It has long been known that adolescents have a harder time than younger children going to sleep early in the evening and thus must sleep later in the morning if they are to attain sufficient sleep. While social factors no doubt play some role, an underlying physiological mechanism is also responsible. Melatonin is the natural substance produced by the pineal gland that helps induce sleep, and among the many hormonal changes that accompany the onset of puberty is a delay in evening melatonin production. A majority of American teenagers’ sleep is insufficient on school nights, with the result that they go to school sleepy and are unable to perform to their optimal level. As a way of addressing this problem, efforts have been underway for several years to convince school boards and administrators to adjust start times.
School start times are determined by many factors, including custom, transportation issues, and demands on afternoon after-school activities. To achieve maximum efficiency, many school districts run two shifts of school bus routes, an early one for high school students and a later one for students in lower grades. School bus routes can be very long, especially in rural areas where bus rides of up to an hour are not uncommon. Children are often outside waiting for a bus before dawn, and have awakened long before dawn to go through their morning routine to be ready for pick-up.
The argument to start school later is based on the following:
All of the above statements are true, yet there has not been a well-designed comprehensive study demonstrating that pushing back school start times for adolescents has a measureable significant effect on academic performance…until now. In a study done by Finley Edwards reported in Economics of Education Review this year, there is direct evidence that has been lacking up to now. Conducting a true experimental design to measure the effects of earlier versus later school start times is not feasible. One would need to randomly assign students to the two conditions, match them on many confounding variables, and assure that the teachers were unaware of the condition their school and students are in. Edwards studied a situation often termed a “natural experiment”, in a large North Carolina school district. Using a three-tier system, schools start times there vary from 7:30 in Tier 1, 8:25 in Tier 2, and 9:15 in Tier 3. Tiers 1 and 2 are composed primarily of middle and high schools, and Tier 3 is entirely elementary schools. Due to a growing population and accompanying strains on the transportation system, a fortuitous situation arose in that some schools were switched to earlier, and some to later start times. Using standardized achievement test scores of Tier 1 and 2 schools, Edwards demonstrates that later start times are associated with higher scores in math and reading. Moreover, the effect of later start times is greater for the bottom half of the distribution, children performing at the lowest levels. This finding is consistent with the hypothesis I have put forward, namely that differences in sleep may account in part for the SES achievement gap that has been so resistant to change in spite of many interventions over decades. The effects on achievement are relatively small for younger children in the study, but begin to increase around age 13, consistent with expectations based on the onset of puberty discussed above.
For some time I have believed that if concrete evidence of improved test scores were available, many more school districts would seriously consider later start times for middle and high schools. And given our predilection to cast schools into competition with one another, once some school districts are perceived to be gaining a further competitive edge, more and more school districts will follow suit. Changing school start times is not a panacea for all the problems associated with academic underachievement, but it is rare that an opportunity so simple and relatively inexpensive presents itself.
Edwards, F. (2012). Early to rise? The effect of daily start times on academic performance. Economics of Education Review, 31, 970-983.