Delayed high school start times and graduation and attendance rates over 4 years: the impact of race and socioeconomics

Abstract

Study Objectives:

The current study was designed to examine if improvements in high school graduation and attendance rates persist 4 years after start times later than 8:30 am. Socioeconomic status and race were accounted for in the graduation analysis.

Methods:

School-wide attendance and graduation records (n = 28 schools) from 1999–2017 were analyzed in August of 2021 using a repeated-measures analysis of variance at 5 points in time, 1 year predelay and 4 consecutive years postdelay start times.

Results:

Graduation rates for students in grade 12 (n > 47,000) after postdelay year 4 increased from 80% to 90% (P < .001) and attendance rates in grades 9–12 (n > 80,000) improved from 90% to 93% (P = .012).

Conclusions:

Attendance and graduation rates for all students improved after 4 years following a delayed start. Economically disadvantaged students and Black students showed significant changes in graduation rates postdelay year 4 and White students significantly improved postdelay years 2, 3, and 4. Adolescent students, especially economically and racially disadvantaged students, have a vulnerability for obtaining adequate sleep. Delaying high school start times is a cost-effective intervention that could promote sufficient sleep and boost graduation and attendance rates.

Citation:

McKeever PM, Dodd R, O’Sullivan DM. Delayed high school start times and graduation and attendance rates over 4 years: the impact of race and socioeconomics. J Clin Sleep Med. 2022;18(11):2537–2543.

BRIEF SUMMARY

Current Knowledge/Study Rationale: Later school start times are associated with significant improvements in attendance and graduation rates. This retrospective analysis extends previous research to examine if changes persist in graduation and attendance rates following a delay in high school start times of later than 8:30 am over 4 years.

Study Impact: The evidence from this study could be used to support uniform policies that endorse delayed high school start times. Significant changes in graduation post–year 4 in groups of students identified as economically disadvantaged and African-American suggest that this cost-effective policy could increase the likelihood that disadvantaged students graduate from high school.

INTRODUCTION

Efforts to narrow the adolescent sleep gap between hours of sleep needed and hours of sleep obtained remain challenging. It is estimated that 78% of high school adolescents are not getting the 8–10 hours of sleep per night recommended.^1,2 Short sleep duration has been associated with early rise times and late sleep onset. Studies find that the greatest predicator of wake times are school start times.^3,4 Sleep researchers have scientifically established that the phase delay that occurs during puberty affects the homeostatic drive to sleep, resulting in a tendency to stay up later in the evening and rise later in the morning.^5,6 In addition to the natural biological trigger, increased indoor light exposure affects the circadian rhythm by delaying melatonin production, linking environmental cues to sleep drive.⁶ Over time, sleep debt accrued by adolescents results in a tendency for teens to feel sleepier during the day.^7–9 According to the American Academy of Sleep Medicine, sleep deprivation occurs when a person does not get the amount of sleep needed to feel rested. The American Academy of Sleep Medicine literature emphasizes that there is no substitute for sufficient sleep, stating that the only real coping strategy to reverse the effects of sleep deprivation is to increase nightly sleep.⁷

Adverse educational and mental health outcomes have been associated with insufficient sleep, including an increase in depressive symptoms and suicidal ideation.¹⁰ Existing research provides evidence that schedules that do not accommodate later sleep onset and offset are not aligned with the adolescent sleep-wake cycle, increasing the risk for poorer attendance, decreased school performance, lower graduation rates, and increased car accidents.^11–14 Findings from studies link an increased risk in high school suspensions (especially for disadvantaged students), driving under the influence, weapon carrying, fighting, contemplated suicide, and attempted suicide with early start times.^10,15,16

Factors that affect sleep vary. Researchers found an association between later start times and longer sleep duration in adolescents.^3,17 Along with school factors that endorse early start times, sociodemographic factors such as socioeconomic status and race have been found to contribute to negative outcomes related to sleep.^18–21 Sleep disparities associated with African-American race include shorter nocturnal sleep duration, later bed time, and fragmented sleep compared with White students.^22,23 Previous research on adolescent sleep includes an association between economically disadvantaged students and shorter sleep duration and self-reported perceptions of obtaining adequate sleep.¹⁹ It is estimated that early school start times are as harmful to disadvantaged students as having a highly ineffective teacher. Delaying start times by 1 hour would result in increased lifetime earnings at a benefit-to-cost ratio of 9:1. This positive effect is comparable to raising teacher quality by 1 standard deviation.²⁴

A lack of sufficient sleep is associated with suboptimal academic performance. Scientists at the Centers for Disease Control and Prevention state that, “Delaying school start times is a demonstrated strategy to promote sufficient sleep among adolescents.”²⁵ In addition to quantity and quality of sleep, racial and socioeconomic factors have been found to be the primary predictor for academic performance on school-wide and district-wide standardized tests, with a negative relationship increasing over grade levels for economically disadvantaged students.²⁶ Graduation completion is jeopardized when successful navigation through the complex demands of passing standardized assessments fails. High school graduates rely on the use of skills hampered by sleep loss, such as emotional regulation, risk taking, memory, attention, and executive functioning.²⁷ Prior studies link later high school start times to improved graduation rates, justifying implementation of an intervention that increases the likelihood for greater student success and improved quality of life in the future.^14,28 The association between negative outcomes and earlier high school start times punctuates the need for school officials and policy makers to remain health conscious and to make decisions based on adolescent sleep research.^{11–14,29–31}

Even with a heightened awareness about the biological mechanisms that regulate adolescent sleep, establishing policies to endorse later start times is not clear cut. Charged with the responsibility to prioritize student well-being, Boston School Board members and school officials adopted a policy to delay start times for over 17,000 high school students.³² In December of 2017, members of the Boston School Committee voted in favor of delaying high school start times. The change required later starts for older students and earlier bell times at the elementary level. After thousands of stakeholders expressed strong opposition, the decision was reversed within 2 weeks of endorsement.³² In contrast, lawmakers from California focused on adolescent sleep science research, including topics such as the biological underpinnings of age-dependent sleep patterns and the need to decrease sleep debt. The research presented during the debates in both the Senate and Assembly led to the enactment of education policy in California that mandates delayed start times for public middle and high schools by July 2022.³³ Lawmakers persevered and continued to lobby after the proposed bill was defeated 2 consecutive years before it was signed by the governor. California state law will require approximately 1,700,000 high school students to start school at 8:30 am or later and over 900,000 middle school students to start school at 8:00 am or later.^33,34

The current study extends previous research that examined the effects of school start times on graduation and attendance rates 2 years after implementation of a delayed high school start of later than 8:30 am.¹⁴ While prior studies focus on the importance of short-term benefits after delayed school start times, the literature lacks evidence regarding the long-lasting impact, with the exception of 1 study that examined student performance over a prolonged period.³⁵ Among the studies reviewed, most compare outcomes 1 year predelay to a few years postdelay.^14,25,36,37

The purpose of the study is 2-fold to examine the benefits of delayed school start times for all students and disadvantaged students. First, to quantify attendance and graduation rates for all students each year over 4 consecutive years following a delay in high school start times later than 8:30 am. Second, to investigate the differences in graduation rates after the implementation of delayed high school start times over 4 consecutive years with the inclusion of race and socioeconomic status.

METHODS

This study was considered a program evaluation and was exempt from institutional review board review. Participants in this analysis include public high school students in grades 9–12 (n >80,000). Data from 28 schools in 8 school districts from 7 states are included in this study. Data on graduation and attendance rates were collected at 5 time periods: baseline predelay and 4 years postdelay. A census of the participating schools comes from a previous investigation that examined the impact of school start times on graduation and attendance rates and a study from the Children’s National Medical Center’s Division of Sleep Medicine.^8,35 Convenience sampling was used in the present study due to limited data sources. This model is used frequently when examining delayed school start benefits.³⁶ Data sources for this study also include state repositories. These sources of data are used by local, state, and federal grant agencies and are considered reliable and valid.

School start times with pre and post changes and October membership for grade 12 (graduation) and grades 9–12 (attendance) are listed in Table 1. All schools met criteria for predelay start times of 8:30 am or earlier, postdelay start times of later than 8:30 am, and all schools were operational for 5 consecutive years.

Table 1.

List of schools and time changes.

State (School Number)	Predelay Time	Postdelay Time	Total October Membership Grade 12 (Graduation) (% of Grades 9–12)	Total October Membership Grades 9–12 (Attendance)
Virginia	Before 2013: 8:30 am–3:00 pm	2013–2017: 8:55 am–3:35 pm
1			1,670 (25.0)	6,814
2			1,224 (28.0)	4,421
3			1,229 (27.0)	4,615
Florida	Before 2000: 7:30 am–2:15 pm	2000–2004: 8:45 am–3:30 pm
1			1,802 (22.0)	8,161
2			1,156 (25.0)	4,619
3			1,377 (19.8)	6,926
4			2,442 (21.1)	10,885
5			2,786 (24.1)	11,543
6			2,143 (23.1)	9,257
7			1,788 (22.4)	7,952
8			2,314 (23.8)	9,702
9			2,428 (22.8)	10,628
New York	Before 2006: 8:00 am–2:37 pm	2006–2010: 8:55 am–3:32 pm
1			1,830 (21.6)	8,434
North Carolina	Before 2012: 8:00 am–3:00 pm	2012–2016: 9:00 am–4:00 pm
1			745 (24.9)	2,991
2			2,612 (25.2)	10,381
3			1,014 (16.4)	6,178
Oregon	Before 1999: 7:30 am–2:20 pm	1999–2003: 8:45 am–3:20 pm
1			2,031 (25.0)	8,133
Arkansas	Before 2012: 7:30 am–2:40 pm	2012–2016: 8:35 am–3:45 pm
1			1,039 (25.0)	4,161
2			619 (24.9)	2,481
3			836 (25.0)	3,346
4			1,004 (21.2)	4,746
5			883 (25.0)	3,538
Florida	Before 2006: 8:00 am–2:45 pm	2006–2010: 9:15 am–3:15 pm
1			1,908 (23.5)	8,106
2			1,919 (19.9)	9,601
3			2,138 (22.4)	9,542
4			2,234 (22.7)	9,812
Minnesota	Before 2009: 7:35 am–2:05 pm	2009–2013: 8:35 am–3:05 pm
1			2,236 (24.5)	9,105
2			2,459 (26.2)	9,377

An analysis was conducted of enrolled public high school students in grades 9–12 during the first year prior and 4 years following a delay in school start times in the summer of 2021. These data were analyzed for attendance and graduation rates from academic years 1999–2017.

Statistical analysis

Graduation and attendance rates were examined longitudinally 1 year before and 4 years following a delay in start times during the regular school year. Attendance rates and graduation rates were evaluated under 2 conditions—predelay time (1 year) and postdelay time (4 consecutive years)—in districts that delayed school start times later than 8:30 am and reported as percentages (0–100%).

Data were analyzed using SPSS version 26 (2019; IBM Corporation, Armonk, NY). Homogeneity of variance was evaluated with Levene’s test and normality of distribution was tested with a Shapiro-Wilk test. Given that years did not necessarily comprise the same students, a paired t test was used to test each postdelay time against the predelay time. For the same reason, no multiplicity adjustment was made. All results with a P value < .05 were deemed statistically significant.

RESULTS

Graduation rates

Table 2 summarizes graduation rate changes during each measurement window. Later school start times in the first postdelay period compared with the predelay year were associated with an increase in graduation rates by 7%; in the second postdelay period an increase of 8%; in the third postdelay period an increase of 7%; and by the fourth and final postdelay period, an increase of 10% was noted.

Table 2.

Descriptive statistics (graduation rates).

Year	All Students		African-American Students		Economically Disadvantaged Students		White Students
	Mean	SD	Mean	SD	Mean	SD	Mean	SD
Predelay year 0	79.93	11.77	79.35	12.07	73.00	11.21	81.83	12.24
Postdelay year 1	86.57	8.32	79.13	13.28	72.40	9.08	84.20	9.54
Postdelay year 2	87.93	8.98	80.11	13.00	77.46	11.22	85.52	10.95
Postdelay year 3	87.43	9.39	78.41	11.48	75.19	13.01	86.81	9.43
Postdelay year 4	89.75	6.88	82.46	10.23	80.44	8.50	88.89	9.42

SD = standard deviation.

In support of delaying school starts, the predelay graduation rates were significantly lower (P < .001) than postdelay years 1–4 for students enrolled in grade 12, with an increase from 80% to 90% postdelay year 4. To better understand the impact from demographic differences, groups were examined separately in Table 2 and Table 3. Mean changes postdelay year 4 for African-American students ranged from 80% to 82% (P = .034), economically disadvantaged students from 73% to 80% (P = .006), and White students from 82% to 89% (P = .008). Although significant changes in graduation rates from the predelay year to all postdelay years significantly increased for grade 12 students (all demographics combined), significant changes for African-American students (P = .034) and students identified as economically disadvantaged (P = .006) occurred postdelay year 4. Significant changes in graduation rates for White students occurred postdelay years 2 (P = .046), 3 (P = .004), and 4 (P = .008).

Table 3.

Paired differences (graduation).

	Paired Differences		P Value (Paired t)
	Mean	SD
Graduation (all students)
Predelay year 0 − postdelay year 1	−6.64	6.41	<.001
Predelay year 0 − postdelay year 2	8.00	8.02	<.001
Predelay year 0 − postdelay year 3	7.50	7.22	<.001
Predelay year 0 − postdelay year 4	9.82	9.58	<.001
Graduation (African-American students)
Predelay year 0 − postdelay year 1	−2.93	15.37	.472
Predelay year 0 − postdelay year 2	3.65	10.78	.182
Predelay year 0 − postdelay year 3	2.00	6.14	.198
Predelay year 0 − postdelay year 4	6.81	11.68	.034
Graduation (economically disadvantaged students)
Predelay year 0 − postdelay year 1	−0.38	9.16	.872
Predelay year 0 − postdelay year 2	4.17	8.84	.062
Predelay year 0 − postdelay year 3	2.83	10.12	.251
Predelay year 0 − postdelay year 4	6.56	8.75	.006
Graduation (White students)
Predelay year 0 − postdelay year 1	−1.44	6.08	.359
Predelay year 0 − postdelay year 2	2.50	4.94	.046
Predelay year 0 − postdelay year 3	3.28	4.13	.004
Predelay year 0 − postdelay year 4	4.67	6.58	.008

SD = standard deviation.

Attendance rates

Table 4 and Table 5 summarize the results of attendance rate means in each measurement window for all students in grades 9–12. Later school start times in the first postdelay period were associated with an increase in attendance rate by 2% (P = .016) compared with the predelay year; in the second postdelay period an increase of 3% (P = .007) was noted; in the third postdelay period an increase of 3% (P = .006) occurred; and by the fourth postdelay period measure, an increase of 3% (P = .012) was found when compared with the predelay data.

Table 4.

Descriptive statistics mean (attendance).

Year	All Students
	Mean	SD
Predelay year 0	90.21	6.47
Postdelay year 1	91.96	4.54
Postdelay year 2	93.54	3.28
Postdelay year 3	93.46	3.75
Postdelay year 4	93.04	3.71

SD = standard deviation.

Table 5.

Paired differences (attendance).

Attendance (All Students)	Paired Differences		P Value (Paired t)
	Mean	SD
Predelay year 0 − postdelay year 1	−1.75	3.62	.016
Predelay year 0 − postdelay year 2	3.32	5.96	.007
Predelay year 0 − postdelay year 3	3.25	5.78	.006
Predelay year 0 − postdelay year 4	2.82	5.54	.012

SD = standard deviation.

DISCUSSION

The current study provides evidence that a change to later high school start times is an effective and long-lasting intervention. Findings differ from a previous report that suggests that significant improvements in high school attendance over 4 years did not persist.³⁵ These findings are consistent with previous studies that demonstrated improvement after a delayed start for high school students.^14,36,38 Importantly, the current investigation fills a gap to conduct within-subject experimental design studies to better understand the long-term benefits of a delayed start time on graduation rates for disadvantaged students.⁴⁰

Attendance rates improved for students in grades 9–12 (all demographics combined) postdelay years 1–4. Specifically, the predelay attendance rates were significantly lower (P < .05) than postdelay years 1–4.

Graduation rates before and after the implementation of delayed high school start times over 4 years were examined in the current study to determine sustained benefits. Student demographic characteristics were also accounted for separately. During the first year after the delayed school start time, significant improvements in graduation rates were noted for grade 12 students (all demographics). During the second year following the time change, graduation rates significantly improved for grade 12 students (all demographics) and for White students. After year 3, significant changes in graduation rates were found for grade 12 students (all demographics) and White students. During the final measurement window of year 4, data revealed significant improvements in graduation rates for grade 12 students (all demographics), African-American students (all economic statuses), students identified as economically disadvantaged, and White students (all economic statuses). Graduation rates for students in grade 12 (all demographics combined) improved after the shift to later school start times in high schools during the 4 time points investigated in the current study. Overall, differences in graduation completion for 4 consecutive years as a possible function of high school start time delays were detected, consistent with prior findings.¹⁴ This study extends and replicates past research suggesting that later start times are associated with improved attendance³⁵ and graduation rates¹⁴ for a sustained period following the shift.

Trends detected in the present study varied significantly according to student demographics. Poor sleep quality is found at a greater prevalence in adolescents identified as economically disadvantaged.²⁶ The establishment of a bedtime routine that matches adolescent sleep needs is important. Researchers found that the greatest predictors for bedtime were socioeconomic factors and daytime activities such as computer use and social activities.²² The factors that influenced bedtime included age, sex, African-American race, and household income. Lack of significant changes until post–year 4 in graduation rates for groups of students identified as economically disadvantaged and African-American suggests that prolonged trial periods are needed to evaluate the impact of delayed school start times. It is possible that 4 consecutive years of improved quality and quantity of sleep increase attendance, leading to improved academic performance and a greater likelihood of graduation completion. Potential obstacles related to obtaining optimal sleep unique to disadvantaged families are important to consider in future research and could provide some insight regarding the lag in response to the delayed start intervention. Chaos in the home, lack of consistent bedtime routine, poor sleep environment, and increased stress are factors that have been previously associated with poor sleep quality and quantity for students identified as economically disadvantaged.⁴¹

There were limitations to these data. Districts in this study opted to implement start time changes. A pre-post design is often used to control for school-to-school variability. This design avoids the need to include a host of covariates because many possible alternative explanations are eliminated. Given the design, it is acknowledged that the schools and school districts are not random samples of all high schools, so generalizability of the results may be limited. There may be other schools that endorse later start times that are not included in the data, or there may be some underlying characteristics of these schools that elected to change start times. Despite the nonrandomized intervention, districts with post-start delays over 4 consecutive years are included to ensure a comprehensive treatment effect. One school district was excluded from the study due to lack of operation for 5 consecutive years. District archives did not include the same demographic information for attendance and graduation rates; therefore, socioeconomic status and race could not be accounted for in the analysis of attendance changes. Specifically, attendance rates for individual grades were not available.

Toward flexibility

Sweeping schedule changes with little prior notice were made during the coronavirus disease 2019 (COVID-19) lockdowns in the winter/spring of 2019–2020. The need for sudden school closures provided a unique opportunity to observe stakeholder willingness to swiftly and collaboratively make change and adopt policies that prioritize student health and well-being.⁴² Shifting to a later school start time did not pose the same obstacles during home confinement because concerns about increased transportation costs, the reliance of older students for afternoon daycare, extracurricular scheduling conflicts, and length of morning commute were absent.⁴³

Objective stakeholder perspectives that support the endorsement of policies that prioritize student well-being should be the goal. Empirical evidence is important to consider when school officials and lawmakers advocate for delays in high school start times. Often, adolescent sleep research is written by experts in various scientific fields.³¹ Discipline-specific terminology used in publications may be difficult for stakeholders to comprehend. School officials with a limited fund of knowledge about the biological underpinnings of adolescent sleep could benefit from research reported in clear language about the unique sleep characteristics for this age.⁴⁴ Scholarly manuscripts should be written to target a wide range of readers, with a focus on cross publishing in both sleep research and educational policy journals.³¹

Despite best intentions, decisions based on empirical evidence will not always prevail. Scientific and nonscientific factors influence how federal, state, and local officials vote on educational policy reform. In both instances, sleep science research was used to justify decisions to delay school start times in California and Boston. Information about perceived obstacles forced the Boston school district to ultimately reverse the endorsement by the board to delay high school start times because of nonscientific reasons.³² Elected board member fears related to job security possibly played a role; however, this concern could be minimized by shifting oversight and decision making to federal and state policy makers.⁴⁵ Decisions that are not based in science are problematic because high school start times impact many students. Eighty-four percent of high school seniors graduated from the 13,600 public school districts in the United States during the 2015–2016 school year.⁸ Armed with scientific research, decisions are more likely to remain student-centered.

From a health equity/disparity perspective, lack of significant changes in graduation rates until post–year 4 in groups of students identified as economically disadvantaged and African American suggest that prolonged trial periods are needed to better understand differences in response to start time delay. Preparation for life after high school includes successful completion of critical tasks such as graduating. Countermeasures used to promote academic and social/emotional growth should be implemented for all students, but especially for disadvantaged students.

Findings from the current study have implications for understanding the impact of race and economic status on sleep acquisition and the impact on graduation. The increased prevalence of shorter nocturnal sleep duration, later sleep onset, and fragmented sleep in African-American students suggests that improving sleep for disadvantaged adolescents is critical.^22,23 Poorer sleep quality and quantity are more frequently reported among adolescents from lower socioeconomic backgrounds.⁴⁶ Adequate sleep is needed for optimal school performance; therefore, the differences in ability to obtain sufficient sleep could contribute to socioeconomic and racial disparities in overall health and school performance.^47,48 A minimum of 4 consecutive years of a delayed school start time to narrow an adolescent sleep gap for African-American and economically disadvantaged students could be necessary. Other possible barriers to obtaining sufficient sleep for disadvantaged students include the role of sleep environment, ability to access coping strategies, and the establishment of sleep routines. Significant results provide evidence that delayed starts could be used as a countermeasure to decrease chronic sleep loss symptoms in adolescent students, increasing the likelihood that the skill acquisition necessary to meet the academic and social/emotional demands is possible for all students.

DISCLOSURE STATEMENT

The authors have seen and approved this manuscript. Work for this study was performed at Hartford Healthcare/BHN-Institute of Living. The authors report no conflicts of interest.