The co-existence of so many generations makes it even more important that we recognize what the real differences between generations are – not so we can point fingers, but so we can understand each other better.

This was what I talked about with Oprah on her new online show, based on my book, Generations: The Real Differences between Gen Z, Millennials, Gen X, Boomers and Silents—and What They Mean for America’s Future. (The book is my magnum opus, summarizing my research on generational differences and why generations exist, along with some fun stuff like pop culture memories and list of the famous members of each generation).

I’ll cover two primary topics in this post: 1. What generational differences are people most interested in? and, just as interesting, 2. You really met Oprah? What is she like?

The show was taped in an event space at a high rise in Manhattan, with the studio on an upper floor and the dressing rooms and overflow space on a lower floor. There was a studio audience with other featured guests sitting in the front row, including a woman who makes hilarious videos about being middle-aged (she’s got glasses on her face, on the top of her head, and on her shirt front), a young man who posted a TikTok video about being a Gen Z worker (and why they stare at us), and Amanda Schneider, who wrote a book about workplace culture (we did a fun podcast later, too). I sat next to Oprah on the small stage, with large podcast-style microphones in front of us.

The show was fast-paced and covered a lot of ground. Some of what we talked about:

• The Gen Z stare – in depth. My favorite moment was when a young man (Demetrius) explained the Gen Z stare, which happens when a Gen Z worker just looks someone blankly instead of answering a question or making small talk. Some Gen Z’ers say they do this when a customer asks a dumb question or when, especially in a service job, they are burned out on showing positivity. When a customer asks something like “Can I get a cheeseburger with no cheese?” Demetrius explained, they stare because “It takes us a second to kind of maybe buffer it.” That instantly made me think of a phone or a website that’s loading, with the round icon that spins. “Like your brain is doing this,” I said, moving my finger in a circle. His eyes lit up. “Like loading – literally loading,” he said. What he said next was even more interesting: “We don’t see work as something that’s enjoyable, especially with influencers making tons of money just doing whatever.” And it’s true – Gen Z is less likely to say they expect their chosen work to be satisfying than previous generations did at the same age. And with only a tiny minority able to make a living as influencers, everything else seems boring and like too much work in comparison.

• Work attitudes. There are lots of rumors out there about generational differences in the workplace – that Gen Z only wants jobs that are interesting and meaningful, that they don’t want to work as many hours, that they will start crying when faced by negative feedback. A lot of it is hard to study, especially since any survey of workers done a single time can’t separate age from generation. If Gen Z workers say they want jobs that are interesting, maybe that’s because they are younger, not because of a generational difference. And apparently it is: In a survey that’s been done every year since 1976, Gen Z 18-year-olds were less likely to say they wanted a job that was interesting compared to Boomers and Gen X’ers when they were 18. The biggest generational difference is instead in work-life balance: Especially post-pandemic, Gen Z is less likely to say they are willing to work overtime and less likely to say that they expect work to be a central part of their lives (see Figure).

Figure: Percent of U.S. 12^th graders willing to work overtime to do a good job and percent agreeing work will be a central part of their life, 1976-2024. Source: Monitoring the Future. Analyzed by Jean Twenge for the Generation Tech Substack.

• The slow life strategy. With people living longer, the entire developmental trajectory has slowed down: Kids are less independent, teens are less likely to have a driver’s license or a job, young adults marry and have kids later, and middle-aged people look and feel younger than their parents or grandparents at the same age. Thinking about these generational shifts through this lens helps in a few ways. First, it helps us see these trends aren’t due to economic shifts. That means they’re probably not going to change anytime soon. Second, it helps us understand each other better and maybe even criticize less. Young people are not necessarily immature or “failing to launch” – they are just taking longer to grow to adulthood, just as their parents are taking longer to feel old. This theory does the best job of explaining all of these trends, including why teens aren’t as interested in driving as they used to be.

• Dating and relationships. Some of Gen Z’s delay around dating and sex is because of the slow life strategy – but not all of it. There’s also a gap in social skills, and meeting people to date is the ultimate test of social skills because much of the interaction is unspoken rather than said directly. And if you’re not getting together in person with friends or going to parties, you’re not going to meet as many people. The rise of dating apps and concern around sexual harassment has also shifted the cultural norm around approaching potential dates. Elsbeth Wimpy, one of the featured guests, interviewed Gen Z’ers about dating and concluded that for young men, “approaching a woman is just weird now.”

• Partying Boomers. Everyone got used to older people being Silents, who married young in the 1950s and early 1960s and never did drugs. Boomers smoked a lot of pot when they were young, and they are continuing to smoke a lot of pot now that they’re older; there’s a graph in Generations showing the huge increase in marijuana use among people in their 60s as the group transitioned from Silents to Boomers. Boomers are also continuing the era of sexual liberation they ushered in when they were young, partially because many of them ended up divorced. When Oprah asked me what was going on, I said, “The Boomers had free love in the 1960s, and now they have free love in their 60s.” I came up with that on the fly, surprising even myself – I always include a lot of humor in my talks, but it’s mostly prepared ahead of time rather than improv.

So what is Oprah like? When we went back downstairs after the show, she discovered that a group of high school students and their teacher hadn’t been able to get seats in the taping room, so they’d watched the show from the overflow room. Oprah spent about 15 minutes talking to them – all off-camera and not something she was at all obligated to do. I could see the delight on the teens’ faces.

When Oprah and I talked backstage before the show, she commented to one of her staffers, “I wore blue because Jean wore blue.” (Faint). As we entered the taping, she held my hand. She talked a lot about her TV talk show that ran from 1986 to 2011, and it sounded like she missed those days. I thought about what that must be like for her, and for anyone who does something huge and then retreats because it’s so all-encompassing it’s beyond exhausting … because then what do you do next? And how do you fill that space in your life?

Not surprisingly, Oprah is an extremely skilled interviewer and host. She diffused the nervous energy in the space (OMG, I’m in a room with Oprah!) and before long we were all laughing. When I wanted to say something but didn’t want to interrupt the flow, she signaled me with her eyes that she’d get to me. She asks follow-up questions to understand what’s really going on, and isn’t afraid to disagree or challenge.

Working with the show’s producers in the weeks leading up to the taping, I noticed something: They called her Ms. Winfrey, not Oprah. It almost felt like the more formal title was a sign of closeness: The world knows her as Oprah, but to her staff she’s Ms. Winfrey. It captures the delicate balance she’s struck throughout her career, between being the consummate professional and being emotionally close to her audience.

In Generations, I used Oprah’s career as an example of the impact of the Boomer generation, who lived the changes brought by the Silent-led civil rights and women’s movements and declared that nothing was off-limits for discussion. Oprah was at the forefront of this new openness, showing it was OK to talk about what used to be taboo, like cancer, menopause, and sexual abuse. She brought that same approach to talking about generations on the show, with people of different generations sharing their experiences. You can check it out here.

I’ve been on both sides of this question. At first I found a decline in identifying as LGB among U.S. 18- to 24-year-olds with early data from 2025, but that decline disappeared when the full year of data for 2025 became available.

But that’s among adults. Some of the samples in the “free fall” report were from high school students and entering college students. Now that 2025 data from the Youth Risk Survey of U.S. 9^th to 12^th graders from the CDC has been released, we can get a much better idea of how sexual orientation has changed among youth.

The trends are striking. They are also very different depending on age (or, in this case, grade).

For 12^th graders, 17 and 18 years old, identifying as LGB declines only a few percentage points between 2023 and 2025 (from 18% to 16%). There’s more of a decline for 11^th graders (16 and 17 years old) and more still for 10^th graders (15 and 16 years old) – you can see the lines fan out on the far right of the figure (see Figure 1).

But the headline result is for 9^th graders (14 and 15 years old): Identifying as LGB absolutely plummets between 2023 and 2025, from 17% to 11%. That really does look like a free fall.

Figure 1: Identifying as lesbian, gay, or bisexual, U.S. high school students, by grade, 2015-2025. Source: Youth Risk Behavior Surveillance System, CDC. Note: Because the “other” response varied by year, the denominator is those identifying as lesbian, gay, bisexual, or heterosexual/straight, excluding “I identify another way,” “I am not sure,” and “I don’t understand what this means” responses.

The decline appears to be driven almost entirely by girls. Identifying as LGB dropped from 27% in 2023 to 22% in 2025 among girls, and from 9% to 7% among boys (see Figure 2). The previous increase in identifying as LGB was also larger among girls, with a 128% increase (more than doubling) from 2015 to 2021 among girls and a 99% increase (doubling) from 2015 to 2023 among boys.

Figure 2: Identifying as lesbian, gay, or bisexual, U.S. high school students, by sex, 2015-2025. Source: Youth Risk Behavior Surveillance System, CDC.

As with the combined data in Figure 1, the decline is by far the largest among 9^th grade girls, where identifying as LGB dropped from 30% in 2021 to 17% in 2025, nearly cutting the numbers in half (see Figure 3). The changes among (for example) 12^th grade girls are much smaller, from 26% in 2021 to 28% in 2023 and 25% in 2025.

Figure 3: Identifying as lesbian, gay, or bisexual, U.S. high school girls, by grade, 2015-2025. Source: Youth Risk Behavior Surveillance System, CDC.

Much of the change in LGB identification among girls is a drop in identifying as bisexual. Figure 4 shows the trend for girls by grade. Once again, the 9^th graders show the largest change. While 26% of 9^th grade girls identified as bisexual in 2021, only 13% -- half as many -- did in 2025.

Figure 4: Identifying as bisexual, U.S. high school girls, by grade, 2015-2025. Source: Youth Risk Behavior Surveillance System, CDC.

Of course, identifying as LGB doesn’t reflect actual behavior. Saying you are gay and having sex with partners of the same sex are certainly correlated, but don’t have perfect overlap. The YRBS survey also asks about sex partners, though, so we can see if there have been any changes in behavior as well. We’ll look at the percentage of sexually active teens who have had at least one same-sex partner. Keep in mind that most high school students are not sexually active, so this is only a subsample of the larger group.

For boys, there’s been a fairly steady rise in same-sex partners, not too different from the pattern of identifying as LGB among boys in Figure 2. The story is different for girls, where having non-heterosexual sex peaked in 2021 and then fell steeply (see Figure 5).

Figure 5: Percent of sexually active high school students who have had at least one same-sex partner, 2015-2025. Source: Youth Risk Behavior Surveillance System, CDC.

Once again, the declines are largest among 9^th grade girls (though it’s similar among 10^th graders, and 11^th graders also show a big drop between 2023 and 2025). The drop is much less among 12^th grade girls (see Figure 6). Although the changes in behavior differ somewhat from the trends in identifying as LGB, they also suggest that high school girls, especially younger girls, are no longer as likely to have same-sex partners.

Figure 6: Percent of sexually active high school girls who have had at least one same-sex partner, by grade, 2015-2025. Source: Youth Risk Behavior Surveillance System, CDC.

(It’s also worth noting how high these numbers are for younger teen girls compared to older ones. Remember this is among teens who are sexually active, and that’s a lower number among younger teens. For girls, having sex with each other might be a more common method of sexual debut, but among those who wait to have sex more have sex with boys).

These trends are especially striking given the relative stability of identifying as LGB among young adults ages 18 to 24. Together with these results for teens, the pattern appears to be little change among those 17 and older, but notable declines among those 16 and younger, especially those 15 and under. This is a generational or cohort effect, with different trends over the years for different age groups.

In spring 2025, 9^th graders were born in 2009 or 2010. There’s still debate over when Gen Z ends, but a consensus seems to be forming that Gen Z ends with those born in 2009 and Gen Alpha begins with those born in 2010. So we might be seeing the beginning of a generational shift. The sharp turn downward in identifying (and acting) as LGB among Gen Alpha teens could be a harbinger of a coming wave of change, or it could be a blip — it’s hard to say.

So why are both LGB identification and behavior declining so much among 14- and 15-year-old girls? The increase in identifying as LGB from 2019 to 2021 was large across all four grades; if there was a COVID effect, it affected everyone. But post-pandemic trends, especially between 2023 and 2025, are strikingly different by grade/age. What happened during this time that impacted 9^th grade girls the most, especially their LGB identification and behavior?

I have to confess: I have no idea. I’d love to hear yours.

At that point, though, the 2025 data for the nationally representative Behavioral Risk Factor Surveillance Survey (BRFSS), administered by the CDC, was incomplete, covering only January and February. Another survey had only a sample from June 2025, and it had changed its design from earlier data collections.

To get the data for the whole year, we’d have to wait until the next BRFSS datafile was released. That would be more definitive for three reasons: The sample size would be much larger (and thus less prone to random variation), the BRFSS design stayed the same over the years unlike the other survey with the June data, and the year comparisons would be equal (comparing full years to each other instead of full years to just a few months in 2025).

That moment arrived last week when the BRFSS released its 2025 data.

The results were not what I expected.

While the first two months of data from 2025 had shown a decline in identifying as LGB among young adults 18-24, the full year of data did not. It instead showed a slight uptick, with LGB identification in 2025 similar to to 2022 (see Figure 1).

Now that we have all of the data from 2025, it appears non-heterosexual identification is not declining (at least not anymore). However, it has stabilized and is no longer racing upward as it did between 2014 and 2022, when identifying as LGB nearly tripled among U.S. 18- to 24-year-olds.

Figure 1: Identifying as lesbian, gay, or bisexual (LGB), by age group, U.S. adults, 2014-2025. Source: Behavioral Risk Factor Surveillance Survey. NOTE: 2020 data are excluded as data collection was paused for the pandemic. Survey year is interview year in the dataset, not the year of the datafile. Includes the full year of 2025 data. Analyzed by Jean M. Twenge for the Generation Tech Substack.

Among 25- to 34-year-olds, identifying as LGB also stabilized after more than doubling from 2014 to 2023. Rates continue to slowly rise among those over age 35; identifying as LGB was at all-time highs in 2025 among both 35- to 44-year-olds (7.02%) and those 45 and older (3.07%).

The uptick in identifying as LGB among 18- to 24-year-olds is completely driven by women (see Figure 2). There was a substantial increase in young women identifying as lesbian, which reached an all-time high in 2025 at 6.31% (the next highest was 4.94% in 2022). So more U.S. young women than ever are identifying as lesbian.

Figure 2: Identifying as lesbian, gay, or bisexual, by gender, U.S. adults ages 18 to 24, 2014-2025. Source: Behavioral Risk Factor Surveillance Survey. NOTE: 2020 data are excluded as data collection was paused for the pandemic. Survey year is interview year in the dataset, not the year of the datafile. Includes the full year of 2025 data. Analyzed by Jean M. Twenge for the Generation Tech Substack.

Young women identifying as bisexual also reached an all-time high in 2025, though not by as much (23.17%, compared to 23.08% in 2022).

What about 2026? Given how the early 2025 data looked different from the full year, I decided not to analyze the early 2026 data. The sample size for the first few months of 2026 is also too low, only 1,821 total and 117 who were 18 to 24 years old. The sample size for the whole year for 2025 on the sexual orientation item is also not what it should be: 78,343. That sounds like a lot until you compare it to the sample size for 2024 on this item, which was 250,040. Interestingly, the sample size drop was not as large for the other items on the survey, suggesting the 2025 survey purposely asked fewer people the sexual orientation question. Also, the question about identifying as transgender that was included in the 2014 to 2023 BRFSS surveys was absent from the 2025 datafile, as it was in 2024. So we don’t know whether identifying as transgender increased, decreased, or stayed the same after 2023 — the data are not there.

Here’s what we do know: Identifying as non-heterosexual did not decline in 2025 as we’d previously seen with data early in the year. Instead, the number of American young adults identifying as LGB stabilized after increasing precipitously between 2014 and 2022. However, identifying as bisexual and especially as lesbian is at all-time highs among young women in 2025. We’ll have to wait until next year to know for sure if this trend will continue into 2026.

The results are, in a word, terrible.

The PISA tests of math, reading, and science are normally administered every three years, with the 2021 assessment was pushed to 2022. Between 2018 and 2022, no one was particularly surprised that test scores went down. Students were barely out of pandemic lockdowns; in many countries online school went on for more than a year, with learning losses as a result.

The 2025 data gives us our first true look at academic performance in the post-pandemic era when students had been back at in-person school for several years. Given that, one might expect that test scores would rebound, perhaps to 2018 levels.

Not only did that not happen, but scores were even lower in 2025 than in 2022, especially in the core subjects of math and reading (see Figure 1 for the trend worldwide; this is based on my analysis of the publicly available raw data). And, as I’ve reported previously, the decline in test scores began well before the COVID-19 pandemic, starting around 2012 (that familiar year when teen mental health also began to suffer).

Figure 1: Academic performance in science, math, and reading, all countries, 2006-2025. Source: PISA. Note: Data are unweighted and thus scores may differ from graphs using data weighted by the sample weight or by population. Includes all countries with data, not just OCED countries as in other reports.

On the PISA performance score, 25 points equals about one year of schooling. Between 2012 and 2025, reading scores declined 47 points. That means 2025 students are nearly 2 grade levels below 2012 students in reading. Thirty points of that decline happened since 2018.

Math scores declined 37 points since 2012, so about a grade level and a half. Science scores are down 28 points, about a grade level. Adolescents around the world are much less capable in major academic subjects than they were just 13 years ago.

Given the importance of the topic, it pays to take a deeper and more precise look at this data. Different countries have participated in PISA over the years, so it’s a good idea to do an analysis including only countries that have data for all or most years. Since the declines start in 2012, it’s also important to have data for that year. If different years also have different samples of countries included, that creates the possibility that the trends are due to a shifting sample. Restricting the analysis to countries with data in the relevant years takes care of that.

Also, regions of the world differ in economic, cultural, educational, and technological trends, so it’s also informative to examine trends within regions. Using these requirements and grouping countries into five regions (Western Europe, Eastern/Central Europe, English-speaking, Latin America, and Asia) results in a sample of 36 countries. I haven’t seen anyone else do this type of analysis yet with the PISA data up to 2025, so I crunched the numbers in the data files.

The end result: There are still large declines in test scores, especially in Western Europe, English-speaking countries, and Asia.

Both math and reading scores are down more than a grade level among students in Western Europe. Science scores are down about half a grade level (see Figure 2).

Figure 2: Academic performance in science, math, and reading, Western Europe, 2006-2025. Source: PISA. Note: Data are unweighted and thus scores may differ from graphs using data weighted by the sample weight or by population.

In English-speaking countries (Australia, Canada, Ireland, New Zealand, the UK, and the U.S.), math scores were declining even before 2012 and are now a full grade level lower than they were in 2006. Reading scores held their own for awhile but then fell off a cliff, declining nearly a grade level between 2018 and 2025. Science scores are down about half a grade level since 2006 (see Figure 3).

Figure 3: Academic performance in science, math, and reading, English-speaking countries, 2006-2025. Source: PISA. Note: Data are unweighted and thus scores may differ from graphs using data weighted by the sample weight or by population.

In Asia, test scores peaked in 2012. Reading scores declined 53 points, two entire grade levels, between 2012 and 2025. Math scores declined 35 points, more than a grade level, and science scores declined about a grade level (see Figure 4).

Figure 4: Academic performance in science, math, and reading, Asia, 2006-2025. Source: PISA. Note: Data are unweighted and thus scores may differ from graphs using data weighted by the sample weight or by population.

So why did test scores continue to decline between 2022 and 2025, including in some of the most prosperous nations in the world?

Here are some explanations worth considering:

1. Pandemic learning loss. If the declines were due to the pandemic, one would have expected that academic performance would be better in 2025 than it was in 2022 – students had three or more years of in-person school. But perhaps pandemic learning loss from earlier grades compounded and students weren’t able to catch up. The 15- and 16-year-old students taking these tests in 2025 were 10 and 11 years old when the pandemic hit in 2020. In the U.S. school system, that’s 5^th grade. Most would have spent part of their 6^th grade year in online school as well. If they missed out on fundamental reading and math skills during this time, their post-pandemic years in 7^th, 8^th, 9^th, and 10^th grade might not have been enough.

2. Use of AI. ChatGPT was introduced on November 30, 2022, and within a few months students were using it and other generative AI systems to write their essays and do their math problem sets. Several studies have shown that students who use AI whip through their homework but then bomb tests because they don’t actually learn the material. That would show up on standardized tests like these where students can’t rely on AI.

3. Laptops and phones. Students spend between an hour and two hours using personal devices during school, with nearly all of that time spent using social media and watching videos. (Yet another reason to put off giving your kid a smartphone until they are older; give them a basic phone instead — see Rule 4 in 10 Rules). School-issued devices are an issue as well; many school laptops do not restrict streaming services and games, allowing students to entertain themselves during class instead of paying attention. This cohort of students entered middle school during and after the pandemic when providing laptops or tablets for each student had gone from common to nearly universal. School-issued devices are also distracting at home, where students assigned digital homework are tempted to instead watch YouTube or stream movies or TV on Disney+. (Which I found out when my middle schooler did just that. She’s not a hacker, nor did she find a clever workaround; Disney+ was not blocked at all on her school-issued laptop).

4. Attention spans and trouble focusing. Another possibility is the impact of social media and short-form video on teens’ brains. These activities may have shortened their attention spans and made it harder for them to concentrate. Fewer teens also read for pleasure. Thus, it becomes more and more difficult for students to focus on schoolwork, especially material that requires deep concentration.

I think all of these are possible reasons for the decline in academic performance among teens worldwide. I’d love to hear what you think – are there other possible causes? Which of these four seems most plausible to you? And what can we do to reverse these very concerning declines in academic performance?

We already know that anxiety, low life satisfaction, emotional distress, poor mental health, and loneliness have increased among adolescents and/or young adults in many countries around the world. But critics have often argued that suicide rates are the exception to this trend – that adolescent suicide rates may have increased in the U.S., but not internationally. Studies using the Global Burden of Disease database have muddied the waters as that database contains only estimates of suicide rates, not actual rates.

In a previous post (also cross-posted on After Babel), I gathered recent studies on trends in the adolescent suicide rate in Europe, South Korea, and Japan, showing that the suicide rate was increasing for adolescent girls in most countries.

I was also able to find data on the adolescent suicide rate over time in Australia, Canada, and the UK (specifically, England and Wales). I’ve combined rates in 5-year blocks to make the overall trends more clear; these are lower-population countries with more variability in the suicide rate from year to year. I was not able to include New Zealand as they only provide suicide rate statistics broken down by sex for broad age groups (for example, 15 to 24), so we can’t see the data for adolescents alone. I’ll cover the U.S. in a future post.

A beach at Rottnest Island, off the coast of Western Australia

Let’s start with Australia. The suicide rate of 15- to 19-year-old girls rose 69% from 2005-09 to 2015-19 (before the COVID-19 pandemic) and then declined slightly, still 53% higher than in the late 2000s (see the red line in Figure 1). Suicide rates for boys (the blue line) rose 44% from 2005-09 to 2015-19 but then declined during and after the pandemic, leaving them 17% higher than in the late 2000s.

Figure 1: Suicide rates of 15- to 19-year-olds, Australia, by sex, 2005-2024

So while the suicide rate is retreating among adolescent boys in Australia, rates for adolescent girls are still much higher than they were 15 years ago.

A red sled in the window of an apartment in Toronto, Ontario, Canada

A similar but even more gender-discordant pattern appears for the suicide rates of 10- to 19-year-olds in Canada (the data are from the supplemental tables of this article.) Among girls, suicide rates rose 40% between 2005-09 and 2015-19 and were still 24% higher in 2020-23. For boys, though, suicide rates rose between 2005-09 and 2015-19 before falling substantially into the early 2020s.

Figure 2: Suicide rates of 10- to 19-year-olds, Canada, by sex, 2005-2023

Highclere Castle, Hampshire, England

In England and Wales, the suicide rate of 15- to 19-year-old girls increased 66% between 2005-09 and 2015-19, remaining 60% higher in 2020-24 (see Figure 3). For boys, the suicide rate rose 42% from 2005-09 to 2020-24.

Figure 3: Suicide rates of 15- to 19-year-olds, England and Wales, by sex, 2005-2024

So across three English-speaking countries, the trend in suicide rates for adolescent girls is virtually identical: It rises from the late 2000s to the late 2010s and then declines slightly into the early 2020s. Something is compromising the mental health of adolescent girls, and the pandemic and post-pandemic years produced only a little relief. There is less consistency across countries in the suicide rates for adolescent boys, with increases but substantial backing off in Australia, decreases in Canada, and increases in England and Wales.

This is similar to the data from Europe and Asia, where increases in the adolescent suicide rate were more evident for girls than for boys. In Japan, the suicide rate for adolescent girls is now equal to that for boys, a reversal of a decades-long pattern. Although the suicide rate for boys in Australia, Canada, and the UK is still much larger than it is for girls, boys’ suicide rates have declined in Australia and Canada in the last 5 years. Whether it’s due to improvements in suicide prevention programs or other cultural changes, there’s been good news for the suicide rates for boys. The same is not as true for girls, suggesting we have further to go to help them.

Time for a reset. Any break from school is the perfect opportunity to find out what your kid is like without devices. I’ve always done a version of this over the summer, giving my kids an hour a day of their laptops and then physically taking them back. But now they’re teens with phones, friends, and strong opinions, making a tech detox harder.

Then my 14-year-old daughter asked for an expensive hair dryer straightener combo (given the price, it should make coffee and maybe even doughnuts too). The first time she asked, I said no. Then I realized it was an opportunity – I told her I would buy it for her if she could spend a week almost entirely without her electronic devices. She agreed, even though it was a summer week when not only did we not have plans, but her sister would be out of the country at a high school summer program. Under normal circumstances, it would be a week when she’d be very tempted to be on her phone all day even though it’s a kids phone with no social media (see Rule #4 in 10 Rules for Raising Kids in a High-Tech World: First phones should be basic phones).

I wasn’t sure how it was going to go, but the week was a tremendous success. She found new interests, learned how to spend time without her devices, read an entire book for the first time in months, and talked to me and her dad much more than usual. I think this experiment worked because we prepared ahead of time and because I got her buy-in. Here are my tips for a successful tech detox for tweens and teens:

1. Decide on your incentive. Maybe your kid really wants something you wouldn’t normally buy them. Maybe they want to go to an amusement park. Or maybe you want to cut back their screen time over the summer and you’re just going to lay down the law. Be upfront about whether you’re going to offer an incentive or not. Given my experience with teens, it will probably be a lot easier if you do. My daughter and I didn’t argue at all over her screen time during the detox week, probably because I told her she wouldn’t get the hair dryer/straightener if she fought the rules we’d settled on.

2. Make a list of non-screen activities in advance. For most kids and especially teens, this should be a list of things they would like to do, not necessarily things you would like them to do. That’s OK as long as it doesn’t involve a screen. As one place to start, Let Grow has a list of activities for teens you can get by signing up at their website. My daughter chose diamond painting, cooking, reading, cleaning out her closet, and bedazzling (where you put sparkly stuff on things like bottles and folders). We bought the diamond painting and bedazzling supplies well in advance of her week.

3. Make a plan for out-of-the-house activities. Most of us, teens and adults, never leave the house without our phones. So it’s good to think of workarounds ahead of time. My daughter wanted to spend one of her days hanging out at a friend’s house. I thought this was a great non-screen activity but reminded her she couldn’t bring her phone. So my husband and I communicated with her friend’s mom for the day – really, just for settling on a time to pick her up. My daughter not having her phone was not an issue at all.

4. Set up ground rules for limited screen time. I gave my daughter a half hour with her phone in the morning to text her friends and Google Meet her sister. I set a timer and then collected the phone when the 30 minutes was up. We also agreed that she could watch TV with me in the evenings for 30 to 45 minutes (this is our usual routine anyway). If you are working out of the house and your kid is at home during the day, you can do the 30 minutes of phone time when you’re around in the evening. (And if you’re concerned they won’t be able to reach you and you don’t have a landline phone, consider one of the newer options for these).

After the week was up, my daughter got her prized hair dryer straightener coffeemaker whatever. She was also happy to get her phone and laptop back. I thought she did very well without them and told her so. Her perspective was a little different – she said she felt “yearning” for her devices all week. When I asked her why, she said she missed looking things up, especially on shopping apps. I had to admit that was relatable.

But I liked the week so much that I’m hoping she’ll have a burning desire for another odd but expensive product in the future. She was more present in the moment that week. I smile when she shows me the things she bedazzled. I like that she learned she can go without her phone if she has to. And, as I’m writing this, she’s reading a book even though she has her phone back.

Now that’s a win.

There are many reasons to support a social media ban that have nothing to do with mental health. Many parents don’t want their kids using TikTok at 2am, buying drugs or exchanging explicit photos on Snapchat, or comparing themselves to the perfect bodies on Instagram. There are also many reasons to care about the teen mental health crisis apart from the role of social media.

But understanding the scope of the mental health crisis is useful for knowing where teens are suffering more than in generations past and for determining the cause of the crisis. If school shootings were the primary driver of the teen mental health crisis, for example, we would expect depression, anxiety, and suicide to rise in the U.S. but not in countries without the same level of school violence. The same goes for other U.S.-specific explanations, such as the opioid epidemic.

We already have solid data from many sources that anxiety, low life satisfaction, emotional distress, poor mental health, and loneliness have increased among adolescents and/or young adults in many countries around the world.

But what about suicide rates? Critics have often argued that suicide data, in particular, do not fit the narrative of an international crisis in youth mental health. In his review of The Anxious Generation, for example, Tobias Dienlin writes, “the mental health crisis Haidt describes appears to be more specific to certain regions, particularly the United States, rather than a global phenomenon. Data from large-scale international sources show no consistent decline in youth well-being (Marquez et al., 2024) or suicides (Michalek et al., 2024). This discrepancy suggests that the mental health issues Haidt highlights are influenced by regional factors rather than being universally applicable.”

As we’ve already covered, youth well-being is indeed down (including in most of the datasets discussed in Marquez et al., 2024, the report Dienlin cites). And the source he cites for suicides uses the Global Burden of Disease database, which relies on estimates rather than actual data.

What happens if researchers look at actual suicide data, not modeled estimates?

Several articles published in the last year have taken a more detailed look at suicide rates in many countries across the globe. These analyses have taken two steps that many previous analyses did not: 1) they gathered actual suicide mortality data instead of using the Global Burden of Disease or WHO Global Estimates and 2) they report suicide rates broken down by gender and age group. This is crucial, since the rise of social media impacted teen girls and young women the most; looking at overall suicide rates isn’t good enough.

A caveat: More than other mental health indicators, suicide rates can be influenced by many factors outside of teens’ everyday routines and relationships, including access to firearms, the availability of mental health treatment, and suicide prevention programs and helplines. There are also cultural differences and events unique to certain countries or regions. Thus, we wouldn’t expect suicide rates to show exactly the same trends in all countries. If we can discern a pattern, though, it’s worth paying attention to, as suicide is the most tragic outcome of poor mental health. From a methods standpoint, suicide rates are the mental health indicator least likely to be influenced by self-report issues.

To try to find as many relevant articles as possible, I searched all EBSCO databases (including Medline and PsycInfo) for articles published in the last year with “suicide” in the title and the words “youth” or “adolescents” and the word “trend” anywhere in the article record. I focus here on articles presenting suicide data on adolescents and young adults from Europe and Asia; in a future post I’ll cover four English-speaking countries (England and Wales, Australia, Canada, and the U.S.). I’m most interested in data from 2010 to the present, given the rise of social media during that time period.

Let’s start with Spain. Below is the authors’ graph of suicide rates by gender (males on the left, females on the right) and age group.

Not much happens for the male suicide rate among adolescents (10-19, red line) or young adults (20-29, orange line), though there is an uptick among young men since about 2018. But check out the graph for females, on the right: Suicide rates nearly double among 10- to -19-year-old girls and increase even more among young women in their 20s. The increase for young women is so large it comes close to equaling the suicide rate for women in their 30s.

Another article looked at suicide rate trends across Europe. (This article uses the WHO Mortality Database, which is different from the WHO Global Health Estimates; the authors specifically say they relied only on official statistics). For male adolescents or young adults, they find increases from 2010-2014 to 2020-2022 in Italy, Spain, and the UK in their country-specific analysis and in North Europe in their regional analysis. That’s still noteworthy, given that these are places without the U.S.-specific issues, such as school shootings, often mentioned as a cause of the adolescent mental health crisis. But in many European countries, the suicide rate for teen boys and young men has not increased.

For girls and young women, however, the authors find increases from 2010-2022 (the green bars) much more broadly. Below is their graph for trends in the suicide rate for females by region:

The green bars head to the right (indicating an increase 2010-14 to 2020-22) among all three adolescent and young adult age groups (10-14, 15-19, and 20-24) in North Europe (which includes Denmark, Finland, Iceland, Ireland, Norway, Sweden, and the UK). And the increase is not small – it’s around 50 to 75%. There are also increases in the female suicide rate in all three of these age groups in Western Europe (Austria, Belgium, France, Germany, Luxembourg, the Netherlands, and Switzerland) and the EU as a whole, and in two of the three age groups (10-14 and 15-19) in South Europe (Greece, Italy, Malta, Portugal, and Spain). The only region without increases is Central/East Europe (e.g., Czech Republic, Hungary, Romania, Serbia).

And here is their graph for country-specific analyses for females:

The green bars are on the right (indicating an increase in suicide rates) for all three age groups (10-14, 15-19, and 20-24) in France, Poland, Spain, and the UK, and are on the right for adolescents (10-14 and 15-19) in Germany and Italy. Russia and Ukraine instead show declines. So in 6 out of 8 countries, the suicide rate for teen girls is higher than it was in the early 2010s.

I’ll move next to Asia. I was all set to include an analysis of suicide data from Japan until I saw they got their data from the Global Burden of Disease dataset – and thus used estimates, not actual data. But I did find an article reporting actual suicide data from South Korea. The authors even provided a supplement with the numbers by age group, gender, and year, which meant I could graph it myself. Here is the suicide rate for 10- to 19-year-olds in South Korea since 2010:

Figure: Suicide rate for 10- to 19-year-olds, South Korea, by gender. Source: Oh et al., 2025

For boys, the suicide rate was 30% higher in 2023 than it was in 2010, though the largest increase was between 2020 and 2022, suggesting COVID might play a role. For girls, the suicide rate in 2023 was 71% higher than it was in 2010, with little change during the peak COVID years of 2020 and 2021. Instead, the largest upticks were between 2017 and 2018 and 2022 and 2023.

The trends here are clear: The suicide rate for teen girls has increased since 2010 in most countries and regions in Europe and in South Korea. These are regions that experienced the rise of smartphones and social media just as the U.S. did but did not have regular highly publicized school shootings or lockdown drills. They also didn’t have a widespread opioid epidemic or the same trends in income inequality or politics that occurred in the U.S. Any U.S.-specific (or English-speaking-specific, or even Western-specific, given South Korea) explanation doesn’t work for these data.

As I noted earlier, many factors impact suicide trends. But combining this data with with the raft of evidence on increases in depression and low well-being among teens across many countries strongly suggests that the adolescent mental health crisis is international, particularly among girls. That means the cause is likely to be something that is also international.

The change wasn’t limited to attitudes about others. Fewer Americans identified as transgender or non-binary – even among kids as young as 13. Fewer identified as non-heterosexual, with an especially large decline in bisexual identity.

We still have very little data from 2025, though, given the delays in datasets becoming publicly available. In an earlier post, I found that fewer young adults, especially women, identified as bisexual in 2024 and early 2025 compared to earlier years. The change wasn’t large, though, and for men only appeared in the data from 2025, which was collected at the very beginning of the year as part of the Behavioral Risk Factor Surveillance System (BRFSS) and was included in the 2024 datafile. We don’t have the 2025 datafile for the BRFSS and won’t until August.

But we do have data from April and June 2025 from the Household Trends and Outlook Pulse Survey, a continuation of the Household Pulse Survey collected by the Census Bureau from April 2020 to September 2024. Household Pulse asked about sexual orientation beginning in July 2021. Beginning in late July 2024, they changed the “other” response from “Something else” to “I use a different term.” So we’ll compare July-September 2024 to June 2025, all times whe the sexual orientation question was asked in the same way. (I included only June to get the most recent data; it’s also not clear if the April and June datafiles are the same people since the design transitioned to longitudinal).

What does this data show?

The short answer: Big changes. From summer 2024 to June 2025, the number of 18- to 29-year-olds identifying as non-heterosexual (gay, lesbian, bisexual, or other) dropped from 30% to 18% -- close to half (see Figure 1). Among 30- to 39-year-olds, they dropped from 17% to 12%. There were smaller changes among those in their 40s and those 50 and older.

Figure 1: Percent identifying as non-heterosexual, by age group, summer 2024 vs. June 2025. Source: Household Pulse Survey, U.S. Census Bureau. Data are not weighted.

For young adults (18-29), the declines were largest among bisexuals, going from 24% to 12% among women and from 11% to 5% among men (see Figure 2). The number of young women identifying as lesbians dropped from 5% to 2%. There was little change in identifying as gay among young men.

Figure 2: Percent of 18- to 29-year-olds identifying as bisexual, gay, or lesbian, by sex, summer 2024 vs. spring 2025. Source: Household Pulse Survey, U.S. Census Bureau. Data are not weighted.

These declines are much larger than those in the BRFSS from 2024 to early 2025.

Why were the changes here so big? There are a few possibilities.

First, it’s possible that at least some of the declines are due to sampling (that is, due to who’s in the survey vs. a real change at the population level). The 2025 Household Pulse survey was redesigned to be longitudinal (following people over several surveys). That may have meant a different type of sample even though the rest of the survey procedure was the same as in 2024. However, the pattern of change, with the largest declines among the youngest age group and among bisexual women, is exactly the same as in the BRFSS, which doesn’t have this issue. It’s also interesting that there’s very little change in identifying as gay among young men, but large changes in identifying as bisexual among women. If there were a sampling change one would expect it to impact all groups equally, or at least not this differently.

Second, the cultural and political shift between summer 2024 and spring 2025 may have made LGB people more reluctant to be truthful about their sexual orientation on government surveys. That might be particularly true as the 2025 surveys transitioned to a longitudinal design with people filling out multiple surveys over the course of months. However, it’s not clear why that would impact younger adults and bisexuals more and gay men and those over 50 hardly at all.

Third, this could represent actual change in people’s identities and behavior. The June 2025 data were collected about 5 months later than the early 2025 BRFSS data, well into Trump’s second term and the resulting cultural shift. The large changes in identifying as bisexual among women from 2014 to 2023 suggests a cultural change; the decline may also be a cultural change. There is some preliminary evidence that women are not having sex with women as much as they used to, a behavioral indicator that also points toward a true change rather than an artifactual one.

We’ll know more once the BRRSS data for 2025 is released. For now, it’s looking more and more likely that fewer young adults are identifying as bisexual.

One of the best parenting books I’ve ever read is Michaeleen Doucleff’s Hunt, Gather, Parent. She visited cultures around the world to see how they raise their kids, and came away with a startling but effective idea: Why not involve our kids in the work of the family instead of kids being the work of the parents? It really informed my thinking when I was writing 10 Rules for Raising Kids in a High-Tech World.

Now Michaeleen, who’s a science reporter at NPR, is out with a new book, Dopamine Kids. I devoured it, and I think you will too. Even better, you’ll learn things that will help your parenting. (I know I’ve been guilty of Mistake #2, below!) Please enjoy her guest post, below, and feel free to share it.

Five years ago, I decided to finally fix a long-standing issue: my phone addiction. At age 44, I had a severe problem.

I was so obsessed with my phone that I could no longer enjoy quiet, relaxing moments with my daughter, Rosy. One afternoon, I sat on a beautiful beach near San Francisco, watching Rosy try to build a crumbly sandcastle. She hummed along contentedly as she dug a little moat for the castle. But all I could think about was checking my emails, my texts, and social media accounts. I drew my phone from my pocket, swiped at each app, and then a few minutes later, I’d pull out my phone and launch the same circuit over again.

Inside, I could feel this low, constant hum of anxiety churning in the back of my mind. It felt like a continuous wondering of What’s next? What’s next? What’s next?

I knew my phone obsession affected Rosy and that eventually I would pass my habit onto her. And how could I protect her from the pitfalls and perils of a digital life if I didn’t fix my own problems first?

So I set out on a mission: to figure out why I couldn’t stop thinking about my phone—and use that knowledge to break my addiction.

I read hundreds of studies and interviewed dozens of psychologists and neuroscientists. And eventually, after a few years and a bunch of experimentation, I not only broke my habit; I crushed it. My average phone time now stays well below 15 minutes each day. My sleep improved beyond belief. I’m a happier person—much happier.

In my new book Dopamine Kids, I describe how I did it and how I successfully weaned Rosy off screens, as well. Throughout the process, I discovered something surprising: parenting advice around phones and screens is woefully outdated. It’s based on research that’s 20 to 30 years old. And it doesn’t work well.

As a mom, I was making a few big mistakes with screens and my daughter—mistakes that made regulating her screen time a hard and continual struggle. Once we fixed these errors, limiting screens actually became fun. And in the end, it meant adding more joy and pleasure into our lives.

Mistake No. 1: Screen time = pleasure time

For years, I struggled to limit my phone use because I didn’t want to decrease the amount of “pleasure” in my life. I believe that I spent hours and hours each night, cuddled up with my phone and scrolling in the dark because social media was my “reward” or “treat” at the end of a stressful day. It was my key source of pleasure.

At the same time, I believed that my daughter spent hours watching cartoons on Netflix or YouTube for the same reason: these apps gave her a continual drip of dopamine, and these surges fill her with joy and happiness. (Dopamine is the molecule of happiness and pleasure, right?)

Therefore, I always felt really bad limiting screen time with her—or with myself. I felt like a mean mom, dedicated to stripping away from her only daughter the very activities that she loved more than anything else in the world. Why would a mom do that? And why would a mom do that to herself?

Limiting screens seemed like settling for less pleasure in life. Neuroscience from the past 30 years tells us that this thinking isn’t quite right. In some instances, it’s simply wrong.

Over the past several decades, scientists have accumulated a massive amount of evidence showing that dopamine’s critical role in our brains isn’t giving us pleasure but rather ensuring our survival—to make sure we want, seek out, and obtain what we need to stay alive. And it does that by triggering the feeling of wanting or desire, neuroscientist Kent Berridge explained to me one afternoon at his office at the University of Michigan. It motivates us and gives us a willingness to work.

But it fuels repeated desires. A surge of dopamine tells you to do something now … and then do it again when you can. Press the button again. Comment again. Scroll again. Check again. Dopamine makes you want, over and over again. Dopamine drives obsession.

The “dopamine-is-pleasure hypothesis” is wrong, says Berridge, who has led the way in untangling dopamine’s complex role in our brains. And here’s the critical part: Although most of the time, we want what brings us pleasure, in modern life, we have activities and foods that split up this beautiful synchrony, Berridge says. We can want—even intensely crave–activities that hurt us.

Take, for instance, teenagers and social media. In one study, published in 2022, some teenagers said that social media caused them to feel depressed, and yet they still wanted to use it. They still had to login. “I have repeatedly deleted Instagram in an effort to improve my emotional state but then, I reinstall. Many times,” one teenager told the scientists.

I had a similar experience with social media. Many times I felt worse after I checked my accounts. I would leave social media with a sense of insufficiency, unworthiness or even anger. And yet I would choose using social media over enjoying the beach with my little girl.

This better understanding of my dopamine system empowered me to quit social media, but it also emboldened me as a parent. I began to see my daughter’s obsession with Netflix and YouTube quite differently. I could see all the ways that screen time didn’t bring Rosy pleasure but rather how it triggered an intense motivation in her that, over time, grew frustrating, uncomfortable, and even stressful for her.

So I realized that limiting her screen time—or even getting rid of it—wasn’t depriving Rosy of pleasure, but it was actually a chance to reclaim pleasure in our lives. But before we could do that, I had to stop making mistake No. 2.

Mistake No. 2: “Please, put down the device and go … read a book”

For years, whenever I tried to limit Rosy’s screen time, I replaced scrolling with … well, very little. I’d tell her to go “read a book” or “play outside.” To her, these ideas seemed about as exciting as me saying, “Put down the screen and go be bored” (because she would literally tell me that. “But that’s boring, Mom,” she would whine).

Honestly, a part of me thought that she needed to be bored. Parenting experts and psychologists had often advised me to take this strategy. “Let Kids Be Bored. It’s Good For Them,” a New York Times headline read in 2023.

As I learned more about the science of screens and talked with behavioral psychologists, I realized, Oy! I’m making a huge mistake.

“No, no. It’s not about being bored,” says neuroscientist and psychologist Mateusz Gola at the University of California, San Diego. “It’s about having more excitement [in your life], not less.”

For decades studies have shown that boredom doesn’t typically spark creativity or prompt us to go seek out friendships, as often believed. But instead, it’s associated with an array of problems, including unhealthy eating, attention problems, depressed feelings, frustration, and even political extremism.

In our age of devices and apps, boredom doesn’t work, Gola says. Modern video games, video streaming, and social media apps offer an endless and rapid-fire supply of surprising, edgy, and evocative content. If we suddenly switch to a void of nothingness—or what feels really boring by comparison—we create a horrible discomfort in our brains and bodies. For me, it felt like standing in a cold shower, shivering, as icy droplets of water slid down my back. Time would go by so slowly. And I was stuck there naked and cold. I hated being bored, and I wanted out. What had always been my out? My phone.

So instead of boredom, what does work: Replacing screen time with something that’s (almost) just as exciting and (almost) just as interesting to your child, says behavioral psychologist BJ Fogg at Stanford University. “You have to find something the child already desires. This is the key to lasting change.”

Find activities that light up your child’s dopamine systems and send their motivational juices flowing, Gola says. “So for me, I replace my phone with extreme sports, like paragliding and surfing. These are not boring.”

Clearly, most kids can’t go surfing or paragliding. But kids don’t need that level of danger. For children, including tweens and teens, excitement and motivation comes from simply having more autonomy in life, tackling manageable risks and honing adult skills. For example, Rosy wanted to learn to ride her bike to the market alone so she could buy gum and run errands for the family. She also yearned to use the oven so she could bake on her own. And she had shown interest in learning to make clothes with a sewing machine, because that needle is sharp, and it’s moving fast. It feels dangerous and exciting to Rosy.

So before I limited screen time, I started to help Rosy cultivate these hobbies and help her experience the joy and pleasure of developing these skills. Then when I finally had enough courage to cut out Rosy’s Netflix habit completely, I didn’t need to leave her empty-handed or bored. I had a whole menu of exciting and fun activities that she already knew that she enjoyed. And if she still complained, I had what I think of as the methadone of Netflix: audio books.

Finally, to ensure screens and apps didn’t creep back into our lives, I had to fix one more mistake.

Mistake No. 3: I’ll build up Rosy’s willpower so she can regulate her own screen time

Willpower is the ability to resist a temptation that’s readily available to you. “So the idea of effortful resistance,” says psychologist Marina Milyavskaya at Carleton University, in Ottawa, Canada.

“Fifteen to 20 years ago, it was thought you could train willpower,” she says. So parents could strengthen a child’s willpower, like a muscle and teach them over time to regulate their own screen time. For example, if I let Rosy watch YouTube each day, I can teach her to stop watching after an hour (and to resist her device at every other moment).

“There’s really not a lot of evidence showing that this strategy works, and none that I know of with kids,” Milyavskaya says. Instead, I was simply creating an entrenched habit and boosting her desire for YouTube, says Michael Inzlicht, a professor of psychology at the University of Toronto.

As I explored in a piece for NPR in March, research over the past 15 years suggests that willpower doesn’t work with modern temptations, especially in the long-term. Studies have found that relying on willpower to resist daily temptations only leaves young adults feeling depleted and less likely to achieve their goals. As psychologist Wendy Wood at the University of Southern California points out, “grown-ups don’t have that much self-control, and certainly kids don’t either.”

Instead of relying on kids’ modicum of willpower, Wood says, set up your home and routine so kids don’t have to resist temptations. Set up times and spaces in kids’ lives in which the offline activity is the only choice. “Structure the environment so that certain behavioral options are just kind of the obvious ones. And other ones aren’t available,” Woods advises.

For example, when we finally put an end to Rosy’s nightly screen habit, I didn’t simply place her devices in a drawer and tell her that they weren’t available. Instead, I hid them—-and hid them extremely well. I placed them in the dryer. And I left them there until Rosy created strong, new habits after dinner.

When screen activities became utterly unavailable to her, Rosy started to enjoy and love offline habits much faster than I had anticipated. And these new activities stuck and became habits. Two years later, she zips around town with her friends for hours at a time. She bakes up a storm, including lasagnas for dinner and breads for breakfast. She not only sews, but also crochets and does crafts.

<<>>

By shifting our mindset around screens and learning strategies that work with our neurology instead of against it, I truly believe we can all break our obsessions with technology, no matter our age. And we can do it in a way that doesn’t trigger tantrums and tears, but rather brings more fun, excitement and pleasure into our lives.

The analysis was from Education Scorecard, a Stanford and Harvard research collaboration. The report concludes that the “learning recession” began around 2013. Figure 2 is really stunning:

The declines (orange bars) for reading (lower graph) are especially striking, because the declines were nearly equal 2017-19 (before the pandemic), 2019-2022 (during the pandemic) and 2022-24 (pandemic recovery). Only from 2024 to 2025 was there any post-pandemic recovery, and it paled in comparison to the large declines since 2013. Clearly this is not just COVID — something else is going on.

On X, I posted a figure from the New York Times story covering the report and the declines in reading scores, which show up in nearly every state:

This graph is especially tragic because it shows just how large the decline in reading proficiency has been—nearly an entire grade level in many states. It also shows how pervasive the decline has been. This is not a regional problem; this is a national problem. (Except, go Mississippi!)

The question, of course, is why. When I posted this graph, the immediate response from many on X was that the slide in reading scores must be caused by demographic shifts, especially increases in immigration. Here’s a representative example:

Others posted some variation of “this is what happens when you let in a flood of immigrants.”

If I’m understanding it correctly, the argument is this:

Large numbers of immigrants entered U.S. schools between 2015 and 2025. English wasn’t their first language and/or they came from disadvantaged circumstances, so they did considerably worse on standardized tests of reading, and that’s why reading scores went down so much.

So is that why?

The good news is this theory is testable. We can start with the first assumption, that large numbers of immigrants entered U.S. schools between 2015 and 2025, and go from there.

The data aren’t difficult to find. The U.S. Census has tables showing the estimated number of people who are foreign-born vs. born in the U.S. for those under 18 and over 18 in each year.

Since it’s K-12 students taking the reading tests, the group under 18 is the most relevant. What percentage of those under 18 were foreign-born in 2024 (the most recent data available for this stat)? Answer: 4.36%, or 4 out of 100. Not exactly “mass immigration.”

In 2015, the number was 3.37%, or 3 out of 100. So the number of foreign-born kids in U.S. schools did increase over the decade-long period, but only by one percentage point. Even if we assume that the foreign-born kids performed worse on reading tests –- and that definitely wasn’t true of all of them -- this is too small a shift to explain the dramatic declines in reading scores.

Some might then argue there has been mass immigration of adults who have kids once they arrive in the U.S. If that were true, there would be little change in the foreign-born population of kids, but a large (or at least larger) increase in the number of foreign-born adults.

But that didn’t happen either. The percentage of adults who are foreign-born was 17.61% in 2024 and 16.47% in 2015 – again, a one percentage point increase. There was no “mass immigration” of kids or adults between 2015 and 2024. There was an uptick, but it’s far too small to have caused a grade-level decline in reading scores in nearly every state.

So if immigration is not the reason, what did cause the decline in reading scores?

Certainly the pandemic school closures are responsible for some of it. But, since the declines started in 2013, COVID is clearly not the whole story. The report’s authors conclude that the Obama administration’s scaling back of George W. Bush’s No Child Left Behind (NCLB) program may be partially responsible, as NCLB held schools more strictly accountable for student achievement.

The authors also write, “The decline in achievement which preceded the pandemic was likely partially due to social media exposure.” The timing lines up: Daily social media use soared in popularity beginning in the early 2010s, and became more algorithmic throughout the decade. Social media and other digital media – whether it’s on a phone or a laptop, even a school-issued laptop -- distracts students in class when they should be learning. It’s tempting to think school-issued devices block access to anything but educational resources, but sadly that is not the case. Many can access YouTube, streaming services, even pornography.

Social media is also distracting at home, when scrolling through TikTok is easy and doing your language arts homework is hard. With so many digital distractions, kids and teens are also not reading for pleasure. Even high school English teachers have given up assigning books.

On X, many blamed American education in general. But, as the report emphasizes, reading scores were actually improving between 1990 and 2013. So the educational system was doing a great job until about 10 years ago. (The figure below also shows the implementation of NCLB and the later law, ESSA, which eased up on accountability).

I agree with the authors of the report that social media played a role and that we should be researching the effects of school phone bans. But the problem is bigger than that. There are at least two other changes that would help:

1. Parents should delay giving their kids smartphones until at least high school, preferably until after they get their driver’s license. Until then, get them a flip phone or a basic phone designed for kids, and lock it down to just a few apps. And don’t even get them this until they absolutely need it – parents often get their kids phones so they can reach them. Then their kid does everything on the phone except call their parents. I go over a detailed plan for how to do this in my book 10 Rules for Raising Kids in a High-Tech World.

2. Schools need to stop handing out tablets and laptops to students, especially elementary school students. In middle school, have shared devices at school and don’t send them home. Parents have had it with school-issued devices that their kids use to watch YouTube and streaming when they are supposed to be doing their homework and with classrooms where kids watch videos all day.

But won’t this mean that kids aren’t learning computer skills? Not really. Watching videos is not teaching them computer skills; it’s training their brains to not be able to pay attention to anything longer than 10-20 seconds. They do need to learn how to type, but they can do that in high school – and with AI dictation, they may not even need typing skills by the time they get to the workplace.

But there are some things students do need for college and long-term success: The ability to focus, think critically, and understand complex ideas. Devices undermine each of those. Books teach them all.

We also know that the more time a teen spends using social media the more likely it is he or she is unhappy, low in life satisfaction, or high in symptoms of depression.

There’s a bit of a difference, though: The studies of change over time have shown increases in mental health issues that require treatment, while the studies among individuals mostly focus on low psychological well-being or symptoms of depression measured by questionnaires. This difference has led some to argue that there is no evidence of a link between social media use and mental health disorders that require treatment.

I’ve always been skeptical of this argument given the strong links between self-report well-being or symptom measures and mental health disorders; the self-report measures correlate strongly with clinical diagnoses and are often used to screen for mental health disorders. It’s pretty unusual to find someone with major depression who says they are happy and doing well.

Still, having data showing a link with mental health disorders would be even better. That way, we’d know for sure that social media use was associated with actual mental health disorders – the type that require treatment – not just with unhappiness, low life satisfaction, or elevated symptoms of depression.

My colleagues and I have done a few of these studies. We found a link between social media use and self-harm behaviors in a large sample of UK adolescents, though self-harm was measured via self-report. We also found that digital media use was linked to depression, self-harm, and suicidal thoughts according to clinical criteria among 9- to 10-year-olds in the U.S. in the ABCD dataset. However, social media use was low in this very young population.

Ideally, we’d want a large sample of adolescents with clinically diagnostic measures of mental health disorders as well as a solid measure of social media use. A study that came out last year (Fassi et al., 2025) does just that …

… and it finds an even larger link than studies using well-being measures.

Teens with mental health conditions including generalized anxiety disorder, major depressive episode, eating disorders, oppositional defiant disorder, or ADHD spent more time on social media than those without a mental health condition, g = .46. That means teens with mental health conditions spent nearly a half a standard deviation more time using social media. That’s equivalent to a linear r of .22, almost exactly what another dataset showed for social media time and self-reported depressive symptoms among girls (r = .24).

For teens with internalizing disorders like anxiety or depression – the clinical-level equivalents to low well-being or high depressive symptoms – the effect size was even larger, g = .62. That’s a linear r of .30, higher than any study I know of that looked at social media time and well-being or depressive symptoms.

There were other differences between those with and without mental health disorders as well. Teens with internalizing disorders were also more likely to experience social comparison on social media (feeling they don’t measure up to other people’s profiles or photos), feel a lack of control over the time they are spending on social media, and report a bigger impact of social media feedback on their moods. Even those with externalizing disorders (like conduct disorder) spent more time using social media than those without any mental health disorder.

Above: Selected figures from Fassi et al. (2025)

This dataset still has more to tell us. For one thing, Fassi et al. did not control for sex. That’s very odd, as girls both spend more time on social media and are more likely to have internalizing disorders. The overall effect sizes may be somewhat lower since they didn’t include that control. The set-up of the analyses, comparing estimated social media time in each group, is strange – in my view, it would be better to see the percentage of teens with a mental health disorder at each level of social media use. The gray shading in the figures above does not indicate statistical significance; instead, they reflect g = -.40 to .40, which the authors of the article (pretty arbitrarily) designated as the “smallest effect size of interest.” Given that childhood lead exposure and adult IQ is g = .22, a g of twice as much is going to leave out many important and meaningful effects. Nevertheless, social media time was above their very high threshold.

This study puts to rest the idea that mental health conditions are not associated with social media use. They are, and at an even higher level than measures of well-being or depressive symptoms. That is further evidence that increases in major depressive episode and self-harm among teens after 2012 are linked to the rise of social media and smartphones. Combined with the two recent meta-analyses showing that social media reduction improves depression and well-being, this study is yet more evidence pointing toward severely limiting social media use among adolescents.

In our house, getting a driver’s license prompts another big teen milestone: A first smartphone. We’ve followed Rule #4 in my book 10 Rules for Raising Kids in a High-Tech World that first phones are basic phones, not smartphones. That meant our 14-year-old and 16-year-old had phones designed for kids – an Android phone with built-in parental controls and no way to download social media, gambling, or AI companion apps. But once a teen gets her license, Rule #5 (first smartphone with the driver’s license) kicks in, and she gets the smartphone. My daughter insisted we go to the AT&T store within minutes of coming home after passing the driver’s test.

But I wasn’t just going to hand her the smartphone and hope for the best. Driver’s license or not, she’s still a teen and needs some guardrails. The process of setting up an iPhone for a teen is not as easy as I was hoping it would be, so I’m sharing what I learned in case it’s helpful to other parents. Since these are device-based controls, not a dashboard you control from your own device, you’ll need your teen’s brand-new phone for a few hours to set it up. The good news is it’s free, since the controls are built into the device. Here are the steps you might consider:

1. Downtime at night. If you don’t want your teen playing games or watching Netflix on their phone in the middle of the night, you’ll want to set up Downtime. First, go to the Settings app, then Screen Time, and then Downtime. Toggle on “Scheduled” for every day (I left the default of 10pm to 7am; you can choose different hours, or set different hours for the weekdays vs. weekends). Then go to “Always Allowed” to choose the apps your teen will always have access to, even overnight. Phone calls are automatically always allowed. I also left Maps as always allowed. I deleted Messages (texting) and Facetime from always allowed. If things go well (for example, she always puts her phone downstairs at night), I’ll consider putting these two back in always allowed. For now, they’re only accessible during the day, between 7am and 10pm.

2. Block app downloads. Don’t want your teen downloading TikTok without your permission? Or, even worse, an AI girlfriend? Then you need to block app downloads. Go to Settings, then Under Screen Time, go to Content & Privacy Restrictions and toggle it on. Go to iTunes & App Store Purchases and turn off “Installing Apps.”

3. Set time limits for apps. You can set limits for whole categories (like Games) or for individual apps. I wanted to set limits app by app, and this took me a little bit to figure out -- choose Games (or Entertainment), not the circle next to it, to get to the menu of individual apps. Then click the circle next to the individual app (say, YouTube), click Next, and scroll for the time limit. You’ll want to consider the total time this adds up to. If you’d rather set a total time limit for, say, all Games or all Entertainment, you can do that instead.

4. Block adult websites. Under Screen Time, Content & Privacy Restrictions, and then App Store, Media, Web & Games: Choose the settings appropriate for your kid – for example, you can turn music, podcasts, news, and fitness to “clean” instead of “explicit.” I clicked “don’t allow” for movies and TV shows because I don’t want her watching movies on her phone. Under Apps I chose the age range 16+ (the default is 18+); there are lower age limits as well. Under Web Content, I chose “Limit Adult Websites.” (“Unrestricted” is the default).

5. Lock the passcode. Under Content & Privacy Restrictions, under “Allow Changes to” turn “Passcode & Face ID” to “Don’t allow.” This way your kid can’t change their phone passcode, so you have access to their phone without having to weasel the passcode out of them.

6. Block nude photos. Go to Screen Time, and then Communication Safety to turn on blocking nude photos and videos.

7. Limit who they can communicate with. Under Screen Time, then Communication Limits. Here, you can decide who your kid can get messages and calls from. There’s one section for “During Screen Time” – for our purposes that means during the day. I chose “Contacts & Groups with at least one Contact” here. That means she can communicate with anyone in her contacts and anyone in a group chat with one of her contacts. It blocks single messages from anyone not in her contacts; my thought was this would prevent her from getting unknown number calls and texts. Another section is “During Downtime,” meaning overnight. Emergency numbers are always allowed, and I added the contact for AAA and the numbers of our immediate family here.

8. Set your parental control passcode. Under Screen Time, go to Lock Screen Time Settings. You’ll have to enter your email and the password to your Apple ID (not your email password!) to recover the code if you forget it. This means you need an Apple ID and have to remember your password; make sure you have these before you start. Then you’ll choose a 4-digit number code that you know & your kid doesn’t. This is the most important step – if you don’t set up this code your kid can easily change all of the settings you just set up.

There are lots of other things you can restrict as well. Under Intelligence & Siri, you can block AI “math results” and “writing tools.” For Siri, you can block explicit language.

The whole process took about two hours – twice as long as I anticipated, partially because Apple has expanded their parental controls since our older daughter got her first iPhone. That’s good, but it also means there are a lot of settings I didn’t understand.

And it wasn’t over. A week or so after I’d set up my daughter’s phone, I saw she had TikTok on it, which we had not agreed to. How did she get it when I’d blocked app downloads? She’d gone to the TikTok website and downloaded it there – apparently app downloads are only blocked via the app store, not websites.

[cue primal scream]

So if you want to more completely block kids downloading certain apps, there are three steps you can take:

1. Block specific websites. This setting is pretty hidden, so bear with me. Under Settings, Screen Time, Content & Privacy Restrictions, and then App Store, Media, Web & Games, then Web Content: Click “Limit Adult Websites.” Once you do that, additional menus will appear that say “Always Allow” and “Never Allow.” Click on Add Website and add the URLs of the websites you don’t want your kids to access. That could be social media like Instagram and TikTok, or it could be gambling sites or gaming sites (pornography sites should already be blocked by default). This eliminates the work-around for downloading the associated apps.

2. Disallow apps. Under Screen Time and then Content & Privacy Restrictions, go to Allowed Apps & Features and toggle off the offending apps. You can turn off the web browser (Safari) here if you want to prevent your kid from downloading social media via the web again. Of course then they can’t use the internet with the phone at all, so this might be a step to take temporarily until they regain your trust.

3. Set time limits to zero. Under Screen Time, App Limits, click Add Limit, and then choose the category and then the app by touching the circle; then click next. Then set the time limit to 0 hour, 0 min. This is another step to take if they have already dowloaded an app you didn’t want them to have. Even if they download it again, it won’t work.

At least it shouldn’t. This is the supremely frustrating thing about phones: Kids seem to always find a workaround. This experience made me even more grateful my daughter had a basic phone until now. It also strengthened my view that parents should put off giving their kids smartphones for as long as possible – there are just too many ways kids can do stuff parents don’t want them doing on a smartphone compared to a basic phone.

This experience also highlights why we desperately need more regulation of social media. It just plain should not be this hard to keep kids off social media or other apps and websites. The settings are complicated, and a lot of parents don’t have the time, resources, or fight in them to install them.

But if governments required parental permission for minors to have social media, these issues would go away. Same thing if they required age verification. If the age minimum was 16 and was actually verified, parents could at least postpone this rigamarole for a few years. Or the minimum age for social media could be 18, legal adulthood. After all, what other contract can minors sign other than the social media account disclaimer signing away their data? Seventeen-year-olds need a parent signature to go on a school field trip, but not to use social media.

For now: Please share this post. I’ve made it free from day one because I hope it will help other parents in the same situation. With so little support from government or the social media companies, we parents have to work together to keep our kids safe, happy, and healthy. It’s not fair that this is on us, but, for the most part, it is.

Not completely, of course, but it’s clear they’ve had some impact when the number of teen boys who don’t agree with equal pay for women goes from 12% to 22% (since 2017, with most of that change occurring 2022-2024). In my view, it’s not a coincidence that these attitudes shifted in the age of ubiquitous social media use and the concomitant rise of manosphere and tradwife influencers.

In discussion of the post on X, this comment popped up:

My initial thought was “OK, fine, it’s probably not just time spent – teens could still be influenced by manosphere and tradwife content in an hour or so a day.” An inverse relationship also seemed at least somewhat plausible as I’ve found (with 12^th graders) that more liberal students spend more time on social media. Still, I thought, that might not be true for these particular views around gender … and then I realized: Why speculate when I can go look at the data? And the link between social media use and gender beliefs is an interesting research question.

I used 8th and 10th graders in Monitoring the Future 2018-2024, as the response choices changed for the social media questions in 2018. I looked at hours per day of social media use and the percentage who agreed or mostly agreed that “It is usually better for everyone involved if the man is the achiever outside the home and the woman takes care of the family.” The results are separated for boys vs. girls and controlled for grade level (8^th vs. 10^th), race/ethnicity, and socioeconomic status (mother’s education). Total n was 24,971, so a nice sample size.

So, do heavy social media users have the most progressive views, as Dr. MacDonald posted on X?

They do not – in fact, it’s the opposite. Especially among boys, more social media use means more agreement with rigid gender roles (see Figure 1). And it’s not subtle: Among boys, 43% of the heaviest social media users agree with rigid gender roles, compared to 27% of non-users. That translates to 56% more heavy users (vs. non-users) with traditional views among teen boys.

Figure 1: Percentage of U.S. 8^th and 10^th graders who agree it’s best if the man achieves outside the home and the woman takes care of the family, by hours per day of social media use and gender. Figure shows those who answered agree or mostly agree; data are from 2018-2024 combined. Controlled for race/ethnicity, grade, and socioeconomic status (mother’s education). Source: Data from Monitoring the Future analyzed by Jean M. Twenge for the Generation Tech Substack.

The association is not as strong among girls, but even there the heaviest social media users are 36% more likely to hold traditional views than non-users.

Thus: heavy social media users have the least progressive views, not the most progressive views as Dr. MacDonald claimed.

I’m not sure why he got this wrong. Maybe he thought the item was scored in the other direction? He did say he was looking at only the 2023 and 2024 data, but the pattern looks the same even with only those two years. Or perhaps he didn’t separate the data by sex or control for it. That’s essential, as girls spend more time on social media and are less likely to agree with this question.

So, assuming the graph above is instead correct, what does this mean? More social media use is linked to belief in more rigid and traditional gender roles, supporting the theory that time spent on social media is associated with more traditional views of men and women’s roles. It seems likely that a combination of more hours on social media plus more manosphere and tradwife content is at least partially behind the rise in support for more rigid gender roles.

The more linear and larger association among boys deserves exploring. Perhaps hours per day has a stronger influence on boys’ exposure to the manosphere and tradwife culture than it does for girls, who might be consuming other types of content more often. This is the opposite of what we usually see with social media and mental health, where effects for girls are larger than for boys. Perhaps hours per day on social media has a bigger impact on girls’ mental health and on boys’ beliefs, especially gender role beliefs.

Another conclusion: Don’t believe everything you read on X.

Then, from the late 1960s to the 1980s, the number of women business leaders, TV reporters, lawyers, professors, engineers, and doctors shot upward. For several decades, half of newly minted doctors and lawyers have been women.

Laws and attitudes have followed suit. Equal pay for equal work is the law, and rejection of rigid gender roles skyrocketed among American adults between the 1970s and the early 2020s (see Figure 3.22 in Generations).

Questions about equal pay and gender roles have also been asked of teens – specifically, of 8^th and 10^th graders (ages 13 to 16) from 1991 to 2024. My colleagues and I had previously found that 12^th graders’ attitudes toward women working had gotten more positive between the 1970s and 2013. But a lot has changed, culturally and politically, since then. For one thing, 13- to 16-year-olds are often immersed in social media, some of them in the so-called manosphere promoting traditional gender roles (often along with a heavy dose of misogyny; last year, the Netflix movie Adolescence did an excellent job showing the impact of manosphere influencers on young males). Both Trump’s first term (2017-2021) and second term (2025-present) have coincided with a rightward shift in the country. Still, political shifts don’t always show up in real people’s attitudes, and even in this environment it would be surprising if teens were questioning the idea that women should be paid equally for the same work.

But they are.

Or at least more are. Monitoring the Future asks 8^th and 10^th graders if they agree or disagree with the statement “Men and women should be paid the same money if they do the same work.” From 1991 to 2019, only about 12 to 13% of boys did not agree that women should be paid equally for their work. Then that began to rise. By 2024, the most recent data we have, 22% of boys – more than 1 out of 5 -- did not agree that women should be paid equally for doing the same work as men. That’s an all-time high in the four-decade history of the survey (see Figure 1).

Figure 1: Percent of U.S. 8^th and 10^th graders who do not agree that women should be paid equally for the same work, by gender, 1991-2024. Figure shows those who answered disagree, mostly disagree, or neither. Source: Data from Monitoring the Future analyzed by Jean M. Twenge for the Generation Tech Substack.

In contrast, only about 5% of teen girls don’t agree women should be paid equally, with no consistent changes over the years.

Teens are also asked if they agree that “A woman should have exactly the same job opportunities as a man.” The number of teen boys who did not agree with this statement declined in the 2010s but then nearly doubled between 2018 and 2024, ending at 3 out of 10 (see Figure 2).

Figure 2: Percent of U.S. 8^th and 10^th graders who do not agree that women should have the same job opportunities as men, by gender, 1991-2024. Figure shows those who answered disagree, mostly disagree, or neither. Source: Data from Monitoring the Future analyzed by Jean M. Twenge for the Generation Tech Substack.

Even more stunning, the number of girls who do not agree women should have the same job opportunities as men has also risen – not as much as among boys, but enough to amount to nearly twice as many girls in agreement in 2024 vs. 2018.

Then the changes come home. Teens are also asked if they agree “It is usually better for everyone involved if the man is the achiever outside the home and the woman takes care of the family.” This is the kind of rigid gender role division Boomers fought against in the 1970s and Gen X’ers thought they demolished in the 1990s. Previous research up to 2013 found that 12^th graders’ agreement with this statement declined an enormous amount between the 1970s and the 1990s, with agreement starting to creep back up in the 1990s.

What happened after that? Agreement with rigid gender roles increased slightly among 8^th and 10^th graders between the 1990s and 2010s, but then fell from 47% to 32% among boys and from 26% to 13% among girls up to 2021.

Then, in 2023 for boys and 2024 for girls, attitudes suddenly shift again, with 41% of boys and 23% of girls agreeing that it’s best if the man achieves outside the home and the woman takes care of the family (see Figure 3).

Figure 3: Percent of U.S. 8^th and 10^th graders who agree it’s best if the man achieves outside the home and the woman takes care of the family, by gender, 1991-2024. Figure shows those who answered agree or mostly agree. Source: Data from Monitoring the Future analyzed by Jean M. Twenge for the Generation Tech Substack.

Unlike the questions on women’s pay and job opportunities, here the shift in attitudes among girls is just as large as the shift among boys. By relative statistics, it’s actually larger, with 72% more girls favoring rigid gender roles in 2024 vs. 2021 (for boys, the jump is 29%).

The trend is even more stark if we instead look at agreement and neutrality combined – in other words, those who do not reject rigid gender roles. By 2024, 3 out of 4 boys and half of girls don’t disagree with the idea that it’s best if men are the achievers and women take care of the home and family (see Figure 4). Those numbers are all-time highs in the history of the survey. They’re also sudden changes, shooting up in 2023 and 2024 with a 66% increase among girls and a 22% increase among boys in just three years.

Figure 4: Percent of U.S. 8^th and 10^th graders who do not disagree it’s best if the man achieves outside the home and the woman takes care of the family, by gender, 1991-2024. Figure shows those who answered agree, mostly agree, or neither. Source: Data from Monitoring the Future analyzed by Jean M. Twenge for the Generation Tech Substack.

Why the sudden shift? Online discourse on social media is once again a possibility – “tradwife” accounts promoting exactly this type of gender role division suddenly started to gain traction a few years ago. It’s hard to say if tradwife accounts are a symptom or a cause, though. Perhaps attitudes shifted toward more traditional gender roles, which is why tradwife discourse caught on. Or tradwife discourse caught on and then attitudes shifted toward more traditional gender roles. Manosphere content may also have promoted these ideas. That content has been around for longer, making it more likely it’s a cause instead of a symptom.

This increasing support for women stepping back from work occurred at a time when nearly 60% of four-year college degrees go to women. Women’s incomes have been rising faster than men’s for a very long time, partially because more young men are not working. These traditional attitudes, especially those questioning equal pay and job opportunities, could be a backlash reaction to how well young women are doing relative to young men.

At the same time, the cost of buying a first home is very high, usually requiring two earners, not to mention the spiraling cost of groceries, gas, and education. With those realities as a backdrop, tradwife culture is unattainable for many.

But Gen Z and Gen Alpha are tired of the rat race. They’re not willing to put in the hours previous generations did at work, partially because they are pessimistic and don’t expect hard work to get them anywhere. That might be why girls have increased even more in their support of traditional gender roles: being a tradwife means they could opt out of the workplace.

No matter what the cause, these are big changes in a short period of time. Teens’ increasing support for rigid gender roles is both surprising given the ascendence of young women in the workplace and not surprising given the cultural shift toward the traditional and online discourse in the last few years. Whether Gen Z and Gen Alpha will live differently due to this shift in attitudes is unclear – we won’t have a good picture of that until they get older. For now, there’s a growing segment of teens who are apparently done with gender equality, with consequences unknown.

But are younger adults just less likely to identify as non-heterosexual, or are they also less likely to engage in non-heterosexual sexual activity? In other words, are there trends in behavior, or just identity?

One of the few national surveys that has measured sexual activity over time is the General Social Survey (GSS) — it’s asked American adults about the sex of their sex partners since 1988. Unfortunately the survey didn’t ask all of the questions about sexual activity in 2024, but it did ask people whether they had sex with men, women, or both in the last year.

Since the sample sizes of the GSS are smaller than some other national surveys, I broadened the age range for younger adults to 18 to 39 to make sure there were at least ~500 respondents for each year, and thus at least ~250 or more for each sex.

As always, take these results with the caveat that future years will give us a clearer picture of what’s going on.

So what are the trends in non-heterosexual sex among younger adults in the U.S.?

In short: Fewer women are having sex with women in 2024 than just two years before.

Just like the changes in identity, the recent trends in sexual behavior are the most dramatic for women. Non-heterosexual sex (having sex with another woman in the last year) dropped by more than half between 2022 and 2024 among 18- to 39-year-old women in the U.S. Previously, the number of women having sex with women had doubled between 2010 and 2022 (see the red line in Figure 1). So there was a big increase, and then a big decrease.

Figure 1: Percent of U.S. 18- to 39-year-olds who had sex with someone of the same sex in the last year, 1988-2024. Source: General Social Survey; data analyzed by Jean M. Twenge for the Generation Tech Substack.

The recent pattern is less clear for men, with ups and downs but little consistent trend. More men had sex with men between 1988 and 2000, but then those rates declined before coming up again in 2014 and varying between 4% and 6%. The rates never got above 7% as they did for women in 2018 (7.4%), 2021 (9.4%), and 2022 (10.2%).

By 2024, fewer women were having sex with women than men having sex with men, which had not happened since twenty years before in 2004.

Especially with the lower sample sizes, it will be important to get at least another year of data to determine whether this is a blip or part of a new trend. The 2026 GSS is being fielded right now, so we’ll know more next year. For now, it appears the decline in women identifying as bisexual extends to who they’re having sex with as well.

Kaufmann found that identifying as lesbian, gay, or bisexual was also declining, but most of his sources were college samples. At the time there wasn’t much nationally representative data on sexual orientation for 2024 or 2025, and the little there was had low sample size.

That nationally representative data is now available. So is it true – is identifying as non-heterosexual in decline among American young adults?

The answer appears to be yes, and the changes are even larger among some groups. I drew from the nationally representative Behavioral Risk Factor Surveillance System of U.S. adults administered by the CDC, which gets samples of more than 200,000 people every year. The 2024 data is fully out, and that datafile also has some responses from 2025 (12,000 – much smaller than a full year, but a large sample size compared to many other surveys). The survey asks about sexual orientation with choices of straight, bisexual, gay or lesbian, and “other” or “something else.” (Because those last choices changed over the years, I treated them as missing data).

Among American young adults (ages 18 to 24), identifying as lesbian, gay, or bisexual – something other than straight or heterosexual – peaked in 2022 at 20%, but then fell to just over 15% by 2025. That’s a 21% decline in just 3 years.

Figure 1: Identifying as lesbian, gay, or bisexual (not straight), by age group, U.S. adults, 2014-2025. Source: Behavioral Risk Factor Surveillance Survey. NOTE: 2020 data are excluded as data collection was paused for much of the year due to the COVID pandemic. Survey year is interview year in the dataset, not the year of the datafile.

Among 25 to 34-year-olds, identifying as non-heterosexual kept rising until 2024, dropping 20% in 2025. There’s just a slight hint of a decline among 35- to 44-year-olds, and only a whisper of it among those over 45 years old. With only one year of decline among those in their late 20s to early 40s, we’ll need more data to know for sure if the decline will persist. But for now, this is a classic generational shift, with the changes first and largest among the youngest group.

So which identification (lesbian, gay, or bisexual) has changed the most among young adults? It’s definitely identifying as bisexual, especially among young women. This shift drives almost all of the change seen in Figure 1. As recently as 2015, only 8% of U.S. young women identified as bisexual, but by 2022 23% did — the rate nearly tripled in just 7 years. But by 2025, the percentage of young women identifying as bisexual had dropped to less than 18% (see Figure 2). That’s a 23% decline in 3 years. There was also a decline in young men identifying as bisexual, but not until 2025.

Figure 2: Identifying as lesbian, gay, or bisexual, by gender, U.S. adults ages 18 to 24, 2014-2025. Source: Behavioral Risk Factor Surveillance Survey. NOTE: 2020 data are excluded as data collection was paused for much of the year due to the COVID pandemic. Survey year is interview year in the dataset, not the year of the datafile.

Similar to non-heterosexual identification overall, the percentage of women identifying as bisexual changed the most among the youngest adults, but less and later among those who were older (see Figure 3).

Figure 3: U.S. women identifying as bisexual, by age group, 2014-2025. Source: Behavioral Risk Factor Surveillance Survey. NOTE: 2020 data are excluded as data collection was paused for much of the year due to the COVID pandemic. Survey year is interview year in the dataset, not the year of the datafile.

Combined with the data on declining transgender and non-binary identification, this means fewer U.S. adults identified as members of the LGBTQ+ collation in 2024-25 vs. 2021-22. Fewer identify outside the gender binary, and fewer identify as a sexual minority. The cyclical nature of these changes, especially their appearance among the youngest adults, suggests a cultural shift – one that might grow or fade during the coming years.

But what about during school, where teens spend more than 30 hours a week? Those hours, too, are filled with technology. Sometimes that’s for truly educational purposes – they’re working on an essay for English class, reading a science textbook in an online library, or taking notes in class.

But not always. Even school-issued laptops often allow access to YouTube and streaming (like Netflix, Disney+, and Peacock), allowing students to sit in the back of class and watch endless hours of entertainment. Others play games. Personal smartphones are also a huge distraction: A recent analysis found that American teens spend more than an hour using their phones during the school day, and almost none of that time is spent on educational activities. Instead, teens scroll through social media, watch videos, and play games. Some take videos of their peers without permission, or sneak off to the bathroom to watch TikToks.

Thus, teens are spending about 20% of their time at school not focusing on schoolwork or talking to their peers. That may be one reason why standardized test scores in math, reading, and science have declined since 2012 and why students have increasingly reported feeling lonely at school. Electronic devices are both distracting in the classroom and isolating in the lunchroom. What impact does that have on teens’ learning and on their mental health?

In a recent paper, my students and I looked into these issues in the PISA dataset of 15- and 16-year-olds around the world. In 36 countries, students consistently took standardized tests in math, reading, and science between 2006 and 2022. In 2022, they were asked how much time they spent using electronic devices (like phones, tablets, and laptops) for leisure purposes (like social media or entertainment) during the school day. This varied quite a bit across countries, with students in some countries spending hardly any time on devices for leisure during the school day, and others spending an average of more than two hours.

In countries where students spent a lot of time using devices for leisure during the school day, test scores plummeted between 2012 and 2022. In countries where they spent less time, test scores merely slid. Thus there was a significantly larger decline in scores in the countries where students spent more time using devices for fun during school hours (see Figure 1).

Figure 1: Scores on standardized tests of math, reading, and science for 15- and 16-year-olds in 36 countries, by low or high use of electronic devices for leisure during the school day. Note: Controlled for GDP per capita. Source: Twenge (2025) using data from PISA.

The consequences of device use aren’t just academic – they are also social and emotional, because device use has displaced students talking to each other during lunch and breaks. In countries where students spend more time using devices for leisure during the school day, the percentage of students who agreed “I often feel lonely at school” rose steeply, with the increase much less pronounced in countries with less leisure device use during school (see Figure 2).

Figure 2: Feelings of loneliness at school among 15- and 16-year-olds in 36 countries, by low or high use of electronic devices for leisure during the school day. Note: Controlled for GDP per capita. Source: Twenge (2025) using data from PISA.

These results show the twin impacts of the leisure use of devices during the school day: Declines in test scores and increases in feeling lonely at school. They are another piece of evidence suggesting that schools should restrict students’ use of smartphones from bell to bell – not just during class, but also during lunch, breaks, and passing periods. A school where students are talking to each other is less lonely. I recently visited a school with a bell to bell no phone policy, and students are now talking, playing cards, and “bedazzling” (had to look that up!) with each other instead of being endlessly absorbed in their phones.

Of course, phones are only part of the problem. The next step is to lock down laptops and tablets so they, too, aren’t being used for social media and entertainment during the school day. Or, especially for younger students, it may be time to go back to paper and pencil – old-school, yes, but with the bonus of no binge-watching YouTube videos during chemistry class. Some states are considering bills outlawing or restricting the use of devices for elementary school students — a welcome step.

Sticking with the status quo means lower test scores and more lonely students — not the outcome any of us want.

p.s. I worked with some truly wonderful undergraduates on the PISA project, which at times seemed endless due to the complexity of the tables (data collected over 22 years across 36 countries). My heartfelt thanks to Spencer Deines, Ellah Fessenden, Lauren Gramse, Julia Lima, Elisa Ruiz, Siri Sommer, and M’Lise Venable. (With a special shout-out to M’Lise’s cats for their starring role).

“AI what?” was my first thought. I soon found out: AI is not just for web searching or writing your English class essay anymore. It can be your friend, or even your girlfriend.

Everything is changing so fast it’s tough to keep up. AI companions (like Character.AI, Replika, and a long list of AI girlfriend apps) are conversational chatbots that mimic a relationship with a real human being. Well, sort of: Because companies make the most profit when users spend the most time on them, AI companions are programmed to keep people coming back for more. They praise, they agree, and they try to give the user what they want, leading many observers to describe them as sycophantic.

The apps are also always available, always willing to talk about sex, and don’t have their own needs. In that way, they are as far from human as you can get.

Parents already had enough to worry about with kids and technology, including social media with no age verification and school laptops with unrestricted access to entertainment. And now kids can access AI companion apps, including some that are very, very adult. The image below – which I took from the Pinwheel kids phone list of banned apps -- gives a small and scary taste of what’s out there.

These apps are almost completely unregulated. Some of these platforms say they are designed for users 18 and older, but they don’t verify age. Others market themselves to minors as young as 13.

If you give a child or teen a smartphone and don’t block app downloads, there is nothing to stop your kid from having their first “romantic relationship” with an AI chatbot.

Or it could become their best friend. Replika AI advertises itself as “Your AI Friend – Always here to listen and talk.”

And teens are using these apps – a lot of teens. According to a Common Sense Media survey conducted last year, 72% of U.S. teens ages 13 to 17 have used an AI companion. Half (52%) use them a few times a month or more. Even more troubling, a third of teens have turned to AI companions instead of actual humans for serious conversations.

That’s likely to increase in the coming years. Meta CEO Mark Zuckerberg observed that the average American has about three friends but would like to have 15. AI companions, he says, will fill in that gap. “I think people are going to want a system that knows them well and understands them in the way that their feed algorithms do,” he said. Except, and I don’t know why this even needs to be stated, AI companions are not friends because THEY ARE NOT HUMAN.

The possibility of a population with 12 AI friends is scary enough when those users are adults. For kids, who are still learning how to be a friend and often have no experience with romantic partnerships, it’s the apocalypse for relationships. What will it mean for kids’ friendships if they get used to sycophantic chatbots with no needs of their own? What will it mean for their future romantic partnerships if they fall in love with a chatbot who’s always sexy, always praising, doesn’t get tired or angry, and doesn’t sleep or eat?

There have already been several tragic cases where AI companions have urged teens to take their own lives. Research is only beginning to emerge about AI use and mental health issues, but at least one study of adults has already shown links between generative AI use and depression.

My hope is that Congress will take action to regulate AI companion apps so those under 18 cannot use them. Two weeks ago, I testified to a Senate committee in D.C. about doing just that.

Until then, it’s up to parents to keep their kids off these apps.

What can parents do? My suggestions:

1. Delay the smartphone as long as possible. AI companions are yet another compelling reason to put off giving your kid a smartphone.

2. If there’s a compelling reason your kid needs a mobile phone, give them a flip phone or a phone designed for kids. (Details are in Rule #4 in my recent book, 10 Rules for Raising Kids in a High-Tech World). Phones designed for kids automatically block AI companion apps (as well as gambling apps, dating apps, and social media).

3. Require a parent passcode for app downloads when your teen does finally get a smartphone (I recommend giving the first smartphone when teens get their drivers license or are making their way around town by themselves using public transportation. This is usually around 16). But even 16-year-olds shouldn’t have AI girlfriends, so block app downloads. (Details on parental controls, including the different types and how to implement them, are in Rule #6 in 10 Rules).

4. Install parental controls (probably third-party software) on kids’ laptops and use it to block AI companion platforms. There’s more on this in Rule #9 in 10 Rules.

And if you think none of this should be necessary, you’re right. That’s something else you can do: Write to your representative or senator and ask them to pass legislation to keep minors off AI companion apps. Our kids deserve the chance to learn how to have relationships with real human beings first.

It’s especially tough for the administrators of large national surveys, like Monitoring the Future (MtF), that are done every year. They don’t want to change the survey too much since that makes it difficult to compare responses over the years, but they also want to make sure they are keeping up with the times.

From 2018 to 2023, MtF asked teens “About how many hours on an average DAY do you spend on social networking sites like Facebook, Twitter, Instagram, etc.?” The question was outdated almost as soon as it was asked. Facebook use was already going out of style among teens in 2018 (because Grandma was on Facebook). Twitter has never had considerable market share among teens. Instagram is the only platform listed in the item that a majority of teens use. And of course no one calls them “social networking sites” anymore.

That’s probably why MtF updated its question in 2024 to “About how many hours on an average DAY do you spend . . . on social media sites like TikTok, YouTube, Snapchat, Facebook, Instagram, X, etc.?” At long last, the question includes the platforms teens use the most. Choices ranged from “none” to “9 hours or more.” Now we can finally get an idea of how much time teens are spending on social media in one of the largest and most reliable nationally representative surveys of U.S. teens.

In 2023, before the update, U.S. 8^th and 10^th graders (13 to 16 years old) estimated they spent about 3 hours a day using social media. In 2024, after the question mentioned the platforms teens actually use, their estimate was 4 and a half hours a day. Among girls, it was 5 hours a day. That’s an enormous amount of time.

But that’s not even the crazy part. The crazy part is how many are spending 7 or more hours a day using social media – close to the time commitment of a full-time job.

In 2024, that number was one out of four among U.S. 8^th graders. According to the U.S. Census, there were about 4 million 14-year-olds in the U.S. in 2024. That means about one million 14-year-olds spend 7 or more hours a day using social media. One million.

At this level, social media use is almost certainly interfering with sleep, schoolwork, exercise, and spending time with friends and family face-to-face. The numbers were even higher for 8^th grade girls: One in three spent 7 or more hours a day using social media (see Figure 1). If three girls sit next to each other in an 8^th grade class, the middle girl can look to her left and right and guess that one of them is spending the equivalent of a full-time job on social media. Even apart from the impacts on mental health, this is a tragic loss of potential.

Figure 1: Percent of teens using social media 7 or more hours a day, by gender and grade level, 2024. Source: Monitoring the Future data analyzed by Jean M. Twenge for the Generation Tech Substack. Note: Sample size for 12^th graders on this item is lower and should be interpreted with caution.

Fewer 10^th graders (15 and 16 years old) and even fewer 12^th graders (17 and 18 years old) spent this much time using social media. These data show once again that younger teens are the most vulnerable to social media algorithms. As I argue in 10 Rules for Raising Kids in a High-Tech World, that’s why kids under 16 (and maybe even 18) shouldn’t have social media accounts at all. It’s why Australia raised the minimum age for social media accounts to 16.

These numbers were already staggering. When I analyzed the data within race/ethnicity and socioeconomic status, they got even more jaw-dropping.

Half of Black teen girls spend 7 or more hours a day using social media. Half.

So do more than a third of Hispanic girls, and a fifth of White girls (see Figure 2). Black and Hispanic boys are also much more likely than White boys to spend 7 or more hours a day using social media. All of these numbers are too high, but the difference by race and ethnicity is stunningly large. (I’ve focused on 8th and 10th graders here as their sample size on the social media item is more than sufficient; the sample size is much lower for 12th graders on this item, so the data are not as reliable when broken down by race and ethnicity).

Figure 2: Percent of 8^th and 10^th graders using social media 7 or more hours a day, by gender and race/ethnicity, 2024. Source: Monitoring the Future data analyzed by Jean M. Twenge for the Generation Tech Substack.

The hits just keep coming. Thirty-seven percent of girls whose mothers do not have a college education (indicative of lower socioeconomic status, or SES) spent 7 or more hours a day using social media, compared to 24% of girls whose mothers have a college education (higher SES). There’s a similar difference, though smaller, for boys. So the teens most likely to be poor — or at least not as advantaged — are spending the most time on social media (see Figure 3).

Figure 3: Percent of 8^th and 10^th graders using social media 7 or more hours a day, by gender and socioeconomic status (mother’s education), 2024. Source: Monitoring the Future data analyzed by Jean M. Twenge for the Generation Tech Substack.

We’ve always known that social media companies were making an enormous profit off the time and attention of teens. Now we know they are making even larger profits off the most vulnerable teens.

The official Australian coat of arms depicts a kangaroo and an emu, two native animals that cannot walk backward. They’re also two animals you wouldn’t want to meet in a dark alley if they were snappish.

Yes, the photo is pixilated … because I was not going to get too close to the emu. One does not mess with the emu. (Photo taken at Lone Pine Koala Sanctuary outside of Brisbane, Queensland, Australia).

Australia is going to need the spirit of the kangaroo and the emu as the law goes into effect – they should not move backward and should fight if necessary. Australia had the courage to be first, and that’s going to mean there will be challenges along the way. But they should persevere, because the arguments against the law are weak and sometimes even nonsensical.

Why have a law keeping under-16s from creating social media accounts in the first place? It is abundantly clear that social media is not safe for kids. Social media companies know that children and younger teens spend more time on the apps because their self-control is not as developed (An internal document from TikTok admitted, “TikTok is particularly popular with younger users who are particularly sensitive to reinforcement in the form of social reward and have minimal ability to self-regulate effectively.”) Kids are tempted to open social media again and again through endless notifications, including in the middle of the night. Links between social media use and poor mental health are larger among those under 16 compared to those over 16. Unknown adults can easily contact kids on these platforms, and inappropriate content is rampant even on so-called teen accounts. These are among the reasons why “No Social Media Until Age 16 – Or Later” is Rule #3 in 10 Rules for Raising Kids in a High-Tech World.

In most countries, age isn’t verified when kids sign up for a social media account, so 11-year-olds easily sign up for Snapchat and Instagram even though the law says they need to be 16. But 13 is a terrible age to introduce social media, and teens want to do what their friends are doing.

With a law setting 16 as the minimum age and verifying it, Australian parents will be a lot less likely to hear their kids say, “But Mum, all of my friends use Instagram! If I’m not on it, I’ll be left out!” Kids will have to say, in effect, “But Mum, all of my friends are breaking the law!” – a little less persuasive. As an American parent, I can only wish I lived in a country that would support me in keeping my kids off social media until they were 16. If Australia’s ban goes well, other countries (and perhaps some U.S. states) might follow, so the stakes are high.

My kids & other visitors with the friendly kangaroos at Lone Pine. Still not going to challenge one to a fight, though.

Of course, not everyone is sold on the idea. Here are the most common arguments against Australia’s law – and why I think they don’t hold water.

• Kids and teens will find work-arounds to get on the platforms anyway. It’s completely true it may take some time to make age verification completely effective. But that’s not a reason to give up. Imagine if someone used this argument against age minimums for buying alcohol: “Teens are just going to find a way around it, so the law is stupid.” Yes, some teens get fake IDs so they can buy alcohol. But most don’t. Almost immediately after the drinking age was raised to 21 in the U.S. in the late 1980s, fewer teens drank alcohol. The downward trend then continued as drinking became steadily less popular among 13- to 18-year-olds. The decline in teen alcohol use is also one of the reasons teen deaths in car accidents went down in a similar pattern. The government got better at enforcing the law, and then social norms took over and made it less acceptable to drink while underage. The same will be true for social media: Enforcement will improve, and eventually the social norm will move away from teens feeling they “have” to be on social media.

See those claws? You don’t want to make them mad either. And that squint says, “Don’t even try using a VPN in my house, honey.” (Also taken at the Lone Pine Koala Sanctuary. It’s awesome.)

• Teens in vulnerable groups will be cut off. For example, a CNN article noted, “Experts who work closely with vulnerable and isolated children warn that taking away social media platforms from those children could deprive them of support networks, making them feel even more isolated and alone.” This argument makes two false assumptions. First, it seems to assume that social media is the only way for vulnerable teens to have a support network. That’s just not true. If teens want to seek out information, they have the rest of the internet. If they want to communicate with others like them, they can call, FaceTime, text, or use WhatsApp. They can also use Discord or Pinterest, which are still allowed for under-16s in Australia. Second, this viewpoint takes for granted that social media use necessarily benefits children in vulnerable groups. Where is the evidence that vulnerable teens who spend more time on social media are happier on average than those who don’t use social media or who use it only lightly? I don’t know of any. If anything it’s the opposite. And if the era of social media was so good for teens in vulnerable groups, why have their rates of depression increased just as much or more than those in less vulnerable groups? Plus, vulnerable teens get bullied more on social media. The risks outweigh the benefits.

• Let’s teach digital literacy to kids instead. Good luck. As one Australian dad put it, “I’m an adult and I know what an algorithm is, but it still gets me.” Yes, we should talk to kids about how to stay safe online. But expecting them to self-regulate their social media use is unfair and unrealistic when the social media companies have poured billions into making their apps as addictive as possible. We should do both: Keep kids off social media until they’re older, and teach them digital literacy.

Stay strong, Australia. The rest of us are counting on you.

Personal p.s.: I fell in love with Australia the first time I visited in 2004. I’ve now traveled there 9 times and love it more every time I go. I took the pictures above just a few hours after getting off the plane on my last visit; I am convinced that the best thing to do after stumbling off a 14-hour overnight flight is to pet kangaroos and take pictures of koalas. I’ve wished I lived in Australia many times over the past 21 years, and now with the new social media law I really do.