Every branch of American power is run, on average, by people born between 1956 and 1967. Those are the most lead-exposed birth years in the country's history.

Read this as a cohort argument.

An ecological argument compares groups, not people. It says this birth cohort had high childhood lead as a group, and this birth cohort now holds these positions as a group, so the average exposure history of the people in those positions is the exposure history of the cohort. It says nothing about which individuals were exposed, how much, or with what effect. Two senators born the same year in the same state could have had childhood blood lead that differed by a factor of ten depending on housing, traffic and diet.

The individual-level science underneath is settled. Childhood lead lowers IQ, weakens working memory, and degrades the brain systems that hold impulse in check. That literature is on the IQ page, the ADHD page, and the personality page. What has not been studied is whether the cohort-average deficit shows up as anything measurable in how a legislature, a court, or a board behaves. That is the question this page sits on, and it is open.

The related page The Leaded Generation runs the same cohort logic across nine behavioral time series and keeps the series that do not fit in plain view. One of its rows is the mean age of Congress, and the correlation with cohort childhood lead there is weak, r = +0.43. That number is on this site already and it cuts against reading too much into this page.

The cohort math.

McFarland, Hauer and Reuben combined NHANES blood lead surveys from 1976 onward with leaded gasoline consumption records to estimate the childhood blood lead distribution of every American birth cohort from 1940 to 2015, then applied the Lanphear 2005 pooled dose-response to get IQ points lost per cohort. Their headline is 824,097,690 IQ points lost across the population alive in 2015, an average of 2.6 per person. The loss is concentrated in people born 1951 to 1980 and peaks in the late 1960s. Blood lead above 5 µg/dL was, in their words, nearly universal among those born 1951 to 1980.

Shares from McFarland 2022, Table 2. The three IQ figures are the only cohort-level deficits the paper reports in text. It also reports that 78% of the 1966 to 1970 cohort and 73% of the 1971 to 1975 cohort were above 15 µg/dL, and that about 3 million children in those two cohorts, 7% of them, were above 30 µg/dL, where the estimated loss is 7.4 points.

One thing to hold onto. The McFarland numbers are their model, built on their interpolation of pre-1976 blood lead from gasoline sales. We run our own childhood-lead-by-birth-year series on this site and it does not agree with McFarland on which year peaked or how deep the peak was. This page quotes McFarland only, because it is the published external estimate and the argument should not depend on which model you pick.

Where those cohorts sit now.

The other half of the argument is a set of age tables, most of them published by the institutions themselves. Where an institution publishes an average age, the mean birth year below is that average subtracted from the year of the count. Where it publishes individual birth dates, the mean is computed directly. The rightmost column is the McFarland share for the five-year cohort that mean falls in.

Three cautions on that table. It is a table of means, and the Senate has members born in the 1930s and members born in the 1980s. The federal judiciary and Federal Reserve rows rest on secondary reporting of primary data, and they are marked as such in the citations; the Court, the Congress, the governors, the CEOs and the presidents are from primary or near-primary tables. And the cabinet row is a placeholder for a number that nobody reputable has computed, so it is on the table only to show that even the youngest branch of leadership still sits inside the affected cohorts.

Congress itself has not aged much over the period. The CRS profile of the 110th Congress in 2007 put the average senator at 61.7 and the average representative at 55.9. Eighteen years later the figures are 63.9 and 57.9, about two years older each. What changed is not the age of the seats. It is that the calendar carried the leaded cohorts into them. Anyone who reaches a senior position in late middle age between roughly 2005 and 2035 was, by arithmetic, a child during the fifty years of leaded gasoline. Part of the overlap on this page is mechanical for that reason, and the confounders section comes back to it.

The overlap, drawn.

The bars are McFarland's share of each five-year birth cohort with childhood blood lead above 5 µg/dL. The numbered markers are the mean birth years from the table above. The three labeled bars are the only cohorts for which McFarland reports a per-person IQ deficit in text.

The chart shows that the markers cluster on the tallest bars. It cannot show whether that has any consequence, because nobody has measured the consequence. The same overlay would look the same for hospital department chairs, newspaper editors or four-star officers, and for the same mechanical reason: those jobs go to people in their late fifties and early sixties, and everyone who is in their late fifties and early sixties now was a toddler under leaded gasoline.

What lead damages, specifically, and why that matters for deliberation.

The deficit lead produces is concentrated in particular functions, and those functions are the ones a legislature or a court is supposed to run on. That is why this is a page about institutions and not another page about IQ.

None of these studies looked at anyone in public office. Nigg studied children, Reuben studied New Zealanders born in 1972, Needleman studied schoolboys, Schwaba studied county averages, Lee studied retirees. Getting from there to Congress takes two steps, individual to cohort and cohort to institution, and the second has no data behind it.

What has actually been studied about lead and politics.

Not much, and almost all of it comes from one city. We went looking so the "zero studies" card above would mean something specific.

Flint, and the vote. Pennington and Wiseman (2021, a chapter of a Berkeley economics dissertation, not peer-reviewed) matched the City of Flint's address-level lead test results to the Michigan voter file. Households that received a positive lead test before the November 2015 mayoral election turned out at a rate 3.1 percentage points lower than households that did not, and the effect was concentrated among Black voters, who turned out 14.4 points less. It is the only study anywhere that connects an individual lead measurement to an individual voting record, and the mechanism it proposes is demoralization, not neurotoxicity. López Moskovits (2023, revised 2025, a working paper) finds Flint precinct turnout fell 20 to 30 percent after the crisis relative to neighboring precincts, and reports the result weakens under robustness checks. Several peer-reviewed surveys, Morckel and Terzano 2019, Sobeck and colleagues 2020, Hamm and colleagues 2023, document collapsed trust in local and state government among Flint residents. In all of these the exposure is having lived through the crisis. None of them measures anyone's blood lead as a child and asks how they vote as an adult.

One thesis on childhood lead and vote choice. A 2026 Portland State honors thesis by Kragen Anderson ran cohort-average childhood blood lead against Cooperative Election Study vote data from 2016 to 2024 and reported a small positive association with Trump vote share, framed as an authoritarianism proxy. It is an undergraduate thesis, it is ecological by design, and it cannot separate lead from anything else that differs between birth years. It is on this page because it is the only piece of work we found that uses childhood blood lead as the exposure and a political choice as the outcome, and that says something about how empty the field is.

The indirect chain. Deary, Batty and Gale (2008, Intelligence) showed in the 1970 British Cohort Study that childhood IQ predicts adult voter turnout. Lead lowers childhood IQ. Nobody has closed that loop with lead data.

Officeholders. The one peer-reviewed paper that touches the behavior of rulers is Nriagu's 1983 essay in the New England Journal of Medicine arguing that Roman emperors showed signs of chronic lead poisoning from wine and food vessels. It is a historical argument from diet records, it has been criticized as overstated, and it is the entire literature. No study tests lead exposure against the votes, rulings, decisions or tenure of any modern legislator, judge or executive. That is what the zero on this page refers to.

The confounders, stated plainly.

If you wanted to argue that the current period of institutional dysfunction has nothing to do with lead, you would have several good arguments. They belong here.

The overlap is partly arithmetic. Leaded gasoline ran from 1923 to 1996, and childhood blood lead was above 5 µg/dL for most children born from about 1950 to 1980. Senior positions in every institution on this page go to people in their late fifties and sixties. For the whole period from roughly 2005 to 2035, the people in those positions were necessarily children under leaded gasoline, whoever they are and however they perform. Congress has aged only about two years since 2007. The cohort did not seize the seats; the calendar delivered it to them. This is the strongest argument against the page, and it does not go away.

Polarization has its own documented causes. Party sorting after the civil rights era, primaries that reward the ideological edge, geographic self-sorting, campaign finance and the collapse of local news each have a literature behind them that does not need lead. Roll-call polarization measures have risen steadily since the 1970s. A birth-cohort effect and a period effect can produce the same curve, and no one has run the age-period-cohort decomposition that would separate them for legislative behavior.

Media changed more than the members did. Cable news arrived in 1980, talk radio in the late 1980s, social platforms after 2006. A member born in 1961 legislates in a different information environment than one born in 1931, whatever either breathed as a toddler.

Selection cuts against the cohort average. Even if every cohort number here is correct, the people who reach the Senate or the bench are not a random sample of their birth year. Selection into elite positions plausibly favors people who were less exposed, or less affected, or both. The cohort average could be 5.9 and the average for the subset who became senators could be near zero. Public data cannot tell you. Reuben's Dunedin finding that childhood lead predicted downward social mobility points the same way: the exposed were, on average, less likely to end up in the rooms this page is about.

The personality leg is contested. Fraas and colleagues (2025) re-ran the Schwaba county analysis with birth-year fixed effects and the natural-experiment effect on personality shrank to nothing they could distinguish from zero. The individual-level Dunedin personality result stands, but anyone leaning on the county-level story for a political inference should know it has been challenged in print.

The productivity numbers are not clean. GovTrack counts 761 laws enacted by the 100th Congress (1987 to 1988) and 274 by the 118th (2023 to 2025). That looks like a collapse. GovTrack itself cautions that Congress has enacted roughly four to six million words of new law per two-year session since World War II and that the words have moved into fewer, larger bills. Fewer laws does not by itself mean less legislating. We do not use the laws-enacted series as evidence on this page for that reason.

The site's own test came back weak. On The Leaded Generation, the mean age of Congress correlates with cohort childhood lead at r = +0.43 across 119 observations, the weakest fit on that page. Robbery is +0.93. If lead cohort replacement drove institutional behavior as strongly as it drives street-level outcomes, this number would be higher.

Three claims, one gap.

One. The birth cohorts that hold the senior positions in American government, finance, business and higher education today had, on average, higher childhood lead exposure than any cohort before or since. That is McFarland plus eight public age tables, and it is not in dispute.

Two. Childhood lead exposure produces deficits in response inhibition, working memory, conscientiousness and agreeableness that persist into midlife and old age. That is Nigg, Reuben, Needleman and Lee, also not in dispute, with the county-level personality evidence noted as contested.

Three. No one has tested whether the first two facts combine into anything measurable at the institutional level. The Flint papers are about a crisis, not a cohort. The one thesis on cohort lead and voting is a thesis. That is the gap.

This page does not claim that lead caused any policy outcome, that any officeholder is impaired, or that the age of the current leadership is itself a problem. The people in these cohorts did not choose the exposure. It was put in the air before most of them were born.

The gap is a research question, and a cheap one. Childhood lead exposure by birth year and county is public. The birth year and home county of every federal legislator since the 1940s is public, and so are roll calls, sponsorships and floor transcripts. The Federal Judicial Center publishes the birth year and appointment history of every Article III judge since 1789, and every published opinion is text. A cohort-level analysis of legislative or judicial behavior against childhood lead, with period, incumbency and selection controls, could be run by a graduate student with a laptop. As far as we can find, it has not been.

The article that has not been written.

The age of American leadership gets covered as individual medical events, as a term-limits debate, or as editorial complaint about gerontocracy. Polarization gets covered through primaries, cable news, social platforms, geographic sorting and money. Neither body of coverage has asked what happened to this particular generation in childhood that did not happen to the one before or the one after.

The datasets that would frame the question have been public for years. McFarland's cohort estimates ran in PNAS in 2022 and won the journal's Cozzarelli Prize. The CRS profile of Congress is updated every session and posted for free. The Court publishes its justices' birth dates. The Fed publishes some of its governors'. Crist Kolder and Spencer Stuart publish CEO ages every year. Nobody has put them in the same article. Kevin Drum's lead-and-crime reporting at Mother Jones came closest to a civic argument and stopped at crime. The Atlantic has written at length about gerontocracy without the exposure data. Lead may explain nothing about the current period. The question still has not been asked in print, and that is the part worth noticing.

Related on this site.

Citations