among Black and Hispanic populations in 2020, directly and indirectly related to the pandemic, far outstripped that seen among non-Hispanic White populations and contributed to the decrease in overall US life expectancy that exceeded that seen in peer countries [7]. In light of disparate impacts of COVID-19 itself and the social and economic fallout of the pandemic, research should concentrate on widening inequalities in fertility and family well-being, and their relationship to health disparities.
Corresponding Author: Philip N. Cohen, PhD, Maryland Population Research Center, Department of Sociology, University of Maryland, Parren J. Mitchell Art Sociology Building, College Park, MD 20742 (pnc@umd.edu).
Conflict of Interest Disclosures: None reported.
References
Kahn LG, Trasande L, Liu M, Mehta-Lee SS, Brubaker SG, Jacobson MH. Factors associated with changes in pregnancy intention among women who were mothers of young children in New York City following the COVID-19 outbreak. JAMA Netw Open. 2021;4(9):e2124273. doi:10.1001/jamanetworkopen.2021.24273
Seltzer N. Beyond the great recession: labor market polarization and ongoing fertility decline in the United States. Demography. 2019;56(4):1463-1493. doi:10.1007/s13524-019-00790-6
Cohen PN. Baby bust: falling fertility in US counties is associated with COVID-19 prevalence and mobility reductions. SocArXiv, March 17, 2021. doi:10.31235/osf.io/qwxz3
Hartnett CS, Gemmill A. Recent trends in US childbearing intentions. Demography. 2020;57(6):2035-2045. doi:10.1007/s13524-020-00929-w
Thomeer MB, Yahirun J, Colón-López A. How families matter for health inequality during the COVID-19 pandemic. J Fam Theory Rev. 2020;12(4):448-463. doi:10.1111/jftr.12398
Woolf SH, Masters RK, Aron LY. Effect of the covid-19 pandemic in 2020 on life expectancy across populations in the USA and other high income countries: simulations of provisional mortality data. BMJ. 2021;373(n1343):n1343. doi:10.1136/bmj.n1343
CNN’s Dr. Sanjay Gupta has a podcast called Chasing Life about coming out of the pandemic. Associate producer Grace Walker interviewed me for an episode titled, “Let’s Talk About Making Babies (Or Deciding Not To).” In it reporter Chloe Melas starts with the story of a Black couple (two women, one of them trans) seeking to have children. At about minute 21, she turns to the fertility decline in the US. The transcript of that part is below. This episode would be good for teaching.
Chloe Melas: But we can’t forget – not everyone wants to have children. And that’s OK. According to the CDC, the number of births in the United States fell by 4% last year – the largest annual decline since 1973. Given the global pandemic, for demographers like Philip Cohen of the University of Maryland, this isn’t too surprising.
Philip Cohen: What we’ve learned in the last century or so is that when there are crises birth rates go down. It’s partly deliberate, that is, people decide to hold off on having children, or decide against having children, because they’re unsure about the future, they’re unsure they’ll be able to care for them, they think they might lose their job, they think their mother might lose her job – all the things that go into the calculations of when and whether to have children.
CM: 2020 is not an outlier. Cohen says birthrates have been on a downward trend for quite a while.
PNC: We were sort of focusing on issues like work-family balance, childcare, healthcare, housing, the expenses of raising children, and the difficulty of raising children, which had been putting pressure on people to reduce their number of children. That’s the main reason. At the same time, when people have more opportunities to do other things in their lives, they’re also inclined to have fewer children, or delay having children. So especially for women, when opportunities improve, the number of children they have tends to go down, because on average they’re more likely to choose something else.
CM: Hispanic women in particular are seeing some of the largest declines. From 2007 to 2017 birth rates fell by 31%. Experts attribute this drop to more Hispanic women joining the workforce, and waiting longer to start families than previous generations. Overall, the data doesn’t lie. Fewer people are having kids. That could lead to smaller kindergarten classrooms, as well as larger demands on Social Security, given the aging population. But Cohen and others think there could be positives, too. For example, fewer people means less of an environmental impact on the planet. So it’s really a glass half empty, glass half full kind of situation. The point is, I think this pandemic has really made many of us reflect on what we want our future to look like, including our future families. Some have been inspired to freeze their eggs, some to seek out help for infertility, and some have decided against having kids while others have been inspired to do so.
The text and figures of this short paper are below, and it’s also available as a PDF on SocArXiv, in more citable form. The Stata code and other materials are up as well, here. It’s pretty drafty — very happy to hear any feedback.
Preamble: When Sabrina Tavernise, Claire Cain Miller, Quoctrung Bui and Robert Gebeloff wrote their excellent New York Times piece, Why American Women Everywhere Are Delaying Motherhood, they elevated one important aspect of the wider conversation about falling fertility rates — the good news that women with improving economic opportunities often delay or forego having children because that’s what they’d rather do.
But it’s tricky to analyze this. Consider one woman they quote, who said, “I can’t get pregnant, I can’t get pregnant… I have to have a career and a job. If I don’t, it’s like everything my parents did goes in vain.” Or another, who is waiting to have children till she finishes a dental hygienist degree, who said, “I’m trying to go higher. I grew up around dysfunctional things. I feel like if I succeed, my children won’t have to.” If people can’t afford decent childcare (yet), or won’t have a job that pays enough to afford the parenting they want to provide until they finish a degree — so they delay parenthood while investing in their careers — are they not having a baby because there are promising opportunities, or because of economic insecurity? These are edge cases, I guess, but it seems like they extend to a lot of people right now. That’s what motivated me to do this analysis.
Hard times and falling fertility in the United States
by Philip N. Cohen
Abstract
Recent reports have suggested that falling fertility in the US since the 2008 recession is being driven by women with advantaged status in the labor market taking advantage of career opportunities. This paper takes issue with that conclusion. Although high incomes are associated with lower fertility in general, both in the cross section and over time (within and between countries), economic crises also lead to lower fertility. I offer a new descriptive analysis using data from the American Community Survey for 2000-2019. In the U.S. case, the fertility decline was widespread after the 2008 recession, but most concentrated among younger women. Although women with above average education have long had lower birth rates, the analysis shows that birth rates fell most for women in states with higher than average unemployment rates, especially among those with below average education. This is consistent with evidence that birth rates are falling, and births delayed, by economic insecurity and hardship.
Introduction
A New York Times article by Sabrina Tavernise et al. was titled, “Why American Women Everywhere Are Delaying Motherhood” (Tavernise et al. 2021). Although it did not provide a simple answer to the question, it did offer this: “As more women of all social classes have prioritized education and career, delaying childbearing has become a broad pattern among American women almost everywhere.” And it included a figure showing birth rates falling faster in counties with faster job growth. Reading that article, the writer Jill Filipovic concluded, “the women who are driving this downturn [in fertility] are those who have the most advantage and the greatest range of choices, and whose prospects look brightest” (Filipovic 2021). This paper takes issue with that conclusion.
Clearly, one driver of delayed childbearing is the desire to maximize career opportunities, but there is also the weight of uncertainty and insecurity, especially regarding the costs of parenting. Filipovic (2021) also wrote, “Children? In this economy?” These two tendencies appear to generate opposing economic effects: A strong economy gives mothers more rewarding opportunities that childrearing threatens (reducing fertility), while also providing greater economic security to make parenting more affordable and desirable (increasing fertility). These two pathways for economic influence on fertility trends are not easily separable in research – or necessarily exclusive in personal experience. In what follows I will briefly situate falling US fertility in the wider historical and global context, and then offer a descriptive analysis of the US trend in births from 2000 to 2019, focusing on relative education and state unemployment rates.
Review and context
Historically, economic growth and development have been key determinants of fertility decline (Herzer, Strulik, and Vollmer 2012; Myrskylä, Kohler, and Billari 2009), although by no means the only ones, and with coupling that is sometimes loose and variable (Bongaarts 2017). In the broadest terms, both historically and in the present, higher average incomes at the societal level are strongly associated with lower fertility rates; and this relationship recurs within the United States as well, as shown in the cross section in Figure 1.
Figure 1. Total fertility rate by GDP per capita: Countries and U.S. states, 2019. Note: Markers are scaled by population. US states linear fit weighted by population. Source: World Bank, US Census Bureau, National Center for Health Statistics, Bureau of Economic Analysis.
A lower standard of living is associated with higher birth rates. However, economic crises cause declines in fertility (Currie and Schwandt 2014), and this was especially true around the 2008 recession in the U.S. (Comolli 2017; Schneider 2015) and other high-income countries (Gaddy 2021). The crisis interrupted what had been a mild recovery from falling total fertility rates in high-income countries, leading to a decline from 1.74 in 2008 to 1.57 by 2019 (Figure 2).
Figure 2. Total fertility rate in the 10 largest high-income countries: 1990-2019. Note: Countries with at least $30,000 GDP per capita at PPP. Source: World Bank.
Figure 2 shows that the pattern of a peak around 2008 followed by a lasting decline is widespread (with the notable exceptions of Germany and Japan, whose TFRs were already very low), although the post-crisis decline was much steeper in the U.S. than in most other high income countries. Figure 3 puts the post-crisis TFR decline in global context, showing the change in TFR between the highest point in 2007-2009 and the lowest point in 2017-2019 for each country, by GDP per capita. (For example, the U.S. had a TFR peak of 2.12 in 2007, and its lowest point in 2017-2019 was 1.71 in 2019, so its score is -.41.) Fertility decline is positively associated with per capita income, as low-income countries continued the TFR declines they were experiencing before the crisis. However, among the high-income countries the relationship reversed (the inflection point in Panel A is $36,600, not shown). Thus, the sharp drop in fertility in the U.S. after the 2008 economic crisis is indicative of a larger pattern of post-crisis fertility trends. Globally, fertility is higher but falling in lower-income countries; fertility is lower in high-income counties, but fell further during the recent period of economic hardship or uncertainty. As a result of falling at both low and high ends of the economic scale, therefore, global TFR declined from 2.57 in 2007 to 2.40 in 2019 (by these World Bank data).
Figure 3. Difference in total fertility rate between the highest point in 2007-2009 and the lowest point in 2017-2019, by GDP per capita. Note: Markers scaled by population; largest countries labeled. Source: World Bank.
The mechanisms for these relationships – higher standard of living and rising unemployment both lead to lower fertility – defy simple characterization. The social scale (individual to global) may condition the relationship; there may be different effects of relative versus absolute economic wellbeing (long term and short term); development effects may be nonlinear (Myrskylä, Kohler, and Billari 2009); and the individual or cultural perception of these social facts is important as well (Brauner-Otto and Geist 2018). Note also that, as fertility rates fall with development, the question of having no children versus fewer has emerged as a more important distinction, which further complicates the interpretation of TFR trends (Hartnett and Gemmill 2020).
U.S. recessions
In the case of recent U.S. recessions, the negative impact on fertility was largest for young women. After the 2001 recession, birth rates only fell for women under age 25. In the wake of the more severe 2008 economic crisis, birth rates fell for all ages of women up to age 40 (above which rates continued to increase every year until 2020) although the drop was still steepest below age 25 (Cohen 2018). For the youngest women, births have continued to fall every year since, while those over age 35 saw some rebound from 2012 to 2019 (Figure 4). Clearly, during this period many women postponed births from their teens or twenties into their thirties and forties. The extent to which they will end up with lower fertility on a cohort basis depends on how late they continue (or begin) bearing children (Beaujouan 2020).
Figure 4. Annual change in U.S. births per 1,000 women, by age: 2001-2020. Source: National Center for Health Statistics.
Contrary to the suggestion that fertility decline is chiefly the result of improving opportunities for women, the pattern of delaying births is consistent with evidence that structural changes in the economy, the decline in goods-producing industries and the rise of less secure and predictable service industry jobs, are largely responsible for the lack of a fertility rebound after the 2008 recession, especially for Black and Hispanic women (Seltzer 2019). Lower education is also associated with greater uncertainty about having children among young people (Brauner-Otto and Geist 2018). For women in more precarious circumstances, especially those who are not married, these influences may be observed in the effect of unemployment rates on birth rates at the state level (Schneider and Hastings 2015). The available evidence supports the conclusion that the 2008 recession produced a large drop in fertility that did not recover before 2020 at least in part because the economic uncertainty it amplified has not receded – making it both a short-term and long-term event.
Birth rates recovered some for older women, however – over 30 or so – which is consistent with fertility delay. But this delay does not necessarily favor the opportunity cost versus economic constraint explanations. On one hand are people with higher levels of education (anticipated or realized) who plan to wait until their education is complete. On the other hand are those with less education who are most economically insecure, whose delays reflect navigating the challenges of relationship instability, housing, health care, childcare and other costs with lesser earning potential. This latter group may end up delaying either until they attain more security or until they face the prospect of running out of childbearing years. Both groups are deliberately delaying births partly for economic reasons, but the higher-education group is much more likely to have planned births while the latter have higher rates of unintended or mistimed births (Hayford and Guzzo 2016).
The opportunity cost of women’s childbearing, in classical models, is simply the earnings lost from time spent childrearing – the product of the hours of employment lost and the expected hourly wage (Cramer 1979). Although rising income potential for women has surely contributed to the long-run decline of fertility rates, in the U.S. that mechanism has not been determinative. Women experienced large increases in earnings for decades during which fertility rates did not fall. As the total fertility rate rose from its low point in 1976 (1.74) to the post-Baby Boom peak in 2007 (2.12) – defying the trend in many other high-income countries – the average weekly earnings of full-time working women ages 18-44 rose by 16% in constant dollars (Figure 5).
Figure 5. Median weekly earnings of full-time employed women ages 18-44, and total fertility rate. Source: Current Population Survey Annual Social and Economic Survey, and Human Fertility Database.
Clearly, other factors beyond lost earnings calculations are at work. However, there is no simple way to distinguish those who make direct cost comparisons, where investments in time and money take away from other needs and opportunities, from those who delay out of concern over future economic security, which weighs on people at all income levels and generates reluctance to make lifelong commitments (Pugh 2015). But the implications of these two effects are opposing. For people who don’t want to lose opportunities, a strong economy with abundant jobs implies lower fertility. For people who are afraid to commit to childrearing because of insecurity about their economic fortunes, a weak economy should decrease fertility. The experience of the post-2008 period provides strong evidence for the greater weight of the latter mechanism.
US births, 2000-2019
If opportunity costs were the primary consideration for women, one might expect an inverse relationship between job market growth and fertility rates: more jobs, fewer babies; fewer jobs, more babies. This is the pattern reported by Tavernise et al. (2021), who found that birthrates after the 2008 crisis fell more in counties with “growing labor markets” – which they attribute to the combination of improving opportunities for women and the high costs of childcare. However, their analysis did not attend to chronological ordering. They identified counties as having strong job growth if they were in the top quintile of counties for labor market percent change for the period 2007 to 2019, and compared them with counties in the bottom quintile of counties on the same measure with regard to birth rates (author correspondence). Thus, their analysis used a 2007-2019 summary measure to predict birth rates for each year from 1990 to 2019, making the results difficult to interpret.
In addition to using contemporaneous economic data, whereas Tavernise et al. (2021) used county-level birth rates, in this analysis I use individual characteristics and state-level data. I construct indicators of individual- and state-level relative advantage during the period before and after the 2008 economic crisis, from 2000 to 2019. Individual data are from the 2000-2019 American Community Survey (ACS) via IPUMS (Ruggles et al. 2021). I include in the analysis women ages 15-44, and use the fertility question, which asks whether they had a baby in the previous 12 months. I analyze this as a dichotomous dependent variable, using ordinary least squares regression. Results are graphed as marginal effects at the means, using Stata’s margins command. The sample size is 9,415,960 million women, 605,150 (6.4%) of whom had a baby in the previous year (multiple births are counted only once).
In models with controls, I control for age in five-year bins, race/ethnicity (White, Black, American Indian, Asian/Pacific Islander, Other/multiple-race, and Hispanic), citizenship (U.S.-born, born abroad to American parents, naturalized, and not a citizen), marital status (married, spouse absent, separated, divorced, widowed, and never married), education (less than high school, high school graduate, some college, and BA or higher degree), as well as (in some models) the state unemployment rate (lagged two years), and state fixed effects. State unemployment rates are from Local Area Unemployment Statistics (Bureau of Labor Statistics 2021). ACS person weights are used in all analyses.
For states, I use the unemployment rate in each state for each year, and divide the states at the median, so those with the median or higher unemployment for each year are coded as high unemployment states, and low unemployment otherwise (this variable is lagged two years, because the ACS asks whether each woman has had a birth in the previous 12 months, but does not specify the month of the birth, or the date of the interview). For individuals, the identification of economic advantage is difficult with the cross-sectional data I use here, because incomes are likely to fall in the year of a birth, and education may be determined endogenously with fertility as women age (Hartnett and Gemmill 2020), so income and education cannot simply be used to identify economic status. Instead, I identify women as low education if they have less than the median level of education for women of their age in their state for each year (using single years of age, and 26 categories of educational attainment), and high education otherwise. Thus, individual women in my sample are coded as in a high or low unemployment state relative to the rest of the country each year, and as having high or low education relative other women of their age and state and year. Using the ACS migration variable, I code women into the state they lived in the previous year, which is more likely to identify where they lived when they determined whether to have a baby (which also means I exclude women who were not living in the U.S. in the year before the survey).
Figure 6 shows the unadjusted probability of birth for women in high- and low-unemployment states for the period 2000-2019. This shows the drop in birth rates after 2008, which is steeper for women who live in high-unemployment states, especially before 2017. This is what we would expect from previous research on the 2008 financial crisis: a greater falloff in birth rates where the economy suffered more.
Figure 6. Probability of birth in the previous year: 2000-2019, by state unemployment relative to the national media (marginal effects at the means). Women ages 15-44. Based on state of residence in the previous year; unemployment lagged two years.
Next, I split the sample again by women’s own education relative to the median for those of the same age, year, and state. Those less than that median are coded as low education, those at or higher than the median are coded as high education. Figure 7 shows these results (again, unadjusted for control variables), showing that those with lower education (the top two lines) have higher birth rates throughout the period. After 2008, within both the high- and low-education groups, those in high-unemployment states had longer and steeper declines in birth rates (at least until 2019). The steepest decline is among low-education, high-unemployment women: those facing the greatest economic hardship at both the individual and state level. Finally, Figure 8 repeats the model shown in Figure 7, but with the control variables described above, and with state fixed effects. The pattern is very similar, but the differences associated with state unemployment are attenuated, especially for those with low education.
Figure 7. Probability of birth in the previous year: 2000-2019, by education relative to the age-state median, and state unemployment relative to the national media (marginal effects at the means). Women ages 15-44. Based on state of residence in the previous year; unemployment lagged two years.
Figure 8. Probability of birth in the previous year: 2000-2019, by education relative to the age-state median, and state unemployment relative to the national media, with controls for age, race/ethnicity, citizenship, marital status, and state fixed effects (marginal effects at the means). Women ages 15-44. Based on state of residence in the previous year; unemployment lagged two years.
Discussion
Although birth rates fell for all four groups of women in this analysis after the 2008 recession, these results reflect that paradoxical nature of economic trends and birth rates. Women with higher education (and greater potential earnings) have lower birthrates, consistent with the opportunity cost reasoning described in Tavernise et al. (2021) and elsewhere. However, women in states with higher unemployment rates – especially when they have high relative education – also have lower birthrates, and in these states saw greater declines after the 2008 crisis. This is consistent with the evidence of negative effects of economic uncertainty and stress. And it goes against the suggestion that stronger job markets drive down fertility rates for women with higher earning potential, at least in the post-2008 period. In the long run, perhaps, economic opportunities reduce childbearing by increasing job market opportunities for potential mothers, but in recent years this effect has been swamped by the downward pressure of economic troubles. US birth rates fell further in 2020, apparently driven down by the COVID-19 pandemic, which raised uncertainty – and fear for the future – to new heights (Cohen 2021; Sobotka et al. 2021). We don’t yet know the breakdown of the shifts in fertility for that year, but if the effects were similar to those of the 2008 economic crisis, we would expect to see greater declines among those who were most vulnerable.
References
Beaujouan, Eva. 2020. “Latest-Late Fertility? Decline and Resurgence of Late Parenthood Across the Low-Fertility Countries.” Population and Development Review 46 (2): 219–47. https://doi.org/10.1111/padr.12334.
Brauner-Otto, Sarah R., and Claudia Geist. 2018. “Uncertainty, Doubts, and Delays: Economic Circumstances and Childbearing Expectations Among Emerging Adults.” Journal of Family and Economic Issues 39 (1): 88–102. https://doi.org/10.1007/s10834-017-9548-1.
Bureau of Labor Statistics. 2021. “States and Selected Areas: Employment Status of the Civilian Noninstitutional Population, January 1976 to Date, Seasonally Adjusted.” 2021. https://www.bls.gov/web/laus/ststdsadata.txt.
Cohen, Philip N. 2018. Enduring Bonds: Inequality, Marriage, Parenting, and Everything Else That Makes Families Great and Terrible. Oakland, California: University of California Press.
———. 2021. “Baby Bust: Falling Fertility in US Counties Is Associated with COVID-19 Prevalence and Mobility Reductions.” SocArXiv. https://doi.org/10.31235/osf.io/qwxz3.
Comolli, Chiara Ludovica. 2017. “The Fertility Response to the Great Recession in Europe and the United States: Structural Economic Conditions and Perceived Economic Uncertainty.” Demographic Research 36 (51): 1549–1600. https://doi.org/10.4054/DemRes.2017.36.51.
Cramer, James C. 1979. “Employment Trends Ofyoung Mothers and the Opportunity Cost of Babies in the United States.” Demography 16 (2): 177–97. https://doi.org/10.2307/2061137.
Currie, Janet, and Hannes Schwandt. 2014. “Short- and Long-Term Effects of Unemployment on Fertility.” Proceedings of the National Academy of Sciences 111 (41): 14734–39. https://doi.org/10.1073/pnas.1408975111.
Gaddy, Hampton Gray. 2021. “A Decade of TFR Declines Suggests No Relationship between Development and Sub-Replacement Fertility Rebounds.” Demographic Research 44 (5): 125–42. https://doi.org/10.4054/DemRes.2021.44.5.
Hartnett, Caroline Sten, and Alison Gemmill. 2020. “Recent Trends in U.S. Childbearing Intentions.” Demography 57 (6): 2035–45. https://doi.org/10.1007/s13524-020-00929-w.
Hayford, Sarah R., and Karen Benjamin Guzzo. 2016. “Fifty Years of Unintended Births: Education Gradients in Unintended Fertility in the US, 1960-2013.” Population and Development Review 42 (2): 313–41.
Herzer, Dierk, Holger Strulik, and Sebastian Vollmer. 2012. “The Long-Run Determinants of Fertility: One Century of Demographic Change 1900–1999.” Journal of Economic Growth 17 (4): 357–85. https://doi.org/10.1007/s10887-012-9085-6.
Myrskylä, Mikko, Hans-Peter Kohler, and Francesco C. Billari. 2009. “Advances in Development Reverse Fertility Declines.” Nature 460 (7256): 741–43. https://doi.org/10.1038/nature08230.
Pugh, Allison J. 2015. The Tumbleweed Society: Working and Caring in an Age of Insecurity. 1 edition. New York, NY: Oxford University Press.
Ruggles, Steven, Sarah Flood, Sophia Foster, Ronald Goeken, Jose Pacas, Megan Schouweiler, and Matthew Sobek. 2021. “IPUMS USA: Version 11.0 [Dataset].” 2021. doi.org/10.18128/D010.V11.0.
Schneider, Daniel. 2015. “The Great Recession, Fertility, and Uncertainty: Evidence From the United States.” Journal of Marriage and Family 77 (5): 1144–56. https://doi.org/10.1111/jomf.12212.
Schneider, Daniel, and Orestes P. Hastings. 2015. “Socioeconomic Variation in the Effect of Economic Conditions on Marriage and Nonmarital Fertility in the United States: Evidence From the Great Recession.” Demography 52 (6): 1893–1915. https://doi.org/10.1007/s13524-015-0437-7.
Seltzer, Nathan. 2019. “Beyond the Great Recession: Labor Market Polarization and Ongoing Fertility Decline in the United States.” Demography 56 (4): 1463–93. https://doi.org/10.1007/s13524-019-00790-6.
Sobotka, Tomas, Aiva Jasilioniene, Ainhoa Alustiza Galarza, Kryštof Zeman, Laszlo Nemeth, and Dmitri Jdanov. 2021. “Baby Bust in the Wake of the COVID-19 Pandemic? First Results from the New STFF Data Series.” SocArXiv. https://doi.org/10.31235/osf.io/mvy62.
This post has been updated with the final signing statement and a link to the form. Thanks for sharing!
I have objected to the use of “generation” divisions and names for years (here’s the tag). Then, the other day, I saw this introduction to an episode of Meet the Press Reports, which epitomized a lot of the gibberishy nature of generationspeak (sorry about the quality).
OK, it’s ridiculous political punditry — “So as their trust in institutions wanes, will they eventually coalesce behind a single party, or will they be the ones to simply transform our political system forever?” — but it’s also generations gobbledygook. And part of what struck me was this: “millennials are now the largest generation, they have officially overtaken the Baby Boom.” Well-educated people think these things are real things, official things. We have to get off this train.
If you know the generations discourse, you know a lot of it emanates from the Pew Research Center. They do a lot of excellent research — and make a lot of that research substantially worse by cramming into the “generations” framework that they more than anyone else have popularized — have made “official.”
After seeing that clip, I put this on Twitter, and was delighted by the positive response:
What if we got a large number of social scientists, demographers, and journalists to sign a very collegial, lightly footnoted appeal to Pew to stop using generation labels, to never name another generation. Would it work? Would you sign it?
So I wrote a draft of an open letter to Pew, incorporating some of the comments from Twitter. But then I decided the letter was too long. To be more effective maybe it should be more concise and less ranty. So here’s the long version, which has more background information and examples, followed by a signing version, with a link to the form to sign it. Please feel to sign if you are a demographer or other social scientist, and share the link to the form (or this post) in your networks.
Maybe if we got a lot of signatories to this, or something like it, they would take heed.
Preamble by me
Pew’s generation labels — which are widely adopted by many other individuals and institutions — encourage unhelpful social science communication, driving people toward broad generalizations, stereotyping, click bait, sweeping character judgment, and echo chamber thinking. When people assign names to generations, they encourage anointing them a character, and then imposing qualities onto whole populations without basis, or on the basis of crude stereotyping. This fuels a constant stream of myth-making and myth-busting, with circular debates about whether one generation or another fits better or worse with various of its associated stereotypes. In the absence of research about whether the generation labels are useful either scientifically or in communicating science, we are left with a lot of headlines drawing a lot of clicks, to the detriment of public understanding.
Cohort analysis and the life course perspective are important tools for studying and communicating social science. We should study the shadow, or reflection, of life events across people’s lives at a cultural level, not just an individual level. In fact, the Pew Research Center’s surveys and publications make great contributions to that end. But the vast majority of popular survey research and reporting in the “generations” vein uses data analyzed by age, cross-sectionally, with generational labels applied after the fact — it’s not cohort research at all. We shouldn’t discourage cohort and life course thinking, rather we should improve it.
Pew’s own research provides a clear basis for scrapping the “generations.” “Most Millennials Resist the ‘Millennial’ Label” was the title of a report Pew published in 2015. This is when they should have stopped — based on their own science — but instead they plowed ahead as if the “generations” were social facts that the public merely failed to understand.
This figure shows that the majority of Americans cannot correctly identify the generational label Pew has applied to them.
The concept of “generations” as applied by Pew (and many others) defies the basic reality of generations as they relate to reproductive life cycles. Pew’s “generations” are so short (now 16 years) that they bear no resemblance to reproductive generations. In 2019 the median age of a woman giving birth in the U.S. was 29. As a result, many multigenerational families include no members of some generations on Pew’s chart. For example, it asks siblings (like the tennis-champion Williams sisters, born one year apart) to identify as members of separate generations.
Perhaps due to their ubiquitous use, and Pew’s reputation as a trustworthy arbiter of social knowledge, many people think these “generations” are official facts. Chuck Todd reported on NBC News just this month, “Millennials are now the largest generation, they have officially overtaken the Baby Boom.” (NPR had already declared Millennials the largest generation seven years earlier, using a more expansive definition.) Pew has perhaps inadvertently encouraged these ill-informed perspectives, as when, for example, Richard Fry wrote for Pew, “Millennials have surpassed Baby Boomers as the nation’s largest living adult generation, according to population estimates from the U.S. Census Bureau” — despite the fact that the Census Bureau report referenced by the article made no mention of generations. Note that Chuck Todd’s meaningless graphic, which doesn’t even include ages, is also falsely attributed to the U.S. Census Bureau.
Generations are a beguiling and appealing vehicle for explaining social change, but one that is more often misleading than informative. The U.S. Army Research Institute commissioned a consensus study report from the National Academies, titled, Are Generational Categories Meaningful Distinctions for Workforce Management? The group of prominent social scientists concluded: “while dividing the workforce into generations may have appeal, doing so is not strongly supported by science and is not useful for workforce management. …many of the stereotypes about generations result from imprecise use of the terminology in the popular literature and recent research, and thus cannot adequately inform workforce management decisions.”
As one of many potential examples of such appealing, but ultimately misleading, uses of the “Millennial” generation label, consider a 2016 article by Paul Taylor, a former executive vice president of the Pew Research Center. He promised he would go beyond “clichés” to offer “observations” about Millennials — before describing them as “liberal lions…who might not roar,” “downwardly mobile,” “unlaunched,” “unmarried,” “gender role benders,” “upbeat,” “pre-Copernican,” and as an “unaffiliated, anti-hierarchical, distrustful” generation who nevertheless “get along well with their parents, respect their elders, and work well with colleagues” while being “open to different lifestyles, tolerant of different races, and first adopters of new technologies.” And their “idealism… may save the planet.”
In 2018 Pew announced that it would henceforth draw a line between “Millennials” and “Generation Z” at the year 1996. And yet they offered no substantive reason, just that “it became clear to us that it was time to determine a cutoff point between Millennials and the next generation [in] order to keep the Millennial generation analytically meaningful, and to begin looking at what might be unique about the next cohort.” In asserting that “their boundaries are not arbitrary,” the Pew announcement noted that they were assigning the same length to the Millennial Generation as they did to Generation X — both 16 years, a length that bears no relationship to reproductive generations, nor to the Baby Boom cohort, which is generally considered to be 19 years (1946-1964).
The essay that followed this announcement attempted to draw distinctions between Millennials and Generation Z, but it could not delineate a clear division, because none can be drawn. For example, it mentioned that “most Millennials came of age and entered the workforce facing the height of an economic recession,” but in 2009, the trough year for that recession, Millennials by Pew’s definition ranged from age 13 to 29. The other events mentioned — the 9/11 terrorist attacks, the election of Barack Obama, the launch of the iPhone, and the advent of social media — similarly find Millennials at a range of ages too wide to be automatically unifying in terms of experience. Why is being between 12 and 28 at the time of Obama’s election more meaningful a cohort experience than being, say, 18 to 34? No answer to this is provided, because Pew has determined the cohort categories before the logical scientific questions can be asked.
Consider a few other hypothetical examples. In the future, we might hypothesize that those who were in K-12 school during the pandemic-inflicted 2020-2021 academic year constitute a meaningful cohort. That 13-year cohort was born between 2003 and 2015, which does not correspond to one of Pew’s predetermined “generations.” For some purposes, an even narrower range might be more appropriate, such as those who graduated high school in 2020-2021 alone. Under the Pew generational regime, too many researchers, marketers, journalists, and members of the general public will look at major events like these through a pre-formed prism that distorts their ability to pursue or understand the way cohort life course experiences affect social experience.
Unlike the other “generations” in Pew’s map, the Baby Boom corresponds to a unique demographic event, painstakingly, empirically demonstrated to have begun in July 1946 and ended in mid-1964. And being part of that group has turned out to be a meaningful experience for many people — one that in fact helped give rise to the popular understanding of birth cohorts as a concept. But it does not follow that any arbitrarily grouped set of birth dates would produce a sense of identity, especially one that can be named and described on the basis of its birth years alone. It is an accident of history that the Baby Boom lasted 18 years — as far as we know having nothing to do with the length of a reproductive generation, but perhaps leading subsequent analysts to use the term “generation” to describe both Baby Boomers and subsequent cohorts.
The good researchers at Pew are in a tough spot (as are others who rely on their categories). The generations concept is tremendously appealing and hugely popular. But where does it end? Are we going to keep arbitrarily dividing the population into generations and giving them names — after “Z”? On what scientific basis would the practice continue? One might be tempted to address these problems by formalizing the process, with a conference and a dramatic launch, to make it even more “official.” But there is no scientific rationale for dividing the population arbitrarily into cohorts of any particular length for purposes of analyzing social trends, and to fix their membership a priori. Pew would do a lot more to enhance its reputation, and contribute to the public good, by publicly pulling the plug on this project.
Open letter to the Pew Research Center on generation labels
We are demographers and other social scientists, writing to urge the Pew Research Center to stop using its generation labels (currently: Silent, Baby Boom, X, Millennial, Z). We appreciate Pew’s surveys and other research, and urge them to bring this work into better alignment with scientific principles of social research.
Pew’s “generations” cause confusion.
The groups Pew calls Silent, Baby Boom, X, Millennial, and Z are birth cohorts determined by year of birth, which are not related to reproductive generations. There is further confusion because their arbitrary lengths (18, 19, 16, 16, and 16 years, respectively) have grown shorter as the age difference between parents and their children has lengthened.
The division between “generations” is arbitrary and has no scientific basis.
With the exception of the Baby Boom, which was a discrete demographic event, the other “generations” have been declared and named on an ad hoc basis without empirical or theoretical justification. Pew’s own research conclusively shows that the majority of Americans cannot identify the “generations” to which Pew claims they belong. Cohorts should be delineated by “empty” periods (such as individual years, equal numbers of years, or decades) unless research on a particular topic suggests more meaningful breakdowns.
Naming “generations” and fixing their birth dates promotes pseudoscience, undermines public understanding, and impedes social science research.
The “generation” names encourage assigning them a distinct character, and then imposing qualities on diverse populations without basis, resulting in the current widespread problem of crude stereotyping. This fuels a stream of circular debates about whether the various “generations” fit their associated stereotypes, which does not advance public understanding.
The popular “generations” and their labels undermine important cohort and life course research
Cohort analysis and the life course perspective are important tools for studying and communicating social science. But the vast majority of popular survey research and reporting on the “generations” uses cross-sectional data, and is not cohort research at all. Predetermined cohort categories also impede scientific discovery by artificially imposing categories used in research rather than encouraging researchers to make well justified decisions for data analysis and description. We don’t want to discourage cohort and life course thinking, we want to improve it.
The “generations” are widely misunderstood to be “official” categories and identities
Pew’s reputation as a trustworthy social research institution has helped fuel the false belief that the “generations” definitions and labels are social facts and official statistics. Many other individuals and organizations use Pew’s definitions in order to fit within the paradigm, compounding the problem and digging us deeper into this hole with each passing day.
The “generations” scheme has become a parody and should end.
With the identification of “Generation Z,” Pew has apparently reached the end of the alphabet. Will this continue forever, with arbitrarily defined, stereotypically labeled, “generation” names sequentially added to the list? Demographic and social analysis is too important to be subjected to such a fate. No one likes to be wrong, and admitting it is difficult. We sympathize. But the sooner Pew stops digging this hole, the easier it will be to escape. A public course correction from Pew would send an important signal and help steer research and popular discourse around demographic and social issues toward greater understanding. It would also greatly enhance Pew’s reputation in the research community. We urge Pew to end this as gracefully as possible — now.
As consumers of Pew Research Center research, and experts who work in related fields ourselves, we urge the Pew Research Center to do the right thing and help put an end to the use of arbitrary and misleading “generation” labels and names.
My new book, Citizen Scholar, is under contract with Columbia University Press (thanks to the support of editor, and Editorial Director, Eric Schwartz).
Some of the writing I’ve been doing here is part of the book’s development, including the piece on “policy implications,” essays on transparency and accountability in research, as well as talks and materials about preprints, open science and the pandemic, politics and science, and others. It’s time for a book (and also more talks, if you’d like to invite me!). I will post essays and excerpts as I go, here, and I welcome your critiques, suggestions, and ideas. The first post describes my ambitions, and plan, for the book.
I love Family Inequality and everyone here but it seemed awkward to repeatedly post stuff for the new book under this heading. So I set up a blog style page, and I’ll post links here, too (and I’ll figure out you can subscribe, for those who want their blog posts via email).
I am at the end of a three-year term as an elected member of the American Sociological Association (ASA) Committee on Publications, during which time, despite some efforts, I achieved none of the goals in my platform. I would like to say I learned a lot or whatever, but I didn’t. I enjoyed the time spent with colleagues at the meetings (and we did some rubber stamping and selecting editors, which someone had to do), but beyond that it was a waste of time. Here are some reflections.
First, some observations about sociology as a discipline, then about the ASA generally, and then about the situation with the Committee on Publications.
Sociology
Sociology has occupied a rapidly declining presence in US higher education for two decades. The percentage of bachelor’s, masters, and PhD degrees that were awarded in sociology peaked at the end of the last century:
Looking just at the number of PhDs awarded, you can see that among the social sciences, since the 2009 recession (scaled to 0 on this figure), sociology is one of the social sciences disciplines that has slipped, while psychology, economics, business, and political science have grown (along with STEM disciplines).
The American Sociological Association
So, sociology as an academic discipline is in decline. But how is ASA doing — is it fair for me to say “collapsing” in the title of this post? The shrinking of the discipline puts a structural squeeze on the association. In order to maintain its organizational dimensions it would need a growing share of the sociology milieu. The prospects for that seem dim.
On the plus side, the association publishes several prominent journals, which were the ostensible subject of our work on the publications committee. Metrics differ, but two ASA journals are in the top ten sociology journals by five-year citation impact factor in Web of Science (American Sociological Review and Sociology of Education [the list is here]). In the Google Scholar ranking of journals by h5-index, which uses different subject criteria, only one (ASR) is in the top 20 of sociology journals, ranking 20th out of the combined top 100 from economics, social science general, sociology, anthropology, and political science (the list is here). In terms of high-impact research, among the top 100 most cited Web of Science sociology papers published in 2017 (an arbitrarily chosen recent year), seven were published in ASA journals (five in American Sociological Review [the list is here]). The 2020 Almetric Top 100 papers, those gaining the most attention in the year (from sources including news media and social media), includes 35 from humanities and social sciences, none of which were published by ASA (although several are by sociologists). So ASA is prominent but not close to dominant within sociology, which is similarly situated within the social sciences. In terms of publications, you can’t say ASA is “collapsing.” (Plus, in 2019 ASA reported $3 million in revenue from publications, 43 percent of its total non-investment income.)
But in terms of membership, the association is leading the way in the discipline’s decline. The number of members in the association fell 24 percent from 2007 to 2019, before nosediving a further 16 percent last year. Relative to the number of PhDs completing their degrees, as one scale indicator, membership has fallen 42 percent since 2007 — from 26 paying members per PhD awarded to 15. Here are the trends:
Clearly, the organization is in a serious long-term decline with regard to membership. How will an organization of sociologists, including organizational sociologists, react to such an organizational problem? Task force! A task force on membership was indeed seated in 2017, and two years later issued their report and recommendations. To begin with, the task force reported that ASA’s membership decline is steeper than that seen by 11 other unnamed disciplinary societies:
They further reported that only 36 percent of members surveyed consider the value of belonging to ASA equal to or greater than the cost, while 48 percent said it was overpriced. Further, 69 percent of members who didn’t renew listed cost of membership as an important reason, by far — very far — the most important factor, according to their analysis. Remarkably, given this finding, the report literally doesn’t say what the member dues or meeting registration fees are. Annual dues, incidentally, increased dramatically in 2013, and now range from $51 for unemployed members to $368 for those with incomes over $150,000 per year, apparently averaging $176 per member (based on the number of members and membership revenue declared in the audit reports).
Not surprisingly, then, although they did recommend “a comprehensive review of our membership dues and meeting registration fee structures,” they had no specific recommendations about member costs. Instead they recommended: Create new ways for sociologists to create subgroups within the association, “rethink the Annual Meeting and develop a variety of initiatives, both large and small,” remove institutional affiliations from name badges and make the first names bigger, give a free section membership to new members (~$10 value), anniversary- instead of calendar-based annual pricing, hold the meeting in a wider “variety of cities,” more professional development (mechanism unspecified), more public engagement, change the paper deadline a couple of weeks and consider other changes to paper submission, and provide more opportunities for member feedback. Every recommendation was unanimously approved by the association’s elected council. The following year membership fell another 16 percent, with some unknown portion of the drop attributable to the pandemic and the canceled annual meeting.
With regard to the membership crisis, my assessment is that ASA is a model of organizational stagnation and failure to respond in a manner adequate to the situation. The sociologist members, through their elected council, seem to have no substantial response, which will leave it to the professional staff to implement emergency measures as revenue drops in the coming years. One virtually inevitable outcome is the association further committing to its reliance on paywalled journal publishing and the profit-maximizing contract with Sage, and opposing efforts to open access to research for the public.
Committee on Publications
But it is on the publications committee, and its interactions with the ASA Council, that I have gotten the best view of the association as a perpetual stagnation machine.
I can’t say that the things I tried to do on the publications committee would have had a positive effect on ASA membership, journal rankings, majors, or any other metric of impact for the association. However, I do believe what I proposed would have helped the association take a few small steps in the direction of keeping up with the social science community on issues of research transparency and openness. In November I reported how, more than two years ago now, I proposed that the association adopt the Transparency and Openness Promotion Guidelines from the Center for Open Science, and to start using their Open Science Badges, which recognize authors who provide open data, open materials, or use preregistration for their studies. (In the November post I discussed the challenge of cultural and institutional change on this issue, and why it’s important, so I won’t repeat that here.)
The majority of the committee was not impressed at the beginning. At the January 2019 meeting the committee decided that an “ad hoc committee could be established to evaluate the broader issues related to open data for ASA journals.” Eight months later, after an ad hoc committee report, the publications committee voted to “form an ad hoc committee [a different one this time] to create a statement regarding conditions for sharing data and research materials in a context of ethical and inclusive production of knowledge,” and to, “review the question about sharing data currently asked of all authors submitting manuscripts to incorporate some of the key points of the Committee on Publications discussion.” The following January (2020), the main committee was informed that the ad hoc committee had been formed, but hadn’t had time to do its work. Eight months later, the new ad hoc committee proposed a policy: ask authors who publish in ASA journals to declare whether their data and research materials are publicly available, and if not why not, with the answers to be appended in a footnote to each article. And then the committee approved the proposal.
Foolishly, last fall I wrote, “So, after two years, all articles are going to report whether or not materials are available. Someday. Not bad, for ASA!” Yesterday the committee was notified by ASA staff that, “Council is pleased that Publications Committee has started this important discussion and has asked that the conversation be continued in light of feedback from the Council conversation.” In other words, they rejected the proposal and they’ll tell us why in another four months. There is no way the proposal can take effect for at least another year — or about four years after the less watered-down version was initially proposed, and after my term ends. It’s a perpetual stagnation machine.
Meanwhile, I reviewed 24 consecutive papers in ASR, and found that only four provided access to the code used and at least instructions on how to find the data. Many sociologists think this is normal, but in the world of academic social science, this is not normal, it’s far behind normal.
I don’t know if the Council is paying attention to the Task Force on Membership, but if they were it might have occurred to them that recruiting people to run for office, having the members elect them based on a platform and some expertise, having them spend years on extremely modest, imminently sensible proposals, and then shooting those down with a dismissive “pleased [you have] started this important discussion” — is not how you improve morale among the membership.
Remember that petition?
While I’m at it, I should update you on the petition many of you signed in December 2019, in opposition to the ASA leadership sending a letter to President Trump against a potential federal policy that would make the results of taxpayer-funded research immediately available to the public for free — presumably at some cost to ASA’s paywall revenues. At the January 2020 meeting the publications committee passed two motions:
For the Committee on Publications to express opposition to the decision by the ASA to sign the December 18, 2019 letter.
To encourage Council to discuss implications of the existing embargo and possible changes to the policy and to urge decisionmakers to consult with the scientific community before making executive orders.
We never heard back from the ASA Council, and the staff who opposed the petition were obviously in no rush to follow up on our entreaty to them, so it disappeared. But I just went back to March 2020 Council minutes, and found this profoundly uninformative tidbit:
Council discussed a recent decision of ASA’s authorized leadership to sign a letter expressing concern about an executive order related to scientific publishing rumored to be coming out with almost no notice or consultation with the scientific community. A motion was made by Kelly to affirm ASA’s policy for making time sensitive decisions about public statements and to confirm that the process was properly followed in this instance. Seconded by Misra. Motion carried with 16 for and 2 abstentions.
This doesn’t mention that the substance of the dispute, that the publications committee objected to the leadership’s statement, or the fact that more than 200 people signed a letter that read, in part: “We oppose the decision by ASA to sign this letter, which goes against our values as members of the research community, and urge the association to rescind its endorsement, to join the growing consensus in favor of open access to scholarship, including our own.” To my knowledge no member of the ASA leadership, whether elected sociologists or administrative staff, has responded publicly to this letter. Presumably, the terrible statement sent by the ASA leadership still represents the position of the association — the association that speaks for a rapidly dwindling number of us.
Side note: An amazing and revealing thing happened in the publications committee meeting where we discussed this statement in January 2020. The chair of the committee read a prepared statement, presumably written by the ASA staff, to introduce the voting on my proposal:
The committee has a precedent that many of you are already aware of, of asking people to leave for votes on the proposals they submitted…. This practice is designed to ensure that the committee members can have a full and open discussion. So, Philip, I’d like to ask you to recuse yourself now for the final two items, which you can simply do by hanging up the phone…
Needless to say, I refused to leave the meeting for the discussion on my proposal, as there is no such policy for the committee. (If you know of committee meetings where the person making a proposal — an elected representative — has to leave for the discussion and vote, please let me know.) It was just an attempt to railroad the decision, and other members stepped in to object, so they dropped it. The motion passed, and council ignored it, so seriously who cares, but still. (The minutes for this meeting don’t reflect this whole incident, but I have verbatim notes.)
You will forgive me if, after this multi-year exercise in futility, I am not inclined to be optimistic regarding the Taskforce on Membership’s Recommendation #10: “Enhance and increase communications from ASA to members and provide opportunities for ASA members to provide ongoing feedback to ASA.” I have one more meeting in my term on the publications committee, but it doesn’t seem likely I’ll be there.
My YouTube career may have peaked in 2015, with the now-classic video, Total Fertility Rate, which has been viewed almost 30,000 times (124 likes!). Since then the technical quality has improved, but not the viewership.
Recently I heard someone say (sorry, I can’t remember who) that they were looking for a short video explaining what life expectancy is. This was after the CDC reported that US life expectancy in the first half of 2020 decreased by 1 year, which generated some confusion. Outside of secondary effects, the pandemic did reduce life “expectancy” for people it didn’t kill, and here we are (still alive, so far) reading about it, so how could our life expectancy have been affected? did last year’s deaths mean people would live less long in the future? I said somewhere on twitter than “life expectancy” is a bad name for this common statistic, and I think it is. I don’t have a better name for it, though, and it’s probably to late to change anyway.
So, to help meet the current need, and to try to reach my past video glory, yesterday I produced, “What is life expectancy?“, a 6-minute explainer, using 3 graphs, to help people understand. I didn’t discuss the recent COVID results so as not to date the video, and I hope it will be useful in the future (that it has a long life expectancy).
