Thirty-Seven Years of Educational Fraud

Executive Summary

Thirty-seven years ago, a West Virginia physician named John Cannell discovered that every state in America was reporting its elementary school students performed above the national average on standardized achievement tests. This statistical impossibility—since named the “Lake Wobegon Effect” after Garrison Keillor’s fictional town where all the children are above average—was documented in reports published in 1987 and 1988, confirmed by the U.S. Department of Education, covered by the Washington Post, New York Times, Wall Street Journal, and 60 Minutes, and led to congressional hearings.

Nothing changed.

Today, the same structural fraud continues. State tests still report flattering results that collapse when compared against the National Assessment of Educational Progress (NAEP)—the only secure, independently administered national benchmark. Mississippi claims 75% reading proficiency; NAEP finds 16%. New York City reports 49% ELA proficiency on state tests; NAEP reports 28%. Cleveland reported record reading gains in 2022 while NAEP showed fourth-grade reading proficiency falling from 13% to 7%. The numbers change. The mechanism does not.

The United States remains the only high-performing nation in the 2022 Programme for International Student Assessment (PISA) without a single national, secure, independently administered examination with school-level and student-level scores. U.S. educators pick the test, administer the test, and report the results—the equivalent of asking taxpayers to audit their own tax returns and submit the results to the IRS.

This white paper summarizes the history of this fraud, documents its continuation in the present, and proposes two concrete reforms that would cost a fraction of what the current corrupt system wastes annually.

I. The Discovery: West Virginia, 1985

In 1985, West Virginia newspapers announced that all 55 of the state’s counties had tested above the national average on commercial standardized achievement tests. This announcement was celebrated by politicians, administrators, and newspapers across the state.

At the time, I was practicing general medicine in Flat Top, West Virginia, in the heart of Appalachian coal country. My patients included adolescents whose school records showed above-average scores on the Comprehensive Test of Basic Skills. Many of these patients could not read. Their teachers, who were also my patients, explained the mechanism directly: the test questions were the same every year, teachers were explicitly instructed by test publishers to review the test before administration, and by the following year the questions had been taught directly to students.

West Virginia—then as now among the lowest-performing states on any honest measure—was above average because “above average” was a fiction constructed by the commercial testing industry and ratified by a school system with every financial and political incentive to accept it.

The obvious question was whether this was unique to West Virginia. In 1987, I founded Friends for Education and dispatched my clinic staff—a nurse, an x-ray technician, and a lab technician—to call the education departments of all 50 states and request their aggregate scores on commercially available norm-referenced tests. The returns were unambiguous: every state reported above-average elementary achievement. Ninety percent of school districts claimed above-average performance. Approximately 70% of individual students were told they performed above the 50th percentile on a normally distributed scale.

I named this the Lake Wobegon Effect. When CTB/McGraw-Hill threatened to sue me if I published the findings, I dipped into personal savings and published anyway.

II. The Mechanism: How the Fraud Works

The Lake Wobegon Effect is not primarily a story of individual teachers cheating, although that occurs. It is a story of structural corruption—a testing infrastructure designed, whether intentionally or through decades of institutional drift, to produce flattering results regardless of actual student learning.

The Norming Problem

Commercial norm-referenced tests compare current students to a “norm group”—a sample of students tested when the test was standardized. When that norm group is five to seven years old, current students will naturally outscore it even without genuine improvement, because curriculum alignment with the test increases over time. Test publishers sold school districts the ability to choose their norm group. A publisher representative explained to me in 1985, when I posed as a school administrator, that choosing a lower-performing norm group would produce high initial scores, while a “tougher” norm group would look bad in year one but produce flattering annual gains thereafter. The publisher knew that unchanged test questions guaranteed rising scores regardless of actual learning.

The Security Problem

Unlike college entrance examinations, medical licensing examinations, or any professional credentialing test, public school achievement tests in the United States have historically been administered by the very educators whose performance they purport to measure. CTB/McGraw-Hill’s own test coordinator handbook explicitly instructed teachers to review all test questions before administration. Teachers who memorized questions taught the answers to students the following year. The questions did not change. Scores rose. Learning did not.

The Exclusion Problem

Districts systematically excluded special education students and limited-English-proficiency students from testing pools used for public reporting. Removing the lower tail of the distribution inflates reported averages without any genuine educational improvement.

The Preparation Materials Problem

At a 1990 pre-filming meeting, 60 Minutes correspondent Morley Safer showed me “test preparation” materials obtained from school districts across the country. They contained the actual test questions with answers. Some of the same academic consultants who designed the tests and argued publicly that score inflation was caused by “high-stakes pressure”—not publisher fraud—had profitable side businesses selling these materials to school districts.

III. The Evidence: State Tests vs. NAEP

The National Assessment of Educational Progress is the only honest benchmark in American public education. It is administered by an independent contractor, uses secure questions that change each cycle, and is not controlled by the educators whose students it tests. The gap between state test results and NAEP results is the direct measure of the Lake Wobegon Effect in any given state.

The pattern is consistent across four decades: state-controlled tests show improvement or proficiency; NAEP shows stagnation or decline. The most parsimonious explanation is the one I identified in 1987: the state tests are invalid because their security has been compromised, their norm groups are manipulated, and the educators being evaluated control the evaluation instrument.

California’s experience is instructive. From 1998 to 2002, California administered the Stanford-9 using the same test booklets and questions for five consecutive years. Fourth-grade reading scores rose from the 40th to the 50th percentile nationally. In 2003, California switched to the California Achievement Test-6, and scores immediately plummeted to below where they had been in 1998—because teachers encountered unfamiliar questions. NAEP showed no improvement during the entire period. The five-year “gain” was an artifact of question familiarity, not learning.

IV. Current Evidence: The Fraud Continues

The structural fraud documented in 1987 and 1988 has not been reformed. The following incidents, drawn from public records and major news sources, demonstrate that identical mechanisms continue to operate in American public schools.

Mississippi: The Largest Documented Gap

Mississippi’s state testing system reports 75% reading proficiency. The National Assessment of Educational Progress finds 16%—a 59-percentage-point gap. This is the most extreme documented discrepancy between state-reported and independently verified proficiency in the United States. Source: Mississippi DOE; NAEP Data Explorer.

New York City: 21-Point Gap in the Nation’s Largest District

New York City public schools report 49% ELA proficiency on New York State tests. NAEP reports 28%. The 21-point gap in the largest school district in the United States affects more than one million students. Source: NYS Education Department; NAEP.

Cleveland: Gains That Weren’t

Cleveland Metropolitan School District reported record reading gains in 2022. NAEP simultaneously showed fourth-grade reading proficiency falling from 13% to 7%—nearly cutting the already-dismal rate in half. Source: NAEP Data Explorer.

Texas: Structural Fraud in Teacher Certification

Texas prosecutors identified more than 430 fraudulent teacher certifications since 2020, with 162 cases under active investigation as of February 2025. Two uncertified individuals were charged with child predation. The mechanism—a credentialing process with inadequate security and self-administered components—is structurally identical to the student testing fraud documented in 1987. Source: Texas Tribune, October 29, 2024.

Ohio: Record Falsification

A 2023 investigation found that a Westerville, Ohio, principal altered 63 student records to artificially boost graduation rates. Source: Columbus Dispatch, August 23, 2023.

V. The International Context

The United States ranked 34th in mathematics among PISA 2022 participants—behind Singapore, Japan, South Korea, Estonia, Finland, Canada, the United Kingdom, Australia, and 25 other nations. Every top-ranked nation in PISA maintains a national, secure, independently administered examination with individual student and school-level scores accessible to parents.

In Singapore, Japan, Estonia, and Finland, educators do not have advance access to examination questions. Test preparation materials sold commercially cannot contain actual test items because the items are secure. The comparison group is current and national, not a five-year-old sample chosen by the test publisher for commercial advantage.

The United States is unique among high-performing PISA nations in having no such instrument. NAEP is the closest approximation, but it samples rather than tests every student and does not produce individual scores. Parents in the United States cannot learn from any secure, independently validated source how their child actually performs relative to national norms. The scores they receive from state testing systems are, as this paper documents, structurally unreliable.

School choice—the central education reform project of many policymakers—cannot function without honest measurement. A parent choosing between a district school and a charter school, or between a public school and a voucher-funded private school, requires reliable comparative data. The current testing system does not provide it. School choice built on Lake Wobegon test scores is a market built on fraudulent price signals.

VI. The Political Economy of Dishonest Testing

The Lake Wobegon Effect persists because it serves the interests of every adult in the educational system. Politicians can report that schools are improving. Administrators can demonstrate they are doing their jobs. Teachers can show their students are succeeding. Test publishers can sell the same questions year after year at premium prices. Academic consultants can sell “test preparation” materials that are functionally crib sheets. Newspapers can run flattering local stories. Parents can be reassured.

The only constituency that does not benefit from this arrangement is the child who is told she is performing at grade level while she cannot read.

The resistance to reform has been consistent and has come from predictable sources. When Secretary Bennett held his 1988 meeting, academic consultants who were simultaneously advising test publishers and selling preparation materials argued that score inflation was caused by “high-stakes pressure”—a formulation that deflected responsibility from publishers and administrators and placed it on an abstraction. This framing has persisted for four decades. When NAEP results are published showing lower performance than state tests, state superintendents explain that they have “not aligned their curriculum with NAEP”—as if NAEP, not the state test, were the outlier requiring explanation.

VII. Current Opportunity: School Choice and Honest Data

Federal and state interest in school choice creates a genuinely cross-partisan opportunity for testing reform.

Those who support school choice require honest comparative data for the market to function. A voucher program in which parents choose between schools using Lake Wobegon test scores is not a market; it is a lottery conducted with fraudulent information.

Those who support educational equity require honest data to identify which schools are failing which students. NAEP data shows that the students most harmed by inflated state test scores are those in the lowest-performing schools and districts, who are told they are succeeding while they fall further behind.

The United States spends approximately $1.7 billion annually on commercial standardized testing that produces systematically inflated scores. The federal government is a major purchaser of commercial testing services through Title I funding requirements. It is purchasing a product that does not perform its stated function.

VIII. Proposed Reforms

Reform 1: Expand NAEP to Test Every Student in Grades 4, 8, and 12

The National Assessment of Educational Progress is currently administered as a sample survey. Expanding NAEP to a census—testing every student in grades 4, 8, and 12—would provide parents, for the first time, with an honest, independently validated measure of their child’s performance relative to national norms. NAEP’s security infrastructure is already established. The marginal cost of expanding from a sample to a census is substantially lower than the current expenditure on commercial testing systems that provide less reliable information.

Reform 2: Establish an Independent National Testing Agency

The fundamental structural problem is that American educators audit their own performance. The solution is to remove educator control from the assessment process entirely, as every high-performing PISA nation has done. An independent national testing agency—modeled on the institutions that administer college entrance examinations, medical licensing examinations, and bar examinations—would develop questions, maintain security, administer tests through independent contractors, and report results directly to parents.

This agency should be prohibited by statute from selling test preparation materials, from reusing questions across consecutive years without replacement, and from allowing educators access to examination content before administration.

IX. Conclusion: Thirty-Seven Years Is Enough

In April 1990, I confronted Governor Bill Clinton at an Education Writers Association meeting in Chicago, charging that Arkansas test scores were fraudulent. Clinton called me back within the week, spent thirty minutes asking what could be done, and Arkansas announced testing security improvements. On March 25, 1990, 60 Minutes aired “Teacher is a Cheater.” Congress held hearings. The U.S. Department of Education commissioned a replication study that confirmed every finding.

Nothing changed.

Mississippi still reports 75% reading proficiency while NAEP finds 16%. New York City still reports 49% ELA proficiency while NAEP finds 28%. Cleveland reports gains while NAEP shows collapse. The mechanism is the same mechanism I documented in Flat Top, West Virginia, in 1985. The children paying the price are different children. The cost is the same.

The reforms proposed here are not ideological. They do not favor any pedagogical approach, any curriculum, or any political party. They require only that the instrument used to measure whether American children are learning should actually measure whether American children are learning.

Thirty-seven years is enough.

About the Author

John Jacob Cannell, MD, is a board-certified psychiatrist and forensic psychiatrist who practiced medicine for 42 years in eight states. In 1987, he founded Friends for Education, which documented the Lake Wobegon Effect across all 50 states. His findings were confirmed by the U.S. Department of Education, covered by the Washington Post, New York Times, Wall Street Journal, 60 Minutes, and NBC News, and contributed to the reconfiguration of NAEP to produce state-by-state comparative data. He is the founder of the Honest Education Alliance. Contact: alex@honesteducationalliance.com.

Also see: About the Author page · Wikipedia

Key Sources

• Cannell, JJ. Nationally Normed Elementary Achievement Testing in America’s Public Schools: How All 50 States Are Above the National Average. Friends for Education, 1987. (ERIC # EJ374428) — on this site · PDF
• Cannell, JJ. How Public Educators Cheat on Standardized Achievement Tests: The “Lake Wobegon” Report. Friends for Education, 1988. (ERIC # ED314454)
• Cannell, JJ. “Lake Woebegone, Twenty Years Later.” Independent Education Review / Nonpartisan Education Review, Vol. 2, No. 1, 2006.
• Feinberg, L. “Above Average Has Become Testing Norm.” Washington Post, February 8, 1988.
• Fiske, E. “Standardized Test Scores: Voodoo Statistics.” New York Times, February 17, 1988.
• Putka, G. “Classroom Scandal.” Wall Street Journal, November 2, 1988.
• 60 Minutes, CBS. “Teacher is a Cheater.” March 25, 1990.
• NAEP Data Explorer. National Center for Education Statistics. nationsreportcard.gov / nces.ed.gov/nationsreportcard
• Texas Tribune coverage of certification fraud, October 29, 2024.
• Columbus Dispatch coverage of Westerville record falsification, August 23, 2023.

For clickable links and additional reading, visit the Resources page.

Download PDF Join the mailing list