Subscribe Today
Home Articles Reader Opinion Editorial Book Reviews Discussion Writers Guide About TCRecord
transparent 13

The Effect of Early vs Delayed Start Times of Classes on Grades of High School Students

by Karen Tarnow - October 10, 2005

Early high school start times have been implicated in adolescent sleep insufficiency; however the effect on school grades has not been investigated, except by student self-report. Therefore, the purpose of this study was to compare the grades of high school students in academic classes that ended before 9:30 am to the grenade of students taking the same course taught by the same teacher later in the day. The grade averages from 1218 paired classes (61,000) students were evaluated for difference. Statistical analysis of the data of all subject courses grouped together revealed a trend, with morning grades being lower than the grades from the same class later in the day.

Research on sleep and its role in the lives of adolescents has been evaluated from many perspectives.  Studies show that teenagers need, on average, 9.2 hours of sleep per night (National Sleep Foundation, 2000; Wolfson & Carskadon, 1998) and experience physical differences involving sleep more than other age groups (Andrade, Benedito-Silva, et al., 1993; Carskadon, 1990; Mercer, Merritt, & Cowell, 1998; Thorpy, Korman, et al., 1998).  Besides the continued need for adequate sleep, teens experience more daytime sleepiness than adults or children and, as a result, experience reduced levels of concentration and alertness (Carskadon, 1990).  Some studies demonstrate that most adolescents are still technically asleep until after 8:00 am (Wahlstrom, 2000) and that daytime sleepiness experienced by teens is most troublesome in the morning (Thorpy, et al., 1998).

As young teens reach puberty, a change in body chemistry causes a shift in the circadian rhythm that affects sleep/wake time within the 24-hour cycle (Carskadon, 1990; Thorpy, et al., 1998).  This shift makes it nearly impossible for teens to fall asleep much before 11:00 pm or midnight (Wolfson & Carskadon, 1998).  Because of early school start times during the week, teenagers experience large disparities between weekend and school night sleep schedules; and they face a sleep deficit that is greater on school nights than weekends.  This asynchronous sleep pattern mimics jet lag, in that the body is awake but responses, alertness, and function are impaired because of the consequences of sleep deprivation (Andrade, et al., 1993).  Consequently, early high school start times deprive the students of a full nights sleep and adverse effects become evident.  However, the effect of sleep deprivation on school grades has not been investigated, except by student self-report in surveys.  Therefore, the purpose of this study was to evaluate the affect of early versus later start times of classes on the grade point averages of high school students.


The grades of high school students in academic classes that started and ended before 9:30 am were compared to the grades of students taking the same course in the same school taught by the same teacher later in the school day.  

To access information about adolescents school performance, it was determined that only public high schools encompassing grades 912 would be contacted for information.  A crucial part of the study was to use information from courses in which there were at least two sections of the same course taught by the same teacher in each school.  It was determined that a high school with a grade 912 population of over 600 students would likely have over 100 students in each grade.  With this grade size, it was most likely that two or more sections of many courses would be offered, which would give information for comparison.

A letter was mailed to the principals of 1160 high schools in 8 different states (Alaska, Alabama, Arkansas, Arizona, California, Colorado, Connecticut, and Michigan).   The contact list began alphabetically with the intent to contact every school on the list.  However, after 1000 letters were sent out it was determined that the scope was too voluminous; and so the last four states were chosen at random from each of the areas of the United States not yet represented.

The contact letter explained the purpose of the study and requested data from the school that would meet the criteria for the study.  Contact information for the researcher was given for fax, email, phone, and address so that school personnel could have questions answered and so that the requested information could be delivered in any format to facilitate participation.

Over the course of 3 months approximately 130 of the schools replied.  Forty-five of the schools could not or chose not to participate for various reasons.  Eighty-two out of the 130 high schools responded with grade information.  The data obtained came in many different forms.  Schools used either a letter grade (A, B, C), a number grade (80, 85, 90) or a grade point average (2.0, 3.0, 4.0).  Because of this variability, a simple grade scale was developed to standardize the information.  Of the 82 respondents, the data from 11 schools could not be used either because it was illegible or because it was not clear that the courses being compared were taught by the same teacher.

A total of 2436 different class average grades obtained were available for comparison from the 71 high schools with usable data.  This represents 1218 (paired) classes of the same course taught by the same teacher early in the morning and then later in the day and corresponds to the grade records of nearly 61,000 high school students from various parts of the United States.  The semester grade averages from all of the students in these classes were averaged together to determine an average grade for the class.  The grades were recorded so that individual subject areas could be analyzed separately to determine differences in the affect that time of day has on learning in different fields of study.  Separate categories were kept for math, science, history, English, and language courses.  Only courses in these 5 academic subjects were assessed.  A table was designed so that data entries were uniform and easily comparable from school to school.  The averages from the 1218 paired classes were subjected to various levels of analysis.


The entire sample was subjected to several levels of analysis and tests were performed for individual subject areas.  Paired t tests were performed for the whole sample (grades of all subjects together) and then with each subject area.  A paired t test compares the average of two paired scores to determine whether there is a statistically significant difference between those averages.  Paired t tests for the entire sample (omnibus test) revealed an increase in grade averages for the later classes as compared to the early ones.  Although the results were not statistically significant for the whole sample, they did approach significance, supporting the hypothesis (Table 1).

Table 1 Paired t-tests of PM minus AM grades









 2 tailed significance

1 tailed


















































# significance level below .10 but above .499999

* significance level below .05

Paired t tests were then performed with data in each subject area.  In science and language the increase in grade average from early to later classes showed statistical significance, confirming the hypothesis.  English and history results approached significance, again supporting the hypothesis that classes held in the early morning adversely affect the grades of high school students.  Math scores demonstrated a reverse in the expected result, with grade averages higher in the early morning classes than in the later classes.  This result, however, was not statistically significant (Table 2).

Table 2 Wilcoxon Paired-Rank tests




   PM Mean



2 tailed


1 tailed significance








* significance level below .05

Because some of these results were of borderline significance in the paired test, a Wilcoxon paired-ranked test, a more robust and exact test, was done to control for experimental variability and was used to more precisely test the probability of the paired t tests.  The Wilcoxon test did demonstrate statistical significance of the increase in class grades from early to later in the entire sample. (Table 2).

In assessing the overall results of this study using different types of analysis, it is possible that the time of day that a class is offered in high school may affect the grades of the students in that class.  


The purpose of this research study was to evaluate whether the time of day that an academic course is offered in high school affects the grades of students in that class.  Issues unique to high school age students such as sleep deprivation and sleep phase delay indicate that high school students may not be awake enough to learn at their fullest capacity before 9:00 am (Andrade, et al., 1993; Dahl, 1999; Epstein, Chillag, & Lavie, 1998; Randazzo, et al., 1998; Harrison & Horne, 1998, 2000).  Other researchers have shown that adolescent students are often still in a sleep phase until 8:00 am (Wahlstrom, 2000).  Many are often struggling to stay awake or may be too groggy to concentrate (Epstein, et al., 1998).  Though grades are only one parameter by which to measure learning in school, it is one of the most objective and predictable methods commonly used in high schools across the United States.  By looking at grades in academic courses taught before 9:30 am and comparing those grades with those of students in the same school taking the same course taught by the same teacher later in the day, it may be possible to get a sense of whether high school students can be as successful at learning academics early in the day as they can later.  This study allowed evaluation of whether high school aged students, who may still be in a sleep phase during early morning classes, were able to rouse themselves to a state of alertness for learning equal to the state of learning by students in later classes.

The high schools that responded to the request for data had an average class size of 25 students.  By looking at 2436 different class averages, the grade records of nearly 61,000 high school students from schools in 8 different states were included.  In addition, data were collected from schools in different parts of the country with mixed demographic populations, varied school sizes, and different starting times.

Analysis demonstrated a statistically significant difference in the grade averages from early morning to later for the combined course sample, implying that high school students learn better and, therefore get better grades, in classes scheduled after 9:30 am (see Graph 1).  When individual course subjects were evaluated separately, there was statistical difference showing that in certain subjects, such as science and language, students also had lower grade scores when taking the course in the early morning than students who took the same course taught by the same teacher later in the day.

By comparing grades only of the exact course taught by the same teacher in the same school, an attempt was made to limit the variable in this study to only time of day.  However, there are several other factors that should be discussed as potential influences of the study results.

The times of day that the classes were offered and attended were carefully monitored.  Regular class time of the early morning courses began and ended before 9:30 am.  The later class time began and ended after 9:30 am, but before the end of the school day.  Therefore, we know the time of day that most of the instruction took place. What isnt clear is whether all of the testing for the course took place at the same time of day as the class.  In many schools, midterm and final exams, the grades from which a considerable percentage of the course grade may be determined, are sometimes scheduled at a time that does not correspond with the usual class time.  Though the expectation is that most of the subject information was learned during the scheduled class period, the results of testing at a different time of day might have an effect on the grades of the students.

Similarly, though the teacher-presented subject material was presented at a given time of day, student study and self-teaching of the subject at other times of the day, evenings, and weekends could affect the amount of material learned and therefore the grade attained in the class.  In other words, a diligent student might fill in the gaps of knowledge resulting from an early morning class experience.

Another possible variable of test grade evaluation is the effect that adrenaline might have on sleepiness.  Students taking tests are often pumped with energizing hormones, which could diminish the effect of drowsiness in a test situation.  Some sleep deprivation literature has postulated that short, novel and stimulating tests would not be expected to be sensitive to sleep loss, but newer studies dispute the excitement effect and show impairment from sleep deficiency despite the conditions (Harrison & Horne, 1998).  

Class scheduling in high schools might be completely random within the grade or, depending on schedule conflicts or limited availability of some courses, might be heavily skewed to one end of the grade spectrum or the other.  For instance, if only one section of an advanced course is available, for example AP Physics, many of the academically strong students might be placed in that course during the period that it is offered.  That could then necessitate placement of these same strong students in the same section of a general course of another subject, (i.e., World History) which could weight the class average to a higher level than random course assignment might.  Or conversely, students assigned to take a remedial math course might then have only one choice of when to take English, thereby weighting the grade average to a lower score.  In the research study of early versus late class grades, there was no information that disclosed this kind of scheduling skew, so some results might be affected by this kind of variable.  With such a large sample of courses being evaluated in this study, however, this kind of effect should be minimized.


The issue of early start times has been discussed in educational circles for many years, but changes are still not being implemented on a large scale basis.  In fact, probably only a small percentage of high schools have actually taken steps toward the needed schedule changes.

The present study sought to add one more objective way of evaluating this issue.  Does taking the same course taught by the same teacher early in the morning versus later in the day make a difference in grade scores for the class?  This study, looking at 1218 paired classes from around the country clearly demonstrated a trend that grade averages are lower in high school classes taken in the early morning than the same courses taught by the same teacher taken later in the school day.  Wilcoxon analysis showed a statistical difference between the lower grades in the early morning class and the higher grades in the later classes for all courses combined.  Further analysis demonstrated a statistical difference between the early and later grades in the subjects of science history and language.  English, on the other hand, did not show any statistical difference in grade point average whether it was attended early or later.  It was also interesting that the only course that showed a reverse in the trend was mathematics, where grade point average was somewhat higher for early morning classes than later classes although this did not reach statistical significance.  This research showed that the early start time of classes may adversely affect the grades of high school students in the subjects of science, history, and language.  The results of this study should bring attention to high school start times and to the times of day that courses in different subject areas are scheduled because of the trends shown between the grades achieved at different times of day.

More prospective, randomized research is needed to verify this trend using average grades, as well as other objective criteria to determine whether students perform as well in early morning classes as they do in the same classes later in the day.


Andrade, M. M., Benedito-Silva, A. A., Domenice, S., Arnhold, I. J., Menna-Barreto, L. (1993).  Sleep characteristics of adolescents:  A longitudinal study.  Journal Adolescent Health, 5, 401­406.

Carskadon, M. A. (1990).  Patterns of sleep and sleepiness in adolescents.  Pediatrician, 17, 512.

Dahl,R.E. (1999).  The consequences of insufficient sleep for adolescents:  Links between sleep and emotional regulation.  Phi Delta Kappan, 80(5), 354359.

Epstein,R., Chillag, N., & Lavie, P. (1998).  Starting times of school: Effects on daytime functioning of fifth grade children in Israel. Sleep, 3, 250256.


Harrison, Y., Horne, J. (2000).  Sleep loss and temporal memory.  Quarterly Journal Experimental Psychology A, 1, 271279.   

Harrison,Y., & Horne, J. (1998).  Sleep loss impairs short and novel language tasks having a prefrontal focus.  Sleep Research, 7, 95100.

Mercer,P.W., Merritt, S. L., & Cowell, J. M.  (1998). Differences in reported sleep need among adolescents.  Journal Adolescent Health, 5, 259-263.  

National Sleep Foundation.  (2000). Adolescent sleep needs and patterns research report and resource guide. Washington DC: National Sleep Foundation.

Randazzo, A. C., Muelback,M. J., Schweitzer, P. K., & Walsh, J. K.  (1998).Cognitive function following acute sleep restriction in children.  Sleep, 8, 861868.

Thorpy, M. J., Korman, E., Spielman, A. J.  (1998). Glovinsky:  Delayed sleep phase syndrome in adolescents.  Journal Adolescent Health Care, 1, 2227.  

Wahlstrom, K.  (November 1998).  School start time and teen sleep. High School Magazine, 7(9), 4041

Wolfson, A., & Carskadon, M.  (1998). Sleep schedules and daytime functioning in adolescents. Child Development, 4, 875-887.

Cite This Article as: Teachers College Record, Date Published: October 10, 2005
https://www.tcrecord.org ID Number: 12217, Date Accessed: 1/17/2022 2:41:57 PM

Purchase Reprint Rights for this article or review
Article Tools
Related Articles

Related Discussion
Post a Comment | Read All

About the Author
Member Center
In Print
This Month's Issue