Free Community College

Pay the tuition of every community college student through a federal-state partnership, with Washington covering three quarters of the bill and states the rest.

AI evaluation · not yet reviewed by a human

This evaluation was produced and sourced by an AI model; a human review is still pending. Figures and conclusions may still change. The review log is at the foot of the page.How review works →

Community college tuition is about 4,050 dollars a year, which grants already cover for the poorest students and nobody covers for the ones just above them. The bill before Congress would waive it for everyone: participating states charge eligible students nothing for two years, the federal government pays 75 percent of what that costs and rises to 80 percent for states that join early, and grant aid is then free to cover rent, transport and books rather than fees. Participation is voluntary for states, and institutions have to keep spending per student at least where it was. Nothing in it changes admission, which at a community college is open. This evaluation looks fifteen years ahead, because the effect on earnings only appears once a cohort has finished and worked.

Balance

Better for the future · 0.62 previous scale

Balance on the previous scale. The Bilanz 2.0 simulation is not yet available for this evaluation. The category comes from the share of the debate on the pro side (r).

For 46 · 62 % Against 28 · 38 %

Size class: medium Scale of this evaluation: Normalised Impact — unitless, calibrated to this topic. For comparison: one point here is worth roughly 1 billion euro per year. Tennessee's college-going rate rose from 58.6 to 64.4 percent in the first year of its free-tuition programme, and that jump is the number this evaluation is built on. What it does not say is how much of it was people who would not otherwise have studied at all and how much was people who would have started at a four-year college instead. Forty percent is treated here as diverted rather than new; treat all of it as new and the measure comes out clearly positive, treat most of it as diverted and it comes out level. How we score →

Arguments for

Arguments against

6 arguments evaluated · Scoring v1.3 Δ absolute +18

Arguments — For

3 arguments

People who finish something

23of 100

Tennessee's college-going rate rose by nearly six percentage points when its community colleges went free, and the students who came were more than twice as likely to leave with a credential as other school leavers.

Value 8 · Life chancesImpact 4.9Plausibility 6

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is what a qualification does to the rest of a person's life: the jobs it opens, the wages attached to them, and the chance to go further. This site places life chances a class above ordinary money, because what changes is the range of lives available rather than the size of a bank balance. Later earnings are the measuring rod because they are the only thing anyone has followed for long enough, but they are the measurement and not the good. Read as ordinary money at the middle of the scale, this argument would be a third smaller. Nothing is counted here for the students who would have enrolled anyway and now pay nothing, which is money and appears as its own argument. The value is a class above money because the stream is the range of lives available to a school leaver, with later earnings standing in as the long measurement.

About 3.4 million Americans leave school each year, and if states holding seventy percent of them take part, 2.4 million are covered. Tennessee's college-going rate rose from 58.64 to 64.4 percent in the first year of its programme, which is 5.8 percentage points, or about 138,000 additional entrants a year at national scale ^[4]. Not all of that is new study: some of it is students who would have started at a four-year college and go to a community college instead, which is booked against this measure in its own argument. Sixty percent is treated as genuinely additional, in a range from 40 to 90 percent, leaving about 83,000. Community college completion within six years runs near 35 percent, giving roughly 29,000 additional credentials a year. What one is worth over a working life is put at 120,000 euro, in a range from 50,000 to 250,000: the gap between a two-year credential and a school leaving certificate is real and it is not the gap a bachelor's degree opens. Weighting by where these families sit gives 1.4. The Impact is the second largest here, and the number that carries it is how much of Tennessee's enrolment jump was new study rather than moved study.

▸ Show calculation ▾ Hide calculation

	School leavers in participating states [6]	3.4 million a year, 70 percent of them covered	2.38 million a year
×	Rise in the share going on to study Tennessee's college-going rate rose from 58.6 to 64.4 percent in the first year of its programme [4]	5.8 percentage points	138,000 additional entrants a year
×	Share that is new study rather than moved study Setting, range 40 to 90 percent: the rest would have started at a four-year college and is booked against this measure separately	60 %	82,800 entrants a year
×	Share who leave with a credential community college completion within six years [3]	35 %	28,980 credentials a year
×	Lifetime value of a credential Setting, range 50,000 to 250,000 euro: the gap over a school leaving certificate, which is not the gap a bachelor's degree opens	120,000 euro each	3.48 billion euro a cohort
×	Weight of a euro in these families community college students are concentrated below the middle of the income distribution	1.4	4.87 billion euro a year
÷	Normalised Impact scale of this evaluation	1 billion euro a point	4.87

Score 4.87 Impact × 8 Value × 6 Plausibility ÷ 10 = 23 of 100

The counterfactual is Tennessee before its programme and the states around it, which is what the comptroller's evaluation and the published studies of promise programmes use. The design is quasi-experimental: an announced statewide policy with a sharp start date, compared against neighbouring states and against Tennessee's own trend, which is weaker than a lottery and stronger than a cross-section. Kalamazoo's programme, which was allocated by residence rather than by choice, gives the same direction with a cleaner comparison — credential attainment 9 to 12 percentage points higher at six years ^[5]. The confounder that matters is that Tennessee's programme arrived with a mandatory application, an adviser and a community service requirement, so what is measured is a package rather than a price; a federal programme that waived tuition without the rest would not reproduce it. That is named and unresolved. Reverse causation does not arise from a legislated start date. The Plausibility is at the middle to upper part: the enrolment response is well measured in two states, and what produced it may be the advising rather than the money.

evidence basis: Multiple precedents · P ceiling 7 identification: Quasi-experimental · rung ceiling 8

Counterfactual: Tennessee before the programme and neighbouring states over the same years; Kalamazoo residents against non-residents. Design: quasi-experimental — announced statewide policy with a sharp start date, plus a residence-based comparison in Kalamazoo [4][5]. Confounder: Tennessee's programme bundled free tuition with a mandatory application, an adviser and community service, so the package rather than the price may carry the effect; named and unresolved. Direction: reverse causation does not arise from a legislated start date. Ceiling: quasi-experimental 8.0 less one point because what is transferred is a package, not the price alone.

Tuition families stop paying

20of 100

Community college tuition is about 4,050 dollars a year. Grants cover it for the poorest students and nobody covers it for the ones just above them, which is where most of the relief lands.

Value 5 · Household budgetsImpact 7.3Plausibility 5.5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is money families no longer pay, priced at the middle of the scale like all money. The federal budget pays it instead, which is the argument standing opposite, and the difference between the two is what a euro is worth to a household at the income where a four-thousand-dollar bill is the reason somebody does not enrol. Only the part of the federal outlay that replaces a bill somebody is paying today is counted here; the rest pays for students who would not have enrolled at all, and what they gain is the first argument rather than money. Nothing is counted for the reduction in anxiety about paying, which is the same fact as the money. The value is the middle of the scale, and only the part of the public money that replaces a bill somebody already pays appears here.

The federal share of the programme is about 9 billion dollars a year, from a ten-year cost of 90 billion, or 7.76 billion euro ^[7]. Fifty-five percent of it replaces tuition somebody is paying today, in a range from 40 to 70 percent; the rest pays for students who are newly enrolled and for the states' administrative costs. That is 4.27 billion euro a year. The households concerned are concentrated just above the point where grant aid runs out — too well off for a full grant, not well off enough to write the cheque without thinking — which gives a weight of 1.7, in a range from 1.3 to 2.2. The relief is smaller than the headline because a large share of community college students already pay nothing after grants, which is the reason free tuition is often criticised as reaching the wrong people. That criticism is about who gains and not about whether anyone does. The Impact is the largest here, and it is smaller than the programme because much of what it buys is study that is not happening today.

▸ Show calculation ▾ Hide calculation

	Federal share of the programme [7]	9 billion dollars a year, from a ten-year cost of 90 billion	7.76 billion euro a year
×	Share that replaces tuition somebody pays today Setting, range 40 to 70 percent: the rest pays for students who are newly enrolled and for administration	55 %	4.27 billion euro a year
×	Weight of a euro in these households Setting, range 1.3 to 2.2: the relief lands just above the point where grant aid runs out	1.7	7.25 billion euro a year
÷	Normalised Impact scale of this evaluation	1 billion euro a point	7.25

Score 7.25 Impact × 5 Value × 5.5 Plausibility ÷ 10 = 20 of 100

The cost is a projection with a known structure: a federal share of a per-student tuition figure, multiplied by an enrolment that is known and a participation rate that is not. The counterfactual is the present grant system. The chain has one behavioural link that carries the quantity, which is how many states join: participation is voluntary, the state has to find a quarter of the cost, and the states with the lowest tuition and the weakest budgets are the ones for whom the arithmetic works worst. The confounder that matters is the split between relief and new enrolment, which is set here rather than measured, and which moves this argument and the first one in opposite directions. That is named and unresolved. Reverse causation does not arise. The Plausibility is a little above the middle: the arithmetic is simple, the participation rate and the relief share are both assumed.

evidence basis: Projection · P ceiling 6 identification: Mechanistic · rung ceiling 6

Counterfactual: the present grant system, under which the poorest students already pay no tuition. Design: mechanistic — the federal share is statutory, but state participation and the split between relief and new enrolment carry the quantity and neither is measured. Confounder: the relief share moves this argument and the credentials argument in opposite directions; named and unresolved. Direction: no reverse causation. Ceiling: projection 6.0 binds and mechanistic gives the same.

Loans that are never taken

3.3of 100

Community college borrowers default more often than any other group of students, because they borrow small amounts and frequently leave without the credential that would have paid for them. Free tuition removes the reason to borrow at all.

Value 5 · Household budgetsImpact 1.3Plausibility 5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is the wreckage of a defaulted student loan: a damaged credit record, wages garnished, tax refunds withheld, and a debt that follows a person who never got the qualification it paid for. It is priced at the middle of the scale like all money, because what is measured is what the default costs the borrower in money and money-equivalent terms. It is separate from the tuition relief above: a student who borrows and repays without difficulty gains only from that argument, and one who defaults gains from this one as well. The debt itself is not counted here, since a loan repaid is not a loss to anybody. Nothing is counted for the government's collection costs, which are small beside the borrower's. The value is the middle of the scale, and what is priced is the damage of a default rather than the debt.

About 1.5 million community college students borrow in a year. Free tuition removes the reason for roughly 40 percent of that borrowing, in a range from 20 to 60 percent — small loans taken to cover fees rather than rent — which is about 600,000 students a year who do not borrow. Default rates among community college borrowers run near 15 percent within three years, the highest of any sector, so about 90,000 defaults a year do not happen. What one default costs the borrower is put at 8,000 euro, in a range from 3,000 to 20,000: garnished wages, a credit record that raises the price of everything bought on credit for seven years, and fees added to the balance. Weighting by where these borrowers sit gives 1.8, higher than for the tuition relief because defaulters are the poorest part of an already poor group. The Impact is a sixth of the relief above, which is the honest size of an effect that reaches only the students who would have borrowed and then failed.

▸ Show calculation ▾ Hide calculation

	Community college students who borrow in a year [2]		1.5 million students
×	Share whose reason to borrow disappears Setting, range 20 to 60 percent: small loans taken to cover fees rather than rent	40 %	600,000 students a year
×	Share who would have defaulted the three-year default rate among community college borrowers, the highest of any sector [2]	15 %	90,000 defaults a year
×	Cost of one default to the borrower Setting, range 3,000 to 20,000 euro: garnished wages, seven years of a damaged credit record, and fees added to the balance	8,000 euro	720 million euro a year
×	Weight of a euro for these borrowers defaulters are the poorest part of an already poor group	1.8	1.3 billion euro a year
÷	Normalised Impact scale of this evaluation	1 billion euro a point	1.3

Score 1.3 Impact × 5 Value × 5 Plausibility ÷ 10 = 3.3 of 100

The counterfactual is the same students under the present system, borrowing to cover fees. The chain is named — no tuition bill, no small loan, no default — and the default rates at the end of it are measured rather than assumed. What is not measured is the first link: how much community college borrowing is for tuition rather than for living costs, which is what decides whether free tuition removes the loan or merely shifts what it pays for. That is the confounder and it points against the argument, since surveys of community college students report rent and transport as the larger burden; it is named and unresolved and is why the share is set at 40 percent rather than higher. Reverse causation does not arise. No study has measured default rates in a state before and after free tuition, so the number rests on a complete chain rather than on a finding. The Plausibility is at the middle: the default rates are counted and the share of borrowing that free tuition actually removes is assumed.

evidence basis: Mechanism · P ceiling 6 identification: Mechanistic · rung ceiling 6

Counterfactual: the same students borrowing to cover fees under the present system. Design: mechanistic — chain named (no tuition bill, no small loan, no default) with default rates measured and the share of borrowing removed assumed. Confounder: community college students report rent and transport as the larger burden, so free tuition may shift what the loan pays for rather than removing it; named, unresolved, and the reason the share is set at 40 percent. Direction: no reverse causation. Ceiling: mechanistic 6.0 binds.

Arguments — Against

3 arguments

Nine billion dollars a year

21of 100

The federal share comes to about 9 billion dollars a year, with states finding a quarter more. It is a small programme by federal standards and it is still real money.

Value 5 · Public financesImpact 7.8Plausibility 5.5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is federal money spent, priced at the middle of the scale as public money always is. The whole federal outlay stands here, not only the part that relieves a family bill, because the treasury pays all of it; what the rest buys stands opposite as study rather than as money. The states' quarter is not added, because a state that declines to find it simply does not join, so the federal figure is the binding one. Nothing is added for the taxes that would have to rise, which is the same euro under a later name. Nothing is deducted for the income tax the additional graduates would eventually pay, which arrives fifteen years later and is small beside the outlay. The value is the middle of the scale, and the whole federal outlay stands here while what it buys stands opposite.

The programme has been costed at 90 billion dollars over ten years, of which 80 billion is the community college partnership and 10 billion is support for minority-serving institutions ^[7]. That is about 9 billion dollars a year, or 7.76 billion euro. Public money carries the standard weight of 1.0. The figure assumes every state joins; each state that stays out lowers the cost and the benefit in the same proportion, which is why the participation rate does not move the balance of this evaluation, only its size. The cost would rise if enrolment responds more strongly than Tennessee's did, since the programme pays per student. Against the 4.27 billion that relieves existing bills, about 3.5 billion buys study that is not happening today. The Impact is the largest single cost here and it is modest by federal standards — about a tenth of what the child care entitlement in the neighbouring area would cost.

▸ Show calculation ▾ Hide calculation

	Ten-year cost of the programme [7]	80 billion for the community college partnership, 10 billion for minority-serving institutions	90 billion dollars
÷	Per year	over ten years	9 billion dollars a year
÷	In euro, at the standard weight for public money exchange rate used throughout this evaluation	1.16 dollars to the euro, weight 1.0	7.76 billion euro a year
÷	Normalised Impact scale of this evaluation	1 billion euro a point	7.76

Score 7.76 Impact × 5 Value × 5.5 Plausibility ÷ 10 = 21 of 100

The direction is certain and the level depends on participation and on the enrolment response, both of which are behavioural. The counterfactual is the present grant system, which spends about 30 billion dollars a year on grants of all kinds. The chain from statute to outlay is short: a federal share of a tuition figure, times enrolment. What makes it a projection rather than an accounting identity is that enrolment is not fixed — the programme pays for the students it attracts — and that states choose whether to take part. The confounder that would push the cost up is exactly the enrolment response the first argument counts as a benefit, which means the two move together and a reader should not treat one as certain and the other as speculative. Reverse causation does not arise. The Plausibility is a little above the middle: the cost per student is known, the number of students is the thing the programme is designed to change.

evidence basis: Projection · P ceiling 6 identification: Mechanistic · rung ceiling 6

Counterfactual: the present grant system, about 30 billion dollars a year in grants of all kinds. Design: mechanistic — federal share times enrolment, with enrolment itself the thing the programme changes. Confounder: a larger enrolment response raises the cost and the benefit together, so the two cannot be treated with different confidence; named. Direction: no reverse causation. Ceiling: projection 6.0 binds and mechanistic gives the same.

Students who start lower

6of 100

A free two-year college is also a cheaper alternative to a four-year one, and some students take it. Starting at a community college lowers the chance of finishing a bachelor's degree, and the gap between the two qualifications is larger than the gap this programme closes.

Value 8 · Life chancesImpact 1.5Plausibility 5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is the same one the first argument counts, seen from the other end: a student who would have taken a bachelor's degree and takes a two-year credential instead has a narrower range of lives afterwards, not a wider one. It carries the same value class for the same reason, and later earnings stand in as the measurement in the same way. What is counted is only the difference between the two paths for the students who switch, not the whole value of the degree they might have taken, since most of them still qualify for something. Nothing is counted for the four-year institutions that lose the students, which is their revenue and not anybody's life. The value is a class above money for the same reason as the first argument, because it is the same stream running the other way.

Forty percent of the enrolment increase is treated here as students who move rather than students who are new, in a range from 10 to 60 percent — about 55,000 a year. The transfer literature finds that beginning at a community college lowers the probability of completing a bachelor's degree by something in the order of 15 percentage points against similar students who start at a four-year institution, even when the intention to transfer is the same. Applied to those 55,000, that is about 8,300 people a year who do not finish a degree they would otherwise have finished. The gap between a bachelor's degree and a two-year credential over a working life is put at 150,000 euro, in a range from 80,000 to 300,000. Weighting gives 1.2, lower than for the first argument because students choosing between a two-year and a four-year college sit higher in the distribution than those choosing between studying and not. The Impact is a third of the credentials gained on the other side, and the two are the same enrolment jump split between two readings of it.

▸ Show calculation ▾ Hide calculation

	Additional entrants who moved rather than newly enrolled Setting, range 10 to 60 percent: Tennessee's own evaluation reports more transfers to bachelor's programmes, which points the other way [4]	40 percent of 138,000	55,200 students a year
×	Lower chance of finishing a bachelor's degree measured against matched students who begin at a four-year institution [3]	15 percentage points	8,280 people a year
×	Lifetime gap between a degree and a two-year credential Setting, range 80,000 to 300,000 euro	150,000 euro	1.24 billion euro a year
×	Weight of a euro in these families students choosing between two-year and four-year colleges sit higher in the distribution than those choosing between studying and not	1.2	1.49 billion euro a year
÷	Normalised Impact scale of this evaluation	1 billion euro a point	1.49

Score 1.49 Impact × 8 Value × 5 Plausibility ÷ 10 = 6 of 100

The counterfactual is students of similar attainment and background who begin at a four-year institution, which is what the transfer literature constructs — and constructs by matching rather than by any exogenous variation, because nobody randomises where a student starts. The design is therefore controlled rather than quasi-experimental, and the confounder is the obvious one: students who choose a community college differ from those who choose a four-year college in ways that also predict finishing, and matching on grades and income does not remove it. That is unresolved and it is why the plausibility sits at the middle. Reverse causation is present in the raw comparison and is exactly what the matching tries to remove. The diversion share itself is a setting rather than a measurement: Tennessee's own evaluation reports a rise in transfers to bachelor's programmes, which points the other way and is the reason the range runs down to 10 percent. The Plausibility is at the middle: the completion gap is real and consistently found, and neither it nor the share of students who divert is identified.

evidence basis: Converging studies · P ceiling 7 identification: Controlled · rung ceiling 7

Counterfactual: matched students of similar attainment and background beginning at a four-year institution. Design: controlled — matching on observed characteristics, with no exogenous variation in where a student starts. Confounder: unobserved differences between students who choose a two-year and a four-year college that also predict finishing; unresolved. Direction: reverse causation is present in the raw comparison and is what the matching attempts to remove. Ceiling: controlled 7.0 binds below the converging-studies ceiling of 9.0. The diversion share is carried in its own range, which runs down to 10 percent because Tennessee reports more transfers to bachelor's programmes, not fewer.

What free tuition does not pay

1.3of 100

Tuition is the smaller half of what a community college student spends. Where it has been waived, institutions have raised the fees that sit next to it, and those are not covered.

Value 5 · Household budgetsImpact 0.6Plausibility 4.5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is money students still pay after the programme has done its work: laboratory and technology fees, course materials, parking, and the charges that a college can rename once tuition is fixed at zero. It is priced at the middle of the scale like all money and weighted like the relief above, because it lands on the same households. It is not the same as the living costs the programme never claimed to cover, which are unchanged and are therefore not an argument here. Nothing is counted for the colleges receiving the fees, which is a transfer inside the sector. The value is the middle of the scale, and only the charges that rise because tuition is waived are counted, not the ones that were there anyway.

About 4.27 billion euro of tuition relief reaches households. Where public tuition has been capped or waived, institutions have recovered part of it through fees that are not called tuition; eight percent is used here, in a range from two to twenty percent, which is 342 million euro a year. Weighting by where these households sit gives the same 1.7 as the relief itself. That is 580 million euro. The mechanism has a limit in the bill, which requires participating institutions to maintain their spending per student and bars them from cutting the aid they already give, but nothing in it fixes what may be charged for a laboratory or a parking permit. The size would be larger in states whose colleges are already underfunded and smaller where the state's contribution is generous. The Impact is the smallest here, about a thirteenth of the relief it eats into.

▸ Show calculation ▾ Hide calculation

	Tuition relief reaching households [7]	from the second argument	4.27 billion euro a year
×	Share recovered in charges that are not called tuition Setting, range 2 to 20 percent: studies of capture in college aid find about a quarter in a different market	8 %	342 million euro a year
×	Weight of a euro in these households the charges land on the same households as the relief	1.7	581 million euro a year
÷	Normalised Impact scale of this evaluation	1 billion euro a point	0.58

Score 0.58 Impact × 5 Value × 4.5 Plausibility ÷ 10 = 1.3 of 100

The counterfactual is the same colleges under the present fee structure. The chain is complete: tuition fixed at zero, a revenue line removed, other charges available and unregulated. The counter-mechanism is named and it is in the statute — the maintenance-of-effort requirement and the bar on cutting existing aid — and it constrains spending rather than charging, so it limits the mechanism without closing it. That is unresolved. What is missing is any measurement: no study has compared fee schedules before and after a promise programme, although the data to do it are published by every college. Studies of capture in college aid find pass-through of roughly a quarter, which is a different market and a larger figure than the one used here. Reverse causation does not arise. The Plausibility is below the middle: the mechanism is ordinary and nothing has measured it in this market.

evidence basis: Mechanism · P ceiling 6 identification: Mechanistic · rung ceiling 6 band: Chain closed, unevidenced · P 4–5

Counterfactual: the same colleges under their present fee structure. Design: mechanistic — chain named (tuition at zero, revenue line removed, other charges unregulated) with no measurement in this market. Confounder: the bill's maintenance-of-effort requirement, which constrains spending rather than charging and therefore limits the mechanism without closing it; named and unresolved. Direction: no reverse causation. Ceiling: mechanistic 6.0 binds. Band: chain closed but unevidenced — links named, counter-mechanism stated, only the size missing.

Nothing measured argues against the claim; what is missing is any comparison of fee schedules before and after a promise programme. The counter-mechanism — the maintenance-of-effort requirement in the bill — is named and unresolved. Read back: about half the time, colleges recover roughly the share of waived tuition assumed here through other charges.

Open: Every college publishes its fee schedule each year, and the states with promise programmes started in known years. Comparing schedules across the start date against colleges in neighbouring states would measure this directly and could carry it above 6.

Summary

This comes out clearly on the positive side, and it does so on a smaller budget than any other measure of comparable weight on this site. Nine billion dollars a year buys about 83,000 additional people starting a course and roughly 29,000 additional credentials a year, plus tuition relief for the students just above the point where grant aid runs out, who are the ones a four-thousand-dollar bill actually stops. The strongest objection is not the cost but the direction of travel: some of the students who show up were on their way to a four-year college, and starting at a community college makes finishing a degree less likely. Forty percent of the enrolment jump is treated here as that kind of movement, which is the single number that decides how large the gain is. What the evaluation cannot settle is whether Tennessee's result came from the free tuition or from the adviser and the mandatory application that came with it — a federal programme that sends only the money may buy less than Tennessee did.

Outlook — effect over time

Better for the future · 0.62 previous scale

Sources

The Century Foundation: How America's College Promise would reshape the free college landscape. tcf.org
U.S. Department of Education, Federal Student Aid: Federal student aid portfolio and cohort default rates. studentaid.gov
National Center for Education Statistics: Digest of Education Statistics: enrolment, completion and transfer. nces.ed.gov
Tennessee Comptroller of the Treasury, Office of Research and Education Accountability: Tennessee Promise evaluation. comptroller.tn.gov
Bartik, Hershbein and Lachowska, Journal of Human Resources: The Effects of the Kalamazoo Promise Scholarship on College Enrollment and Completion. jhr.uwpress.org
National Center for Education Statistics: Public high school graduates and immediate college enrolment rates. nces.ed.gov
The Century Foundation: Congress's free community college plan could benefit 8 million students, with the right funding formula. tcf.org

Last reviewed by Claude Opus 5 · September 6, 2026 · 2× AI, not yet reviewed by a human

September 6, 2026AI reviewClaude Opus 5record updated
i_spanne und normalisierung an allen 6 (nur globaler Anker eur-nennwert), Paar pro-2 ⇄ con-1 verdrahtet, Gruppe enrolment-response-is-real, massstab_hinweis ohne r. Die Aufteilung neu/verlagert skaliert zwei Argumente mit umgekehrtem Vorzeichen und bleibt deshalb Szenario statt Spanne. Kategorie bleibt Besser (P(D>0) 0,80). Offen fuer Julian: count-once-Verletzung der Gewichte 1,4/1,2 auf V-8-Bildungsstroemen und ein einbeiniger Brutto-Transfer in con-3.
September 6, 2026AI reviewClaude Opus 5First evaluation
First evaluation: Tennessee's enrolment jump split between new study and study moved from four-year colleges, which is what decides the result.

Evaluations are produced with AI support and reviewed on a schedule for new developments; human passes are marked separately.How we review →