11/12/15

1

The SAT

SAT scores range from 200 to 800 with a median of 500. The standard deviation for a first-time test taker with an

expected score of 500 is ~33. How large a difference between two individual scores is

necessary for statistical significance?

The SAT

The expected value of two SAT scores must differ by more than 100 points to be significantly different.

0 0.02 0.04 0.06 0.08

0.1 0.12 0.14

400 450 500 550 600 650 700 750 800

Prob

abili

tiy

SAT

Comparing Two SAT Scores

Expected 500 Expected 610

The SAT Interpreted

The SAT score should use the AP’s scale: 1 to 5. The SAT norms grades across high schools. There are no correct answers to SAT questions. Test prep increases scores by gaming the test.

4

K12 What works, what doesn’t.

Malcolm Getz Vanderbilt Economics

Two Philosophies

1.  Education fills empty vessels. 2.  Education sparks curiosity.

The Goal

The education of other people’s children has significant value.

A.  People are more productive in groups of educated people.

B.  Markets and democracy work better with educated people.

11/12/15

2

7

Five Questions

1.  Why focus on education now? 2.  Will more spending improve outcomes? 3.  Have tests produced gains? 4.  How do we educate poor children? 5.  Is privatizing effective?

8

#1 Why education now?

•  The returns to schooling are high.

•  Meritocracy emphasizes position.

•  We spend less time at home.

9

The college-high school gap grew. Real Household Mean Income by Education

Mea

n H

H In

com

e (2

013

$s)

50,000

100,000

150,000

200,000

1990 1995 2000 2005 2010 2015Year

under 9 9 to 12 yrHi SchoolBAMA

ProfDoc

10

We seek relative position.

•  A positional race can be enervating. •  We live in a meritocracy where educational

performance defines us. •  Tournament play may discourage all but the top.

11

We spend less time at home. Female Labor Force Participation

Perc

enta

ge in

Wor

k Fo

rce

0102030405060708090

1960 1970 1980 1990 2000 2010Year

Totl SingTotl Mard

Sing ch6-17Mard ch6-17

Sing ch0-6mard ch--6

12

1. Demand is rising.

We have powerful motives for wanting more quality in education.

11/12/15

3

13

#2 Will more spending improve outcomes?

What can we buy? What Pays?

14

Spending per student grew. Real Expenditure per Pupil

$/pu

pil (

2012

$s)

$0

$2,000

$4,000

$6,000

$8,000

$10,000

$12,000$14,000

1960 1980 2000year

Smooth

15

Teacher salaries haven’t grown. Average Teacher Salary US $2013

Elem

enta

ry &

Sec

onda

ry

$0

$10,000

$20,000

$30,000

$40,000

$50,000

$60,000

1960 1980 2000Year

ElementarySecondary

16

Employment 1970-2000

Women were such a small share of engineers, pharmacists, physicians, and lawyers in 1970 that the Bureau of Labor Statistics didn’t report a count. They are 780,000 females in 2000.

17

Average class size has fallen. Pupil Teacher Ratio, Public Schools

Pupi

l/Tea

cher

0

5

10

15

20

25

30

1955

1961

1963

1965

1967

1969

1971

1973

1975

1977

1979

1981

1983

1985

1987

1989

1993

1995

1997

1999

2001

2003

2005

2007

2009

2011

1955 to 201218

Reading scores are steady. Natl Assmnt Educatnl Progress

Rea

ding

4th

& 2

mor

e

200

225

250

275

300

1990 1995 2000 2005 2010 2015Year

Reading 4thReading 8thReading 12th

11/12/15

4

19

Math scores rose. Natl Assmnt Educatnl Progress

Mat

h:4t

h, 8

th, 1

2th

100

150

200

250

300

1995 2000 2005 2010 2015Year

Math 4thMath 8thMath 12th

20

Long-term Impact of Teachers

• Students of an average teacher yields $250,000 more in earnings for pupils as adults compared to a 5th percentile teacher.

Chetty et.al AmEconRvw 2014

21

Spending improves outcomes.

• To improve quality, we could spend more: pay teachers more.

• With rising demand, why aren’t we spending more?

22

#3 Have tests produced gains?

Good principals and teachers make good schools, not good tests.

23

Tests measure limited outcomes.

•  Measure a limited number of skills. •  Teachers teach to the test. •  The same test questions year after year •  Focus on tests makes school boring. •  Multiple choice for art, music, PhysEd.

24

Incentive pay doesn’t work.

Peabody study found no significant difference in gains in test scores with large incentives to teachers.

11/12/15

5

25

Finland has the best schools.

•  Teaching is a valued profession. •  Teachers are well paid. •  Teachers colleges are highly selective. •  Teachers design the curriculum. •  Teachers drive most testing. •  There are few children from poor families.

26

The case for no-stakes tests.

•  Good principals make good schools. •  Good teachers make good schools. •  When incentives are based on tests, we

invite teaching to the test and cheating. •  To raise mean test scores, teach only to the

middle. •  Test scores rise while learning drops.

27

#4 How do we educate the poor?

•  Everyone learns by building on what they know.

•  Experienced, capable, professional teachers design a school program that takes students where they are and moves them forward.

28

A higher percentage of children are poor.

0

5

10

15

20

25

30

35

1965 1975 1985 1995 2005 2015

Perc

enta

ge P

oor

Year

Incidence of Poverty

All ages Under 18 years

29

Kirkpatrick Elementary

East Nashville: Gross family income in Cayce Homes: $8,140 (2012) 720 units, 2,100 people Rest of neighborhood: $28,000. 30% to 48% of pupils enter and exit during the year.

30

% 4-year olds in preschool

11/12/15

6

31

Educate all children.

Providing effective education for children from poor families is challenging.

32

#5 Privatize?

Contractors Charters Vouchers Online Teach for America

There are many ways to privatize.

33

Choose a World?

World A: Households average $50,000 of income; you get $45,000. World B: Households average $30,000 of income; you get $35,000.

34

Three Problems with Private

1.  Exaggerates a positional race. 2.  Exaggerates the role of tests. 3.  Undermines community.

35

Is contracting effective?

•  A school board can write a contract with third party to operate a local school. The contractor’s profit motive gives strong incentives for good performance.

•  From Fran Tarkenton in the 1970s to Chris Whittle in the 1990s, to the charter movement today, contracting has not delivered systematic gains.

36

Are charter schools effective?

•  In recruiting and expelling students, charters create a positional race.

•  Many charters depend on dense testing. •  Charters generally increase segregation. •  Do charters increase community responsibility

for the schools?

11/12/15

7

37

Charters show no general gains.

•  With random assignment of students to charters or regular schools, the charters show outcomes similar to regular schools.

•  Charters often attract some, repel others.

Examples of careful studies

Angrist et al.: Massachusetts charters: a. low-income children in inner city in highly

structured charters: Higher test scores b. Charters in suburban areas: Lower test

scores.

39

Vouchers?

•  A voucher is a payment that the State provides to schools chosen by parents.

•  Parents may be required to pay additional tuition. •  In some settings, schools may choose among

applicants. •  Voucher schools generally need not meet State

standards for teachers, curriculum, buildings, transport, special education, and other services.

40

Online Instruction

Florida and Idaho require it. The online operators get paid without review. Online schooling is a cash cow for investors that provides little education.

Dell, Microsoft, Apple, News Corp, and venture capitalists promote

public spending online.

41

Teach For America http://nepc.colorado.edu/publication/teach-for-america Vasquez Heilig and Jez

TFA teachers perform slightly better than other uncertified teachers, but worse than certified teachers with the same experience. Inexperienced teachers are much less effective than experienced teachers. It costs $70,000 to support a new TFA. Fifty percent leave after two years, 80% after three. TFA is an expensive way to build a corps of experienced teachers.

42

Privatizing: not a clear path.

•  Contracting hasn’t worked. •  Incentive pay doesn’t work. •  Voucher and charter plans depend on the

details of design. Gains are idiosyncratic. •  Online tools are not a substitute for

teachers.

11/12/15

8

43

Do we understand the fundamentals?

•  If declines in home-life and declining teacher quality are important, are charters and vouchers a plausible response?

•  Any choice system will cause sorting with likely gains for some at the expense of others.

44

Progress

•  More support for mothers and children. •  More quality pre-schooling (OK, GA) •  Expect rich, full curriculum. •  Employ great principals. •  Let school systems choose most tests.

45

Conclusions

1.  Demand for school quality is rising. 2.  Spending more on teachers will help. 3.  Good management is essential. 4.  Careful education of poor children pays. 5.  Privatization is not a clear path. 6.  Stratification discourages all but the top.

Oxter