Mike Males, Principal Investigator, YouthFacts.org| August 2026
Not ANOTHER media- and politician-pandering “expert” wildly distorting data to berate teenagers and screens… and, of course, winning uncritical accolades.
“A leading neuroscientist explains why Generation Z is the first generation to be less intelligent than their parents,” the MSN headline (one among dozens over days and days) blares of the testimony of Harvard University’s Jared Cooney Horvath to a U.S. Senate committee.
“They’re the first generation in modern history to score lower on standardized academic tests than the one before it,” Horvath declared. “And to make matters worse, most of these young people are overconfident about how smart they are. The smarter people think they are, the dumber they actually are.”
Horvath channels Socrates, who 2,400 years ago accused books, writing, and art of creating shallow young people sporting a “spurious appearance of intelligence” because they could lazily read and view information instead of memorizing it. “They will seem to be men of wide knowledge, when they will usually be ignorant.”
Now, Horvath recites the same criticism of Gen Z: “‘Most of them have too much confidence in their own intelligence’ because they have constant access to facts online. This creates an illusion of knowledge where people think they are smart because they can find information quickly.”
Horvath favors cliches, and (no surprise) just released an alarmist book needing market hype. His book summary: “Schools, once alive with deep learning born of human connection, are now dominated by screens.”
Yesterday’s schools were “alive,” “deep,” “born of human connection”? Did he go to Summerhill? Entire libraries could be filled with books and studies detailing how stifling schools of the past were, with titles like Death at an Early Age, A Nation at Risk, Growing Up Absurd, The Vanishing Adolescent, The Closing of the American Mind, on and on.
Gen Z is not less intelligent than their parents
Cursory examination shows Horvath’s testimony presented to senators is junk, even after he selectively cherry-picks and omits data to show only what he wants. Detailed analysis respectful of Horvath’s credentials as a “leading” Ivy League expert suggests something worse.
If one is going to compare generations over time periods, it’s crucial to pick criterion-referenced tests (those that subject all generations to scoring by the same standards over time) rather than norm-referenced tests (those that change over time and are scored by comparing each generation to itself).
Nearly all major tests of intelligence and student knowledge such as IQ, Stanford-Binet, Wechsler, SAT, ACT, etc., are norm-referenced, change radically from generation to generation, lag behind modern multiple and adaptive intelligence concepts, and thus are worse than useless for generational comparisons.
Even AI like Chatbot can summarize the overwhelming research consensus on the “standardized tests” Horvath relies on: “IQ tests are not ideal for directly comparing intelligence across time because they are influenced by cultural, educational, and technological changes, as well as shifts in how intelligence is defined.” Likewise, standardized tests like the SAT “are not a reliable or direct measure for comparing intelligence across time due to changes in test content, scoring, and the influence of preparation and background factors.”
Does Horvath not know this, or is he deliberately misleading?
My analysis focuses on the two best of the few criterion-referenced tests that are consistent over time: the National Assessment of Educational Performance (NAEP), and the Programme for International Student Assessment (PISA). Horvath barely mentions the former (wonder why?) and grossly distorts the latter with the ideologue’s usual truncated graphs and omissions.
PISA’s 2024 report summarizes the latest international student test scores. Horvath mashes together the wildly divergent test-scores of 80-plus nations, many supplying just one year of data, and then blames negatives on social media and school tech.
Examination of the report reveals that national scores go in all different directions. They plummeted in Slovakia, Iceland, and France; rose in Columbia, Macao, Qatar, and the United States; stayed flat in Mexico, Ireland, United Kingdom, and Japan; and went up-down in Germany, Latvia, Montenegro, Panama, and Portugal; etc. Did Gen Z get dumber in some places and smarter in others?
Since student social media use advanced rapidly in all countries during the 2000s, how can screens be causing radically divergent student test-score trends?
American Gen Z students of all races perform better than their parents
Horvath is cajoling U.S. legislators and educators to implement drastic curbs on teenagers’ use of laptops, smartphones, and social media. Yet, Horvath’s testimony cites NO data from the U.S. other than a few trivial effect sizes for educational interventions that range from nothing to small.
His testimony fails to evaluate the biggest potential influences on U.S. student test scores – changing student demographics, controversial Common Core curriculum standards, the eruption of parent-age drug/alcohol abuse, and COVID-19’s severe disruptions, to cite just four.
Figures 1 to 5 will suggest why ideologues never cite comprehensive facts. The figures below present all the data for all years criterion-referenced tests show. Like Horvath, I truncate the y-axis, but only because using a full scale would obliterate the small trends he’s striving to highlight.
Where data are available, Gen Z performance generally rose during the social media era (2009 forward) compared to the pre-social-media era (1990s). The exception was 2022, amid COVID-19’s severe school quarantines.
Let us look first at U.S. data from PISA, for which Horvath cites no U.S. numbers.
Figure 1. Programme for International Student Assessment data show that from the pre-social-media era, 2000, to the present, 2022, UNITED STATES’ 15-year-olds’ (all races) reading and science scores rose, while math scores fell.
Source: PISA, 2024.
Scores bounce around (highly visible when the y-axis is truncated), and the PISA site shows some years are unavailable. Reading and science scores generally are up from the first years surveyed, 2000/03, in which few teens had access to social media, to 2018, when over 90% did. (The best surveys (generally Pew Research’s) show daily social media use among U.S. high schoolers rose from 0% in 1990 and 4% in 1995 to 93% by 2010 and 96% today. Similarly, student cellphone/smartphone use rocketed from 0% in 2005 to 75% by 2010 and 95% by 2024.)
The exception, math scores, would have remained stable (483 in 2003; 474 in 2006; 478 in 2018) if 2022’s COVID-era school-shutdown score (465) was eliminated. One could just as easily cite the declines in test scores from 2000 to 2006 as proof that Millennials are the dumb ones.
Gen Z’s achievements by race/ethnicity and discipline
Now, let’s examine the largest criterion-referenced data set available, NAEP, which Horvath mentions briefly but does not present U.S. trends, the only ones relevant to his testimony.
Figures 2-5 show why. They present all the scores for all years available and subjects going back to 1971 tabulated in the most recent Digest of Education Statistics, focusing on the same age group (8th graders) and controlling for demographic changes.
Figure 2. U.S. 8th graders, all races, National Assessment of Education Performance reading and math scores, 1971-2022.
Horvath has not been honest with senators, media reporters, or us. He cites the drop in reading and math scores from 2013 to 2019-2022 as proof that Gen Z is less intelligent than their Gen X parents were around 1990 (assuming a 30-year time lag between generations).
Even on its face, that claim is false. Gen Z’s scores in 2019, the year before COVID-19 closed many schools, were higher for reading (263 vs. 260 in 1990) and substantially for math (282 versus 270 in 1990). Even after COVID-19 quarantines shifted most schools to less effective remote learning, Gen Z scores in 2022 remained equivalent for reading and higher for math than their parents’ scores in 1990 or 1992.
Horvath blames social media and Ed Tech for the imaginary test score decline he invents, which is also illogical: 2013 students were already heavily exposed to both. He fails to note that the pre-social-media schools he praised as “alive,” “deep,” and “born of human connection” had substantially lower reading and math test scores even though test-takers back then were dominated by higher-scoring demographics.
That’s a big problem. Horvath fails to account for a big factor in average student scores: changing demographics. Socioeconomic disadvantage is far more important than anything attributable to screen time. In 1990, just 27% of 8th graders were Black or Hispanic, the lowest-scoring demographics, compared to 40% in 2022. Figures 3-5 show that separating 8th graders’ scores by demographic groups makes Horvath’s deceptions even clearer.
Figure 3. READING
Figure 4. MATH
Figure 5. HISTORY
Source for figures 2-5: NAEP, Digest of Education Statistics, 2025, 2000.
By every measure and for every race, Gen Z (through 2018-19, with 90% social media use and heavy Ed Tech investment, but before COVID-19 disruptions) is strongly outperforming its parents (near-zero social media use and very little school technology in the early 1990s).
Comparing 2019 to 1990, reading scores improved 10 points for White students, 14 points for Hispanics, and 2 points for Blacks. Math scores rose 18 points for White students, 13 points for Hispanics, and 11 points for Blacks.
History scores are available only after 1994, but with expanded races. Through 2019, they rose 6 points for Whites, 10 points for Hispanics, 18 points for Blacks, and 13 points for Asians. That’s also important, given Horvath’s claim that internet-savvy Gen Z doesn’t really know much.
Scores did fall in 2022 during the COVID-19 school closures and educational disruptions. Even so, they generally remained substantially higher than their parents’ scores of the early 1990s.
So, by multiple measures, it’s a flat-out falsehood that Gen Z is “less intelligent” than its parents.
Horvath: “more screen time, lower performance”
Horvath rushes to blame the Gen Z test score crisis he concocts on social media and screens, led by EdTech.
Again, on its face, the blame-screens premise is nuts. Back in the pre-social-media days Horvath glorifies as educationally superior, Boomer and Gen X teenagers spent 25-30 hours a week watching TVs (the worst, passive kind of screen), plus many hours on landline phones and listening to radio. Monitoring the Future shows Gen X teens were just as (un)happy and (un)lonely as Gen Z teens.
Horvath’s position is purist:
“Higher daily screen exposure consistently corresponds to lower scores in reading, mathematics, and science. The relationship is monotonic: more screen time, lower performance.”
Not exactly. Pro-Futuro’s international team’s more detailed analysis of PISA data, separating learning time from leisure time on computers both at school and outside, reaches a more complex, useful conclusion:
“Moderate and regulated use of digital devices in schools is associated with better learning outcomes.
“The findings of PISA 2022 reveal that 15-year-old students who use digital devices moderately for learning in school tend to perform better and report a greater sense of belonging in school. Students who spent one to five hours daily learning on digital devices achieved better learning outcomes than those who did not use them.
“PISA data revealed another interesting trend: students who use their devices recreationally for up to an hour a day achieve better results than those who do not. Beyond one hour of recreational use, educational outcomes begin to decline. On average, in OECD countries, students who spent up to an hour a day on leisure activities on digital devices in school scored 49 points higher in mathematics than their peers who spent five to seven hours daily on their screens, even after adjusting for the socioeconomic profiles of students and schools.”
No one is arguing for the “indiscriminate digital expansion” Horvath criticizes. The best information indicates 1-5 hours of computer use in schools is the ideal, but even here, nuances intrude. Moderation derived from ongoing research and recognition of individuality – not dogmatic purism – is the key for implementing every type of learning strategy, including digital.
Nostalgic atavism always seems to trump complex reality
Every elder generation has indulged the same we’re-so-smart-they’re-so-dumb self-backpatting Horvath and yet another critically-thoughtless, amnesiac media and political class worships yet again.
In 2009, Emory University’s Mark Bauerlein accused Millennial teens of being The Dumbest Generation, “stupefied” by the “digital age.” In fact, Bauerlein’s incoherent claim turned out to be that Millennials weren’t really dumber; they were just less smart than he thought they should be.
In 1983, a US Department of Education commissioned report, A Nation at Risk, accused Gen X and Boomer youth of being the worst ever, all but ineducable, the student detritus of schools that threatened the country worse than enemy invasion. We already mentioned 1960s education critics, and Socrates, who would have fit right in.
Instead of Horvath’s recitation of atavistic fears that each new technology and media innovation is a horrific threat, evolving concepts of multiple intelligences, practical intelligence, and adaptive intelligence, which helps “to make the world a better place by achieving a broad, common good,” hold promise for assessing our species’ ability to meet emerging challenges.
Certainly, Gen Z’s ability to “find information quickly” (a skill that horrifies Horvath) in an age of billions of terabytes would seem a crucial skill in meeting these emerging challenges. Critical, evidence-based thinking – the ability and imagination to see patterns in chaos and chaos in patterns – is another. Some evolving measures may prove as faulty as those of the past, but they are pointing to vital new ways of thinking.