Corrected: An earlier version of this article incorrectly said U.S. secondary-level teachers report spending more hours teaching than students spend in class. It should have said primary-level teachers.
It鈥檚 a statistic that has echoed for years in global policy discussions about education: U.S. teachers are in front of their classes 50 percent to 73 percent more than their peers in other countries, including nations鈥攍ike Finland and Japan鈥攚hose students outperform Americans on international tests.
That striking statistic has become common wisdom as part of the Organization for Economic Cooperation and Development鈥檚 regular Education at a Glance reports, but a new study suggests it鈥檚 significantly overblown.
Teachers in the United States still lead the world in instructional time, but the analysis released last week by the Center for Benefit-Cost Studies of Education鈥攂ased at Columbia University鈥攆inds their edge considerably slimmer:
鈥� Primary school teachers in grades K-6 spend 12 percent, not 50 percent, more time leading class each year than the average in the 34 OECD member countries.
鈥� Teachers in grades 7-9 teach 14 percent, not 65 percent, longer than their global peers.
鈥� Upper-secondary teachers spend 11 percent, not 73 percent, longer on instructional time.
鈥淚 do not want to say in any way that U.S. teachers don鈥檛 work so hard; they work super-hard,鈥� said Samuel E. Abrams, the author of the study and the director of the National Center for the Study of Privatization in Education, at Teachers College, Columbia University in New York. 鈥淏ut when you have such a glaring difference in teaching time鈥攚hen Americans are apparently teaching 73 percent more than their OECD counterparts鈥攕cholars start to focus on that.鈥�
One in 4 teachers report leading class longer than the length of the school day, according to a new analysis of a national survey. Teaching times vary among states.
鈥淭he Mismeasure of Teaching Time,鈥� Center for Benefit-Cost Studies of Education
鈥淏ut it is a phantom problem,鈥� he said, 鈥渁nd it has obscured the critical differences between U.S. teachers and teachers abroad.鈥�
The data problems may cause headaches for researchers and policymakers who have based studies and policies on the OECD鈥檚 Education at a Glance reports and the federal study on which U.S. numbers are based, the Schools and Staffing Survey, in use since 1987.
鈥淏enchmarking education systems in terms of how they work, as well as their outcomes, has become a common practice around the world,鈥� said Pasi Sahlberg, a visiting professor of practice at the Harvard Graduate School of Education and the author of the 2011 book Finnish Lessons. He was one of many researchers who relied heavily on the OECD鈥檚 Education at a Glance statistics to analyze teaching and learning in his native Finland and the United States.
The new study, Mr. Sahlberg said, 鈥渞aises questions about how reliable 鈥榖ig data鈥� in education actually is.鈥�
How did a study so long-standing get it so wrong?
Difficult Calculation
The question of instructional time is a tricky calculation that is required about two-thirds of the way through the 44-page Schools and Staffing survey, which asks teachers and principals to report on school conditions and teaching practices.
The question asks teachers to report the number of hours they spent in the last week 鈥渄elivering instruction to a class of students,鈥� rounded to the nearest hour. It is supposed to collect information on general class instruction and is not intended to include the time teachers spend on study halls or test administration.
Instead, Mr. Abrams found, teachers often rounded each 45-minute class to the hour, rather than only rounding the weekly total. That meant teachers reported being in front of a class 5陆 to 6陆 hours a day, not counting breaks.
After Mr. Abrams reported his data analysis to the National Center for Education Statistics, the arm of the U.S. Department of Education that runs the Schools and Staffing Survey, Thomas D. Snyder, the program director for the annual reports and the U.S. representative to the OECD, conducted a follow-up study.
Mr. Snyder found that 1 in 4 teachers who responded to the survey said they taught longer than the school day ran. The difference between the reported teaching time and the school day ranged from as little as three minutes in Utah to about an hour in New Hampshire.
At the primary school level, according to OECD data, U.S. teachers spend 164 more hours teaching each year than students spend learning in class.
That鈥檚 only possible, Mr. Abrams says in the study, if teachers are lecturing in empty classrooms, have no lunch breaks, team-teach, or teach students in overlapping shifts. While the last two do happen鈥攔arely鈥攖he first is ridiculous on its face, and the second would typically run against teachers鈥� contracts.
U.S. teachers report spending far more hours each year on 鈥渃ontract time"鈥攈ours spent directly instructing a class鈥攖han teachers in other countries in the Organization of Economic Cooperation and Development, but a new study suggests teachers may be confused about the question.
*Note: There no data point for 1999 because there was a change in calculation that year.
SOURCE: Center for Benefit-Cost Studies of Education, Columbia University; OECD iLibrary, Education at a Glance 2014
Always Working?
Those points don鈥檛 necessarily mean such a heavy time commitment is out of the question, said Segun Eubanks, the director of the teacher-quality department for the National Education Association.
鈥淚t is often the case,鈥� he said, that teachers are instructing students directly throughout the entire school day. 鈥淓ven when teachers are not in front of a class, they are doing ... other instructional activities, individual tutoring. ... They are virtually on the clock all day long.鈥�
In seeking a more accurate picture of how U.S. teachers spend their time, Mr. Abrams found that when reporting to the OECD, the Schools and Staffing Survey did not correct for early-release days, professional-development days, and the weeks of class time eaten up by testing.
鈥淭he main problem in the U.S. education [system],鈥� Mr. Sahlberg said, 鈥渋s not teachers鈥� teaching time per se. The real issue is the central role that external testing has as a time-consuming element in the U.S. education system.
鈥淲hen teachers and administrators say that anything between 20 percent to 35 percent of actual teaching and learning time goes to preparing and taking all kinds of standardized tests,鈥� said Mr. Sahlberg, 鈥渋t makes no sense to me.鈥�
The focus on overinflated hours of direct class time, Mr. Abrams said, obscures deeper differences in how American schools structure their time and support their staffs, in comparison to other countries.
鈥淭he real problem is, teachers do not get paid what their counterparts get paid, and the structure of the school day is very different ... in other countries,鈥� he said.
Mr. Abrams pointed to the often-highlighted Finland, where the school day is roughly the same鈥� 345 minutes, just nine minutes less than in the United States. But Finnish secondary students have 270 minutes of instruction and 75 minutes of break time, while U.S. students have 336 minutes of instruction and only 18 minutes of break.
While efforts to extend the school day, week, or year have gained ground among policymakers, Mr. Eubanks, of the NEA, said adding instructional time is less effective without also building in more collaborative-planning time for teachers.
鈥淲e鈥檙e trying to just squeeze these [improvement] ideas in, rather than trying to restructure the schedule in comprehensive ways,鈥� Mr. Eubanks said. 鈥淚t鈥檚 not just about extending the time we have, but about using the time we have in better and more effective ways.鈥�
Search for Solutions
In the meantime, both the NCES and the OECD are trying to correct the data problems.
It鈥檚 slow-going. Mr. Abrams, the author of the new study, first flagged the problems with the teacher data in 2012, but the matter was not noted in later editions of Education at a Glance and ancillary reports.
Jean Yip, an analyst for the Paris-based OECD鈥檚 education and skills directorate, said in an email that the directorate relies on individual countries to supply and check their own data, and that it 鈥渇ully observed鈥� the standard procedures for processing the data, which include notifying the countries if someone finds miscalculations.
鈥淯nfortunately, this does not preclude the possibility that errors may be found in the data at a later time,鈥� Ms. Yip said. She said the directorate is 鈥渓ooking into鈥� the data problems.
While many other OECD countries calculate teaching time using administrative data, Mr. Snyder, of the NCES, said that information is available only at the state level in the United States, not for the country as a whole, 鈥渨hich would be a really big limitation.鈥�
Rather, Mr. Snyder said, the NCES is field-testing new ways to ask teachers about their instructional time. The statistics agency aims to correct the data before the next administration of the Schools and Staffing Survey, in the 2015-16 school year.
鈥淲e realized what we reported is too high, but we鈥檙e trying to get a sense of how much too high,鈥� Mr. Snyder said. 鈥淲hat we鈥檙e doing is evaluating alternative sources in the short term.鈥�
Among those sources is the OECD Teaching and Learning International Study, or TALIS. 鈥淭he TALIS report statistics came from surveys, in this case, teachers鈥攊ncluding those in the U.S.鈥攁nswering the same survey questions about their use of time,鈥� said Stanford University education professor Linda Darling-Hammond, who based a recent analysis of teaching conditions on the TALIS, 鈥渟o I don鈥檛 think they suffer from the same problem as the SASS data or the international comparisons where NCES supplied data to OECD directly.鈥�
Yet Mr. Abrams found even the TALIS overestimates class instructional time by about 200 hours a year, likely also because respondents count test administration and nonclass instructional time, he said.
