Greenspun

https://washingtonmonthly.com/features/2007_17ya/0709.careycascadia.html:

At the other college, however, things are … different. “Harder.” First of all, her professor never seems to explain anything. Instead, he’s constantly posing questions that seem deliberately vague, then he “tells you to go find the materials and figure out the answer for yourself.” She can’t skip class, even if it’s been a long day selling popcorn, because she’s part of a group of students who are all doing hands-on research and wrestling with tricky questions together; she doesn’t want to let them down. She feels like she’s learning a lot, sure, but she didn’t realize college would be so much work.

Yet Hayley’s experience of the comparative advantage of Cascadia (which is located next to the University of Washington) is borne out by hard data. Although its enrollees typically have less promising academic backgrounds than UW freshman, Cascadia graduates who then continue at UW earn better grades than their peers. It’s hard to imagine a clearer indication that the education students receive at Cascadia is superior.

Indeed, other measures of teaching quality suggest that Cascadia is the best community college in America. Using data from a well-respected survey of educational best practices, the Community College Survey of Student Engagement, the Washington Monthly has created the first-ever list of the nation’s top two-year colleges. (See “America’s Best Community Colleges”) Cascadia places number two overall, and in those measures most closely correlated with high grades and graduation rates -the extent to which teaching is “active and collaborative” - Cascadia tops the list.

Cascadia’s success is extraordinary. But the difference doesn’t depend on funding: the money spent per pupil at Cascadia is typical among community colleges, and about half that spent at the University of Washington. Nor is the college’s achievement the result of some secret formula not known to other educators. Not explaining things and making students work in teams to discover answers turn out to be precisely the kinds of teaching practices that decades of research say help students learn most. Yet the vast majority of four-year colleges and universities don’t teach their undergraduates this way. Instead, they rely far too often on the same old teaching methods nobody thinks are any good.

Unfortunately, there was a problem: the old model turned out to be a terrible way to teach most undergraduates. The standard lecture did little to engage students or push them to do the hard, hands-on work necessary to truly grasp college-level material. The doctoral programs that produced the nation’s college professors offered little or no instruction on the theory or practice of teaching. Instead, they trained and tenured PhDs in narrow areas of scholarship, who were then hired and promoted based wholly on their research, not their aptitude in the classroom.

The sharpest observers realized the mistake in expanding a system ill-suited for its primary mission, educating undergraduates. In 1963_61ya, Clark Kerr, the legendary architect of the California higher education system, delivered a historic lecture series at Harvard where he warned of the “cruel paradox” that “a superior faculty results in an inferior concern for undergraduate teaching.” As he later explained, the emphasis on research and the emphasis on teaching “were not as compatible as we first assumed … the German Humboldt model assumed that teaching is always and in all ways improved by engagement with research. It is not.” The upshot, as Kerr foresaw and others later came to realize, was that “educational policy for undergraduates was neglected.”

About the same time that the great expansion of higher education leveled off in the 1970s, a new wave of researchers studied and defined teaching methods superior to what most undergraduates actually received. Among the most famous was a seminal 1987_37ya paper by researchers Arthur Chickering and Zelda Gamson, “Seven Principles for Good Practice in Undergraduate Education.” Synthesizing years of cognitive science and educational research, Chickering and Gamson mapped out the fundamental principles of effective teaching: The more students actively engage with subject matter, the better they master material and develop critical skills. Undergraduates learn most when they’re asked to solve problems, perform original research, work collaboratively - and receive regular feedback from the professor and their peers. The passive, impersonal lecture turned out to be the worst of all possible worlds.

It’s not often that a giant flaw in a vital public institution is known but almost completely ignored for decades on end. But that’s exactly what’s happened - Clark Kerr’s words ring as true today as they did in 1963_61ya. Chickering and Gamson’s seven principles were published twenty years ago, and now colleges like Cascadia offer proof positive that the ideas work in practice - not just in a class here or there, but college-wide. Yet poor teaching still abounds. As former Harvard President Derek Bok recently said, “Colleges and universities, for all the benefits they bring, accomplish far less for their students than they should.” That institutions built to educate and discover the truth refuse to implement the successful teaching practices that they themselves have discovered is a bitter and consequential irony.

Fixing this won’t be easy. New colleges and universities aren’t built very often, and we can’t just tear down the ones we have and replace them. There’s no reason, moreover, to believe that our institutions of higher education will voluntarily change on their own.

But there is at least one proven way to make many college presidents stand up, take notice, and rapidly implement reforms: alter their reputation in the marketplace. The U.S. News college rankings may be terribly flawed, but they’re undeniably influential. When the magazine began including alumni giving rates in the rankings equation, hundreds of call centers sprang up across the land to start bugging people at dinnertime for donations. If institutional reputations hung on measures of quality teaching, higher education leaders would finally have a strong reason to make the difficult choices they have for decades managed to avoid. Reliable measures of educational excellence for four-year schools do exist, but right now college administrators are the only ones who ever see them. Students and parents need information before they can exert pressure for reform, and Washington should mandate that we all have access to it.

https://philip.greenspun.com/teaching/universities-and-economic-growth:

It made a lot of sense for professors to lecture in the 11th Century. What other means of broadcasting information from 1 person to 100 existed? Printing was very expensive and cumbersome. Having monks make 100 copies of a textbook by hand was not economically feasible.

The university incorporated an important quality control mechanism: student associations paid professors according to how well they taught, how many students were attracted to their lectures, and whether they showed up on time.

It made sense for students to show up to lecture and to do their homework. A student’s lodging might not have been heated. It might make sense to come to lecture simply to get warm. Students in 1088_936ya had no television, no radio, no Internet, no email, no instant messaging, no mobile phone. A student might come to lecture for entertainment.

What about homework? Students in a pre-technological university would do homework either in the library or at home. Both places lacked television, video games, email, etc.

How has this changed the way classes are conducted? We still have lectures and homework, just as in 1088_936ya. What other industry could survive without adopting at least some of the technologies of the last 1000_1,024ya years into its core processes?

Improved technology has rendered the traditional university instructional method far less effective. The student has a warm cozy apartment and will find sleeping late an attractive alternative to attending a lecture (or watching Good Morning America). The student sitting in lecture has some sort of device capable of browsing the Web, sending and receiving text messages, supporting games, displaying photos or video to an adjacent student.

Focusing on homework has become much tougher. A modern dorm room has a television, Internet, YouTube, instant messaging, email, phone, and video games. The students who get the most out of their four years in college are not those who are most able, but rather those with the best study habits.

No company would rely on this system for getting work done, despite the potential savings in having each employee work from home. Companies spend a fortune in commercial office space rent to create an environment with limited distractions and keep workers there for most of each day.

I was asked by Neumont University, a startup for-profit computer science school in Utah, for advice on how to structure the school. They didn’t know how unworkable these ideas were so they adopted most of them. Because their student body consists of kids from middle class families, there is no need for a long summer break for students to join their parents on a grand tour of Europe. Nor need the students take a month off in the winter to yacht around the Caribbean. Simply by being in session Monday through Friday, 8-5, for most of the year, Neumont is able to graduate CS majors in about 2.5 years. That’s 1.5 years in which the kid is not coming home to ask his parents for more money.

Neumont also adopted the idea of making most learning project-based. Neumont freshmen start with substantially lower SAT scores and high school grades than University of Utah freshmen. With brighter and better-prepared students plus a 150-year headstart, how does U. of Utah’s faculty do compared to Neumont’s? The graduate of U of U’s traditional lecture-and-homework CS program will be 1.5 years older than a Neumont graduate, start at a lower salary, and have fewer job offers.

https://philip.greenspun.com/blog/2007_17ya/08/23/improving-undergraduate-computer-science-education/:

How about lectures? You need to broadcast some information to 100 people. Printing was expensive and cumbersome in 1865_159ya. Telephone, television, and Internet did not exist. A lecture was probably indeed the most efficient way of getting some information to a large group, despite the fact that humans can read 3X faster than they can listen. Compare to 2007_17ya, however, when you could simply email a list of those 100 people or provide them with a URL.

• Lecturing has been found to be extremely ineffective by all researchers. The FAA limits lectures to 20 minutes or so in U.S. flight schools.

• Lab and project work are where students learn the most. The school that adopted lab/projects as the core of their approach quickly zoomed to the first position among American undergrad schools of engineering (https://www.olin.edu/).

• Engineers learn by doing progressively larger projects, not by doing what they’re told in one-week homework assignments or doing small pieces of a big project

• Everything that is part of a bachelor’s in CS can be taught as part of a project that has all phases of the engineering cycle, eg. teach physics and calculus by assigning students to build a flight simulator

• It makes a lot of sense to separate teaching/coaching from grading and, in the Internet age, it is trivial to do so. Define the standard, but let others decide whether or not your students have met the standard.

https://philip.greenspun.com/teaching/undergrad-cs https://philip.greenspun.com/teaching/teaching-software-engineering

Summer break badly affects learning and the learning of lower class kids the most https://www.rand.org/pubs/monographs/MG1120_904ya.html https://www.rand.org/content/dam/rand/pubs/monographs/2011/RAND_MG1120_904ya.pdf:

During summer vacation, many students lose knowledge and skills. By the end of summer, students perform, on average, one month behind where they left off in the spring. Of course, not all students experience “average” losses. Summer learning loss disproportionately affects low-income students. While all students lose some ground in mathematics over the summer, low-income students lose more ground in reading, while their higher-income peers may even gain. Most disturbing is that summer learning loss is cumulative; over time, the difference between the summer learning rates of low-income and higher-income students contributes substantially to the achievement gap.

• Cooper, Harris, Kelly Charlton, Jeff C. Valentine, and Laura Muhlenbruck, with Geoffrey D. Borman, Making the Most of Summer School: A Meta-Analytic and Narrative Review, Monographs of the Society for Research in Child Development, Vol. 65, No. 1, Malden, Mass.: Blackwell Publishers, 2000

• Cooper, Harris, Barbara Nye, Kelly Charlton, James Lindsay, and Scott Greathouse, “The Effects of Summer Vacation on Achievement Test Scores: A Narrative and Meta-Analytic Review,” Review of Educational Research, Vol. 66, No. 3, Fall 1996_28ya, pp. 227-268.

• Cooper, Harris, Jeffrey C. Valentine, Kelly Charlton, and April Melson, “The Effects of Modified School Calendars on Student Achievement and on School and Community Attitudes,” Review of Educational Research, Vol. 73, No. 1, Spring 2003_21ya, pp. 1-52.

We modeled our literature search approach after the work of Cooper, Nye, et al (1996_28ya), using keyword searches of computerized reference databases, sifting through reference lists for relevant sources, and leveraging the expertise of education researchers who are leaders in the out-of-school-time (OST) field. Cooper, Nye, et al (1996_28ya) provide a rigorous summary of the early evidence of summer learning loss through an extensive meta-analysis of the research published between 1975_49ya and 1994_30ya. The literature included in Cooper, Nye, et al (1996_28ya) was found through the computerized reference databases ERIC (Education Resources Information Center) and PsychLIT using the following keywords: summer loss, summer vacation, summer break, summer intercession, summer school, and summer variations.

We identified the work by Cooper, Nye, et al (1996_28ya), Heyns (1978_46ya, 1987_37ya), and Entwistle & Alexander 1992_32ya as the foundational studies on summer learning loss. We searched for studies that had referenced these pieces. Google Scholar indexed 294 publications that cited Cooper, Nye, et al (1996_28ya); similar searches were performed for each of the other key articles. We reviewed each indexed publication citing one or more of these key articles for inclusion in our study, considering whether (1) the students represented in the research were between kindergarten and eighth grade and (2) whether summer loss was measured for an academic content area.

We also searched several computerized databases for articles published since 1994_30ya. The databases included in our search were ERIC, JSTOR, ISI Web of Knowledge, and Google Scholar. Wherever possible, we made use of “thesaurus terms,” such as “summer programs,” pairing them with the keywords “loss,” “slide,” or “gap”; ERIC identified 41, 47, and 29 publications, respectively (for a total of 117 citations). Through Google Scholar, we found 69 references that matched a search for “summer program” and “academic achievement” with the same sequence of loss-related keywords, and 23 of these references had been published between 2000_24ya and 2010_14ya. JSTOR indexed 38 articles related to “summer loss,” and the ISI Web of Knowledge was used to find 19 articles that had been published since 2000_24ya. We reviewed the abstracts to determine whether each article contained some information or assessment of summer learning loss and whether it fit our inclusion guidelines. Chapter Two presents a more detailed discussion of the extent of summer learning loss, its cumulative effects, and differences by subject and grade level.

The loss of knowledge and educational skills during the summer months is cumulative over the course of a student’s career and further widens the achievement gap between low- and upper-income students, according to a RAND Corporation study issued today.

The study confirms that students who attend summer programs can disrupt the educational loss and do better in school than peers who do not attend the same programs.

…“It is becoming increasingly clear that the conventional six-hour, 180-day school year is insufficient to give many disadvantaged students the education they deserve,” said Nancy Devine, director of communities at The Wallace Foundation. “This long-awaited and timely RAND study, ‘Making Summer Count,’ confirms the disproportionate impact of the ‘summer slide’ on low-income students, and suggests that high-quality summer learning programs, though challenging to develop, are a promising path forward.”

But “The impact of year-round schooling on academic achievement: evidence from mandatory school calendar conversions”, by McMullen & Rouse 2012_12ya suggests redistributing vacation may not work (undermining the spaced repetition interpretation):

In 2007_17ya, 22 Wake County, NC traditional-calendar schools were switched to year-round calendars, spreading the 180 instructional days evenly across the full year. This paper exploits this natural experiment to evaluate the impact of year-round schooling on student achievement. We estimate a multi-level fixed effects model to separate the impact of year-round schooling from the confounding impacts of other school, family, and individual characteristics. Results suggest year-round schooling has essentially no impact on academic achievement of the average student. Moreover, when the data is broken out by race, we find no evidence that any racial subgroup benefits from year-round schooling.

…Proponents of YRS calendars argue that they are beneficial to students because they help alleviate human capital loss during the long summer break (“summer learning loss”). Supporters further contend that the long break is particularly harmful for low-income, low-performing students who are less able to afford supplemental learning opportunities in the summer (Von Drehle, 2010_14ya). These assertions are largely supported by a wide literature on summer learning loss, which has found that student achievement stagnates over the summer, and that for low achieving and disadvantaged students especially, achievement can often decline while not in school (Cooper et al 1996_28ya; Jamar 1994_30ya; Alexander et al 2007_17ya).5 Alexander et al 2007_17ya finds that by the end of ninth grade, almost two-thirds of the socioeconomic achievement gap can be explained by differential summer learning loss. It is important to note, however, that the ability of YRS to address this problem depends crucially upon the nature of the human capital accumulation process. In this paper, we present a simple model that illustrates YRS can only improve achievement if learning loss accelerates with the number of days out of school or if there are diminishing returns to learning.⁶ Thus, even if disadvantaged students lose more human capital than their wealthier counterparts over summer, YRS cannot alleviate the problem unless there are specific non-linearities in the human capital process. If YRS acts largely as a remedy for summer learning loss, the impact should be no greater than the documented negative impact of a summer vacation away from school, which is rarely larger than a loss of 0.1 standard deviations of student achievement per year, and often close to zero (Downey et al 2004_20ya; Cooper et al 1996_28ya).

Our study adds to a body of literature, primarily coming from outside of the field of economics, that is well-summarized by the meta-analysis performed by Cooper et al 2003_21ya. The general consensus coming out of that review is that the impact of year-round education on student achievement is, on average, nearly negligible. On the other hand, the evidence suggests the modified calendar does benefit low performing and economically disadvantaged students. McMillan (2001_23ya) finds similar results using a cross-sectional dataset from North Carolina. The primary drawback of these early studies is their failure to account for non-random student and school selection. The studies included in Cooper et al 2003_21ya do not adequately control for student and school characteristics, and none attempt to control for both unobserved student and school heterogeneity. Cooper et al 2003_21ya thus concludes that it “would be difficult to argue with policymakers who choose to ignore the existent database because they feel that the research designs have been simply too flawed to be trusted (p. 43).” …Most recently, Graves (2010_14ya) uses detailed longitudinal school-level data from California to estimate the impact of the multi-track year-round calendar on academic achievement. By including school fixed effects and school-specific time trends, Graves is able to mitigate concerns over non-random year-round calendar implementation. In contrast to much of the prior research on YRS, Graves finds achievement in multi-track year-round schools is 1 to 2 percentile points lower than that in traditional calendar schools. However, without student-level data, she is not able to control for non-random student selection into YRS or to estimate the impacts separately by race.

…Results from these models are presented in columns (c) and (f) of Table 6. The results from these model specifications tell a very different story. In contrast to the results reported in columns (a), (b), (d) and (e), which indicate a positive, statistically significant impact of YRS on both math and reading test score levels and growth, estimates from columns (c) and (f) imply that YRS has essentially no impact on either math or reading achievement. Nearly every estimate is close to zero in magnitude and in all growth models is slightly negative. The primary conclusion coming out of these models is that failure to control for school-level unobserved heterogeneity leads to estimates that largely over-state the impact of YRS on student achievement. This is consistent with the hypothesis that year-round schools are placed in high-growth and possibly high achieving areas.

Homework is wasted? “Are we wasting our children’s time by giving them more homework?”

Following an identification strategy that allows us to largely eliminate unobserved student and teacher traits, we examine the effect of homework on math, science, English and history test scores for eighth grade students in the United States. Noting that failure to control for these effects yields selection biases on the estimated effect of homework, we find that math homework has a large and statistically meaningful effect on math test scores throughout our sample. However, additional homework in science, English and history are shown to have little to no impact on their respective test scores.

It is a sad story indeed when the astonishing linguistic incapacity of U.S. military forces and intelligence organizations is contrasted with the abundance of American civilians who speak all known foreign languages, and the brilliant record of foreign-language education in the U.S. Army and Navy, which used to produce as many good Chinese and Japanese speakers as they wanted by selecting for natural aptitude in the recruit pool, giving them a year of intensive courses (eight hours a day, six days a week), and quickly sending away those who failed to keep up with their classes. Nothing prevents the military from doing the same for Arabic, Persian and, say, Azeri now, except for an unwillingness to invest in the future, and probably a lack of disciplined volunteers willing to learn a language eight hours a day, six days a week, for a whole year or more.

https://harpers.org/archive/2007_17ya/02/0081384/

Public school teachers are in much the same position as prison wardens. Wardens’ main concern is to keep the prisoners on the premises. They also need to keep them fed, and as far as possible prevent them from killing one another. Beyond that, they want to have as little to do with the prisoners as possible, so they leave them to create whatever social organization they want. From what I’ve read, the society that the prisoners create is warped, savage, and pervasive, and it is no fun to be at the bottom of it….If I could go back and give my thirteen year old self some advice, the main thing I’d tell him would be to stick his head up and look around. I didn’t really grasp it at the time, but the whole world we lived in was as fake as a Twinkie. Not just school, but the entire town. Why do people move to suburbia? To have kids! So no wonder it seemed boring and sterile. The whole place was a giant nursery, an artificial town created explicitly for the purpose of breeding children. And as for the schools, they were just holding pens within this fake world. Officially the purpose of schools is to teach kids. In fact their primary purpose is to keep kids locked up in one place for a big chunk of the day so adults can get things done. And I have no problem with this: in a specialized industrial society, it would be a disaster to have kids running around loose. What bothers me is not that the kids are kept in prisons, but that (a) they aren’t told about it, and (b) the prisons are run mostly by the inmates. Kids are sent off to spend six years memorizing meaningless facts in a world ruled by a caste of giants who run after an oblong brown ball, as if this were the most natural thing in the world. And if they balk at this surreal cocktail, they’re called misfits.

–Paul Graham, “Why Nerds Are Unpopular”

Spaced Repetition

Spaced repetition is not in use in school systems except accidentally or by primitive experience & intuition. Even though tremendous amounts of lower education is memorization (and some areas - like foreign languages - are almost entirely based on memorization), and the principle has been scientifically proven for literally centuries.¹

School Hours

That school schedules are so perverse in a way obvious to any teenager - the people who need to get up last have to get up first - is one of the best proofs, I feel, that quality education is only a secondary goal of public education.

https://www.overcomingbias.com/p/do-larks-repress-owlshtml

Ask any current college student: many dread 8 AM classes. If you’re a lark, try looking in on some 8 AM classes at 8:30 or 9 or so, and see how many of the students struggle to pay attention or stay awake. You don’t see very many faces flat on the desk in afternoon classes… https://www.overcomingbias.com/p/do-larks-repress-owlshtml#comment-443555

In my school, there were two offered justifications for the obscene 7:40 starting time:

1. Students have to travel for extracurricular activities, such as sports, scheduled in the early afternoon. If the school starts later, they have to miss class to participate in them.

2. Students prefer to leave school earlier so they can get to their jobs.

Both prompt a big WTF from me.

https://www.overcomingbias.com/p/do-larks-repress-owlshtml

monophasic sleep highly artificial: https://historycooperative.org/a-short-history-on-sleep-before-the-industrial-revolution/

Stampi quote on primitives not monophasic: https://web.archive.org/web/20110101132820/http://blog.myzeo.com/forum/polyphasic-sleep-experiment-discussion/evolutionary-sleep-more-polyphasic/

we revert to biphasic naturally: https://www.jdmoyer.com/2010_14ya/03/04/sleep-experiment-a-month-with-no-artificial-light/

hard to make up deficits: http://www.futurepundit.com/archives/007378.html

kids who go to bed late the smarter ones: https://www.quora.com/Are-night-owls-generally-more-intelligent-than-other-people (so early school hurts those who could benefit most from schooling) https://www.dailymail.co.uk/sciencetech/article-1173028/How-night-owls-cleverer-richer-people-rise-early.html

lack of sleep harmful: http://ngm.nationalgeographic.com/print/2010_14ya/05/sleep/max-text

later school hours help grades: http://news.bbc.co.uk/2/hi/uk_news/education/8579951.stm drop-out rates, savings (!), easy schedule change, parental support etc.: “High Schools Starting Later to Help Sleepy Teens” and help mood, health, and sleep quality: https://jamanetwork.com/journals/jamapediatrics/fullarticle/383436

correlation: >9 hours is best for 6-7-year-olds?

highly durable circadian rhythm changes: https://www.thefreelibrary.com/Sleepy+teens+haven%27t+got+circadian+rhythm.-a0134623686 https://www.sciencenews.org/article/sleepy-teens-havent-got-circadian-rhythm

TODO: what was on http://blog.myzeo.com/back-to-school-sleep-college-edition/ ?

By the time they enter sixth grade, many middle-class children sleep so little during the school week that daytime drowsiness may compromise their ability to pay attention and learn, a new study suggests.

This situation derives from a combination of factors, say psychologist Avi Sadeh of Tel Aviv University in Israel and his colleagues. Children tend to fall asleep at increasingly later times as they move from the second to the sixth grade, while continuing to be awakened at the same time for school.

“Our study suggests that the sleep behavior of the older children may not be in accordance with their physiological needs,” they contend. “These children are thus at risk of being chronically sleep deprived.”

What’s more, Sadeh’s team finds that nearly 20 percent of kids in the second, fourth, and sixth grades have serious sleep problems that typically aren’t perceived by either the children or their parents. In the study, sleep disturbances consisted of regularly being awake for at least 10 percent of the night after falling asleep or waking three or more times during the night for at least 5 minutes each time.

Children fell asleep at later times as they got older regardless of their supposed bedtimes. Sixth graders drifted into slumber slightly more than 1 hour after second graders did and about 25 minutes after fourth graders did, the researchers report in the May DEVELOPMENTAL PSYCHOLOGY. Sleep quality, such as the number of night wakenings and length of sleep periods, was similar at all grades.

shallow sleep linked with worse glucose processing & diabetes: https://www.sciencenews.org/view/generic/id/9302/title/Sleep_disruption_and_glucose_processing

little sleep linked with obesity (elementary school): https://www.sciencenews.org/article/too-little-sleep-may-fatten-kids

A modest but constant sleep shortage undermines alertness and other mental faculties in a matter of days, according to according to a report in the March 15 Sleep. Moreover, people who get by on a modest sleep deficit are often not aware of their shrinking thinking capabilities and don’t feel particularly drowsy drowsy, say Hans P.A. Van Dongen https://www.thefreelibrary.com/Sleep+debt+exacts+deceptive+cost.+%28Behavior%29.-a0100110931 https://www.sciencenews.org/article/sleep-debt-exacts-deceptive-cost

“Job Market Paper: Early to Rise? The Effect of Daily Start Times on Academic Performance”, Edwards 2012_12ya:

This paper uses data on all middle school students in Wake County, NC from 1999^–₇2006_18ya to study the impact of start times on academic performance. … The differences in start time across schools is generated by bus scheduling concerns, while the differences within schools are driven by population growth. … I find that a one hour later start time increases standardized test scores on both math and reading test by three percentile points. Since start times may be correlated with other determinants of test scores, I also estimate the effect using only variation in start times within schools over time and find a two percentile point improvement. The effect of start times on academic performance is robust to different specifications and sources of variation. The magnitude of the effect is similar to the difference in test scores for one additional year of parental education.

The impact of later start times on test scores is persistent. Conditional on a high school fixed effect, a one hour later start time in grade eight is associated with an increase in test scores in grade ten similar in magnitude to the increase in grade eight. … The impact of start times is greatest in grade eight (who are more likely to have begun puberty than those in the sixth or seventh grade). … Students who begin school later have fewer absences and spend more time on homework each week. … Over the seven years examined in this paper, [this school district] grew from 20,530 student enrolled in twenty-two middle schools … to 27,686 students enrolled in twenty-eight middle schools

practical challenges to shifting school times: https://www.ithacacityschools.org/ review of research: http://www.icsd.k12.ny.us/redistricting/startend/starttimesummary.pdf

overview: http://www.icsd.k12.ny.us/redistricting/startend/starttimesummary.pdf list of some schools who have changed: http://www.icsd.k12.ny.us/redistricting/startend/StartTimeChanges-v3.pdf

Attendance rates for all students in grades 9, 10, and 11 in the district have shown to have improved statistically significantly in the years from 1995^–₅2000_24ya. The greatest rate of improvement is for 9th grade students, where the daily rate of attendance went from 83%-87% after the later start was initiated. The probability that this would occur by chance is less than one in a thousand.

…Given the numerous obstacles to obtaining “clean data”, it required nearly a year of time to conduct this analysis. The ultimate findings from the analysis of the letter grades earned by students in grades 9-12 in the three years prior to the change (starting time of 7:15 AM) versus the grades earned in the three years after the change (starting time of 8:40 AM) reveal a slight improvement in grades earned overall, but the differences were not statistically significant. A finding from this time-consuming and intensive data analysis is that the difficulty of making comparisons and subsequent judgments is likely to be a problem for any district attempting to judge the efficacy of a change using the letter grades earned as the primary indicator.

…Minneapolis high school students continue to get an hour’s more sleep each school night or obtain five more hours’ sleep per week than students whose high schools begin an hour earlier than Minneapolis schools. This finding supports the medical researchers’ finding that nearly all teenagers become sleepy at about 11:00 PM. It also lays to rest the fears and expectations that a later start would mean that Minneapolis students would just end up staying up an hour later on school nights.

https://www.cehd.umn.edu/CAREI/Reports/summary.html#SchoolStart, from executive summary

[abstract] In the early 1990s, medical research found that teenagers have biologically different sleep and wake patterns than the preadolescent or adult population. On the basis of that information, in 1997_27ya the seven comprehensive high schools in the Minneapolis Public School District shifted the school start time from 7:15 a.m. to 8:40 a.m. This article examines that change, finding significant benefits such as improved attendance and enrollment rates, less sleeping in class, and less student-reported depression. Policy implications are briefly discussed, acknowledging this to be a highly charged issue in school districts across the United States.

…[page 10] Before the later school start time was instituted, many parents and administrators expressed a fear that students would merely use the later morning start time as an excuse to stay up an hour later on school nights. The data, however, show that this did not happen. Students continued to go to bed at the same time (approximately 15 minutes before 11 p.m.). This finding makes sense from a biological perspective, as it is likely that nighttime circadian rhythms were contributing to feelings of sleepiness around 11 p.m, regardless of what time the students woke up in the morning. Minneapolis students slept about an hour more each school night (due to the later school start time) than their peers whose school began at 7:30 a.m.

…[page 12] The students whose high schools started at 8:30 a.m. or later reported statistically significant less depressive feelings on those questions than did the early start students (p ranged from < .05 to < .001.)

Similarly, scores on questions measuring daytime sleepiness, the struggle to stay awake in class, and sleepiness while doing homework all showed statistically significant better outcomes for the students whose school day started later. For example, students in late-start schools reported being less likely to arrive late to class because of oversleeping, or to fall asleep in a morning or afternoon class, or to feel sleepy while taking a test. They also reported statistically significant fewer feelings of sleepiness when at a computer, reading, or studying.

…Many of the benefits of the later start time were similar for both urban and suburban students, with their actual scores being nearly identical despite the differences in their local economic conditions. Again, if the need for and the benefits of more sleep are a biological phenomenon of the human body during the adolescent years, then one would expect those kinds of results, which are not related in any way to socioeconomic status.

“Up All Night: The science of sleeplessness”, New Yorker:

According to Roenneberg, age also has a big influence on chronotype. Toddlers tend to be larks, which is why they drive their parents crazy by getting up at sunrise. Teen-agers are owls, which is why high schools are filled with students who look (and act) like zombies. Roenneberg advocates scheduling high-school classes to begin later in the day, and he cites studies showing that schools that delay the start of first period see performance, motivation, and attendance all increase. (A school district in Minnesota that switched to a later schedule found that the average S.A.T. scores for the top ten per cent of the class rose by more than two hundred points, a result that the head of the College Board called “truly flabbergasting.”) But, Roenneberg notes, teachers and school administrators generally resist the change, preferring to believe that the problem is insoluble.

“Changing Times: Findings From the First Longitudinal Study of Later High School Start Times”

Between childhood and adulthood, we go through puberty and adolescence. While the end of puberty is defined as the point of cessation of bone growth (epiphyseal closure; girls: 16 y; boys: 17.5 y), the end of adolescence (∼19 y) is defined less clearly, by a mixture of physical, psychological, social, and mental measures [[1]]. One conspicuous property of adolescence is the apparently unsaturable capacity to stay up late and to sleep in. Investigating ‘chronotypes’ we observed an abrupt change in the timing of sleep at around the age of 20 and propose this change as the first biological marker of the end of adolescence.

https://www.cell.com/current-biology/fulltext/S0960-9822(04)00928-5

“Most studies show a fairly consistent 9 1/4 hours sleep requirement,” says Emsellem. “So there’s a huge gap between what they’re getting on an average school night and what they require.”

An adolescent’s biology bears some of the blame for this sleep problem. As teens progress through puberty, unprecedented growth occurs in body and brain that requires a lot of sleep.

In addition, something else is changing: The very brain chemical that makes one feel sleepy - a hormone called melatonin - is released later and later in the evening as teens get older.

Because of this shift in the onset of melatonin, teenagers don’t feel sleepy until later at night, says Stephanie Crowley, a sleep researcher at Rush University Medical Center in Chicago.

“A 16- or 17-year-old might be able to stay awake later compared to a 10-year-old who will likely fall asleep on the couch watching TV,” Crowley explains.

…But this sleep-wake pattern makes things worse for the teen, not better, Emsellem says.

“Even if you catch up by sleeping in late on your weekend mornings,” she says, “by doing so, it makes it harder for you to fall asleep by 10 or 10:30 on Sunday night. And you start all over again, sleep restricted.”

https://www.npr.org/2011_13ya/05/16/136275658/late-to-bed-early-to-rise-makes-a-teen-sleepy

The 18 studies were performed in different contexts. Five studies were done only on school days,53,58,60-62 3 studies were performed on week ends or during summer time,54,52,59 and 2 studies included both school days and nonschool days.57,64 Eight studies did not specify the time of the year.

…In children and adolescents, the relation between age and TST [total sleep time] was moderated by the recording methods; studies that used in-laboratory PSG found significantly larger correlations than those using actigraphy (z statistic for contrast: -7.92; P < .0001). Similarly, the relation between age and TST was moderated by the time of recording. Studies that took place during school days (z statistic for contrast: -7.60; P < .0001) had larger correlations than those that were done on nonschool days. The results showed that TST decreased with age only when recordings took place on school days. On nonschool days, TST remained the same from childhood to the end of adolescence.

“Meta-Analysis of Quantitative Sleep Parameters From Childhood to Old Age in Healthy Individuals: Developing Normative Sleep Values Across the Human Lifespan” 2004

In response to recent sleep studies with students, Battle Ground schools will start and end the school day 30 minutes later beginning this fall. Multiple experts and studies show teenagers need more sleep at night to be successful learners in the morning. Kyla Wahlstrom, director of the Center for Applied Research and Education Improvement in St. Paul, Minn., says, “From the onset of puberty until late teen years, the brain chemical melatonin, which is responsible for sleepiness, is secreted from approximately 11 p.m. to 8 a.m.”

https://www.kptv.com/story/15084456/battle-ground-schools-let-students-sleep-in/

Recent sleep research finds that many adolescents are sleep-deprived because of both early school start times and changing sleep patterns during the teen years. This study identifies the causal effect of school start time on academic achievement by using two policy changes in the daily schedule at the US Air Force Academy along with the randomized placement of freshman students to courses and instructors. Results show that starting the school day 50 minutes later has a significant positive effect on student achievement, which is roughly equivalent to raising teacher quality by one standard deviation.

“A’s from Zzzz’s? The Causal Effect of School Start Time on the Academic Achievement of Adolescents”

“To Study or to Sleep? The Academic Costs of Extra Studying at the Expense of Sleep”, Gillen-O’Neel et al 2012_12ya:

This longitudinal study examined how nightly variations in adolescents’ study and sleep time are associated with academic problems on the following day. Participants (N=535, 9th grade Mage=14.88) completed daily diaries every day for 14 days in 9th, 10^th, and 12th grades. Results suggest that regardless of how much a student generally studies each day, if that student sacrifices sleep time in order to study more than usual, he or she will have more trouble understanding material taught in class and be more likely to struggle on an assignment or test the following day. Because students are increasingly likely to sacrifice sleep time for studying in the latter years of high school, this negative dynamic becomes increasingly prevalent over time.

…Socializing with peers and working for pay, for example, both increase across the course of high school (Shanahan & Flaherty, 2001_23ya; Wight, Price, Bianchi, & Hunt, 2009_15ya). As they advance through high school, adolescents’ academic obligations also intensify and often require more time and effort (Eccles, et al 1993_31ya). As a result, many high school students end up with irregular study schedules, often facing nights in which they need to spend substantially more time than usual studying or completing school work.

• Shanahan, M. J., & Flaherty, B. P. (2001_23ya). Dynamic patterns of time use in adolescence. Child Development, 72, 385-401

• Wight et al 2009_15ya, “The time use of teenagers”. Social Science Research, 38, 792-809

• Eccles, J. S., Midgley, C., Wigfield, A., Buchanan, C. M., Reuman, D., Flanagan, C., et al (1993_31ya). Development during adolescence: The impact of stage-environment fit on young adolescents’ experiences in schools and in families. American Psychologist, 48, 90-101. doi: 10.1037_987ya/0003-066x.48.2.90

…In high school, sacrificing sleep to study may be especially problematic because, in general, high-school age adolescents are chronically sleep-deprived (Carskadon, 1990_34ya). Although the optimal amount of sleep varies somewhat across individuals, most adolescents need just over 9 hours of sleep each night (Wolfson & Carskadon, 1998_26ya). Only about 9% of high school students, however, sleep for at least the requisite 9 hours per night (National Sleep Foundation, 2006_18ya). One-fourth of high school students get a borderline amount of sleep (between 8 and 9 hours per night), and the vast majority of high school students (62%) get insufficient sleep (less than 8 hours per night; National Sleep Foundation, 2006_18ya).

• Carskadon, M. A. (1990_34ya). Patterns of sleep and sleepiness in adolescents. Pediatrician, 17, 5-12.

• Wolfson, A. R., & Carskadon, M. A. (1998_26ya). Sleep schedules and daytime functioning in adolescents. Child Development, 69(4), 875-887. doi: 10.2307/1132351

• National Sleep Foundation. (2006_18ya). Sleep in America poll. Retrieved February 18, 2011_13ya, from http://www.sleepfoundation.org

…Across the course of high school, the biologically-needed hours of sleep remain constant, yet the average amount that students sleep declines (Carskadon, Acebo, & Jenni, 2004_20ya). In 9th grade, the average adolescent sleeps for 7.6 hours per night, and this time decreases to 7.3 hours in 10^th grade, 7.0 hours in 11th grade, and 6.9 hours in 12th grade (National Sleep Foundation, 2006_18ya). Thus, adolescents start high school sleeping for fewer hours than they need, and this sleep deprivation worsens over the course of high school (Fukuda & Ishihara, 2001_23ya).

• Carskadon, M. A., Acebo, C., & Jenni, O. G. (2004_20ya). Regulation of adolescent sleep: Implications for behavior. Annals of the New York Academy of Sciences, 1021_1,003ya, 276-291. doi: 10.1196_828ya/annals.1308_716ya.032

• Fukuda, K., & Ishihara, K. (2001_23ya). Age-related changes of sleeping pattern during adolescence. Psychiatry and Clinical Neurosciences, 55, 231-232

…this troublesome association becomes even stronger in 12th grade. The association between [more] study time and [more] academic problems occurs regardless of whether or not students have a test coming up, and therefore, is not simply an artifact of studying for and taking a difficult test. Although we expected that nights of extra studying might not be as effective as students suppose (Pilcher & Walters, 1997_27ya), it was somewhat surprising that nights of extra studying would be associated with worse academic functioning the following day. This surprising finding, however, made more sense once we examined extra studying in the context of adolescents’ sleep. As other studies have found, our results indicate that extra time spent studying cuts into adolescents’ sleep on a daily basis (Adam et al 2007_17ya). This tradeoff between studying and sleeping occurs in 9th grade and becomes more dramatic in the latter years of high school. Our mediation results suggest that the reduced sleep that tends to occur on nights of extra studying is what accounts for the increase in academic problems that occurs the next day.

• Adam, E. K., Snell, E. K., & Pendry, P. (2007_17ya). Sleep timing and quantity in ecological and family context: A nationally representative time-diary study. Journal of Family Psychology, 21, 4-19. doi: 10.1037_987ya/0893-3200.21.1.4

• Adam, E. K., Snell, E. K., & Pendry, P. (2007_17ya). Sleep timing and quantity in ecological and family context: A nationally representative time-diary study. Journal of Family Psychology, 21, 4-19. doi: 10.1037_987ya/0893-3200.21.1.4

“Total Sleep Time Severely Drops during Adolescence”, Leger et al 2012_12ya: g > Data from 9,251 children aged 11 to 15 years-old, 50.7% of which were boys, as part of the cross-national study 2011_13ya HBSC were analyzed…A serious drop of TST [total sleep time] was observed between 11 yo and 15 yo, both during the schooldays (9 hours 26 minutes vs. 7 h 55 min.; p < 0.001) and at a lesser extent during week-ends (10 h 17 min. vs. 9 h 44 min.; p < 0.001). Sleep deprivation concerned 16.0% of children aged of 11 yo vs. 40.5% of those of 15 yo (p < 0.001). Too short sleep was reported by 2.6% of the 11 yo vs. 24.6% of the 15 yo (p < 0.001).

“Adolescent sleep and fluid intelligence performance”, Johnstone et al 2010_14ya; abstract:

Fluid intelligence involves novel problem-solving and may be susceptible to poor sleep. This study examined relationships between adolescent sleep, fluid intelligence, and academic achievement. Participants were 217 adolescents (42% male) aged 13 to 18 years (mean age, 14.9 years; SD = 1.0) in grades 9-11. Fluid intelligence was predicted to mediate the relationship between adolescent sleep and academic achievement. Students completed online questionnaires of self-reported sleep, fluid intelligence (Letter Sets and Number Series), and self-reported grades. Total sleep time was not significantly related to fluid intelligence nor academic achievement (both p > 0.05); however, sleep difficulty (eg. difficulty initiating sleep, unrefreshing sleep) was related to both (P < 0.05)…

“Dissimilar Teen Crash Rates in Two Neighboring Southeastern Virginia Cities with Different High School Start Times”, Vorona et al 2011

VB high schools start 75-80 minutes earlier than Chesapeake’s. We hypothesized that VB teens would manifest a higher crash rate than Chesapeake teens…For VB and Chesapeake, teen drivers’ crash rates in 2008_16ya were 65.8/1,000 and 46.6/1,000 (p < 0.001), respectively, and in 2007_17ya were 71.2/1,000 and 55.6/1,000. Teen drivers’ crash peaks in the morning occurred one hour earlier in VB than Chesapeake, consistent with school commute time. Congestion data for VB and Chesapeake did not explain the different crash rates….Based on our own 2008_16ya results, we estimate that 16 crashes could be prevented yearly if the crash rate in Virginia Beach approximated the rate in Chesapeake…We did not assess crash severity. Future studies might explore whether the earlier high school start times in Virginia Beach were also related to vehicular crashes marked by increased injury or mortality rates.

…Early high school start times could contribute to insufficient sleep in teenagers¹⁰ and increased motor vehicle crashes. One study found start time to be the main determinant of wake times in adolescents.¹¹ A recent study revealed that a 30-min delay in high school start time was associated with 45 min of additional sleep on weekday nights and reduced sleepiness.¹²…previous data indicate that earlier rise times in teens are not correlated with earlier bedtimes.³…One recent study by Danner and Phillips did demonstrate that delaying high school start times reduced vehicle crashes in teens. In Lexington Kentucky, a 1-h delay in high school start times was associated with a 16.5% decline in teen crashes in the ensuing 2 years.¹³ 3,10-13

• \3. Carskadon MA, Wolfson AR, Acebo C, Tzischinsky O, Seifer R. Adolescent sleep patterns, circadian timing, and sleepiness at a transition to early school days. Sleep 1998_26ya;21:871-81

• \10. Hansen M, Janssen I, Schiff A, Zee PC, Dubocovich ML. The impact of school daily schedule on adolescent sleep. Pediatrics 2005_19ya;115:1555-61.

• \11. Knutson KL, Lauderdale, DS. Sociodemographic and behavioral predictors of bed time and wake time among US adolescents aged 15 to 17 years. J Pediatr 2009_15ya;154:426-30.

• \12. Owens JA, Belon K, Moss P. Impact of delaying school start time on adolescent, sleep, mood and behavior. Arch Pediatr Adolesc Med 2010_14ya;164:608-14.

• \13. Danner F, Phillips B. Adolescent sleep, school start times, and teen motor vehicle crashes. J Clin Sleep Med 2008_16ya;4:533-5.

TODO: how many such teen crashes are fatal?

“Morningness-eveningness and educational outcomes: The lark has an advantage over the owl at high school”, Preckel et al 2013_11ya:

…Sample. The sample comprised 272 students attending 9th and 10^th grades at five German high schools. Data was also obtained from 132 parents of these students. Method. Students were assessed in class via self-report questionnaires and a standardized cognitive test. Parents filled out a questionnaire at home. The incremental validity of chronotype was investigated using hierarchical linear regression. Validity of the chronotype questionnaire was assessed by correlating student ratings of their chronotype with behavioral data on sleep, food intake, and drug consumption and with parent ratings of chronotype.

Results. Eveningness was a significant (negative) predictor of overall grade point average (GPA), math-science GPA, and language GPA, after cognitive ability, conscientiousness, need for cognition, achievement motivation, and gender were held constant. Validity evidence for the chronotype measure was established by significant correlations with parent-ratings and behavioral data.

…Recent research has documented statistically meaningful relationships between chronotype and academic performance and demonstrated that eveningness and academic performance are negatively related, whereas morningness and academic performance are positively related (eg. Giannotti, Cortesi, Sebastiani, & Ottaviano, 2002_22ya; Kirby & Kirby, 2006_18ya; Randler & Frech, 2006_18ya; for a meta-analysis see Preckel, Lipnevich, Schneider, & Roberts, 2011_13ya)…Children usually have elevated morningness relative to other age groups. During adolescence a delay of phase preference is usually observed (Carskadon, Wolfson, Acebo, Tzischinsky, & Seifer, 1998_26ya; Crowley, Acebo, & Carskadon, 2007_17ya) reaching a maximum in ‘lateness’ at around the age of 20 (Roenneberg et al 2004_20ya). After the age of 50, studies document a fast increase in morningness (Diaz-Morales & Sorroche, 2008_16ya; Roenneberg et al 2007_17ya). In regards to gender, it appears that women tend to have more morning characteristics than men (see Kerkhof, 1985_39ya; Tankova, Adan, & Buela-Casal, 1994_30ya for reviews). Thus, a recent meta-analysis suggests a weak but significant effect of gender on morningness consistent with this assertion (Randler, 2007_17ya).

…Researchers consistently show that eveningness and academic performance are strongly and inversely related, whereas morningness and performance in school are positively related. These patterns hold for both school children (Giannotti, Cortesi, & Ottaviano, 1997_27ya; Giannotti et al 2002_22ya; Randler & Frech, 2009_15ya) and university students (eg. Besoluk, 2011_13ya; Besoluk, Onder, & Deveci, 2011_13ya; Randler & Frech, 2006_18ya). Preckel et al.’s (2011_13ya) meta-analysis also found small but significant and homogenous correlations between morningness and academic achievement (r = .16, 13 studies) and eveningness and academic achievement (r = −.14, 6 studies).

…Students with a proclivity towards eveningness are likely to collect sleep debts over the week (Gau & Soong, 2003_21ya; Gau et al 2004_20ya): Students with a proclivity towards eveningness go to bed later than students with a proclivity towards morningness but they all have to get up at the same time due to the school schedule. Therefore, students with a proclivity towards eveningness report greater daytime sleepiness, which is by itself associated with lower school achievement (Kirby & Kirby, 2006_18ya; Meijer, 2008_16ya).

…Klein (2004_20ya) investigated 850 seventh to ninth grade students in Israel and found a gradual increase for the level of academic performance from the morning to the afternoon hours. Similarly, Wahlstrom (2002_22ya) conducted a 4-year longitudinal study of the impact of changing start time in seven comprehensive high schools from 7:20 a.m. to 8:40 a.m., without changing the length of the school day. The study revealed that students gained an hour’s more sleep each school night, with improvements related to daytime sleepiness and attendance. Grades improved, but not significantly. School districts in other US states have implemented similar start time changes, with similar and consistently positive outcomes (Fairfax County School Board Transportation Task Force, 2008_16ya). However, starting school later is often not possible due to the organization of the work life in our society and concerns related to student participation in extracurricular activities.

“Extra hour in bed ‘boosts pupils’ exam results’: A school experiment that allowed pupils an extra hour in bed has boosted exam results and slashed absenteeism”, The Telegraph 21 Mar 2010_14ya (TODO followup):

The usual 9am start for lessons at Monkseaton High School, North Tyneside, was pushed back to 10am as part of the trial…“The data are only preliminary but show lateness has dropped eight per cent and long-term absence 27 per cent because of the changes to the start of the school day. Our GCSE results in maths and English in January are significantly improved on the scores in January 2009_15ya,” Paul Kelley, the headmaster, told The Sunday Times…Prof Foster and other academics found that teenagers have a biological predisposition to go to bed late and get up late, and may not begin to function fully until 10am, two to four hours later than adults. The research also suggests that the most difficult lessons should take place in the afternoon, when pupils will be at their most alert.

“Lack of sleep blights pupils’ education” (see also http://timss.bc.edu/timss2011_13ya/ ):

Other countries with the most sleep-deprived youngsters were New Zealand, Saudi Arabia, Australia, England, Ireland and France. High-performing Finland is also among the most lacking in sleep…The researchers uncovered regional trends that bucked expectations. Asian countries are the highest-performing in maths tests - and Mr Minnich says this has often been associated with long hours and cramming in after-school classes. “One would assume that they would be extremely tired,” he said. “And yet when we look at the sleep factor for them, they don’t necessarily seem to be suffering from as much sleep deprivation as the other countries.”…There are also big changes as pupils get older. Younger pupils in South Korea have among the lowest levels of sleep deprivation in the world, but in secondary school they have some of the worst problems.

“The Science of Sleepy Teenagers: School schedules make them grouchy, impulsive, and humorless”, New Scientist

So what are the facts about teenage slumber, and how should society adjust to these needs? The biology of human sleep timing, like that of other mammals, changes as we age. This has been shown in many studies. As puberty begins, bedtimes and waking times get later. This trend continues until 19.5 years in women and 21 in men. Then it reverses. At 55 we wake at about the time we woke prior to puberty. On average this is two hours earlier than adolescents. This means that for a teenager, a 7 a.m. alarm call is the equivalent of a 5 a.m. start for people in their 50s. Precisely why this is so is unclear, but the shifts correlate with hormonal changes at puberty and the decline in those hormones as we age. However, biology is only part of the problem. Additional factors include a more relaxed attitude to bedtimes by parents, a general disregard for the importance of sleep, and access to TVs, DVDs, PCs, gaming devices, cellphones, and so on, all of which promote alertness and eat into time available for sleep.

…Mary Carskadon at Brown University, who is a pioneer in the area of adolescent sleep, has shown that teenagers need about nine hours a night to maintain full alertness and academic performance. My own recent observations at a U.K. school in Liverpool suggested many were getting just five hours on a school night. Unsurprisingly, teachers reported students dozing in class. Evidence that sleep is important is overwhelming. Elegant research has demonstrated its critical role in memory consolidation and our ability to generate innovative solutions to complex problems. Sleep disruption increases the level of the stress hormone cortisol. Impulsive behaviors, lack of empathy, sense of humor, and mood are similarly affected.

“Melatonin Treatment Effects on Adolescent Students’ Sleep Timing and Sleepiness in a Placebo-Controlled Crossover Study”, Eckerberg et al 2012_12ya:

During the last few decades, the incidence of sleep-onset insomnia, due to delay of circadian phase, has increased substantially among adolescents all over the world. We wanted to investigate whether a small dose of melatonin given daily, administered in the afternoon, could advance the sleep timing in teenagers. 21 students, aged 14-19 yrs, with sleep-onset difficulties during school weeks were recruited. The study was a randomized, double blind, placebo (PL)-controlled crossover trial, lasting 5 weeks…Post hoc analysis showed that reported sleep-onset times were advanced after melatonin school weeks compared with PL school weeks (p<.005) and that sleep length was longer (p<.05). After the last melatonin school week, the students fell asleep 68 min earlier and slept 62 min longer each night compared with the baseline week. Morning melatonin values in saliva diminished compared with PL (p<.001) and evening values increased (p<.001), indicating a possible sleep phase advance. Compared with PL school weeks, the students reported less wake up (p<.05), less school daytime sleepiness (p<.05) and increased evening sleepiness (p<.005) during melatonin weeks. We conclude that a small dose of melatonin given daily, administered in the afternoon [4-6:30PM], could advance the sleep timing and make the students more alert during school days even if they continued their often irregular sleep habits during weekends…Only two participants noted tiredness [side effects] 1 h after melatonin intake and one noted headache a few times following melatonin intake…In a short inquiry form at the end of the study 20 of 21 participants answered “yes” to the question “Did you benefit from the study - can you go to sleep earlier now?” and one answered “Don’t know.”

Also of interest:

A dose of 3 mg is most effective when given about 5 h before DLMO, whereas .5 mg is best given 2-4 h before DLMO (Burgess et al 2010_14ya). As early as 1991_33ya, Dahlitz et al. showed that melatonin, given for some weeks 5 h before habitual sleep-onset time, could advance the sleep timing more than 1 h in patients with DSPD (Dahlitz et al 1991_33ya). This was confirmed in a recent meta-analysis including nine randomized controlled trials, comparing melatonin with placebo (PL; Van Geijlswijk et al 2010_14ya). Four of the studies included children (Smits et al 2001_23ya, 2003_21ya; Van Der Heijden et al 2007_17ya; Weiss et al 2006_18ya). None of them included teenagers, however, and to the best of our knowledge, no such controlled studies exist.

…Nowadays, a consensus exists that melatonin is a safe treatment with appropriate dosage and timing (Arendt et al 2008_16ya). There might still be some concern, however, about its long-term use during childhood and adolescence, as melatonin is known to be involved in the seasonal regulation of animals’ reproduction (Srinivasan et al 2009_15ya). The gonadotrophic effects differ among species, however (Scherbarth & Steinlechner, 2010_14ya), and in a recent placebo-controlled study, no treatment-related changes of hormones, including testosterone and estradiol, were detected after long-term treatment of humans (Wade et al 2010_14ya). The only study to follow pre-pubertal children with regard to their development (Van Geijlswijk et al 2010_14ya) found that puberty onset seemed to be undisturbed after 3.1 yrs of melatonin usage.

…The students reported less sleepiness at awakening during M2 than PL and were also [statistically-]significantly more alert during morning hours. As in the discussion above, this may have been effects of both a phase advance of melatonin and a diminished sleep debt. A tendency to more afternoon alertness may only be explained by diminished sleep debt, though. The results are in accordance with some earlier studies (Kayumov et al 2001_23ya; Nagtegaal et al 1998_26ya). The overall link between sleep duration and school performance is well established (Wolfson & Carskadon, 2004_20ya). As daytime sleepiness has been shown to be even closer connected to poor school performance than short sleep duration (Anderson et al 2009_15ya), the increased alertness during school days is perhaps the most important finding of this study.

https://www.northjersey.com/news/222120551_High_schools__early_starts_deprive_students_of_crucial_sleep.html/?page=all:

About 40 percent of U.S. public high schools start before 8 a.m., according to the U.S. Department of Education’s National Center for Education Statistics, with just 15 percent beginning at 8:30 a.m. or later. Logistics for extracurricular activities and busing often create the early schedules. In districts where early starts are necessary because the same bus does multiple runs for high school, middle school and elementary students, teens often get the early shift.

…Research has shown that melatonin levels in the blood of teenagers naturally rise later at night than in most children and adults, according to the National Sleep Foundation. Rising melatonin levels promote sleep, and the delay in teens makes it difficult for them to go to bed early. High school-age students like Shea need about eight to nine hours of sleep each night, Carbone said. However, only 8 percent of high schoolers are getting that much sleep, according to a 2010_14ya study published in the Journal of Adolescent Health. The study’s authors found that 10 percent of adolescents sleep only five hours and 23 percent sleep only six hours on an average school night. “Many students will say, ‘I only get six hours of sleep, but I’m fine,’” Carbone said. “But many studies are showing that they’re not fine.”

“Later School Start Time Is Associated with Improved Sleep and Daytime Functioning in Adolescents”, Boergers et al 2014:

Method: Boarding students (n = 197, mean age = 15.6 yr) attending an independent high school completed the School Sleep Habits Survey before and after the school start time was experimentally delayed from 8:00 a.m. to 8:25 a.m. Results: The delay in school start time was associated with a significant (29 min) increase in sleep duration on school nights. The percentage of students receiving 8 or more hours of sleep on a school night increased to more than double, from 18% to 44%. Students in 9th and 10^th grade and those with lower baseline sleep amounts were more likely to report improvements in sleep duration after the schedule change. Daytime sleepiness, depressed mood, and caffeine use were all significantly reduced after the delay in school start time. Sleep duration reverted to baseline levels when the original (earlier) school start time was reinstituted.

…Chronic insufficient sleep has become a virtual epidemic among adolescents in the United States.¹ A host of adverse health outcomes, including drowsy driving-related crashes,² obesity,^3,4 cardiovascular disease,⁵ and metabolic dysfunction,6 have been identified as potential consequences of deficient sleep in adolescents. Adolescents achieving insufficient sleep are vulnerable to impairments in mood, behavioral control,⁷ attention, memory, and executive function.⁸ Not surprisingly, sleep loss is also associated with negative impacts on academic performance, increased tardiness and absenteeism, and decreased motivation to learn.^9-12

1. Eaton DK, McKnight-Eily LR, Lowry R, et al. “Prevalence of insufficient, borderline, and optimal hours of sleep among high school students-United States, 2007”. J Adolesc Health. 2010_14ya;46: 399-401.

2. Dahl RE. “Biological, developmental, and neurobehavioral factors relevant to adolescent driving risks”. Am J Prev Med. 2008_16ya;35(3 suppl):S278-S284.

3. Hasler G, Buysse DJ, Klaghofer R, et al. “The association between short sleep duration and obesity in young adults: a 13-year prospective study”. Sleep. 2004_20ya;27:661-666.

4. Cappuccio FP, Taggart FM, Kandala NB, et al. “Meta-analysis of short sleep duration and obesity in children and adults”. Sleep. 2008_16ya;31: 619-626.

5. Gangwisch JE, Malaspina D, Babiss LA, et al. “Short sleep duration as a risk factor for hypercholesterolemia: analyses of the National Longitudinal Study of Adolescent Health”. Sleep. 2010_14ya;33:956-961.

6. Verhulst SL, Schrauwen N, Haentjens D, et al. “Sleep duration and metabolic dysregulation in overweight children and adolescents”. Arch Dis Child. 2008_16ya;93:89-90.

7. Pasch KE, Laska MN, Lytle LA, et al. “Adolescent sleep, risk behaviors, and depressive symptoms: are they linked?” Am J Health Behav. 2010_14ya;34:237-248.

8. Carskadon MA. “Sleep’s effects on cognition and learning in adolescence”. Prog Brain Res. 2011_13ya;190:137-143.

9. Fredriksen K, Rhodes J, Reddy R, et al. “Sleepless in Chicago: tracking the effects of adolescent sleep loss during the middle school years”. Child Dev. 2004_20ya;75:84-95.

10. Curcio G, Ferrara M, De Gennaro L. “Sleep loss, learning capacity and academic performance”. Sleep Med Rev. 2006_18ya;10:323-337.

11. Wolfson AR, Carskadon MA. “Understanding adolescents’ sleep patterns and school performance: a critical appraisal”. Sleep Med Rev. 2003_21ya;7:491-506.

12. Perkinson-Gloor N, Lemola S, Grob A. “Sleep duration, positive attitude toward life, and academic achievement: the role of daytime tiredness, behavioral persistence, and school start times”. J Adolesc. 2013_11ya;36:311-318.

Factors contributing to chronic sleep deficits are complex and interrelated and include circadian biology and environmental influences, such as homework, jobs, extracurricular activities, and use of technology.^13,14 The pubertal shift in circadian phase preference from “morning” to “evening” type and an attenuation of the homeostatic “sleep drive” during adolescence both contribute to significant difficulty falling asleep at an earlier bedtime.¹⁵ Sleep needs, however, do not decline significantly across the transition to adolescence, and optimal sleep duration remains at 8.5 to 9.5 hours per night.16,17 Thus, the typical teen has difficulty falling asleep before 11:00 p.m. and should wake up about 8:00 a.m. to obtain enough sleep,15 although most American high schools start before 8:00 a.m.¹⁸

13. Cain N, Gradisar M. “Electronic media use and sleep in school-aged children and adolescents: a review”. Sleep Med. 2010_14ya;11:735-742. 14. Knutson KL, Lauderdale DS. “Sociodemographic and behavioral predictors of bed time and wake time among US adolescents aged 15 to 17 years”. J Pediatr. 2009_15ya;154:426-430. 15. Carskadon MA, Acebo C, Jenni OG. “Regulation of adolescent sleep: implications for behavior”. Ann N Y Acad Sci. 2004_20ya;1021_1,003ya: 276-291. 16. Carskadon MA, Harvey K, Duke P, et al. “Pubertal changes in daytime sleepiness”. Sleep. 1980_44ya;2:453-460. 17. National Sleep Foundation. Sleep in America Poll. Washington, DC: National Sleep Foundation; 2006_18ya. 18. Wolfson AR, Carskadon MA. “A survey of factors influencing high school start times”. NASSP Bull. 2005_19ya;89:47-66.

The recognition of the high cost of sleep loss in adolescents has led to a quest to identify potentially modifiable factors such as early school start times. Numerous studies comparing middle and high schools with early (before 8:00 a.m.) versus later start times demonstrate adverse consequences, such as shorter sleep duration, increased sleepiness, difficulty concentrating, absenteeism, lower academic performance,^19-22 and motor vehicle crashes.²³

19. Wolfson AR, Spaulding NL, Dandrow C, et al. “Middle school start times: the importance of a good night’s sleep for young adolescents”. Behav Sleep Med. 2007_17ya;5:194-209. 20. Dexter D, Bijwadia J, Schilling D, et al. “Sleep, sleepiness and school start times: a preliminary study”. WMJ. 2003_21ya;102:44-46. 21. Carrell SE, Maghakian T, West JE. “A’s from ZZZZ’s? The causal effect of school start time on the academic achievement of adolescents”. Am Econ J. 2011_13ya;3:62-81. 22. Edwards F. “Early to rise? The effect of daily start times on academic performance”. Econ Educ Rev. 2012_12ya;31:970-983. 23. Vorona RD, Szklo-Coxe M, Wu A, et al. “Dissimilar teen crash rates in two neighboring southeastern Virginia cities with different high school start times”. J Clin Sleep Med. 2011_13ya;7:145-151.

In response to these findings, a small but growing number of school districts have undertaken initiatives to delay start times and systematically examine the impact on students. In a large study, more than 18,000 high school students were assessed before and after the district’s start time changed from 7:15 to 8:40 a.m.^24,25 After the change, bedtimes did not shift to a later time, and as a result, students obtained nearly an hour more of sleep on school nights. In another study of a school district that delayed high school start times by 1 hour (7:30 to 8:30 a.m.), the percentage of students who reported 0.8 hours of sleep increased from 37% to 50%.²⁶ A recent study of adolescents at an independent high school that delayed start time by 30 minutes (8:00 to 8:30 a.m.) reported bedtimes actually shifted on average 20 minutes earlier, school night sleep duration increased by 45 minutes, and the percentage of students reporting at least 8 hours of sleep increased from 16% to 55%.²⁷ In addition to increased sleep, these studies have demonstrated a wide range of other benefits, including lower drop-out rates,^24,25 improvements in standardized reading and math scores,²² lower rates of depression symptoms,²⁷ and lower rates of car crashes.²⁶ Recent research also demonstrates that delaying school start times for middle school students is associated with similar positive outcomes, including more sleep on school nights, decreased daytime sleepiness and tardiness, and better performance on computerized attention tasks.&19,28&

24. Wahlstrom K. “Accommodating the sleep patterns of adolescents within current educational structures: an uncharted path”. In: Carskadon MA, ed. Adolescent Sleep Patterns: Biological, Sociological, and Psychological Influences. Cambridge, England: Cambridge University Press; 2002_22ya. 25. Wahlstrom K. “Changing times: findings from the first longitudinal study of later high school start times”. NASSP Bull. 2002_22ya;86:3-21. 26. Danner F, Phillips B. “Adolescent sleep, school start times, and teen motor vehicle crashes”. J Clin Sleep Med. 2008_16ya;4:533-535. 27. Owens JA, Belon K, Moss P. “Impact of delaying school start time on adolescent sleep, mood, and behavior”. Arch Pediatr Adolesc Med. 2010_14ya;164:608-614. 28. Lufi D, Tzischinsky O, Hadar S. “Delaying school starting time by one hour: some effects on attention levels in adolescents”. J Clin Sleep Med. 2011_13ya;7:137-143.

“Sleep restriction worsens mood and emotion regulation in adolescents”, Baum et al 2014:

…50 healthy adolescents, ages 14-17, completed a 3-week sleep manipulation protocol involving a baseline week, followed by a sleep restriction (SR) condition (6.5 hr in bed per night for five nights) and healthy sleep duration (HS) condition (10 hr in bed per night for five nights). The study used a randomized, counterbalanced, crossover experimental design. Participants’ sleep was monitored at home via self-report and actigraphy. At the end of each condition, participants and their parents completed questionnaires of mood and mood regulation. To assess for expectancy effects, we also analyzed parent and teen ratings of hyperactivity/impulsivity, which prior research suggests is not sensitive to SR in adolescents. Wilcoxon Signed Rank tests compared questionnaire outcomes across the two conditions. Results: Participants averaged 2.5 more hours of sleep per night during HS relative to SR. Compared with HS, adolescents rated themselves as significantly more tense/anxious, angry/hostile, confused, and fatigued, and as less vigorous (p = 0.001–0.01) during SR. Parents and adolescents also reported greater oppositionality/irritability and poorer emotional regulation during SR compared with HS (p < 0.05). There were no cross-condition differences in depression or hyperactivity/impulsivity (p > 0.05).

“Organizing Schools to Improve Student Achievement: Start Times, Grade Configurations, and Teacher Assignments”, Jacob & Rockoff2011

In this paper, we describe three organizational reforms that recent evidence suggests have the potential to increase K-12 student performance at modest costs: (1) Starting school later in the day for middle and high school students; (2) Shifting from a system with separate elementary and middle schools to one with schools that serve students in kindergarten through grade eight; (3) Managing teacher assignments with an eye toward maximizing student achievement (eg. allowing teachers to gain experience by teaching the same grade level for multiple years or having teachers specializing in the subject where they appear most effective). We conservatively estimate that the ratio of benefits to costs is 9 to 1 for later school start times and 40 to 1 for middle school reform.

…As any parent knows, it is very difficult to wake a sleeping teenager. Not only is it difficult to rouse them early in the morning, there is mounting evidence that it is also difficult to educate them early in the day. The earliest school start times are associated with annual reductions in student performance of roughly 0.1 standard deviations for disadvantaged students, equivalent to replacing an average teacher with a teacher at the sixteenth percentile in terms of effectiveness. Moving school start times later in the morning appears to improve sleep and ameliorate the biological barriers to learning.

…School days start early in the morning. According to the National Household Education Survey (NCES 2001_23ya), roughly half of middle schools start at or before 8:00 a.m., and fewer than 25 percent start at 8:30 a.m. or later. High schools start even earlier. Wolfson & Carskadon 2005_19ya, surveying a random sample of public high schools, found that more than half of the schools reported start times earlier than 8:00 a.m. In 2005_19ya, two thirds of high schools in Kansas started at 8:00 a.m. or earlier, and more than 99 percent started at 8:30 a.m. or earlier. In the school year 2010–2011_13ya, roughly 10 percent of high schools in New York City started at 7:30 a.m. or earlier and more than 80 percent started at 8:30 a.m. or earlier.

…Important changes in the circadian rhythm during adolescence shift children’s internal clocks to later bed and wake times (see, for example, Carskadon, Vieira, and Acebo 1993_31ya; Crowley, Acebo and Carskadon 2007_17ya; Wolfson and Carskadon 1998_26ya). As noted by Carrell, Maghakian, and West (2011_13ya), melatonin levels peak at roughly 7:00 a.m. for adolescents and at 4:00 a.m. for adults, so waking a teenager at 7:00 a.m. is similar to waking an adult at 4:00 a.m. Ideally, teenagers faced with early school start times would just go to bed early. However, a voluminous body of research demonstrates that earlier school start times lead teenagers to sleep less (see, for example, Dexter, Bijwadia, Schilling, and Applebaugh 2003_21ya; Hansen, Janssen, Schiff, Zee, and Dubocovich 2005_19ya; Wolfson and Carskadon 2003_21ya). In a recent study in Minnesota, students in schools where classes started at 8:30 a.m. reported getting about one hour more sleep than students in schools where classes started at 7:25 a.m. or 7:15 a.m. (Wahlstrom, Davison, Choi, and Ross 2001_23ya). Another survey found that high school students slept an average of two fewer hours on school nights than they slept on weekends or on weekdays during the summer (Hansen et al 2005_19ya).

…The strongest evidence on the impact of early start times on academic performance comes from a recent study by Carrell and colleagues (2011_13ya) who focus on first-year students in the U.S. Air Force Academy. These students have no choice over their course schedules and, during the two years of the study, are assigned start times ranging from 7:00 a.m. to 8:50 a.m.¹ Moreover, unlike most high schools, all first-year Air Force students take the same classes and the same standardized course exams, providing a consistent objective outcome measure. Carrell and his colleagues (2011_13ya) found that students assigned to start classes prior to 8:00 a.m. performed worse not only in their first-period course, but in all of their courses. Moreover, the size of the effect was substantial, with a one-hour delay associated with a 0.15 standard deviation increase in performance. For sake of comparison, note that the achievement gap between black and white students in the United States is roughly 1.0 standard deviation.

…Two other recent studies are more directly related to the K-12 setting. A recent study of school start times for middle schools takes advantage of changes to the tiered busing system in Wake County, North Carolina. Due to rapid enrollment growth and changes to school attendance boundaries, many of the district’s middle schools experienced changes in start times, with some students being picked up earlier and others later in the morning. An analysis by Edwards (2011) indicates that school start times shifted one hour later increase reading test scores by 0.03 to 0.10 standard deviations and math test scores by 0.06 to 0.09 standard deviations. Disadvantaged students benefited the most, with effects roughly twice as large as advantaged students; the effects persisted into high school. …In his study of Wake County, North Carolina, Edwards (2011) estimates that it would cost roughly $150 per student to allow all students to start at 9:15 a.m.⁵ If we aggregate costs of $150 per student per year over the thirteen years a student is in a K-12 system, we arrive at an increase in transportation costs of $1,950 over the student’s school career.

…A study of high school students in the Chicago Public Schools provides some additional insight. Cortes, Bricker, and Rohlfs (2009_15ya) examine how high school students perform in morning versus afternoon classes in Chicago, where school typically starts by 8:00 a.m. and students are commonly tardy. The authors find students are absent roughly six more days per year in first period relative to other periods. Moreover, they find that student grades and test score performance are notably lower for their first-period courses. For example, students assigned to a math class in first period do systematically worse on the end-of-year standardized math exam, whereas students assigned to English during first period do worse on the English exams.

…One piece of countervailing evidence on the issue comes from a careful reanalysis of changes that Minneapolis made to its school schedule in 1997–1998_26ya (Hinrichs 2011_13ya). Minneapolis and several nearby suburban districts moved high school start times from 7:15 a.m. to 8:40 a.m. while St. Paul and other suburban districts maintained their 7:30 a.m. start time. Hinrichs finds that neither student attendance nor ACT scores changed in these districts that shifted their start times later relative to those districts who maintained a consistent, earlier start time.

• “When the Bell Tolls: The Effects of School Starting Times on Academic Achievement”

…In order to translate this achievement gain into a dollar value, we follow the methodology used in Krueger (2003). A 1.0 standard deviation rise in test scores raises future earnings by 8 percent, which we match to the age-earnings profile in the Current Population Survey (2008) and assume a 1 percent growth rate for real wages and productivity. We calculate the present value of lost wages for an increase in achievement of 0.175 standard deviations and a discount rate of 4 percent. Doing so, we estimate that moving start times one hour later would result in roughly $17,500 in increased future earnings per student in present value.

“A Longitudinal Assessment of Sleep Timing, Circadian Phase, and Phase Angle of Entrainment across Human Adolescence”, Crowley et al 2014:

Ninety-four adolescents participated; 38 (21 boys) were 9-10 years (“younger cohort”) and 56 (30 boys) were 15-16 years (“older cohort”) at the baseline assessment. Participants completed a baseline and then follow-up assessments approximately every six months for 2.5 years. At each assessment, participants wore a wrist actigraph for at least one week at home to measure self-selected sleep timing before salivary dim light melatonin onset (DLMO) phase - a marker of the circadian timing system - was measured in the laboratory. Weekday and weekend sleep onset and offset and weekend-weekday differences were derived from actigraphy. Phase angles were the time durations from DLMO to weekday sleep onset and offset times. Each cohort showed later sleep onset (weekend and weekday), later weekend sleep offset, and later DLMO with age. Weekday sleep offset shifted earlier with age in the younger cohort and later in the older cohort after age 17. Weekend-weekday sleep offset differences increased with age in the younger cohort and decreased in the older cohort after age 17… This descriptive study indicated that circadian phase and self-selected sleep delayed across adolescence, though school-day sleep offset advanced until no longer in high school, whereupon offset was later.

…The transition through adolescence (the second decade) is often accompanied by a shift toward later timing of sleep/wake behavior. Survey studies from around the globe report later bedtimes on both school and non-school nights and later wake-up times on non-school or vacation mornings as youngsters age [1], [2], [3], [4], [5], [6], [7]. Cross-sectional and longitudinal studies using actigraphically-estimated sleep support these findings [8], [9]. A shift toward “eveningness” also emerges as youngsters age [10], and this shift appears to be linked to pubertal development [11], [12]. For American students, this delay of sleep behavior is often concurrent with the earliest school start times, reducing the opportunity for sleep on school nights for many. Cross-sectional and longitudinal studies of adolescent sleep length reflect this circumstance, showing a consistent age-related reduction of total sleep time [1], [9], [13], [14], [15], [16], [17], [18], [19]. Restricted weekday sleep during the school year is often compensated by over-sleeping on weekends, primarily through later wake-up times [1], [4], [20], [21]… Of the few studies reporting longitudinal patterns of adolescent sleep/wake timing, two examined young adolescents (~10-13 years) [2], [9]. Laberge and colleagues reported an age-related delay in parent-reported bedtimes on school and weekend nights, and a delay in weekend wake-up time over four years in young Canadian adolescents. Sadeh and colleagues found a similar age-related delay in sleep onset time measured by actigraphy in a group of young adolescents (9.9 to 11.2 years at baseline) followed over 3 years, and this change in sleep timing predicted changes to self-assessed puberty ratings. Andrade and colleagues [20] reported a delay in self-reported weekend wake-up time (on average between 34 and 39 minutes) in Brazilian adolescents (12-16 years); however, measurements were taken over one year only. Weekday and weekend bedtime and weekday wake time did not change over this year. The last longitudinal study focused on the transition from high school to college [23]. Urner and colleagues measured actigraphically-estimated sleep/wake patterns in Swiss students while in high school (aged 17 to 19 years) and again 5 years later when in university. Bedtime, wake-up time, and mid-sleep time shifted on average 34, 52, and 44 minutes later on school days after the transition to college, but sleep timing did not change on non-school days. In summary, previous longitudinal studies focused on early [2], [9] or late adolescence [23] and the only longitudinal study that focused on a relatively wide age range, followed youngsters for one year only [20].

“The Great Sleep Recession: Changes in Sleep Duration Among US Adolescents, 1991^–₂₁2012_12ya”, Keyes et al 2015:

METHODS: Data were drawn from Monitoring the Future, a yearly, nationally representative cross-sectional survey of adolescents in the United States 1991^–₂₁2012_12ya (n = 272 077)representing birth cohorts 1973^–₂₇2000_24ya. Adolescents were asked how often they get >=7 hours of sleep and how often they get less sleep than they should. Age-period-cohort models were estimated.

RESULTS: Adolescent sleep generally declined over 20 years; the largest change occurred between 1991^–₄1995_29ya and 1996^–₄2000_24ya.

https://www.washingtonpost.com/national/health-science/blue-light-from-electronics-disturbs-sleep-especially-for-teenagers/2014/08/29/3edd2726-27a7-11e4-958c-268a320a60ce_story.html http://op-talk.blogs.nytimes.com/2014/08/27/how-students-can-get-more-sleep/?_php=true&_type=blogs&_r=0

https://www.newyorker.com/science/maria-konnikova/the-walking-dead

“Synchronizing education to adolescent biology: ‘let teens sleep, start school later’”, Kelley et al 2014

Arne Duncan, US Secretary of State for Education, tweeted in 2013_11ya: ‘let teens sleep, start school later’. This paper examines early starts and their negative consequences in the light of key research in the last 30 years in sleep medicine and circadian neuroscience. An overview of the circadian timing system in adolescence leading to changes in sleep patterns is given and underpins the conclusion that altering education times can both improve learning and reduce health risks. Further research is considered from education, sleep medicine and neuroscience studies illustrating these improvements. The implementation of later starts is briefly considered in light of other education interventions to improve learning. Finally, the impact of introducing research-based later starts synchronized to adolescent biology is considered in practical and policy terms.

…During adolescence biological changes dictate both a sleep duration of nine hours and later wake and sleep times, a phenomenon found in other mammals (Hagenauer et al 2009_15ya; Rüger et al 2012_12ya). At its peak the combination of these two biological changes leads to a loss of two to three hours sleep every school day. Thus, a 07:00 alarm call for older adolescents is the equivalent of a 04:30 start for a teacher in their 50s. Failure to adjust education timetables to this biological change leads to systematic, chronic and unrecoverable sleep loss. This level of sleep loss causes impairment to physiological, metabolic and psychological health in adolescents while they are undergoing other major physical and neurological changes (Hansen et al 2005_19ya; Giedd 2009_15ya; Giedd et al 2012_12ya; Sawyer et al 2012_12ya; Sørensen et al 2012_12ya; Foster et al 2013_11ya).

• Hagenauer, M. H., J. I. Perryman, T. M. Lee, and M. A. Carskadon. 2009_15ya. “Adolescent Changes in the Homeostatic and Circadian Regulation of Sleep.” Developmental Neuroscience 31 (4): 276-284. doi:10.1159_865ya/000216538

• Rüger, M., M. A. St Hilaire, G. C. Brainard, S. B. Khalsa, R. E. Kronauer, C. A. Czeisler, and S. W. Lockley. 2012_12ya. “Human Phase Response Curve to a Single 6.5 h Pulse of Short-Wavelength Light.” The Journal of Physiology 591 (1): 353-363. doi:10.1113_911ya/jphysiol.2012_12ya.239046

• Hansen, M., I. Janssen, A. Schiff, P. C. Zee, and M. L. Dubocovich. 2005_19ya. “The Impact of School Daily Schedule on Adolescent Sleep.” Pediatrics 115 (6): 1555^–₆1561_463ya.

• Giedd, J. N. 2009_15ya. “Linking Adolescent Sleep, Brain Maturation, and Behavior.” Journal of Adolescent Health 45 (4): 319-320. doi:10.1016_1,008ya/j.jadohealth.2009_15ya.07.007

• Sawyer, S. M., R. A. Afifi, L. H. Bearinger, S. J. Blakemore, B. Dick, A. C. Ezeh, and G. C. Patton. 2012_12ya. “Adolescence: A Foundation for Future Health.” The Lancet 379 (9826): 1630^–₁₀1640_384ya. doi:10.1016_1,008ya/S0140-6736(12)60072-5

• Sorensen, K., A. Mouritsen, L. Aksglaede, C. P. Hagen, S. S. Mogensen, and A. Juul. 2012_12ya. “Recent Secular Trends in Pubertal Timing: Implications for Evaluation and Diagnosis of Precocious Puberty.” Hormone Research in Paediatrics 77 (3): 137-145. doi:10.1159_865ya/000336325

• Foster, R. G., S. N. Peirson, K. Wulff, E. Winnebeck, C. Vetter, and T. Roenneberg. 2013_11ya. “Sleep and Circadian Rhythm Disruption in Social Jet-lag and Mental Illness.” Progress in Molecular Biology and Translational Science 119: 325-346. doi:10.1016_1,008ya/B978-0-12-396971-2.00011-7

…The sleep-wake cycle arises from the interplay of the circadian rhythm generated by the SCN pacemaker (alertness) and a homeostatic sleep cycle which increases the pressure to go to sleep with increasing time awake (sleepiness), and dissipates this pressure with increasing time asleep. These processes work in opposition to maintain consolidated wake during the day, and a consolidated sleep during the night. Their interaction also generates a ‘wake maintenance zone’ (WMZ) which, paradoxically, occurs several hours before sleep is typically initiated (Dijk and Czeisler 1995_29ya; Shekleton et al 2013_11ya). This is a major factor in the sleep difficulties experienced by adolescents. In the early evening, the homeostatic drive for sleepiness is usually too low to fully counteract the circadian drive for alertness, resulting in several hours where it is very difficult to fall asleep. In adolescence the timing of the circadian clock shifts later, delaying the phase at which sleep can be initiated, and pushing the WMZ later into the evening.

• Dijk, D. J., and C. A. Czeisler. 1995_29ya. “Contribution of the Circadian Pacemaker and the Sleep Homeostat to Sleep Propensity, Sleep Structure, Electroencephalographic Slow Waves, and Sleep Spindle Activity in Humans.” The Journal of Neuroscience 15 (5): 3526-3538.

• Shekleton, J. A., S. M. Rajaratnam, J. J. Gooley, E. Van Reen, C. A. Czeisler, and S. W. Lockley. 2013_11ya. “Improved Neurobehavioral Performance During the Wake Maintenance Zone.” Journal of Clinical Sleep Medicine 9 (4): 353-362. doi:10.5664/jcsm.2588

…The homeostatic regulation of sleep also changes such that the build-up of sleep pressure becomes slower, meaning that it takes a longer time to reach the critical threshold required to initiate sleep, which also contributes to a later sleep time. The WMZ is the reason why we, including adolescents, cannot simply choose to go to sleep earlier than normal, as the brain is still promoting wakefulness at that time. These biological changes in the timing of sleep propensity underlie the conflict with education start times; the brain will not allow students to go to sleep early but education times still require adolescents to wake (or be woken) too early in their circadian cycle, systematically restricting the time available for sleep and causing severe and chronic sleep loss. In turn, short sleep duration and chronic sleep deprivation are sleep disruptions that have been linked to negative impacts on cognition, emotional and physical health (Lockley et al 2004_20ya).

• Lockley, S. W., J. W. Cronin, E. E. Evans, B. E. Cade, C. J. Lee, C. P. Landrigan, J. M. Rothschild, et al 2004_20ya. “Effect of Reducing Interns’ Weekly Work Hours on Sleep and Attentional Failures.” The New England Journal of Medicine 351 (18): 1829^–₈1837_187ya.

…A recent review cited studies showing that restricted sleep was associated with impaired immune response, metabolic disorders, diabetes, hypertension, anxiety, depression and obesity (Luyster et al 2012_12ya). The general conclusion of this review was that sleep of less than six hours’ duration was potentially harmful. This conclusion was supported in a recent elegant experiment showing the negative impact on the body of a single week with 5.7 hours sleep a night (compared to controls sleeping 8.5 hours), identifying 711 genes up- or down-regulated by insufficient sleep, and a highly significant impact on genes with circadian expression profile (Möller-Levet et al 2013_11ya).

• Luyster, F. S., P. J. Strollo Junior, P. C. Zee, and J. K. Walsh. 2012_12ya. “Sleep: A Health Imperative.” Sleep 35 (6): 727-734. doi:10.5665/sleep.1846

• Möller-Levet, C. S., S. N. Archer, G. Bucca, E. E. Laing, A. Slak, R. Kabiljo, J. C. Lo, et al 2013_11ya. “Effects of Insufficient Sleep on Circadian Rhythmicity and Expression Amplitude of the Human Blood Transcriptome.” Proceedings of the National Society of Sciences of the United States of America 110 (12): E1132-E1141_883ya. doi:10.1073_951ya/pnas.1217154110

…Adolescents also show the impact of short sleep duration and chronic sleep deprivation. Adolescent studies found poor communication, decreased concentration and cognitive performance, unintended sleeps, decreased motor performance, increased risk taking and changes in mood pattern, specifically depression (Millman 2005_19ya; Knutson et al 2007_17ya; Hagenauer et al 2009_15ya; Foster et al 2013_11ya; de Souza and Hidalgo2014). This is hardly surprising, given that older teenagers show the largest discrepancy in sleep duration between free days and workdays of any age group, combined with a need for a longer sleep duration than adults. Other studies have shown that short sleep duration has a pervasive negative impact on their learning and academic performance (Curcio, Ferrara, and De Gennaro 2006_18ya), including poor grade performance and poor long-term memory encoding (Carskadon 2011_13ya). These studies reflect research with adults summarized in Table 1 (Foster and Wulff 2005_19ya; Wulff et al 2011_13ya).

…There are other demonstrations that short sleep duration in adolescence on this scale impacts on alertness and performance. Adolescent drivers have a peak rate of accidents in the morning (Czeisler 2009_15ya), unusual in older drivers, which illustrates the functional impact of waking at an adverse circadian phase and without enough sleep. When a school district in Kentucky moved school times an hour later students reported sleeping an hour longer, and not staying up later. The impact on school day car crashes involving 17-18-year-old drivers was significant: while the state-wide rate went up 7.8%, the rate for drivers in the school district went down 16.5% (Danner and Phillips 2008_16ya), a finding replicated in Virginia (Vorona et al 2011_13ya). From a wider perspective, in modern societies sleep loss and sleep deprivation have reached epidemic levels. A recent review of the difference between work/education schedules and biological time quantified the extent of this sleep deficit at different ages (Foster et al 2013_11ya). In an international survey (N ∼ 150,000), questionnaire-based assessments of chronotype show a rapid rise in eveningness throughout adolescence, occurring earlier in girls and coincident with earlier onset of puberty. These changes slowly reverse in early adulthood (starting at mean of 20.2 years), when chronotype begins to advance again, and wake times start to become earlier (Roenneberg et al 2004_20ya). when there is no interference from education, adolescents’ sleep during the week matched their weekend sleep (Hansen et al 2005_19ya; Jenni, Achermann, and Carskadon 2005_19ya). The cause of this decline in sleep duration is shown in the difference between sleep on education/work days and free days. On education days the difference reaches almost three hours in late puberty (Figure 1). This scale of sleep loss cannot be recovered at weekends, as illustrated by simple arithmetic: losing three hours sleep per day in the week would need more than 15 hours additional hours of sleep per night on the weekend which is both physically and practically impossible.

…A study in Spain showed for students aged 12-16 years evening types performed significantly worse than other chronotypes (Escribano et al 2012_12ya). In Germany, evening-type students in high school showed significantly more use of stimulants and depressants associated with managing sleep such as coffee, cola, alcohol and nicotine when compared with morning-type students (Preckel et al 2013_11ya). In Baden-Württemberg, a study found highly significant differences in Abitur scores (used for German university entrance) with evening types having lower performance (Randler and Frech 2006_18ya). Other studies show that evening-type adolescents may have increased risk of health and mental health problems (Giannotti et al 2002_22ya; Hsu et al 2012_12ya). Such findings are consistent with other research on adolescents as a group showing the very rapid impact of shorter sleep duration on neurobehavioural performance (Möller-Levet et al 2013_11ya), and the high incidence of clinical diagnoses of mental health issues during adolescence (Wulff et al 2011_13ya). For example, in circadian rhythm sleep disorders (ICSD-II), the prevalence of delayed sleep phase type has been calculated as 7-16% of US adolescents and young adults (15-24 years of age), or more than 2.9 million cases (Kelley and Lockley 2013_11ya).

…Synchronizing education start times to adolescent biology is the obvious way to address the problem of chronic sleep deprivation currently experienced by adolescents on school days. Astronomical time data and changes in sleep patterns from international studies show at the age of 10 biological wake time is about 06:30, so synchronized school starting times would be 08:30-09:00. At the age of 16 biological wake time is about 08:00, and synchronized school start times 10:00-10:30, and at 18 biological wake time is about 09:00, and synchronized education start times 11:00-11:30 (Kelley and Lockley 2013_11ya). Unfortunately almost all previous studies of later school start times used times before 09:00, an approach that significantly underestimated the scale of change needed. Yet even these small changes to times before 09:00 have shown benefits that have led many researchers to call for a change to later start times (for example, Hansen et al 2005_19ya; Jenni, Achermann, and Carskadon 2005_19ya; Czeisler 2009_15ya; Giedd 2009_15ya; Hagenauer et al 2009_15ya; Foster et al 2013_11ya). The strength of the evidence is such that it has led to later start times in some education institutions, and significant advocacy activity by organizations and individuals for systemic change in start times in schools. Kirkby, Maggi, and D’Angiulli (2011_13ya) in a comprehensive review of later start times concluded that the benefits of later starts included better sleep, health and learning, and the benefits were an ‘enormous potential payoff’ for the organizational costs involved. Jacob & Rockoff 2011_13ya in a review of later start times came to a similar conclusion, asserting they offered major learning improvements from a very well-researched intervention with little or no associated costs.

…In a study of data from Wade County, North Carolina, it was possible to compare outcomes within schools with variation in start times across schools because of changes in transport arrangements in the county (Edwards 2012_12ya). Analysing large cohorts (n > 4000), a start time of an hour later had a significant two-percentile point gain in state-wide maths test scores and similar changes in reading. There were two distinctive findings in the study: first, the benefits were greater for those students at the lower end of distribution of test scores; second, the two-percentile gain in grade 8 scores remained two years later in grade 10. In one of the first large studies based on sleep science research, seven comprehensive high schools in the Minneapolis Public School District moved the school start time from 07:15 to 08:40 (Wahlstrom 2002_22ya). This study used data on more than 50,000 students collected before and after the time changed, and was recently replicated in schools in three different states (Wahlstrom et al 2014). The self-reported student evidence indicated that students liked the change, slept an hour longer compared to students in two other similar school districts, and reported their attendance, achievement, behavior and mood improved. Unlike most studies, the assessment included data on a very wide range of variables, including detailed responses from parents. Parents were very positive about the later start, with 92% liking the change and reporting that their children were easier to live with and families had more conversation time (CAREI 2002_22ya). The impact of early starts was the subject of a particularly compelling case study of the United States Air Force Academy where researchers looked at academic performance of first-year students (18-19 years old) during a three-year period when the start times changed each year in a 07:00-07:50 range (Carrell, Maghakian, and West 2011_13ya). The academy used randomized placement of students to instructors and classes, and only some students would have a class at the start times on any particular day, while the rest had additional free time (n = 6165). The later the start times of courses for individual students, the better those students performed in the course overall. The significance of student improvement was equivalent to 1 standard deviation improvement in teacher quality. Interestingly researchers found that students assigned to start classes prior to 08:00 performed worse not only in their first period course, but in all of their courses on that day. Moreover, the size of the effect was substantial, with a 1-hour delay associated with a 0.15 standard deviation increase in performance. The researchers contextualized this change comparing it to the 1.0 standard deviation difference between black and white students in the USA.

…There have been schools and universities that have moved starting times to 10:00 am and later (for example, London University’s School of Media, Arts and Design starts at 10:30) although most have not been the subject of formal studies. One school in New Zealand changed the start times for students aged 16 or over to 10:30, and analysis showed 14-15-year-old students in the same school (starting at 09:00) reported more sleep loss, more difficulty getting up (p < .001) and were significantly less likely to report waking after 07:00 on weekdays (p < .0001) than older students starting at 10:30 (Borlase, Gander, and Gibson 2013_11ya). Sakarya University in Turkey has two full academic sessions a day (08:00-14:50 and 15:00-21:50), and academic achievement varies by session and chronotype. Most examinations are at 09:30 and both teaching and test start times impact on academic performance, with eveningness types achieving better outcomes in the later session and worse in examinations held early in the day (Bes ̧oluk, Onder, and Deveci 2011_13ya). Surprisingly, Shekleton et al 2013_11ya found general performance was better in the WMZ than earlier in the day. These data suggest that exploration of an upper time limit could be valuable in determining the appropriate end of the education day or educational provision for groups of adolescents. The success of Asian and other education systems appear to have in classes in the early evening may also be an indication that times up to 20:00 are potentially very productive in adolescent learning.

“Later school start times in the U.S.: An economic analysis”, Hafner et al 2017:

Research Questions:

• What would be the economic impact of a state-wide universal shift in U.S. school start times to 8:30 a.m.?

• Would a state-wide universal shift in U.S. school times to 8:30 a.m. be a cost-effective policy measure?

Numerous studies have shown that later school start times are associated with positive student outcomes, including improvements in academic performance, mental and physical health, and public safety. While the benefits are well-documented in the literature, there is opposition against delaying school times across the U.S. A major argument is the claim that delaying school start times will result in significant additional costs due to changes in transportation, such as rescheduling bus routes. This study investigates the economic implications of later school start times by examining a policy experiment and its subsequent state-wide economic effects of a state-wide universal shift in school start times to 8:30 a.m.

Using a novel macroeconomic modeling approach, the study estimates changes in the economic performance of 47 U.S. states following a delayed school start time, which includes the benefits of higher academic performance of students and reduced car crash rates. The benefit-cost projections of this study suggest that delaying school start times is a cost-effective, population-level strategy which could have a significant impact on public health and the U.S. economy. From a policy perspective, the study’s findings demonstrate the significant economic gains resulting from the delay in school start times over a relatively short period of time following the adoption of the policy change.

Key Findings

• The study suggested that delaying school start times to 8:30 a.m. is a cost-effective, population-level strategy which could have a significant impact on public health and the U.S. economy.

• The study suggested that the benefits of later start times far out-weigh the immediate costs. Even after just two years, the study projects an economic gain of $8.6 billion to the U.S. economy, which would already outweigh the costs per student from delaying school start times to 8:30 a.m.

• After a decade, the study showed that delaying schools start times would contribute $83 billion to the U.S. economy, with this increasing to $140 billion after 15 years. During the 15 year period examined by the study, the average annual gain to the U.S. economy would about $9.3 billion each year.

• Throughout the study’s cost-benefit projections, a conservative approach was undertaken which did not include other effects from insufficient sleep, such as higher suicide rates, increased obesity and mental health issues - all of which are difficult to quantify precisely. Therefore, it is likely that the reported economic benefits from delaying school start times could be even higher across many U.S. states.

“Is 8:30 a.m. Still Too Early to Start School? A 10:00 a.m. School Start Time Improves Health and Performance of Students Aged 13–16”, Kelley et al 2017 (quasi-experiment):

While many studies have shown the benefits of later school starts, including better student attendance, higher test scores, and improved sleep duration, few have used starting times later than 9:00 a.m. Here we report on the implementation and impact of a 10 a.m. school start time for 13 to 16-year-old students. A 4-year observational study using a before-after-before (A-B-A) design was carried out in an English state-funded high school. School start times were changed from 8:50 a.m. in study year 0, to 10 a.m. in years 1–2, and then back to 8:50 a.m. in year 3. Measures of student health (absence due to illness) and academic performance (national examination results) were used for all students. Implementing a 10 a.m. start saw a decrease in student illness after 2 years of over 50% (p < 0.0005 and effect size: Cohen’s d = 1.07), and reverting to an 8:50 a.m. start reversed this improvement, leading to an increase of 30% in student illness (p < 0.0005 and Cohen’s d = 0.47). The 10:00 a.m. start was associated with a 12% increase in the value-added number of students making good academic progress (in standard national examinations) that was significant (<0.0005) and equivalent to 20% of the national benchmark. These results show that changing to a 10:00 a.m. high school start time can greatly reduce illness and improve academic performance. Implementing school start times later than 8:30 a.m., which may address the circadian delay in adolescents’ sleep rhythms more effectively for evening chronotypes, appears to have few costs and substantial benefits.

“3.4 million real-world learning management system logins reveal the majority of students experience social jet lag correlated with decreased performance”, Smarr & Schirmer2018:

Misalignments between endogenous circadian rhythms and the built environment (ie. social jet lag, SJL) result in learning and attention deficits. Currently, there is no way to assess the impact of SJL on learning outcomes of large populations as a response to schedule choices, let alone to assess which individuals are most negatively impacted by these choices. We analyzed two years of learning management system login events for 14,894 Northeastern Illinois University (NEIU) students to investigate the capacity of such systems as tools for mapping the impact of SJL over large populations while maintaining the ability to generate insights about individuals. Personal daily activity profiles were validated against known biological timing effects, and revealed a majority of students experience more than 30 minutes of SJL on average, with greater amplitude correlating strongly with a significant decrease in academic performance, especially in people with later apparent chronotypes. Our findings demonstrate that online records can be used to map individual- and population-level SJL, allow deep mining for patterns across demographics, and could guide schedule choices in an effort to minimize SJL’s negative impact on learning outcomes.

“Academic achievement across the day: Evidence from randomized class schedules”, Williams & Shapiro2018:

This study expands our understanding of how school day schedules affect achievement. We focus on three aspects related to scheduling: student fatigue, time of instruction, and instructor schedules. Data cover five academic years at the United States Air Force Academy, where schedules are randomized, grading is standardized, and there is substantial variance in schedule structure. Analyzing over 180,000 student-course outcomes, we find causal evidence of cognitive fatigue brought on by scheduling multiple courses in a row. The expected performance of two students in the same class may differ by as much as 0.15 standard deviations simply owing to their prior schedules. All else equal, students perform better in the afternoon than in the early morning. We also note that instruction improves with repetition. Heterogeneous effects by ability suggest that overall gains are possible. Prioritizing certain schedules would be equivalent to improving teacher quality by one-third of a standard deviation. A reorganization of students’ daily school schedules is a promising and potentially low-cost educational intervention.

“Interplay of chronotype and school timing predicts school performance”, Goldin et al 2020:

Most adolescents exhibit very late chronotypes and attend school early in the morning, a misalignment that can affect their health and psychological well-being. Here we examine how the interaction between the chronotype and school timing of an individual influences academic performance, studying an unique sample of 753 Argentinian students who were randomly assigned to start school in the morning (07:45), afternoon (12:40) or evening (17:20). Although chronotypes tend to align partially with class time, this effect is insufficient to fully account for the differences with school start time. We show that (1) for morning-attending students, early chronotypes perform better than late chronotypes in all school subjects, an effect that is largest for maths; (2) this effect vanishes for students who attend school in the afternoon; and (3) late chronotypes benefit from evening classes. Together, these results demonstrate that academic performance is improved when school times are better aligned with the biological rhythms of adolescents.

Backlinks