Category: Data Source
Overreaching causal language in the social sciences
Across the social sciences, many studies use cross-sectional designs that reveal associations but are generally unable to support direct causal claims, yet authors of such articles may make or imply causal claims anyway. Here, to examine the prevalence of such ‘overreaching’ causal language, we analysed 194,631 cross-sectional articles using large language models. Over the period 1980–2024, an average of 46% of articles contained causal language in their titles or abstracts, where the annual rate has risen almost threefold since 2000 from 20% to 60%. To examine the effects of such language, we conducted a human-subjects experiment (N = 1, 105), finding that readers frequently indicate abstracts with this phrasing provide causal evidence but that methodological labels (β = −0.4, 95% confidence interval −0.56 to −0.19) and associational wording (β = −0.3, 95% confidence interval −0.43 to −0.07) reduce this tendency. Experiments with five LLMs revealed that model summaries of these articles (N = 100 each) can amplify causal overstatement, removing hedges and introducing causal claims where articles used strictly associational phrasing; however, prompting caution diminishes this pattern.
That is from a recent paper by Calvin Isch, Timothy Dörr, Neil Fasching, Grace Jennings & Duncan J. Watts. Note that Isch is on the job market this year, working with Tetlock and Watts.
Common sense in charge
…children are increasingly seen as interfering with the freedom of parents; views on whether mothers of young children should work have become markedly more progressive and account for a substantially larger share of the decline among the tertiary-educated; and fewer people believe that women or men need children to lead a fulfilled life. This last attitude changed most over the decade and is the largest contributor to the fall in intended fertility, above all among less-educated women.
That is from a new NBER working paper by
Childhood Exposure to Joint Custody Reforms and Adult Family Formation
Joint custody reforms are among the most consequential family-law changes for children, yet little is known about their long-term effects. Exploiting staggered adoption across US states and 13 million ACS observations, we show that childhood exposure reduces adult fertility by 7 percent, symmetrically for women and men and operates primarily through lower parenthood and couple formation. Effects are sub-additive within couples, while exposed individuals assortatively match. Our findings point to the formation of family preferences during childhood and align with a US fertility decline increasingly driven by rising childlessness and declining couple formation rather than smaller families among parents.
That is from a recent research paper by Daniel Fernández-Kranz and Sébastien Fontenay. Via the excellent Samir Varma.
The End of Friday Nights with Friends
This study examines how Americans’ time with friends has changed over the week, focusing not on how much social interaction has declined but on when it has. Using American Time Use Survey data from 2003 to 2024 (N = 243,095), I map hourly patterns of “friend time” across the days of the week. In the early 2000s, social life followed a clear weekly rhythm: Modest weekday interaction built toward pronounced peaks on Friday and Saturday nights. Since the mid-2010s, that rhythm has collapsed. Friday night, once a central site of social activity, now looks like a typical weeknight. Saturday remains elevated but less so than before. Overall, time with friends has fallen by more than half. The mix of social activities, however, has remained stable. Americans have not replaced one form of interaction with another; they have reduced social interaction across the board. These findings point to a temporal reorganization of social life marked by the disappearance of the “night out.”
That is by Neal Caren, via the excellent Kevin Lewis. So what did you go last evening?
Green shoots for the UK?
Early signs of tech-driven improvements in productivity growth could herald a sustained strengthening in the UK’s economic outlook, analysts have said, in a turnaround after years of underperformance. Private sector productivity grew by 1.8 per cent in the second quarter compared with a year earlier, up from 1.2 per cent previously, according to analysis of official data by investment bank Morgan Stanley.
The rise extended gains since 2024 and reduced the growth gap with the US. The reasons behind the upsurge are heavily contested, but some analysts point to increasing AI adoption in sectors including information technology and business services.
If the recent productivity acceleration can be sustained over years, it could bolster incomes and help alleviate some of the strains on Britain’s public finances.
Did UBI make people happier?
Eh, only in the short run:
We study the causal impacts of income on a rich array of employment outcomes, leveraging an experiment in which 1,000 low-income individuals were randomized into receiving $1,000 per month unconditionally for three years, with a control group of 2,000 participants receiving $50/month. We gather detailed survey data, administrative records, and data from a mobile phone app. The transfer caused total individual income excluding the transfers to fall by about $1,900/year relative to the control group and a 4.2 percentage point decrease in labor market participation. Participants reduced their work hours as a result of the transfers by 1-2 hours/week and participants’ partners reduced their work hours by a comparable amount. Among other categories of time use, the greatest increase generated by the transfer was in time spent on leisure. Despite asking detailed questions about amenities, we find no impact on quality of employment, and our confidence intervals can rule out even small improvements. Treated participants broadly increase expenditures, led by spending on non-durable goods and services, with smaller increases in spending on durable goods and human capital. We observe no significant effects on degree attainment, though the magnitudes of the estimated effects generally appear larger among younger participants. Measures of subjective well-being are higher among treated participants in the first year of the transfers but then revert to control group levels. Overall, our results suggest a moderate labor supply effect that does not appear offset by other productive activities.
That is from the QJE by Eva Vivalt, Elizabeth Rhodes, Alexander Bartik, David Broockman, Patrick Krause, and Sarah Miller. Via Matt Yglesias.
Declining Occupations and Career Outcomes in the United States
This strikes me as somewhat less of a problem than I might have thought:
We study long-run career consequences of initial employment in an occupation that subsequently declines. Linking the 2000 Decennial Census to US administrative employment and earnings records through 2020, we follow more than 2.4 million workers. Employment in an occupation that contracts by at least 25 percent is associated with about 5 percent lower cumulative earnings despite slightly more quarters worked. The earnings differential closely matches evidence from Sweden and Norway, although employment adjustment differs. Occupational mobility is substantial but incomplete, while children’s later occupational destinations are much less tied to their household heads’ 2000 occupational-growth categories.
That is from a new NBER working paper from
Some fertility and AI forecasts
The 2024 forecast is particularly pessimistic about China’s fertility prospects. Both projections produce very substantial global aging, a major global capital glut producing very low long-run real capital returns. The latest forecast entails 10% lower global GDP in 2100 and far higher payroll tax rates to fund old-age benefits. Most important, it entails a major change in the course of economic hegemony with China’s 2100 global GDP share falling from 25.6% to 14.9% and the US share rising from 11.2% to 14.4%. Our results are sensitive. Should the US eliminate all future immigration, its 14.4% global 2100 GDP share would drop to 9.2%. And were global fertility to follow the UN’s low variant, 2100 world output would be one third, not one tenth lower. The level and division of global output is also highly sensitive to the speed at which AI expands frontier technologies. Accelerated AU/AI – 4x faster-than-recent growth in capital’s share through 2050 – or Transformative AU/AI – 10x faster capital-share growth – reinforce demographic forces, ensuring long-run US economic hegemony. Indeed, Transformative AI combined with 2024 demographics implies US and Chinese 2100 global GDP shares of 25.3% and 16.9%, respectively.
That is from a new NBER working paper by
How economics is changing
As a proportion of the literature, research about econometric theory, monetary policy and corporate governance has slumped. Meanwhile, papers on development, crime and gender are on the up. This could perhaps be because of changing intellectual interests, or maybe demand-side pressures, such as policymaker priorities and external funding…
Research from Prashant Garg, a postdoctoral researcher at Bocconi University, and Thiemo Fetzer, economics professor at Warwick University, finds causal claims in economics have jumped. In 1990, 7.7 per cent of claims made in the literature were causal. In 2023, that hit 32.6 per cent. Additionally, papers with more causal claims are more likely to receive citations and wind up in top five journals, the research suggests.
Here is more from Harvey Nriapia at the FT.
Intergenerational mobility of immigrants in 15 destination countries
We estimate intergenerational mobility of children of immigrants in fifteen receiving countries. Children of immigrants have somewhat lower income than children of local-born parents. Around half of this gap can be explained by differences in parental income, with the remainder due to differences in mobility parameters. The daughters of immigrants enjoy higher absolute mobility than daughters of locals in most destinations. Absolute mobility of sons of immigrants is higher outside Europe and lower in Europe compared to sons of locals. Cross-country differences in absolute mobility are not driven by parental country-of-origin, but instead by destination labor markets and immigration policy.
Here is the paper, by Leah Boustan, et.al. Via the excellent Samir Varma.
What makes people happy?
Measuring politically sensitive attitudes in authoritarian settings is difficult because expressions of political support may reflect repression and social pressure as much as genuine belief. We propose a complementary approach: using responses to questions that are not themselves politically sensitive to provide indirect evidence about politically sensitive attitudes. Using subjective well-being (SWB), we ask whether within-person changes in life satisfaction, after accounting for changes in material and personal circumstances, can provide evidence of incorporation into a wartime national project. Drawing on more than a decade of panel data from the Russian Longitudinal Monitoring Survey (2013–2024), we find that life satisfaction rises significantly following Russia’s 2022 invasion of Ukraine. Gains are initially larger among ethnic Russians and smaller in Moscow and St. Petersburg, consistently larger in military-industrial regions, and sharply lower among those nearest the front. This theoretically patterned heterogeneity is difficult to reconcile with generalized social-desirability bias and is consistent with differentiated wartime incorporation. By 2024, rising non-response and attenuating subgroup differences suggest that the informational value of SWB may deteriorate over time.
That is from recent research by Sinikka Parviainen and William Pyle. Via Pippa Norris.
Immigrant Earnings Assimilation, 1981–2021
In this paper, we characterize trends in the earnings assimilation of immigrant workers from 1981 to 2021. We use administrative longitudinal data that contain the earnings of workers beginning in their first year of residence in the United States and in each year thereafter, allowing us to identify immigrants who eventually leave the United States (referred to here as return migrants). We use those data to produce the first examination of trends in earnings assimilation over a 41-year period and to estimate earnings assimilation separately for return migrants versus those who stay in the United States. We document several new facts about immigrants who arrived between 1981 to 2010. First, roughly one-fifth to one-third of immigrants return migrate from the United States within 10 years after arrival. Second, return migrants have entry earnings similar to those of permanent migrants but experience slower rates of earnings growth. Third, earnings assimilation occurs relatively quickly for cohorts arriving since the mid-1990s: The earnings of permanent immigrants converge, or come close to converging, with those of native-born people within 10 years after arrival. Migrants from earlier arrival cohorts experience significant earnings growth but generally do not converge to that of the native born. We discuss how changes in the labor market quality of immigrant cohorts (measured by their relative earnings upon entry) and selective return migration play an important role in determining whether repeated cross-sectional data over- or underestimate earnings assimilation.
That is from a new NBER working paper by
How well does AI peer review work?
Claude and I planted 100 known errors into 10 open-access psychology papers and then ran them through frontier models and two commercial AI review tools. In brief:
- The best single system caught 71 of 100 errors, while the worst caught 30.
- Pooling every system’s output caught 93 of 100. Models are only partly correlated in the errors they find, making ensembling a big lever for finding issues in papers. Check your papers against multiple models!
- Seven errors could not be caught by any system. All were omissions — information deleted from a paper rather than mistakes inserted into it.
- Refine.ink contributes more unique catches than any other single system, though it’s expensive.
- I didn’t measure false positives and I don’t know how this error distribution compares to the distribution of errors in real papers.
- I’ve made the papers, errors, model outputs, and the full experiment log public. I hope people can build on this work to create a comprehensive eval benchmark across disciplines.
That is from Paul Litvak, here is more. Note that is not even using the very latest generation of models.
Who enters Congress?
We trace the family origins of Members of Congress (MC) born between 1830 and 1950, linking each MC, their parents, and their brothers to the complete-count censuses. Future MCs have always been economic outliers. By 1940, close to 60 percent of MCs aged 18–40 held a college degree, against under 5 percent of comparable young men outside Congress. In the 19th-century censuses, MCs as adults held roughly three times the wealth of demographically matched controls. They also come from economically elite families: their fathers earn more, attain more education, and hold more wealth. The brothers of future MCs sit between the population and the MCs themselves. Across the socioeconomic measures we compare, we calculate that roughly half of an MC’s adult premium is shared with his brother and the other half is his own. Even as the American economy and political system have changed, the family background gap has widened over the last century and a half. Wealth, household servants, and four separate occupation-based scores all show the gap holding or growing across cohorts. We study four Progressive-Era reforms in a triple-difference design—women’s suffrage, the secret ballot, direct primaries, and the direct election of senators—and none produce detectable shifts in who reaches Congress. Our results document a Congress drawn from the most fortunate families in American society in every cohort we observe.
That is from a new NBER working paper by
Do Faculty Affect Student Partisanship?
We study whether Democratic college professors make their students more liberal. We link voter data to salary records from 33 state flagships and show that faculty skew Democratic, especially in the humanities and social sciences. We then use student transcripts from one flagship to estimate causal effects. Students become more liberal during college regardless of major and sort toward ideologically similar instructors. Exploiting plausibly random variation in when instructors teach courses, we find no effect of faculty partisanship on student partisanship. Text analysis shows exposure to liberal topics reflects student demand rather than instructor supply, leaving little room for indoctrination.
That is from a new paper by Micah Baum, Joaquín Endara, and Annaliese Paulson. Via Jay Fan Bavel.