1) The study finds that additional instructional time has a positive effect on student achievement, though there are diminishing returns.
2) The benefits of additional time depend on the quality of the learning environment - time is less beneficial in classrooms with more disruptions or lower teacher quality.
3) While more time improves scores, the costs of implementing longer school days, such as reduced time for other subjects, must also be considered.
INEE. Ponencia Prof. Rivkin. Universidad Ilinois. Horas clase en materias instrumentales y resultados PISA.
1. Instruction Time, Classroom
Quality, and Academic
Achievement
Steven G. Rivkin and Jeffrey C. Schiman
February 2013
2. Policy Background
• Increase in the quantity of an input would be
expected to increase output in virtually all
circumstances
– Evidence on school resources suggests the
presence of extensive inefficiencies in public
schools
– Raises questions about the benefits of increased
instructional time
3. Longer Instruction Time and Higher
Achievement
• Poland substantially increased time devoted
to 9th grade mathematics instruction in 2001
by over 80 minutes per week on average
• Instruction time at KIPP Charter schools is 60%
higher than the US average
• In each case a number of other factors also
differ, raising uncertainty about the effects of
instructional time
4. Existing Literature
• Supports positive effect of instructional time
• Most research designs fail to account for the
non-random allocation of instructional time
• Higher achieving students attend schools with
a greater emphasis on academics
• Lavy (2010) is an exception
– Method uses within student and school time
variation by subject
– Deficiencies of 2006 categorical time data
5. Conceptual Framework
• Allocation of instruction time depends upon
many factors including ability
• Benefit of additional time likely depends on
classroom quality
– Teacher skills
– Curriculum
– Student behavior
• Behavior and instruction quality may vary with
time
6. Empirical Framework
• Use panel data methods to account for
differences in ability and quality of instruction
that could be related to instructional time
• Aggregate instructional time information to the
school-grade-subject level to mitigate problems
introduced by student class assignments
• Use survey questions to generate measures of
learning environment that will be used to
investigate heterogeneity in time effects
7. Structure of 2009 PISA data
• Tests in mathematics and language arts
• 15 year old students are in multiple grades
• School representative reports class length
– Typically samefor all grades and subjects
• Students report number of classes attended
• School representative answers questions on
classroom environment
8. Empirical model
A sgjc M sgjc sgjc
sgjc s g j c sgc
sj gj sgjc
9. School by grade Fixed Effects
• School by grade fixed effects accounts for
student and school differences common
across subjects (similar to student fixed
effects)
• Subject specific factors that vary at the
individual or school level could introduce bias
10. Accounting for subject specific factors
• Country by grade by subject fixed effects
account for factors such as national curricula,
investments, or emphasis
• School by subject fixed effects account for
differences among subjects in ability, teacher
skill, or quality of curriculum
11. Identification with school by subject
fixed effects
• Cumulative nature of learning attenuates
estimates
– Assumption that 9th grade instructional time has
no effect on 10th grade achievement likely violated
• Uses within subject instructional time
differences between grades
– Assumes cohort differences in ability and grade
differences in the quality of instruction are not
related to grade differences in instructional time
12. Presentation of Results
• Distribution of instructional time and
corresponding test score differences
• OLS and Fixed Effect Estimates of weekly
minutes and weekly classes effects
• Non-linear estimates of instructional time
effects
• Construction of classroom quality variable
• Estimates of interaction with quality
13. Joint Distribution of Classes Per Week
Mathematics
Language Arts 0-2 3 4 5 6+
0-2 15,776 7,690 3,943 2,580 1,687
3 6,565 32,226 19,054 6,942 2,385
4 5,433 20,260 68,412 23,464 5,623
5 1,930 4,410 27,530 62,870 11,384
6+ 1,081 1,586 9,746 16,800 36,011
14. Mathematics minus reading score difference
by numbers of classes
Mathematics
Language Arts 0-2 3 4 5 6+
0-2 -2.3 3.5 -2.2 5 11.5
3 1.4 -1 4.6 3.2 6.6
4 -4.3 -1.4 0.4 0 9.2
5 -12.2 -3.7 -9.6 -1.3 5.6
6+ -14.7 3.1 -0.2 0.2 10.5
15. Table 3. Estimated Effects of Weekly Instructional Minutes and
Classes per Week on Achievement
Panel A:
Weekly Minutes of
Instruction 0.072*** 0.030*** 0.018*** 0.006
(0.008) (0.005) (0.007) (0.009)
Panel B:
Weekly Number of
Classes 5.597*** 2.426*** 1.142*** 0.270
(0.495) (0.335) (0.482) (0.482)
School-by-grade
fixed effects N Y N Y
16. Table 4. Estimated Effects of Weekly Minutes
Weekly Minutes 0.069*** 0.108***
(0.014) (0.018)
Weekly Minutes Squared -0.00006*** -0.00013***
(0.00002) (0.00002)
School-by-grade fixed
effects Y N
School-by-subject fixed
effects N Y
17. Table 6. Estimated Effects of Classes per Week on
Achievement, By Class Length
40 3.251*** 0.905
(0.821) (0.644)
45 1.850*** 1.071*
(0.531) (0.564)
50 3.263*** 1.910***
(0.60) (0.58)
55 4.017*** 1.988***
(0.94) (0.64)
60 3.250*** 0.946
(1.11) (0.74)
School-by-grade
fixed effects Y N
18. Variation in benefits of instructional
time by classroom quality
• Hindrances to learning
– Disruption
– Lower teacher quality
– Ineffective curriculum
• Quality of learning environment might
decrease as class length increases
19. Measurement of Learning Hindrances
Teachers’ low expectations of students
Student absenteeism
Poor student-teacher-relations
Disruption of classes by students
Teachers not meeting individual student needs
Teacher absenteeism
Students skipping classes
Students lacking respect for teachers
20. Table 8. Estimated Effects of Instructional Time, by
Classroom Hindrances
Weekly Minutes of
Instruction 0.080*** 0.051*
(0.017) (0.028)
Weekly minutes-by-
Classroom Hindrances -0.020*** -0.013
(0.006) (0.010)
Weekly Number of Classes 4.560*** 3.338**
(1.033) (1.692)
Weekly Classes-by-
Classroom Hindrances -0.824** -0.819
(0.369) (0.607)
School-by-grade fixed
effects Y N
School-by-subject fixed
effects N Y
21. Summary and Policy Implications
• Evidence supports the hypothesis that
additional instructional time raises
achievement
– Significant school by subject estimates likely
provide a lower bound on effects
• Modest diminishing returns to additional time
• Evidence that benefits depend on quality of
learning environment
22. Determinants of return to instructional
time
• Quality of learning environment
• Opportunity costs in terms of other subjects
foregone or financial cost of additional time at
school
• Return also depends on other benefits of
longer time in school potentially including
– reduced crime and deviant behavior
– Improvements in other valued outcomes
Notes de l'éditeur
Noisy proxies for share of class time available for learning