You are on page 1of 75

Educating Americas Children

Rachel Mathers
Dr. David Brat
Department of Economics & Business
SURF 2005
RandolphMacon !olle"e# $shland %ir"inia 2&005
'0()52)&5&
$EF Economics
*rof. David Brat is the contact person
d+rat,rmc.edu
ABSTRACT
-he purpose of this paper is to e.amine *resident Bush/s 0o !hild 1eft Behind
$ct and to relate this federal pro"ram to %ir"inia/s Standards of 1earnin" tests. 2ith this
+road literature in mind# 3 4ant to determine 4hat factors drive student performance on
%ir"inia S51 tests. -he standard economic research on this topic +6 7anushe8 concludes
that there are ver6 fe4 polic6 varia+les 4hich can alter student test scores. !ard and
9rue"er have challen"ed this findin"# +ut still no optimistic outcome e.ists. 3 4anted to
run these tests for %ir"inia to reach m6 o4n conclusions +ased on the +est science
possi+le.
3 provide an e.tensive literature revie4 4hich reveals man6 criticisms of 0!1B.
!riticisms include settin" unrealistic "oals# "amin" the s6stem# and teachin" to the test.
$nother criticism is that States either accept 0!1B "uidelines or lose federal fundin". 5n
the other hand# accounta+ilit6 has +een much needed and recent 0$E* scores are "oin"
up over time.
Since test scores are used to determine school :ualit6# it is important to 8no4
4hich varia+les affect these scores. Unfortunatel6# m6 research sho4s that man6 of the
ma;or varia+les# such as the num+er of students receivin" free school lunch# cannot +e
manipulated +6 polic6. Broad economic "ro4th or ma;or redistri+ution 4ould +e
re:uired. 5n the other hand# less ro+ust varia+les# such as truanc6 rates# can +e controlled
throu"h the use of appropriate school polic6. Further research 4ill e.amine 4h6 hi"h
school variation e.plained <R
2
=

is lo4 and 4hether all varia+les relevant to our re"ression
stor6 have +een captured.
I. Introduction
2h6 should 4e 4orr6 a+out holdin" schools responsi+le for the performance of
their students> 7ere/s 4h6?
At Luther Burbank School, students cannot take textbooks home for
homework in any core subject because their teachers have enough
textbooks for use in class onlySome math, science, and other core
classes do not have even enough textbooks for all the students in a single
class to use during the school day, so some students must share the same
one book during class timeThe social studies textbook Luther Burbank
students use is so old that it does not reflect the breaku of the former
Soviet !nion" Luther Burbank is infested with vermin and roaches and
students routinely see mice in their classroomsThe school library is
rarely oen, has no librarian, and has not recently been udated" Luther
Burbank classrooms do not have comutersThe school heating system
does not work well" #n winter, children often wear coats, hats, and gloves
during class to kee warm" $leven of the %& teachers at Luther Burbank
have not yet obtained regular, non'emergency credentials"
(
3s this school in a developin" nation> 0o. -his is a description of 1uther Bur+an8
Middle School in San Francisco# !alifornia. 3n m6 vie4# it is a dis"race that 4e allo4
an6 school to +e in such pitiful condition. 7o4 can this +e>
M6 SURF pro;ect 4ill address educational reform in the United States. 3t 4ill
first e.amine the economics of education literature# and this 4ill serve as an introduction
to the uni:ue mar8et for education. 3t 4ill also e.amine man6 aspects of 0o !hild 1eft
Behind# the federal pro"ram 4hich holds schools accounta+le for student performance.
$t the State level# 3 4ill anal6@e ho4 %ir"inia/s Standards of 1earnin" pro;ect fits under
the Federal 0!1B pro"ram. Finall6# 3 4ill do the research necessar6 to determine 4hich
varia+les trul6 have an effect on student test scores. -his 4ill ena+le me to assess 4hether
and to 4hat e.tent state and federal polic6 ma8ers can trul6 reform education. 7o4 much
A
De+orah Meier and Beor"e 2ood# )any *hildren Left Behind <Boston? Beacon *ress# 200(=# )'.
can polic6 ma8ers reall6 do> 3 4ill close +6 discussin" the polic6 implications of this
research.
II. The Economics of Education
Before 4e can ta8e a loo8 at ne4 education initiatives# 4e must first e.amine and
understand the economics of education. -he ma;orit6 of this e.amination 4ill rel6 on the
4or8 of Economist !aroline 7o.+6 at 7arvard Universit6. 7o.+6 descri+es the 4a6 in
4hich mar8ets and incentives can create true reform in 9A2 education. She +e"ins +6
contrastin" the 9A2 mar8et 4ith the hi"her education mar8et. 7o4 do the6 4or8>
C7i"her education in the United States is competitive# in three si"nificant 4a6s.
!olle"e students have much more decisionma8in" po4er a+out 4here the6 4ill attend
colle"e +ecause the6 are more fle.i+le "eo"raphicall6 than elementar6 and secondar6
students. $lso# private institutions offer man6 more numerous options at the colle"e
level than at the primar6 and secondar6 levels. Finall6# and perhaps most si"nificantl6#
colle"e students and their families shoulder more of the +urden of pa6in" for their o4n
educations than do their counterparts in the primar6 and secondar6 sector. -his
additional responsi+ilit6 encoura"es +etter performance on the part of the students# and
more competition amon" colle"es for the students/ pri@ed dollars.D
2
For this and man6
other reasons# our hi"her education s6stem is the env6 of the 4orld. Ever6 countr6 sends
their +est to the U.S.
C-he avera"e primar6 and secondar6 school in the United States does not face
nearl6 as much competition# for several reasons. -he median metropolitan area in the
2
!aroline M. 7o.+6# E2here Should Federal Education 3nitiatives Be
Directed> 9A2 Education %ersus 7i"her Education#D in M. 9osters# ed. +inancing
*ollege Tuition# 2ashin"ton# D!? $E3 *ress# 200A# 2F.
United States has onl6 four school districtsGthat is# parents can choose amon" onl6 four
possi+le elementar6 and secondar6 school districtsH!learl6# the t6pical school district
does not face as much competition as the t6pical colle"e faces to attract students and
parents. 3n addition# parents are relativel6 constrained in their choices amon" school
districts. Even if man6 districts e.ist# s4itchin" +et4een them usuall6 re:uires a chan"e
of residenceG4hich can +e costl6. !hoosin" an elementar6 and secondar6 school
district is simpl6 a more constrained choice than choosin" a colle"e.D
&
3n other 4ords#
the 4a6 to demand a particular elementar6 or secondar6 school is to move# so there is
much less competition at the elementar6 and secondar6 level than at the hi"her education
level.
CFinall6# the amount of competition in primar6 and secondar6 education has
decreased si"nificantl6 over time in the United States. -he first reason is school district
consolidation. 3n AF50# there 4ere more than '5#000 school districts in the United States.
-he nation no4 has fe4er than A5#000 school districts. -hat amounts to almost a si.fold
decrease in the num+er of school districts in the United States. 5f course# man6 of the
districts that 4ere consolidated 4ere rural# and their consolidation ma6 not have had
much effect on the competitiveness of elementar6 and secondar6 education. But the
num+er of metropolitan school districts fell +6 more than 200 percent since AF50# and
their consolidation has undou+tedl6 affected parents/ de"ree of choice.D
(
5nce a"ain# 4e
see that the demand for primar6 and secondar6 education does not function li8e the
demand for most products.
&
3+id.# &A&2.
(
3+id.# &2.
3n addition# schools do not have to 4or8 to find local approval since most of the
school +ud"et comes from the state# not localities. C3n AF(0# the t6pical school district in
the United States raised ;ust under )0 percent of the mone6 it spent. Most of the mone6
4as raised throu"h local propert6 ta.es# and districts therefore had to +e responsi+le to
local homeo4ners and voters. $ district that used its mone6 inefficientl6 4as li8el6 to
e.perience fallin" housin" prices# lo4 voter support for +ond issues# fallin" propert6 ta.
revenues# and ti"ht school +ud"ets. 3n contrast# the t6pical AFF0s school district in the
United States relies primaril6 on the state for its fundin". !urrentl6# 5) percent of the
avera"e school/s fundin" comes from state and federal "overnments. $s a result# school
districts do not compete so hard for local pu+lic support. 3f a school district 4ants to
o+tain a "enerous +ud"et# it ma6 do +etter +6 focusin" its attention on pleasin" the state
le"islature rather than pleasin" local parents.D
5
$nother issue that affects the 9A2 education mar8et is citi@ens/ mindset a+out
education. CMar8et structure affects ho4 people thin8 a+out education. Empiricall6# 4e
8no4 that people thin8 a+out 9A2 education as an entitlement# +ut thin8 a+out hi"her
education as an investment choice that individuals ma8e. -hese different patterns of
thou"ht are important +ecause there is evidence that students are less en"a"ed in their
schoolin" 4hen the6 re"ard it as an entitlement.D
I
Ultimatel6# these differences contri+ute to the "reater efficienc6 of hi"her
education as opposed to 9A2 education. C-he differin" mar8et structure of the primar6
secondar6 and hi"her education sectors accounts for the fact that $merican colle"es and
universities are si"nificantl6 more successful and efficient at producin" education than
5
3+id.
I
3+id.# &F(A.
are elementar6 and secondar6 schools. -his is one reason 4h6 the hi"her education
sector is the lo"ical focus for federal polic6 that see8s to ma8e the +est use of ta. mone6.
$nother reason is that federal mone6 has more levera"e in hi"her education than in
elementar6 and secondar6 education. 3n sum# the disproportionate 4ei"ht traditionall6
"iven to hi"her education in federal +ud"ets seems appropriate.D
)
7o.+6 concludes that# Cultimatel6# federal polic6 should attempt to help the
hi"her education sector do more than educate and remediateGthat is# provide education
to those 4hose secondar6 schoolin" is inade:uate. 3t should ma8e the hi"her education
sector a lever that puts pressure on elementar6 and secondar6 schools to perform. -his
can +e done +6 allo4in" students to use their secondar6 school mone6 for colle"e
courses# as is done in Minnesota. $lso# colle"es can +e re4arded for settin" tou"her
curricular and achievement standards# since these standards set an e.ample for secondar6
schools.D
'
$s can +e seen from 7o.+6/s 4or8# the mar8et structure of hi"her education is
much different than that of primar6 and secondar6 education. Ultimatel6# it appears that
the primar6 and secondar6 schools need the mechanism of competition in order to force
positive chan"es in the schools. 3n order to do this# the demanders <i.e. parents= can +e
"iven more choice a+out 4hich school fits their children the most. Schools 4ould have
to improve or +e forced out of the mar8et +ecause no parents 4ould choose to send their
children there.
)
3+id.# 50.
'
3+id.# 505A.
II. No Child Left Behind
Educational reform has ta8en shape recentl6 in the form of the 0o !hild 1eft
Behind $ct# commonl6 referred to as 0!1B. C-he *resident/s FJ 200( Bud"et 4ill
result in a (A percent increase <K&.F +illion= in -itle 3 spendin" since the passa"e of the
0o !hild 1eft Behind $ct.D
F
C-he stated purpose of the act is Lto close the achievement
"ap 4ith accounta+ilit6# fle.i+ilit6# and choice# so that no child is left +ehind./ -he
ultimate "oal is to esta+lish a s6stem in 4hich A00 percent of children can meet hi"h
academic standards +6 the 20A&20A( school 6ear.D
A0
7o4 do 4e plan to reach this am+itious "oal> C-4o ma;or components of the act
are that states develop ade:uate 6earl6 pro"ress <$J*= accounta+ilit6 +enchmar8s to4ard
the achievement of the ten6ear "oal for schools# divisions# and as a state# and that
schools emplo6 teachers and paraprofessionals 4ho are hi"hl6 :ualified.D
AA
C-he
forerunner of 0!1B 4as the landmar8 Elementar6 and Secondar6 Education $ct <ESE$=
of AFI5HESE$ and its multiple titles <-itle 3 readin" pro"rams# for e.ample= tar"eted
dollars to communities 4ith the "reatest need.D
A2
C3n order to receive -itle 3 funds# states
had to create Lchallen"in"/ content and performance standards in at least readin" and
math# develop assessments that 4ere ali"ned 4ith those standards# and formulate plans to
assist and ultimatel6 sanction failin" schools.D
A&

5ne of the 8e6 "oals in educational reform has +een to +rid"e the achievement
"ap +et4een the poor and the rich. CESE$ 4as the first federal statute that <as part of
F
,resident Bush-s +. /001 $ducation Budget2 Sending )ore, and Sending it )ore 3isely#
http?MMed4or8force.house."ovMissuesMA0'thMeducationMfundin"M+ud"etfactsheet020(0&.htm.
A0
CRevie4 of Factors and *ractices $ssociated 4ith School *erformance#D Senate Noint Resolution &(F
O200& SessionP# A2.
AA
3+id.
A2
De+orah Meier and Beor"e 2ood# )any *hildren Left Behind <Boston? Beacon *ress# 200(=# viii.
A&
Names E. R6an# C-he *erverse 3ncentives of the 0o !hild 1eft Behind $ct#D 4ew .ork !niversity Law
5eview# Sprin" 200(# F&F.
*resident 16ndon Nohnson/s LBreat Societ6/ pro"ram= provided reall6 su+stantial#
precedentsettin" amount of federal mone6 to local schools so that +etter education could
+e provided to historicall6 underserved student "roups such as minorit6 and lo4 income
students. *rior to AFI5# almost all funds for operatin" our pu+lic schools had come from
local ta. dollars.D
A(

0!1B initiated a ne4 strict era of accounta+ilit6 in the nation/s pu+lic schools.
CSchool accounta+ilit6 is the centerpiece of *resident Beor"e 2. BushQs education
reform.D
A5
2ith this ne4 s6stem of strict accounta+ilit6 of all schools in all states# the
federal "overnment is attemptin" to correct the evident "aps +et4een students of different
ethnic or socioeconomic +ac8"rounds and ensure that each and ever6 student is properl6
educated so that heMshe can ma8e a si"nificant contri+ution to societ6.
II.1 The Basics of NCLB
-he 4ord most often associated 4ith 0!1B is accounta+ilit6. C-he
accountability sections of the ne4 la4 refer to those parts of the le"islation intended to
hold pu+lic school educators directl6 responsi+le for the effectiveness of their
instructional efforts.D
AI
3nherent in this idea of accounta+ilit6 is the fact that actions 4ill
+e ta8en if schools are not meetin" the standards. C$ccounta+ilit6 s6stems have an
overall influence on schools in t4o 4a6s? throu"h definin" areas of particular attention
for schools and throu"h providin" re4ards or punishments.D
A)

7o4 4ill the "overnment 8no4 if a school is performin" 4ell> Standardi@ed
testin" is the "overnment/s ans4er to this :uestion. C0!1B calls for all states to install
A(
2. Names *opham# America-s 6+ailing7 Schools <0e4 Jor8? Routled"eFalmer# 200(=# A(.
A5
David 0. Fi"lio# 3hat )ight School Accountability 8o9, http?MM444.n+er.or"MreporterMfall0AMfi"lio.html.
AI
*opham# America-s 6+ailing7 Schools# A5.
A)
Eric $. 7anushe8 and Mar"aret E. Ra6mond# C1essons a+out the Desi"n of State $ccounta+ilit6
S6stems#D Taking Account of Accountability2 Assessing ,olicy and ,olitics, :arvard !niversity# FAA Nune 2002# A).
annual readin" and mathematics tests in "rades three throu"h ei"ht# and onetime readin"
and mathematics tests in "rades ten# eleven# or t4elve no later than the 2005I school
6ear.D
A'
CStartin" in the 200)' school 6ear# states must also administer science tests at
least once in "rades three throu"h five# si. throu"h nine# and ten throu"h t4elveHthe
tests must +e standardi@edHthe tests should +e dia"nostic# providin" relevant details
a+out 4ea8nesses in the students/ master6 of 4hat 4as supposed to +e learned.D
AF
0otice
that it is up to the states to desi"n and score the tests for their students. CStates are free to
determine their o4n standards# to create their o4n tests# and to determine for themselves
the scores that individual students must receive in order to +e deemed Lproficient/.D
20
-hou"h the federal "overnment does not desi"n the tests# some ver6 "eneral "uidelines
are provided to states. For e.ample# Cthe la4 re:uires each state to clearl6 descri+e at
least three levels of student achievementGnamel6# basic, roficient, and advanced"D
2A

3t is o+vious that test scores are a 8e6 component of 0!1B. -hese test scores are
carefull6 scrutini@ed to determine 4hether schools are improvin" their scores and
4hether each su+"roup of students is improvin". C-est scores are the fuel that ma8es the
0!1B$ run. Scores are ta+ulated for schools in the a""re"ate and must +e disa""re"ated
for a num+er of su+"roups# includin" mi"rant students# disa+led students# En"lish
lan"ua"e learners# and students from all ma;or racial# ethnic# and income "roups. $ll of
these scores are then used to determine 4hether schools are ma8in" Lade:uate 6earl6
pro"ress./D
22
A'
*opham# America-s 6+ailing7 Schools# A5.
AF
3+id.# AI.
20
R6an# C-he *erverse 3ncentives#D F(2.
2A
*opham# America-s 6+ailing7 Schools# 2&.
22
R6an# C-he *erverse 3ncentives#D F(0.
C$de:uate 6earl6 pro"ress <$J*=# in turn# is the linchpin of the 0!1B$.
$de:uate 6earl6 pro"ress is tied to 4hether a sufficient percenta"e of students are
performin" proficientl6 on state tests. -he 0!1B$ re:uires states to +rin" all students to
the proficient level 4ithin t4elve 6ears of the $ct/s passa"e <i.e.# +6 20A(=# and states
must ensure that their definitions of ade:uate 6earl6 pro"ress 4ill ena+le the ultimate
t4elve6ear "oal to +e met. -o accomplish this# states must set a proficienc6 "oal each
6ear# and that percenta"e must rise periodicall6 so that +6 20A(# it hits A00R. For a
school to ma8e ade:uate 6earl6 pro"ress# the student population as a 4hole# as well as
each identified subgrou of students# must meet the same proficienc6 "oal.D
2&
3n other
4ords# all t6pes of student populations must +e improvin".
C$de:uate 6earl6 pro"ress is thus less a+out 6earl6 achievement "ains than it is
a+out hittin" uniform +enchmar8s. $ll states must set a uniform +ar for achievement for
all schools and all su+"roups of students 4ithin a school. -he first +enchmar8s 4ere
+ased on test scores from 200AS2. Usin" these test scores# states had to esta+lish a
startin" point for $J* that 4as the hi"her of the follo4in" t4o values? <A= the percenta"e
of students in the lo4estachievin" su+"roup# state4ide# 4ho 4ere performin"
proficientl6T or <2= the threshold percenta"e of students performin" proficientl6 in the
lo4estperformin" :uintile of schools state4ide. 3f &0R of a state/s poor students# for
e.ample# scored at the proficient level in 200AS2# 4hile (0R of all students in the school
at the t4entieth percentile of achievement scored at the proficient level# the initial $J*
+ar must +e at least (0R for all schools and all su+"roups of students.D
2(
C-he $ct
re:uires all schools 4ithin a state# re"ardless of 4hether the6 receive -itle 3 fundin"# to
2&
3+id.
2(
3+id.# F(A.
ma8e ade:uate 6earl6 pro"ress.D
25
-hus# the initial $J* +ar is set lo4 so that all
su+"roups of students must at least achieve the success achieved +6 the lo4est :uintile of
students.
C-he $ct re:uires a rate of improvement on test scores that 4ill +e incredi+l6
difficult# if not impossi+le# to sustain.D
2I
2hat happens if a school can/t meet the
ri"orous $J* re:uirements> C-hose schools that receive federal fundin" and fail to
ma8e ade:uate 6earl6 pro"ress are identified as in need of improvement. -he6 are also
su+;ect to a ran"e of pro"ressivel6 more serious actions. $fter t4o consecutive 6ears of
failure# schools must develop a plan for improvement and are supposed to receive
Ltechnical/ assistance. Students in those schools are also allo4ed to choose another
pu+lic school# includin" a charter school# 4ithin the same district. $fter three 6ears#
students 4ho have not alread6 departed for "reener pastures must +e provided 4ith
tutorin" services from an outside provider# pu+lic or private. -hose schools that fail to
ma8e $J* for four consecutive 6ears must ta8e one of several measures# includin"
replacin" school staff or institutin" a ne4 curriculum# and those that fail for five 6ears in
a ro4 must essentiall6 surrender control to the state "overnment# 4hich can reopen the
school as a charter school# turn over mana"ement to a private compan6# or ta8e over the
school itself.D
2)

2hat do economists thin8 of school choice> CMan6 economists find school
choice appealin" for t4o reasons. First# it +roadens schoolin" options for families 4hose
choices mi"ht other4ise +e limited <+6 such constraints as lo4 incomes# ;o+ location# and
residential se"re"ation=. Second# the increased competition amon" schools ma6 spur
25
3+id.# F(2.
2I
3+id.# F(5.
2)
3+id.# F(2.
efficienc6 "ains +ecause the loss of student revenues provides schools 4ith an incentive
to improve.D
2'
C$lthou"h schools that do not receive -itle 3 funds must in theor6 meet
$J* and 4ill have their test results reported# the6 do not face the pu+licchoice#
restructurin"# or other accounta+ilit6 provisions that the 0!1B$ imposes on -itle 3
schools. 3f a nonS-itle 3 school accepts -itle 3 transfer students# ho4ever# this mi"ht
convert it into a -itle 3 school. 3t is unclear from the 0!1B$ 4hether this 4ould happen
automaticall6 4ith even one transfer# or 4hether it onl6 4ould happen if enou"h poor
students transfer to +rin" the povert6 level of the chosen school to the re:uisite levelH
Schools that accept transfer students 4ho are poor or $frican $merican simpl6 mi"ht
stop doin" so.D
2F
3f this happens# then these students 4ill +e stuc8 in the same
underperformin" school# and the6 ma6 never receive an ade:uate education.
7o4 4ill 4e 8no4 if a state is trul6 improvin"> C-he 0!1B$ re:uires that the
0ational $ssessment of Educational *ro"ress <0$E*= readin" and math tests +e
administered ever6 t4o 6ears to fourth and ei"hth "raders. -he 0$E* is an e.tensive
testin" pro"ram that has +een used for over thirt6 6ears to collect data a+out student
achievement.D
&0
C*rior to the 0!1B$# participation in the 0$E* 4as voluntar6# +ut no4
all states must participate. 0onetheless# onl6 a random sample of students 4ithin each
state must ta8e the test# and scores are not reported for individual students or individual
schools. -he 0!1B$ does not indicate 4hat is supposed to +e done 4ith the results of
the 0$E*# +ut supporters of the $ct su""est that results on the 0$E* 4ill ensure the
ri"or of standards and tests used in each state.D
&A
C-he validit6 and relia+ilit6 of 0$E*#
often called the Lnation/s report card#/ are 4ell accepted. 3t is a test for 4hich students
2'
David 0. Fi"lio# 3hat )ight School Accountability 8o9, http?MM444.n+er.or"MreporterMfall0AMfi"lio.html.
2F
R6an# C-he *erverse 3ncentives#D FI&.
&0
3+id.# F(&.
&A
3+id.
cannot easil6 +e prepped and# since the performance of individual school districts#
schools# or students is not reported# there is little incentive to cheat or even to prepare for
the test. Since the test 4as adopted +efore the advent of state accounta+ilit6 s6stems# it
also provides a neutral standard for assessin" the effects of state policies. -hus#
improvement there reflects more "eneral learnin"# not ;ust responses to the specific state
testin" instruments.D
&2

5ne of the pro+lems 4ith allo4in" each state to create its o4n test is that 4e have
no 4a6 of simpl6 and accuratel6 comparin" results from state to state. 7o4ever# the
0$E* ma6 serve as a ma8eshift# thou"h not perfect# means of comparison. For e.ample#
if a state claims that it/s scores on its o4n test have increased +6 five percent# 4e 4ould#
naturall6# e.pect some increase in their 0$E* scores. 3f the 0$E* scores have not
increased# it is possi+le that the state has t4ea8ed the test or the scorin" scale so that
more students 4ould pass# or it is also possi+le that the teachers are merel6 teachin" to
the test instead of +roadenin" students/ overall 8no4led"e.
-he provisions of 0!1B reach +e6ond mere testin". C$s for teachers# the
0!1B$ re:uires that -itle 3 schools hire onl6 Lhi"hl6 :ualified/ teachers for all su+;ects
and that veteran teachers in such schools demonstrate that the6 are Lhi"hl6 :ualified/ +6
2005SI. -he $ct also reaches +e6ond -itle 3 schools and re:uires that all teachers of
Lcore academic su+;ects/ in nonS-itle 3 schools must +e Lhi"hl6 :ualified/ +6 2005SI.
*ursuant to the $ct and accompan6in" re"ulations# teachers are considered Lhi"hl6
:ualified/ if the6 are full6 certified and have demonstrated competenc6 in the su+;ects
the6 teach. !ompetence is assumed if the teacher ma;ored in the su+;ect in colle"eT
alternativel6# it can +e demonstrated +6 passin" a state test or# for e.istin" teachers# +6
&2
7anushe8 and Ra6mond# C1essons a+out the Desi"n of State $ccounta+ilit6 S6stems#D 2F.
convincin" state evaluators that the6 8no4 their su+;ect areas.D
&&
2ith these
re:uirements# the "overnment is attemptin" to ma8e sure that all schools have 4ell
e:uipped teachers.
II.2 The Benefits of Accountailit! " NCLB
Despite the man6 criticisms of accounta+ilit6# the rationale +ehind accounta+ilit6
is simple and lo"ical. C-he economic rationale +ehind school accounta+ilit6 s6stems is
that the6 4ill provide schools 4ith an incentive to chan"e and to improve their
performance.D
&(
2hen schools are held accounta+le# the6 have incentives to chan"e
+ecause the6 8no4 that there 4ill +e repercussions for their poor performance. 3n a
sense# this is a 4a6 to force chan"e in schools that have +een performin" poorl6 and
haven/t made the chan"es necessar6 on their o4n.
$nother +enefit of accounta+ilit6 is that it could add more of the competition that
7o.+6 su""ested as an important part of the success of schools. 3f students +e"in usin"
the school choice options lin8ed to poorl6 performin" schools# schools 4ill have an
additional incentive to improve so that the6 don/t lose all of their students. 3n the future
4e could +e seein" a much more competitive mar8et for schoolin". *arents can choose
4hat food their children eat and 4hat t6pe of media the6 e.pose their children to# so 4h6
not have more choice in education> 5f course# the lo"istics of lar"escale choice could
+e mess6# +ut it is unli8el6 that most parents 4ould e.ercise their choice options.
II.# Criticisms of NCLB
!riticisms are man6Gthe6 are centered on si. main claims. First# man6 claim
that the 0!1B sets unrealistic "oals 4hich can/t possi+l6 +e met. Second# "amin" is an
&&
R6an# C-he *erverse 3ncentives#D F(0.
&(
David 0. Fi"lio# 3hat )ight School Accountability 8o9, http?MM444.n+er.or"MreporterMfall0AMfi"lio.html.
inherent ris8 4hen there are unrealistic "oals. -hird# race and povert6 are an issue in
re"ards to school success. Fourth# school choice ma6 not 4or8 in the 4a6 it 4as intended
to 4or8. Fifth# there is concern that some students in su+"roups that routinel6 perform
poorl6 on the tests 4ill dropout. 1astl6# there is concern a+out teachin" to the test.
II.#.A $nrealistic %oals
0!1B has +een critici@ed +6 man6 4ho claim that its "oals are unrealistic and
unattaina+le# thus creatin" incentives for states to cheat the s6stem. C-hose 4ho favor
the $ct emphasi@e its lauda+le "oals and cele+rate its tou"h accounta+ilit6 measures.
-hose 4ho critici@e the $ct lament the heav6 emphasis on testin" and the inevita+le
Lteachin" to the test/ that 4ill follo4. -he6 also chastise the federal "overnment for
interferin" 4ith state and local control over education 4hile failin" to fund all of the costs
associated 4ith the $ct.D
&5
C!ommentators consider the ultimate "oal of achievin" A00R
proficienc6 in t4elve 6ears to +e utterl6 unrealistic.D
&I

2hat is the result of these unrealistic "oals> C-he $ct creates counterproductive
incentives +6 esta+lishin" overl6 am+itious achievement "oals and imposin" si"nificant
sanctions for failin" to meet those "oals. 3t allo4s states to act on these incentives +6
leavin" them free to create their o4n tests and scorin" s6stems. -his odd com+ination of
re"ulator6 strin"enc6 and la.it6Hcould 4ell prove disastrous. 3t 4ill encoura"e states to
lo4er their standards# ma8e their tests easier# or lo4er the scores needed to +e deemed
proficient. 3t 4ill promote "reater se"re"ation +6 class and race. $nd# finall6# it 4ill help
push talented teachers a4a6 from schools li8el6 to +e deemed failin"# or from teachin"
alto"ether.D
&)
C3n an attempt to drive education polic6 4ithout intrudin" too "reatl6 upon
&5
R6an# C-he *erverse 3ncentives#D F&&.
&I
3+id.# F(5.
&)
3+id.# F((.
state authorit6# the federal "overnment has com+ined re"ulator6 strin"enc6 re"ardin"
$J* 4ith re"ulator6 la.it6 re"ardin" the :ualit6 of standards and assessments. -his 4ill
li8el6 prove to +e an un4or8a+le compromise.D
&'

II.#.B %aming
2hile sanctions are intended to prompt poorl6 performin" schools to improve#
the6 ma6# in fact# encoura"e those schools to find 4a6s to "ame the s6stem. C-he federal
"overnment can create all the monetar6 re4ards and sanctions it li8es# +ut if states are the
sole ;ud"es of 4hether their standards are sufficientl6 ri"orous# those re4ards and
sanctions 4ill either +e futile or counterproductive.D
&F
CStates have four options. First#
the6 could direct their ener"6 and resources in an earnest effort to improve achievement#
hopin" a"ainst hope that the6 can pull off a miracle and meet the $ct/s "oalsHSecond#
states could stall +6 settin" annual 6earl6 pro"ress "oals in such a 4a6 as to postpone the
need for lar"e increases until later in the t4elve6ear periodH 3f the $ct is not modified
or repealed# ho4ever# postponement 4ill not +e enou"h. $ third strate"6 4ould +e to
i"nore the federal mandate or to decline -itle 3 fundin"H $lthou"h lar"escale defection
ma6 indeed occur over time# at the moment it appears that all states are attemptin" to
compl6 4ith the letter of the la4. -his leaves one last option? ma8e the tests easier or
lo4er the score needed to +e considered proficient.D
(0

3f a state can manipulate the s6stem so that their schools appear to +e doin" 4ell
and don/t receive sanctions# 4hat/s "oin" to stop them from doin" this> C1ouisiana#
!olorado# !onnecticut# and -e.as have all tin8ered 4ith their scorin" s6stems in order to
increase the num+er of students 4ho 4ill +e deemed proficient for purposes of the
&'
3+id.# F').
&F
3+id.# F''.
(0
3+id.# F().
0!1B$H5ther states ma6 alter not ;ust the scorin" s6stem +ut the tests themselves#
ma8in" them easier to pass.D
(A
CStates on their o4n mi"ht en"a"e in a race to the +ottom
+6 creatin" eas6 tests that are not accurate pro.ies for :ualit6 +ut nonetheless "ive the
impression that their schools are "ood. 7ence the pu@@le? 2h6 4ould a state ever use a
test or scorin" s6stem that does not ma8e most schools loo8 "ood> 2h6 4ouldn/t states
instead tr6 to manipulate tests or scorin" s6stems to "ive the impression that their schools
are e.cellent>D
(2
-hus# the sanctions of 0!1B could merel6 result in easier tests# not
+etter schools.
C*rior to the 0!1B$# test scores 4ere some4hat am+i"uous pro.ies for :ualit6#
+ecause "enerall6 lo4er scores in one state could +e interpreted as a si"n of a trul6
ri"orous test. 1o4er test scores "enerated +6 harder tests 4ere also tolera+le +ecause
sanctions 4ere limited and relativel6 mild# at least for schools. $lthou"h states adopted
accounta+ilit6 measures prior to the 0!1B$# the6 reserved the most severe sanctions for
students. 3n ei"hteen states# students had to <and still must= pass e.it e.ams in order to
"raduate# 4hile in others the6 had to <and still must= pass tests to +e promoted from one
"rade to the ne.t.D
(&

Some states that had "ood tests to +e"in 4ith mi"ht actuall6 reform for the 4orse
in order to spare their schools from punishment. C-he re:uirement that an increasin"
percenta"e of students in ever6 school achieve a certain test score each 6ear is ar+itrar6
and unrealistic# in that it esta+lishes achievement "oals 4ithout an6 reference to past
achievement levels or rates of achievement "ro4th. Man6 schools# includin" some that
are considered effective# 4ill +e una+le to meet these achievement tar"ets. -his 4ill
(A
3+id.# F('.
(2
3+id.# F52.
(&
3+id.# F5&.
create pressure to ma8e the tar"ets easier to meet +6 dum+in" do4n the tests or ma8in"
scorin" s6stems more "enerous. B6 this process# a la4 intended to raise academic
standards ma6 lo4er them.D
((

CSupporters of the $ct nonetheless su""est that states 4ill refrain from lo4erin"
their standards +ecause of the 0$E*. Recall that a sample of fourth and ei"hth "rade
students in ever6 state 4ill have to ta8e the 0$E* readin" and math tests ever6 other
6ear. 3f a state/s students perform 4ell on state tests +ut poorl6 on the 0$E*# the
ar"ument "oes# the state 4ill +e em+arrassed into raisin" its standards. -his ar"ument is
not ver6 persuasive. First# state tests and the 0$E* need not +e identical or even similar.
$ state 4hose students perform poorl6 on the 0$E* could easil6 claim that it prepares
students for the state tests +ut not the 0$E*. Second# 0$E* results are not reported for
individual students or individual schools# +ut instead are reported state4ide# and even
then onl6 for a couple of "rade levels.D
(5
C3t thus seems unli8el6 that state and local
officialsGor their constituentsG4ill +e +othered +6 a "ap +et4een state test results and
0$E* results.D
(I
-herefore# states ma6 +e a+le to lo4er their standards in order to ma8e
it seem as thou"h a lar"e percenta"e of students are passin" the state tests# re"ardless of
0$E* performance.
II.#.C Race and &o'ert!
0!1B ma6 not even help the students it 4as most intended to assist. C*rofessors
9ane and Stai"er concluded that schools that contain an $frican $merican or
economicall6 disadvanta"ed su+"roup are much more li8el6 to fail to ma8e ade:uate
6earl6 pro"ress than those that do not. -o improve the chances that a particular school or
((
3+id.# F&(.
(5
3+id.# F5F.
(I
3+id.# FI0.
schools 4ithin a district ma8e $J*# administrators have an incentive to minimi@e the
num+er of $frican $merican or poor students in a school or district. 3mportantl6#
administrators need not e.clude all such students. -he 0!1B$ onl6 re:uires the
disa""re"ation of scores for a su+"roup if it is sufficientl6 lar"e to 6ield Lstatisticall6
relia+le information./ Because there is no sin"le formula for determinin" this fi"ure# the
0!1B$ allo4s states to determine the minimum si@e of su+"roups.D
()

3nstead of assurin" that minorit6 students are meetin" the standards# states could
simpl6 set a hi"h num+er for the minimum si@e that su+"roups must +e in order for their
scores to affect 4hether a school meets their $J* "oal. 3f this happens# these students
4ill continue to +e ne"lected# thou"h it/s unli8el6 that a state could set an e.tremel6 hi"h
num+er for the minimum 4ithout +ein" noticed.
II.#.( Lac) of Real School Choice emedded in NCLB
CRecall that the 0!1B$ allo4s students in -itle 3 schools that fail to ma8e $J*
for t4o consecutive 6ears to attend another pu+lic school 4ithin the same district. 5nl6
schools that have made $J* are eli"i+le to receive transfer students. 3f there are no such
schools 4ithin the district# the 0!1B$ and its re"ulations encoura"e +ut do not re:uire
districts to arran"e for students to attend school in another district. -he 0!1B$
re"ulations also su""est that lac8 of space in a L"ood/ school 4ithin the same district is
not a sufficient reason to den6 students their ri"ht to choose another school.D
('
C3f
minorit6 and poor students disproportionatel6 do 4orse on standardi@ed tests# -itle 3
schools 4ith such students 4ill +e more li8el6 to fail to ma8e $J*. $s a result# man6
minorit6 and poor students 4ill have the option to transfer. -he schools to 4hich the6
()
3+id.# FI2.
('
3+id.# FII.
transfer are more li8el6 to +e 4hite and middle class if# a"ain# past performance on
standardi@ed tests is an6 indication.D
(F

C$s a result# the operation of the pu+lic school choice provision in the 0!1B$
ma6 promote "reater racial and socioeconomic inte"ration. -his is indeed a possi+ilit6#
and for those 4ho favor "reater inte"ration# it is a 4elcome one. -here are reasons to +e
s8eptical# ho4ever# that the choice provisions 4ill pla6 out in the 4a6 ;ust descri+ed. -o
+e"in# it is important to reco"ni@e that interdistrict choice is not re:uired +6 the 0!1B$.
3n man6 metropolitan areas# se"re"ation occurs +et4een rather than 4ithin districts# and
in these areas the 0!1B$ choice provision offers little hope of promotin" inte"ration.
Second# 4here there is diversit6 4ithin a "iven district# space constraints 4ill surel6 limit
the amount of movement. 3t is inconceiva+le that states and districts 4ill a+ide +6 the
re"ulation that su""ests a lac8 of space is no e.cuse for failin" to "uarantee school
choiceH administrators of successful schools ma6 claim that the6 lac8 much# if an6#
space for transfer students.D
50
C-here is a littlenoticed provision in the 0!1B$ that
ma8es the school choice provision contin"ent on state permission. -he 0!1B$ re:uires
schools to offer choice unless the6 are prohi+ited from doin" so +6 state la4. $lthou"h
this mi"ht +e an e.treme move# it is possi+le that# if nothin" else 4or8s# states 4ill enact
la4s prohi+itin" school choice. -a8en to"ether# all of these o+stacles ma8e it unli8el6
that the 0!1B$ re:uirement of offerin" choice 4ill +e sufficient to overcome the stron"
incentives to maintain or increase racial and socioeconomic se"re"ation.D
5A
-herefore#
there ma6 not +e an6 real choice offered +6 the school choice provision of 0!1B.
II.#.E Incenti'es for (ro*outs
(F
3+id.# FI).
50
3+id.# FI).
5A
3+id.# FI'.
Schools ma6 +e so desperate to "et rid of those students 4ho perform poorl6 that
the6 ma6 actuall6 discoura"e poor and minorit6 students from sta6in" in school.
CSchools# to the e.tent the6 can# 4ill 4or8 to avoid enrollin" those students 4ho are at
ris8 of failin" the e.ams. -he same pressure could lead schools to push lo4performin"
students out# either to another school <if one can +e found that 4ill accept them= or out of
the school s6stem entirel6. -his temptation presuma+l6 4ill +e stron"est at the hi"h
school level# +oth +ecause students most t6picall6 drop out at this sta"e and +ecause lo4
performin" hi"h school students are most li8el6 to +e farthest +ehind. Biven the
connection +et4een performance on tests# socioeconomic status# and race# the students
most li8el6 to +e tar"eted for e.clusion 4ill +e poor andMor racial minoritiesH5ne less
student performin" +elo4 the proficienc6 level increases the overall percenta"e of
students 4ho have hit that +enchmar8.D
52

3nstead of helpin" those students 4ho are most in need of help# schools could
push these students out# leavin" them little hope of o+tainin" a ;o+ that pa6s a+ove
minimum 4a"e and provides mentall6 stimulatin" 4or8. C-he 0o !hild 1eft Behind $ct
provides 4ea8 protection a"ainst this temptation. 3t re:uires that "raduation rates +e
included as part of a school/s determination of $J*# +ut it does not sa6 4hat the rate
must +e# nor does it demand that the rate increase over a certain period of time.
Moreover# "raduation rates can onl6 +e counted against a school 4hen determinin" $J*.
$ school 4ith poor test scores# in other 4ords# cannot point to a relativel6 hi"h
"raduation rate and there+6 ma8e $J*.D
5&

52
3+id.# FIFF)0.
5&
3+id.
C5n the other hand# a school 4ith "ood test scores +ut lo4 "raduation rates could
+e at ris8 of failin" to ma8e $J* if the state sets a hi"h tar"et for "raduation rates. States
thus have little incentive to esta+lish a demandin" "raduation rate. -he lo4er that rate is
set# of course# the easier it is for schools to push out students. -o +e fair# the 0!1B$
does re:uire that information a+out "raduation rates +e made pu+lic. Disseminatin" this
information is far from useless# +ut it remains to +e seen 4hether simpl6 pu+lishin"
"raduation rates 4ill provide sufficient protection for students at ris8 of +ein" pushed out.
3f it does not# and if dropout rates increase# the 0!1B$ could end up further harmin"
those students 4ho o+viousl6 need the most helpGleavin" them# :uite literall6#
+ehind.D
5(
3nstead of receivin" an ade:uate education# students 4ould +e pushed out of
school +efore the6 could "arnish the s8ills needed to succeed.
II.#.+ Teaching to the Test
-he "oals of 0!1B are so difficult to achieve that it is o+vious that much
instructional time 4ill +e devoted to test preparation. C$ s6stem +uilt solel6 on test
scores# for e.ample# filters out ever6thin" e.cept student academic achievement.
Similarl6# if some su+;ects are tested and others are not# it is natural to thin8 that attention
4ill "o more to the tested areas than the untested areas. Related# part of the de+ate a+out
testin" has ar"ued that tests of lo4er order s8ills tend to drive out attention of schools to
hi"her order s8ills.D
55

-his could ta8e some of the ;o6 out of teachin" and discoura"e potential
emplo6ees from enterin" the profession in the first place. CEsta+lishin" standards and
re:uirin" periodic testin" mi"ht protect students a"ainst unmotivated teachers 4ho# "iven
5(
3+id.
55
7anushe8 and Ra6mond# C1essons a+out the Desi"n of State $ccounta+ilit6 S6stems#D A).
the chance# 4ould shir8 their responsi+ilities. $t the same time# ho4ever# reducin" their
autonom6 can ma8e teachin" less attractive to ver6 "ood teachers. -hose teachers 4ho
can +e trusted to motivate and teach their students ma6 find teachin" less re4ardin" the
more the6 are shac8led to state standards and tests. *rotectin" students a"ainst +ad
teachers can thus simultaneousl6 deter "ood ones from enterin" or remainin" in the
profession.D
5I

For e.ample# CEn"land/s e.perience is instructive. $ little more than a decade
a"o# under Mar"aret -hatcher/s leadership# the British "overnment introduced a national
curriculum# 4hich descri+ed in precise detail 4hat each child should learn in each "rade.
-he British "overnment also implemented a series of mandator6 tests in En"lish# math#
and science for students at a"es )# AA# A(# and AI. -he tests 4ere desi"ned as La form of
handcuffs#/ to ma8e sure that teachers follo4ed the national curriculum. -en 6ears later#
En"land is facin" a severe shorta"e of teachers. 5ther factors have contri+uted to the
shorta"e# of course# +ut the same trend e.ists else4here. $s one report o+served# in
countries 4here accounta+ilit6 s6stems have undermined teacher autonom6# li8e
En"land# !anada# and $ustralia# there is a Lrecruitment crisis/.D
5)

2hile the United States still "ives states freedom in the desi"n and scorin" of
their tests# are 4e movin" in the direction of En"land/s polic6> 2hile 3 thin8 this is a
possi+ilit6# it appears to +e a remote one. 3f the federal "overnment used a national test
as the +asis of state educational evaluation# the states/ ri"hts 4ould +e trampled# so this
seems unli8el6 to happen. States 4ould certainl6 protest such a move# so 3 don/t see a
national test as the cause for teachers to leave the profession in the near future. 7o4ever#
5I
R6an# C-he *erverse 3ncentives#D F)2.
5)
R6an# C-he *erverse 3ncentives#D F)2.
the6 could flee for other reasons. CSchools 4ith poor test scores# or even those that
"enerall6 have "ood test scores +ut have one lo4performin" su+"roup# 4ill not ma8e
ade:uate 6earl6 pro"ress. -eachers in those schools 4ill have to suffer the sti"ma of
+ein" associated 4ith failin" schools# 4hich can limit future career opportunities.
-eachers 4ho remain in schools that consistentl6 fail to ma8e ade:uate 6earl6 pro"ress
face the possi+ilit6 of +ein" fired or moved to another school.D
5'

C3n addition to these sanctions# imposed +6 the 0!1B$# some state accounta+ilit6
s6stems also create the possi+ilit6 that teachers in lo4performin" schools 4ill +e fired# or
that the6 4ill face the dispiritin" prospect of 4atchin" their collea"ues receive +onuses
for "ood test results. $ttachin" conse:uences to failure ma6 +e necessar6 to provide
incentives to ta8e the tests seriousl6. But it raises the costs associated 4ith failure# 4hich
ma6 ma8e teachin" even less attractive.D
5F
-hese tests could potentiall6 chase a4a6
some of the nation/s +est teachers from the profession.
II., -hat Changes -ould .a)e NCLB Better/
-hree chan"es that could improve the 0!1B are hi"hli"hted +elo4. -hese
chan"es are the use of "ro4th rates# the use of alternative measures of success# and the
settin" of realistic "oals.
II.,.A +ocus on Test Score %ro0th Rates and not Asolute Le'els
3t/s :uite eas6 to attac8 a piece of le"islation# +ut it is much more difficult to thin8
of constructive 4a6s to ma8e it +etter. 5ne of the most common su""estions is to focus
on the "ro4th rates of test scores rather than the a+solute scores. -his 4ould +e a s6stem
that could re4ard a school for havin" ama@in" improvement even if the school still
5'
3+id.# F)&.
5F
3+id.# F)&.
doesn/t meet the ri"orous standards. Re4ard for improvement could motivate these
schools to 8eep up the "ood 4or8# as opposed to sanctionin" them for not reachin" the
a+solute "oals 6et. CRather than focus on a+solute achievement levels as the +asis for
school accounta+ilit6# R6an ar"ues that the federal "overnment and states should focus
on rates of "ro4th. Doin" so 4ould not onl6 "ive a more accurate picture of school
:ualit6# and thus provide a fairer +asis for school accounta+ilit6T it 4ould also diminish
or eliminate the perverse incentives created +6 the 0o !hild 1eft Behind $ct.D
I0

II.,.B $se Alternati'e .easures of Success
-est scores shouldn/t +e the onl6 measure of a school/s success +ecause there are
man6 factors that could ma8e test scores unrelia+le. CMan6 factors can ma8e a test score
inaccurate. For e.ample# there mi"ht +e am+i"uous items in the test itselfT the child ma6
not +e feelin" 4ell on the da6 the test 4as ta8enT the child ma6 have eaten too little# or
too much# +efore the testT or the child ma6 also have +een e.cessivel6 an.ious a+out the
test. -hat/s especiall6 true for ver6 important tests. -hen# too# on an6 "iven da6 a
student ma6 simpl6 "et luc86 and "uess correctl6 most of the time or# in contrast# ma6
have a +adluc8 da6 and come up 4ith man6 incorrectl6 "uessed ans4ers. Sometimes
studentsGespeciall6 teena"ersGdon/t ta8e tests as seriousl6 as educators hope# so the
student/s don/t tr6 to do their +est.D
IA

C3n short# there are numerous factors that can reduce the accurac6 of students/ test
scoresH$ test score should +e seen onl6 as a rou"h aroximation of a student/s actual
I0
3+id.# F&2.
IA
*opham# America-s 6+ailing7 Schools# 55.
achievement level.D
I2
C$ valueadded s6stem of accounta+ilit6 4ould provide a more
accurate picture of school :ualit6 and 4ould not "enerate the same perverse incentives
that have +een unleashed +6 the 0!1B$.D
I&
C0o attention has +een paid to 4hether the
level of "ro4th demanded is at all feasi+le# nor is much credit "iven to schools that
accomplish more than avera"e levels of "ro4th +ut still fall short of the uniform $J*
+enchmar8 or the safe har+or tar"et. 3f the levels of "ro4th re:uired are indeed not
feasi+le# this creates the pressure# descri+ed a+ove# to ma8e the tests easier or to lo4er the
scores necessar6 to +e deemed proficient.D
I(

C$ll of this su""ests that a more appropriate +asis for accounta+ilit6 4ould +e one
that isolates the :ualit6 of the school. -his is precisel6 4hat socalled Lvalueadded/
methods of assessment attempt to do.D
I5
C$lthou"h this method is fairl6 comple. in its
details# its +asic approach is to focus on achievement "ains over time for the same
individual or "roups of students. -he underl6in" supposition is that if 4e 8no4 ho4
much a student/s achievement has improved from one 6ear to the ne.t# 4e have a much
+etter sense of 4hat value the school has added to the student/s academic performance.
-he reason is fairl6 simple? E.o"enous factors that affect achievement 4ill influence
achievement ever6 6ear that a student is tested. B6 focusin" on "ains made or lost +6 the
same students# rather than overall levels of achievement# those e.o"enous factors are
cancelled out. $t least that is the idea.D
II
C3n attemptin" to isolate a school/s
performance# valueadded approaches "enerall6 i"nore a+solute levels of achievement.
Because the focus is on achievement "ro4th# schools can +e ;ud"ed effective even if
I2
3+id.
I&
R6an# C-he *erverse 3ncentives#D F)'.
I(
3+id.# F'0.
I5
3+id.# F'0.
II
3+id.# F'A.
most or all of their students are performin" +elo4 avera"e# or +elo4 an esta+lished
proficienc6 level. %alueadded s6stems that ta8e into account factors li8e race or
socioeconomic status e.acer+ate this pro+lem# at least politicall6# +ecause the6
essentiall6 esta+lish different "ro4th tar"ets for different t6pes of students. 2hat ma8es
these s6stems potentiall6 more accurate measures of school performance is precisel6
4hat ma8es them politicall6 controversial.D
I)

CBecause valueadded s6stems do not create the same unproductive incentives#
and do not paint as inaccurate a picture of school :ualit6# on +alance the6 seem prefera+le
to the approach codified +6 the 0!1B$. -o +e sure# the6 are an imperfect solution and
carr6 the ris8 of as8in" too little of some students and schools.D
I'
3n effect# some schools
ma6 "ive up on the ultimate "oal of A00 percent proficient if the6 8no4 that the6 4ill not
+e punished for not meetin" $J* "oals as lon" as the su+"roups in their school improve
<i.e. their valueadded is positive=.
CRe4ardin" schools for meetin" "ro4th tar"ets that are realistic ma6 not provide
sufficient incentives for schools to do an6 more than improve mar"inall6 on the status
:uo. Focusin" on "ro4th# moreover# necessaril6 tolerates different levels of a+solute
achievement for different students. $ school 4hose fifth "raders +e"in the 6ear readin" at
the third"rade level and end readin" at the fourth"rade level certainl6 has added value
and done a decent ;o+T +ut those students 4ould still +e a 6ear +ehind in readin". Despite
these shortcomin"s# a valueadded approach seems clearl6 superior to the approach
em+odied in the 0o !hild 1eft Behind $ct.D
IF

I)
3+id.# F'&.
I'
3+id.# F'5.
IF
3+id.# F&5.
7o4 can 4e set "ro4th tar"ets for students> CBecause 4e do not have "ood
information a+out 4hat schools# teachers# and students are capa+le of achievin" over a
certain period of time# an6 accounta+ilit6 s6stem is +ound to as8 for too little or too
much. 3f it as8s for too little# it ma6 +e selflimitin". 3f it as8s for too much# it ma6 +e
selfdefeatin"# for all of the reasons 3 have descri+ed. 3t is temptin" to suppose that a
h6+rid s6stem# 4hich com+ines a focus on a+solute tar"ets and "ro4th in achievement#
could solve the pro+lem of as8in" too much and as8in" too little. But this is a false hope#
primaril6 +ecause ne4 students enter the school s6stem each 6ear.D
)0
II.,.C Realistic %oals
Even the "oals of 0!1B themselves should +e revised. 3t/s ver6 idealistic to
proclaim that all students should pass their state tests at the proficient level +6 20A(# +ut
this is a completel6 unrealistic "oal. -he federal "overnment should set realistic "oals
that can +e met# 8eepin" in mind that it 4ill never +e the case that all students pass
proficient. C-here is a fundamental pro+lem that it is impossi+le to attain A00 percent
proficienc6 levels for students on normreferenced tests <4hen 50 percent of students +6
definition must score +elo4 the norm and some proportion must +6 definition score
+elo4 an6 cut point selected=# 4hich are the 8ind of tests that have +een adopted +6 an
increasin" num+er of states due to the specific annual testin" re:uirements of 0!1BH
Usin" a definition of proficienc6 +enchmar8ed to the 0ational $ssessment of
Educational *ro"ress <0$E*=# one anal6st has calculated that it 4ould ta8e schools more
)0
3+id.# F'&.
than one hundred 6ears to reach such a tar"et in all content areas if the6 continued the
fairl6 +ris8 rate of pro"ress the6 4ere ma8in" durin" the AFF0s.D
)A

III. -hat (o Test Scores Im*l!/ -hat to (o/
5nce 4e have these test scores# ho4 should 4e trac8 and report them> -here are
several options that 4ill +e discussed +elo4. -he first 4a6 to loo8 at test scores is called
the status chan"e model. -he second 4a6 is the cohort "ain model. -he third option is
the individual "ain score model. 1astl6# 3 revie4 the various t6pes of reportin" to
parents.
III.1 Status Change .odel
0o4 that states must test their students for proficienc6# the :uestion is ho4 the
"overnment should "o a+out usin" these scores. 5ne option is the status chan"e model.
C-he status chan"e model for a school is calculated +6 a""re"atin" performance across
tested "radesH2e classif6 this model as crosssectional# +ecause it compares snapshots
of the school scores across 6ears <as opposed to trac8in" the performance chan"es for
)A
Meier and 2ood# )any *hildren Left Behind# A0.
individual students across 6ears=. -he status chan"e model is +6 far the most common
approach to assessin" 4hat is happenin" in schools.D
)2
5ne issue 4ith this approach is
that 4e 4on/t see the pro"ress of individual students in order to assess the actual
contri+ution of the school.
III.2 Cohort %ain .odel
$ +etter option is the cohort "ain model. C$ccounta+ilit6 is :uite different 4hen
it focuses on the pro"ress of students over time# 4hich 4e classif6 as a lon"itudinal
approach. 5ne such approach is the Lcohort "ain/ model. -his approach trac8s the
performance of individual cohorts of students as the6 pro"ress throu"h school. !onsider#
for e.ample# comparin" the scores of third "raders in 200A 4ith those of fourth "raders in
2002. 2ith a sta+le student +od6 <i.e.# 4ith no in or out mi"ration for the school=# the
historical school and nonschool factors 4ould cancel out <+ecause the6 influence a
cohort/s performance +oth in "rade & and "rade (=. -he cohort "ain score 4ould then
reflect 4hat the school contri+uted to learnin" in "rade ( plus an6 differences in
idios6ncratic test factors or measurement errors across the t4o "radesH$s a result# the
cohort model 4ould "enerall6 6ield a closer measure of school inputs than the status
model.D
)&
III.# Indi'idual %ain Score .odel
$n even more refined approach is the individual "ain score model. 2ith this
approach# it is actuall6 possi+le to trac8 the school/s contri+ution to an individual
student/s learnin". -his method also avoids the pro+lem of student mo+ilit6 in the
calculation of a school/s scores. C-he influence of mo+ilit6 su""ests an alternative
)2
7anushe8 and Ra6mond# C1essons a+out the Desi"n of State $ccounta+ilit6 S6stems#D ).
)&
3+id.# FA0.
measure for accounta+ilit6# the Lindividual "ain score/ model. -his approach improves on
cohort chan"e models +ecause it anal6@es data at the student level and can include all
students 4ith "ain scores# not ;ust the students in the ori"inal "roup. 3f 4e follo4
individual students across "rades# an6 historical influences of families and nonschool
factors 4ash out# and the avera"e of individual "ains across "rades 4ould more closel6
reflect school :ualit6 for the "iven "rade.D
)(
III., Re*orting Scores
3n addition to the 4a6 scores are trac8ed# there are t4o 4a6s that an individual
student/s scores can +e reported. -he t4o t6pes of reportin" are relative reportin" and
a+solute reportin". C$ relative score reort indicates ho4 a student/s performance
compares to that of other students. For instance# 4hen the parents of a fourth"rade
student learn that she scored at the I)
th
ercentile in mathematics on a standardi@ed
achievement test# this means she outperformed I) percent of the fourth "raders 4ho 4ere
included in the test/s norm grou. $ norm "roup is a representative sample of students
4ho completed a standardi@ed test soon after it 4as developed. $ standardi@ed test# +6
the 4a6# is simpl6 a test that must +e administered and scored in a standard#
predetermined manner. Because traditional standardi@ed tests invaria+l6 use norm"roup
comparisons to interpret a "iven student/s scores# man6 educators descri+e such tests as
Lnormreferenced./ -hat/s +ecause# to infer 4hat a child/s score actuall6 means#
educators must Lreference/ that child/s score +ac8 to the scores made +6 the test/s norm
"roup.D
)5

)(
3+id.# AA.
)5
*opham# America-s 6+ailing7 Schools# 5&.
CAbsolute score reorts# on the other hand# focus on 4hat it is that the child can or
can/t doH-he report mi"ht sa6 somethin" such as the student has mastered this s8ill at a
Lproficient#/ or perhaps at an Ladvanced#/ level. Educational tests +uilt chiefl6 to provide
a+solute interpretations are sometimes referred to +6 educators as Lcriterionreferenced/
+ecause# 4hen interpretin" scores# educators Lreference/ the student/s performance +ac8
to a clearl6 descri+ed criterion +ehavior such as a 4elldefined s8ill or +od6 of
8no4led"e.D
)I
C$+solute score reportin" schemes are far more usefulGinstructionall6G
to +oth teachers and parents. -hat/s +ecause such reports indicate more clearl6 4hat it is
the child alread6 8no4s or# in contrast# 4hat the child needs to 4or8 on. Relative scores
fail to provide such clarit6# for the6 offer onl6 a comparative picture of the child/s
attainmentsH-6picall6# a+solute score reports are supplied either as Lpercentofmaster6/
reports or Lperformance la+els/ such as basic, roficient, or advanced"D
))
)I
3+id.# 5&.
))
3+id.
I1. 1irginias Standards of Learning
%ir"inia/s tests to satisf6 0!1B re:uirements are called the S51s <Standards of
1earnin"=. CU.S. citi@ens repeatedl6 cite improvin" schools as the num+er one priorit6
for polic6ma8ers. For this reason# the !ommon4ealth of %ir"inia undertoo8 a ma;or
development aimed at creatin" clear# measura+le o+;ectives and promotin" accounta+ilit6
of schools. Based on the 4or8 of over 5#000 individuals representin" parents# teachers#
+usinesses# and education officials# the Board of Education adopted the Standards of
1earnin" <S51= in AFF5. -he S51 formulated a set of re:uirements "overnin" 4hat
teachers 4ould teach and 4hat students 4ould +e e.pected to learn. Emphasis 4ould +e
placed on four core su+;ectsGEn"lish# mathematics# science# and histor6 and social
science. -he S51 4ould +e tested throu"hout a student/s career# 4ith testin" in &
rd
# 5
th
#
'
th
# and A2
th
"rades.D
)'
%ir"inia uses "raduation rates and attendance rates as other
re:uirements <in addition to testin"= necessar6 to meet $J* standards.
-he S51s si"nificantl6 impact +oth schools and students. C200&200( is the first
6ear in 4hich unsatisfactor6 S51 test results 4ill +e used to potentiall6 den6 diplomas to
students 4ho 4ould other4ise :ualif6 +ased on classes ta8en and "rades received.D
)F

-his is a heft6 penalt6 for a student to face. CBraduation from hi"h school is no4
conditioned on passa"e of desi"nated endofcourse S51 tests. Students are re:uired to
have 22 credits in order to receive a standard diploma. Be"innin" 4ith the 2000200A
ninth "rade class <"raduatin" class of 200(=# si. of these 22 credits must +e verified
credits. 3n order to receive verified credits# a student must ta8e and pass an endofcourse
)'
David 1ehr and Melanie Mar8s# C*erformance %ariance on %ir"inia/s Standards of 1earnin"? $
*reliminar6 School1evel Empirical $nal6sis#D Su+mitted to ;irginia $conomic <ournal, <200A=.
)F
CRevie4 of Factors and *ractices $ssociated 4ith School *erformance#D Senate Noint Resolution &(F
O200& SessionP# U.
S51 test. 5f the si. verified credits re:uired to "raduate# t4o verified credits must +e in
En"lish and the remainin" four credits ma6 +e in su+;ects of the student/s choosin".
Be"innin" 4ith the ninth "rade class of 200&200( <"raduatin" class of 200'= and
+e6ond# each hi"h school student must earn t4o verified credits in En"lish# one each in
math# science# and histor6# and one in a su+;ect selected +6 the student. 3n order to
receive an advanced diploma# a student must have 2( total credits# and nine of those must
+e verified throu"h endofcourse S51 tests.D
'0

Schools face losin" their accreditation as a result of poor S51 performance +6
their students. CB6 the end of school 6ear 20020&# a )0 percent passin" rate on the S51
tests 4ill have to +e achieved in order for schools to attain LFull6 $ccredited/ status.D
'A

C$ccordin" to the ;irginia Administrative *ode# schools that are accredited 4ith 4arnin"
are su+;ect to academic revie4 and monitorin" +6 State D5E staff. 3n order to help them
improve their accreditation status# these schools are also re:uired to develop a three6ear
school improvement plan 4ith the assistance of parents and teachers.D
'2
I1.1 2o0 is 1irginia (oing/
Since the S51s 4ere first introduced# %ir"inia schools have "raduall6 +een
improvin". Braphs A# 2# and & from schoolmatters.com displa6 the scores over the past
fe4 6ears. C5ver the course of several 6ears of S51 implementation# S51 test scores
and pass rates have increased su+stantiall6.D
'&
3n addition# Cthe num+er of full6
'0
3+id.# '.
'A
!nderstanding ;irginia-s 5eort *ard, http?MM444.thomas;effersoninst.or"MpdfMarticlesMe.ecutiv.pdf.
'2
5eview of +actors and ,ractices Associated with School ,erformance in ;irginia,
http?MM;larc.state.va.usMSummar6MRpt&05MSch*erfm.htm.
'&
CRevie4 of Factors and *ractices $ssociated 4ith School *erformance#D Senate Noint Resolution &(F
O200& SessionP# 33.
accredited schools has increased su+stantiall6 over the last five 6ears from AA' in AFFF
2000 to A#(A( in 200&200(.D
'(

$n important :uestion is ho4 much of a 6earl6 increase in scores 4ill +e
necessar6 to ensure that %ir"inia meets the deadline for A00R proficienc6. -he RaM*
Up -ar"et is a C"au"e of ho4 much pro"ress a school# district# or state must ma8e each
6ear <on avera"e= in order to reach A00R in 20A(.D
'5
%ir"inia/s RaM* Up -ar"et is 2.(
percenta"e points per 6ear.
5ne additional area of concern is the "raduation rate. C%ir"inia devotes onl6 &.&
percent of its total ta.a+le resources to education S less than (A other states. Moreover#
in 200& the state spent appro.imatel6 KAI per middle and hi"hschool student on
dropout prevention# 4hile spendin" K)F#000 per 6oun" person incarcerated in ;uvenile
prison.D
'I
Somethin" is certainl6 4ron" 4ith this picture. C5f the students in %ir"inia
that entered Fth "rade in AFF)# onl6 )(R completed hi"h school 4ithin four 6ears# 4hich
is a+ove the national avera"e of )0R.D
')
Even thou"h this fi"ure is a+ove the national
avera"e# it doesn/t mean that there isn/t a pro+lem.
I1.2 &redictors of S3L Scores
2hen e.plainin" variation in S51 scores# 4e loo8 to re"ressions to tell us the
impact of certain varia+les on the scores. -hese varia+les are called independent or ri"ht
hand side varia+les. 3t is important to note that not all of these varia+les are under the
control of the state or school polic6# so this implies that not ever6 pro+lem is a direct
result of school success or ne"li"ence. For e.ample# if a student/s povert6 status
'(
3+id.# I.
'5
Statewide $ducation #nsights, http?MM444.schoolmatters.com.
'I
$ndre4 Bloc8# C$ttention to S51s $ppears to Be 2idenin" the Braduation Bap#D 5ichmond Times'
8isatch, Nune '# 2005# sec. $.
')
Statewide $ducation #nsights, http?MM444.schoolmatters.com.
contri+utes to hisMher lo4 S51 score# this is a varia+le that cannot +e easil6 controlled +6
the school or state. 3f most of the ma;or predictors of S51 scores happen to +e varia+les
such as this# this spells trou+le for the school +ecause these are pro+lems that cannot +e
com+ated 4ith polic6 chan"es. 3n this section# 3 4ill e.amine three varia+les 4hich are
not directl6 under the influence of polic6 ma8ers. -hese are income level# sin"leparent
households# and race.
7anushe8 and Ra6mond provide an eas6 4a6 to vie4 educational inputs?
total student achievement = school inuts > other inuts,
4here other inuts = ability > family > eers > history > measurement error.
''

5+viousl6# the school isn/t the onl6 influence on a student/s score. 3n fact# school
varia+le variation ma6 not even +e a ma;or determinant. CBecause schools 4ith lo4 S51
scores 4ill +e considered as Lfailin"#/ it is important to identif6 4hat varia+les impact
these standardi@ed test scores.D
'F
I1.2.A Income Le'el
5ne varia+le not determined +6 polic6 is the income level of the child/s famil6.
Braph ( displa6s the spread of income levels across %ir"inia. C$B3 <$d;usted Bross
3ncome= per capita is a measure of the avera"e personal income level of a communit6/s
residentsH*rincipals indicated that parental income has an effect on test scores and
attri+uted this relationship to the level of resources availa+le to supplement a child/s
education in the home.D
F0
3f a child comes from a less affluent famil6# the famil6 ma6 not
+e a+le to afford to +u6 +oo8s# ta8e their child to museums and other educational places#
''
7anushe8 and Ra6mond# C1essons a+out the Desi"n of State $ccounta+ilit6 S6stems#D &(.
'F
David 1ehr and Melanie Mar8s# C*erformance %ariance on %ir"inia/s Standards of 1earnin"? $
*reliminar6 School1evel Empirical $nal6sis#D Su+mitted to ;irginia $conomic <ournal, <200A=.
F0
CRevie4 of Factors and *ractices $ssociated 4ith School *erformance#D Senate Noint Resolution &(F
O200& SessionP# 5I.
or spend much time instructin" their child# as the parents ma6 need to 4or8 multiple ;o+s.
-he parents themselves ma6 not +e ver6 4ell educated. CSchools in communities 4ith a
smaller proportion of colle"eeducated "raduates tend to also +e located in localities that
are less affluent# and in school divisions that pa6 their teachers less and spend less per
student on instruction.D
FA

C$ recent stud6 +6 *eter -uer8# a doctoral student at the Universit6 of %ir"inia#
dre4 a ver6 stron" connection +et4een the percenta"e of hi"hl6 :ualified teachers in a
school and success on the S51s.D
F2
C-he +road patterns 4hich emer"e from these studies
are? A= famil6 characteristics such as income and education level of parents are
over4helmin"l6 important to school performance# and 2= increased resource spendin"
<includin" spendin" to reduce teacherMpupil ratios# increased facult6 salaries# and per
pupil e.penditures in "eneral= are not si"nificant e.planator6 varia+les.D
F&
-his indicates
that pourin" mone6 on the pro+lem 4ill not solve an6thin". C-he direct relationship
+et4een spendin" and student proficienc6 is not ver6 stron". -his does not mean that
Lmone6 doesnQt matter./ 3t simpl6 means that spendin"# +6 itself# does not determine
performance.D
F(
-herefore# 4e cannot merel6 compensate for povert6filled areas +6
dumpin" more and more mone6 on the schools.
CStatistical anal6sis indicates that povert6 <percenta"e of students participatin" in
the free and reducedprice lunch pro"ram=# race <percenta"e of +lac8 students=# and adult
educational attainment <proportion of adults over 25 in the communit6 4ho hold a
colle"e de"ree= are the three most po4erful predictors of S51 test scoresH-hese three
FA
3+id.# 3%.
F2
$ndre4 Bloc8# C$ttention to S51s $ppears to Be 2idenin" the Braduation Bap#D 5ichmond Times'
8isatch, Nune '# 2005# sec. $.
F&
David 1ehr and Melanie Mar8s# C*erformance %ariance on %ir"inia/s Standards of 1earnin"? $
*reliminar6 School1evel Empirical $nal6sis#D Su+mitted to ;irginia $conomic <ournal, <200A=.
F(
Statewide $ducation #nsights, http?MM444.schoolmatters.com.
factors e.plain almost t4othirds of the variation in test scores across %ir"inia
divisions.D
F5
2h6 does povert6 pla6 such a lar"e role in test scores> C$ccordin" to school
principals# children 4ho are raised in povert6 or live in communities 4ith a small
proportion of colle"eeducated adults ma6 receive less academic support and
encoura"ement from their parents# have less motivation and selfesteem# and receive less
e.posure to learnin" outside of school. -he6 also tend to move more fre:uentl6 and are
e.posed to more crime and violence in their nei"h+orhoods.D
FI
C-he +etter the teacher#
the +etter the student performance# re"ardless of the student/s +ac8"round. Researchers
disa"ree over 4hich teacher characteristics matter the mostGe.perience# education
+ac8"round# su+;ect matter 8no4led"e# or un:uantifia+le traits. But the6 "enerall6 a"ree
that# 4hatever characteristic is chosen# +etter teachers tend to +e found in middle class
schools rather than in hi"hpovert6 schools.D
F)
C$ clear correlation +et4een povert6 and
performance does e.ist.D
F'

7o4 does povert6 or lo4 income affect test scores> C*oor parents often 4or8
multiple ;o+s# 4hich leaves them little time to provide needed support to their children.
Man6 poor parents also as8 their children to ta8e on additional famil6 responsi+ilities that
conflict 4ith school# such as +a+6sittin" 6oun"er si+lin"s or 4or8in" to supplement the
famil6/s income. Moreover# parents 4ith lo4 educational attainment tend to have more
difficult6 assistin" their children 4ith their home4or8# +ecause the6 ma6 lac8 the
F5
5eview of +actors and ,ractices Associated with School ,erformance in ;irginia,
http?MM;larc.state.va.usMSummar6MRpt&05MSch*erfm.htm.
FI
3+id.
F)
R6an# C-he *erverse 3ncentives#D F)A.
F'
Statewide $ducation #nsights, http?MM444.schoolmatters.com.
8no4led"e needed to do soH-eachers +elieve that this lac8 of parental support for
academic achievement creates a si"nificant o+stacle to student performance.D
FF

$lso# C*rincipals indicated that the lac8 of student motivation sometimes results
from the parents/ vie4 of educationHman6 poor parents 4ho did not complete colle"e or
hi"h school ma6 not place much priorit6 on education. -hese parents 4ho vie4 education
ne"ativel6# or place little value on it# tend to have lo4 academic e.pectations for their
children# 4hich results in lo4 student motivationH $lon" 4ith the lac8 of motivation#
another effect of povert6 and lo4 adult educational attainment is lo4 selfesteem.D
A00
C3n
schools 4ith lar"e num+ers of students from poor homes or 4ith parents 4ho have
limited education# lo4 e.pectations are reinforced amon" peers.D
A0A
I1.2.B Single"&arent 2ouseholds
3n addition# students from sin"leparent households ma6 not perform as 4ell on
the tests. C*rincipals intervie4ed +6 N1$R! staff advanced three main reasons 4h6
havin" a lar"e num+er of students comin" from femaleheaded households ma6 +e
associated 4ith lo4er test scores. First# students <particularl6 +o6s= raised in female
headed households ma6 lac8 a positive male influence in the home# a factor that can lead
to +ehavioral issues often associated 4ith lo4er performance. 3n addition# sin"le mothers
fre:uentl6 have less time to supervise their child/s academic pro"ress and help 4ith
home4or8. Finall6# schools ma6 have to provide more social support and services to
students 4ho come from sin"le parent homes# 4hich ma6 divert resources from their
academic mission.D
A02

FF
CRevie4 of Factors and *ractices $ssociated 4ith School *erformance#D Senate Noint Resolution &(F
O200& SessionP# ().
A00
3+id.
A0A
3+id.# ('.
A02
5eview of +actors and ,ractices Associated with School ,erformance in ;irginia,
http?MM;larc.state.va.usMSummar6MRpt&05MSch*erfm.htm.
I1.2.C Race
Race is another hot +utton issue 4hen it comes to test scores. Braphs 5 and I
from schoolmatters.com displa6 the different test scores for su+"roups of students.
C0on4hite is statisticall6 si"nificant <N1$R!= and sho4s that# for our sample# for ever6
AR increase in the num+er of students 4ho are non4hite# a school 4ill on avera"e
e.perience a .0F)R decrease in the num+er of students passin" the e.am.D
A0&
CSeveral
factors that ma6 impact academic performance also tend to occur more fre:uentl6 in
places 4ith hi"her concentrations of +lac8 students# and ma6 lar"el6 e.plain the stron"
relationship that appears to e.ist +et4een race and S51 test scores. Man6 of the factors
previousl6 descri+ed as coincidin" 4ith hi"h levels of povert6 also coincide 4ith a hi"h
proportion of +lac8 pupils# 4hich is to +e e.pected +ecause these t4o demo"raphic
characteristics fre:uentl6 coe.ist.D
A0(
I1.# -ides*read Concerns Aout the S3Ls
Man6 people have reservations a+out the S51s. CSome of the concerns e.pressed
+6 principals and teachers are that the S51s reduce teachin" creativit6# reduce the
opportunit6 for enrichment activities# create too much pressure for the students# and limit
time availa+le for teachin" hi"herlevel critical thin8in" s8ills.D
A05
C3n settin" uniform
academic standards# state polic6ma8ers necessaril6 ta8e discretion a4a6 from teachers.
1ocal teachers# 4ith some direction from local officials# once determined not onl6 how to
teach +ut what to teach.D
A0I
A0&
David 1ehr and Melanie Mar8s# C*erformance %ariance on %ir"inia/s Standards of 1earnin"? $
*reliminar6 School1evel Empirical $nal6sis#D Su+mitted to ;irginia $conomic <ournal, <200A=.
A0(
CRevie4 of Factors and *ractices $ssociated 4ith School *erformance#D Senate Noint Resolution &(F
O200& SessionP# 5&.
A05
5eview of +actors and ,ractices Associated with School ,erformance in ;irginia,
http?MM;larc.state.va.usMSummar6MRpt&05MSch*erfm.htm.
A0I
R6an# C-he *erverse 3ncentives#D F)A.
$nother concern is that schools 4ill find 4a6s to "ame the s6stem instead of
actuall6 improvin". C$ccounta+ilit6 s6stems# no matter ho4 4elldesi"ned# 4ill have
man6 incentives em+edded 4ithin them for schools to L"ame the s6stem./ -he successful
desi"n of accounta+ilit6 s6stems hin"es on the identification and closure of as man6 of
these loopholes as possi+le. 7o4ever# the li8elihood that schools 4ill find other
mechanisms throu"h 4hich the6 can inflate their o+served test performance for the
purposes of accounta+ilit6 su""ests that all a""re"ate test scores should +e ta8en 4ith a
"rain of salt# and not vie4ed as perfect indicators of school productivit6.D
A0)

I1., %aming the S!stem
3n this section# 3 4ill e.amine specific 4a6s that schools can "ame the s6stem and
artificiall6 raise their percenta"e passin" rates. -he three instances 3 discuss here are
teachin" to the test# the use of school suspensions# and the manipulation of student
nutrition.
I1.,.A Teaching to the Test
3f schools do not thin8 it is possi+le for their students to pass the tests# the6 ma6
attempt to find 4a6s to artificiall6 raise the percenta"e of their students 4ho are passin".
C5ne possi+ilit6 is socalled Lteachin" to the test#/ in 4hich schools focus on test
preparation s8ills and tailor their instruction to su+;ects included on the e.amination 4ith
hi"h pro+a+ilit6. 2hile controversial# it is unclear as to 4hether teachin" to the test is
desira+le or undesira+le# especiall6 4hen the test content is ri"orous and 4ideran"in".
$nother e.ample of +ehavior that could tend to reduce the informative si"nal of
a""re"ate test scores involves the assi"nment of students to special education. Several
A0)
David 0. Fi"lio# C-estin"# !rime and *unishment#D 0ational Bureau of Economic Research <!am+rid"e#
M$= 2or8in" *aper 0o. AAAF(# March 2005# AI.
recent authors# includin" !ullen and Re+ac8 <2002=# Fi"lio and Bet@ler <2002= and Naco+
<2002=# have sho4n that schools tend to respond to accounta+ilit6 s6stems and testin"
re"imes +6 classif6in" more mar"inal students as disa+ledH5ne interpretation of these
results is that schools are +ehavin" in an insidious manner# reclassif6in" potentiall6 lo4
performin" students into teste.cluded cate"ories in order to ma8e avera"e test scores loo8
+etter.D
A0'
I1.,.B Sus*ensions
5ne 4a6 that schools could "ame the s6stem is +6 "ivin" students suspensions
durin" the testin" da6s so that the6 can e.clude students the6 +elieve 4ill perform poorl6
on the tests. CDurin" the testin" 4indo4# potentiall6 lo4performin" students could +e
"iven harsher punishments <lon"er suspensions= than potentiall6 hi"hperformin"
students receive for similar infractions# +ecause the school ma6 desire to have as man6
hi"hperformin" students as possi+le in school to ta8e the e.amination +ut at the same
time hopes to have more lo4performin" students sta6 home durin" testin" periods.D
A0F
$ stud6 in Florida found that C4hile schools al4a6s tend to assi"n harsher punishments
to lo4performin" students than to hi"hperformin" students throu"hout the 6ear# this "ap
"ro4s su+stantiall6 durin" the testin" 4indo4. Moreover# this testin" 4indo4related "ap
is onl6 o+served for students in testin" "rades. 3n summar6# schools apparentOl6P act on
the incentive to reshape the testin" pool throu"h selective discipline in response to
accounta+ilit6 pressures.D
AA0
C3n nearl6 si.t6 percent of cases# t4o students suspended for
the same offense receive differential suspensions.D
AAA

I1.,.C Nutrition
A0'
3+id.# A2.
A0F
3+id.# 2&.
AA0
3+id.# (5.
AAA
3+id.# I.
Schools can also "ame the s6stem throu"h the manipulation of the school lunch
pro"ram. -his tactic can +e used to artificiall6 increase test scores. CSeveral studies find
"lucose to improve shortterm co"nitive a+ilit6 <Benton and *ar8er# AFF'T and *ollitt#
!ueto and Naco+6# AFF'=. -he results indicate a su+stantial and si"nificant "ain in
co"nitive a+ilit6 from the consumption of "lucose# or empt6 calories# 4hich provides a
+oost in ener"6. -he increased calories consumed enhanced scores on a ps6cholo"ical
+atter6# and on ver+al intelli"ence.D
AA2
CUsin" detailed dail6 school nutrition data from a
random sample of %ir"inia school districts# 4e find that school districts havin" schools
faced 4ith potential sanctions under %ir"iniaQs Standards of 1earnin" <S51=
accounta+ilit6 s6stem apparentl6 respond +6 su+stantiall6 increasin" calories in their
menus on testin" da6s# 4hile those 4ithout such immediate pressure do not chan"e their
menus. Su""estive evidence indicates that the school districts 4ho do this the most
e.perience the lar"est increases in pass rates.D
AA&

CFirst# school lunch pro"rams tend to tar"et precisel6 the students most Lneeded/
+6 schools to succeed in the accounta+ilit6 s6stem? School accounta+ilit6 s6stems
"enerall6 evaluate schools +ased on the fraction of students attainin" a particular
minimum accepta+le threshold. -he students 4ho do not attain this threshold are
over4helmin"l6 lo4income# and there+6 eli"i+le for su+sidi@ed lunches. $nd since
school meal ta8eup is unsurprisin"l6 ver6 hi"h amon" those identified as eli"i+le to
participate in the pro"ram# the stomachs <and hence# accordin" to the nutrition literature#
AA2
David 0. Fi"lio and Noshua 2inic8i# CFood for -hou"ht? -he Effects of School $ccounta+ilit6 *lans on
School 0utrition#D 0ational Bureau of Economic Research <!am+rid"e# M$= 2or8in" *aper 0o. F&AF#
0ovem+er 2002# 5I.
AA&
3+id.# A.
the minds= of a ver6 lar"e fraction of the mar"inal students in a school are reached +6 the
school nutrition pro"ram.D
AA(

CSecondHthis is one mar"in more easil6 manipula+le +6 the school than most
levers that schools mi"ht choose to affect student outcomes. Most school food service
directors either have de"rees in nutrition or have at least +een certified +6 the $merican
School Food Service $ssociation# and are presuma+l6 familiar 4ith the lin8s +et4een
nutritional content and co"nitive performance# so this mar"in ma6 4ell +e 8no4n to most
school nutrition decisionma8ers. Moreover# unli8e some of the methods of "amin" more
4idel6 su""ested# such as reclassification of students as disa+led# school nutrition is
much more difficult to audit# and importantl6# the hi"her "overnmental a"enc6 that audits
school nutrition is the U.S. Department of $"riculture# rather than the state or federal
education authorities implementin" an accounta+ilit6 s6stem.D
AA5
-herefore# this is one
form of "amin" that schools can <and have= easil6 "otten a4a6 4ith.
I1.4 Ethnicit! and the Self"+ulfilling &ro*hec!
-eacher e.pectations could also pla6 a role in student performance on the e.ams.
-he desire to improve scores amon" all student "roups 4ill not 4or8 if teachers don/t
"ive the same support and encoura"ement to all of their students. 5-here is reason to
+elieve that e.pectations matter? -he recent 4or8 on teacher "radin" standards <Betts#
AFF5T Betts and Bro""er# 200&T Fi"lio and 1ucas# 200(= indicates that hi"her standards
lead to improved student test scores.D
AAI
C-he consistent findin" from this literature is
that teachers ta8e Blac8 students less seriousl6 than the6 do 2hites <Fer"uson# AFF'=.D
AA)

AA(
3+id.# (5
AA5
3+id.
AAI
David 0. Fi"lio# C0ames# E.pectations and the Blac82hite -est Score Bap#D 0ational Bureau of
Economic Research <!am+rid"e# M$= 2or8in" *aper 0o. AAAF5# March 2005# 2.
AA)
3+id.
3f teachers don/t e.pect much from their students# 4hat motivation do the students
have to tr6 harder> C-eachers/ stereot6pes# +oth positive and ne"ative# influence
children/s co"nitive performance. Naco+son and Rosenthal <AFF2= stress the importance
of the selffulfillin" prophec6 in the classroom.D
AA'
C-eachers ma6 e.pect less from
children 4ith names that Lsound/ li8e the6 4ere "iven +6 uneducated parents. -hese
names# empiricall6# are "iven most fre:uentl6 +6 Blac8s# +ut the6 are also "iven +6
2hite and 7ispanic parents as 4ell.D
AAF
David Fi"lio notes# C3 find that teachers tend to
treat children differentl6 dependin" on their names# and that these same patterns
apparentl6 translate into lar"e differences in test scores. -hese results are consistent 4ith
the notion that teachers and school administrators ma6 su+consciousl6 e.pect less of
students 4ith names associated 4ith lo4 socioeconomic statusGnames that are
disproportionatel6 "iven to Blac8 childrenand these e.pectations ma6 possi+l6 +ecome
a selffulfillin" prophec6.D
A20

C-he estimated relationship +et4een names and test scores su""ests that a
reasona+l6 lar"e fraction of the Blac82hite test score "ap can +e e.plained +6
children/s namin" patterns. Because Blac8 children are considera+l6 more li8el6 to +e
"iven names associated 4ith lo4 socioeconomic status than are 2hite children# one can
calculate that around A5 percent of the Blac82hite test score "ap ma6 +e due to
differences in names "iven across the races.D
A2A
-his research a+out namin" certainl6 paints a "loom6 picture for the status of
education and e:ualit6. 3/m sure that man6 teachers pro+a+l6 don/t even reali@e that the6
hold different e.pectations of students 4ith different t6pes of names. 3t/s not necessaril6
AA'
3+id.# &.
AAF
3+id.# (.
A20
3+id.# 5.
A2A
3+id.# 22.
a pro+lem that can/t +e com+ated# thou"h. $ re:uired teachin" seminar dealin"
specificall6 4ith this issue could help alleviate this pro+lem +6# at the ver6 least# +rin"in"
it to the attention of teachers and promptin" them to ta8e a loo8 at their o4n e.pectations
of students and ho4 those e.pectations impact performance.
I1.6 Im*ro'ing S3L Test Scores
2hat are the +est performin" schools doin" to achieve "ood S51 results> -his is
an important :uestion to as8# as the techni:ues used in the school districts that perform
the +est could +e used to help other school districts improve. CBased on intervie4s 4ith
principals that have achieved academic success# in +oth hi"hscorin" and successful
challen"ed schools# there appear to +e nine effective practices that help promote student
achievement on the S51 tests.D 3t is important to note that m6 research findin"s ma6 run
counter to the su""estions +elo4.
D-hese practices are? <A= stron" and sta+le leadershipT <2= an environment
conducive to learnin"T <&= an effective teachin" staffT <(= datadriven assessment of
student 4ea8nesses and teacher effectivenessT <5= curriculum ali"nment and pacin"T <I=
the use of differentiation in teachin" to meet the needs of all studentsT <)= an emphasis on
academic remediationT <'= the use of team4or8 and colla+oration 4ithin "rades and
vertical inte"ration across "rade levels# and <F= the ma.imi@ation of instructional time
throu"h attention to the structure and intensit6 of the school da6.D
A22

3n schools 4ith students 4ho aren/t performin" 4ell# strict chan"es are necessar6
to ma8e a difference. C!hallen"ed schools also have to ta8e additional steps to create an
environment conducive to learnin". -he6 have to focus more on controllin" disruptive
A22
CRevie4 of Factors and *ractices $ssociated 4ith School *erformance#D Senate Noint Resolution &(F
O200& SessionP# I&.
+ehavior and imposin" discipline. Unli8e schools 4ithout demo"raphic challen"es# these
schools also have to "o to "reater len"ths to motivate their students# +uild their self
esteem# and set hi"h academic e.pectations for them# +ecause the students ma6 lac8 all
three.D
A2&

1. -hat 2a'e 3ther States and Locals (one/
-he main focus of this paper is the 0!1B and ho4 %ir"inia has reacted to it# +ut
it is also 4orth4hile to loo8 at 4hat other states are doin". Specificall6# 3 4ill e.amine
the s6stems used in Florida and !hica"o.
1.1 +lorida
A2&
5eview of +actors and ,ractices Associated with School ,erformance in ;irginia,
http?MM;larc.state.va.usMSummar6MRpt&05MSch*erfm.htm.
CFlorida schools have chan"ed their instructional policies# practices# and
allocations of resources 4ithin and +et4een schools as a result of increased
accounta+ilit6.D
A2(
C3n 0ovem+er AFF'# Florida voters elected Ne+ Bush "overnor# and
once he too8 office in Nanuar6 AFFF Bovernor Bush 4or8ed 4ith the state le"islature to
implement his $V education plan that comin" summer. $t its centerpiece is a s6stem of
accounta+ilit6 +ased lar"el6 on student test scores# in 4hich each school 4ould receive a
letter "rade ran"in" from L$/ to LF./ Schools on +oth ends of the spectrum are affected?
students attendin" schools rated LF/ in t4o 6ears out of a four 6ear 4indo4 are eli"i+le
for school vouchers# or Lopportunit6 scholarships#/ that can +e used to send a child to a
private school or alternativel6# that ma8e a child eli"i+le to transfer to a L!/ or hi"her
rated pu+lic school in the same district or an ad;acent district. Schools receivin" "rades of
L$/ <or in su+se:uent 6ears# increasin" their letter "rades from 6ear to 6ear= are eli"i+le
for financial re4ards totalin" a+out KA00 per pupil that can +e spent for purposes such as
hirin" teacher aides or providin" teacher +onuses# etc.D
A25
Florida/s s6stem of
accounta+ilit6 4as a ma;or inspiration +ehind the 0!1B# 4hich is 4h6 man6 of it/s
features are reflected in the act.
1.2 Chicago
!hica"o *u+lic Schools <!hi*S= +e"an their test+ased accounta+ilit6 polic6 in
AFF).
A2I
Naco+ notes# C3 find that math and readin" scores on the hi"hsta8es e.am
increased sharpl6 follo4in" the introduction of the accounta+ilit6 polic6.D
A2)
C-he first
A2(
David 0. Fi"lio# 3hat )ight School Accountability 8o9, http?MM444.n+er.or"MreporterMfall0AMfi"lio.html.
A25
David 0. Fi"lio and Maurice E. 1ucas# C2hat/s 3n a Brade> School Report !ards and 7ouse *rices#D
0ational Bureau of Economic Research <!am+rid"e# M$= 2or8in" *aper 0o. '0AF# 0ovem+er 2000# I.
A2I
Brian $. Naco+# C$ccounta+ilit6# 3ncentives and Behavior? -he 3mpact of 7i"hSta8es -estin" in the
!hica"o *u+lic Schools#D 0ational Bureau of Economic Research <!am+rid"e# M$= 2or8in" *aper 0o.
'FI'# Ma6 2002# 2.
A2)
3+id.# &.
component of the polic6 focused on holdin" students accounta+le for learnin"# +6 endin"
a practice commonl6 8no4n as Lsocial promotion/ 4here+6 students are advanced to the
ne.t "rade re"ardless of a+ilit6 or achievement level. Under the ne4 polic6# students in
third# si.th and ei"hth "rades are re:uired to meet minimum standards in readin" and
mathematics on the 3o4a -est of Basic S8ills <3-BS= in order to advance to the ne.t
"rade. Students 4ho do not meet the standard are re:uired to attend a si.4ee8 summer
school pro"ram# after 4hich the6 reta8e the e.ams. -hose 4ho pass move on to the ne.t
"radeT those 4ho fail this second e.am are re:uired to repeat the "rade.D
A2'
C3n
con;unction 4ith the social promotion polic6# the !hi*S also instituted a polic6 desi"ned
to hold teachers and schools accounta+le for student achievement. Under this polic6#
schools in 4hich fe4er than A5 percent of students scored at or a+ove national norms on
the 3-BS readin" e.am 4ere placed on pro+ation. 3f the6 did not e.hi+it sufficient
improvement# these schools could +e reconstituted# 4hich involved the dismissal or
reassi"nment of teachers and school administrators.D
A2F

C1oo8in" across all "rades and su+;ects# several +road patterns +ecome apparent.
First# students in lo4performin" schools seem to have fared considera+l6 +etter under
the polic6 than compara+le peers in hi"herperformin" schoolsH Second# students 4ho
had +een scorin" at the A0th50th percentile in the past fared +etter than their classmates
4ho had either scored +elo4 the A0th percentile# or a+ove the 50th percentile.D
A&0

CStudents in AFF' 4ere A.) percenta"e points more li8el6 to correctl6 ans4er :uestions
involvin" comple. s8ills in comparison to cohorts in AFF( and AFFI. -he compara+le
improvement for :uestions testin" +asic s8ills 4as &.F percenta"e points# su""estin" that
A2'
3+id.# 5.
A2F
3+id.# 5I.
A&0
3+id.# AIA).
under accounta+ilit6 students improved more than t4ice as much in +asic s8ills as
compared 4ith more comple. s8ills.D
A&A
CUnli8e math# it appears that the improvements
in readin" performance 4ere distri+uted e:uall6 across :uestion t6pe. -his anal6sis
su""ests that test preparation ma6 have pla6ed a lar"e role in the math "ains# +ut 4as
perhaps less important in readin" improvement.D
A&2
1I. Conclusion of Literature Re'ie0
2ithout understandin" the literature and national conversation a+ove# it 4ould +e
ver6 difficult to leap into m6 re"ression stud6. 0o4 that 4e/ve revie4ed the literature of
the field# +oth federal and state# that 8no4led"e can form the +ac8drop for m6 o4n
anal6sis 4hich is to come. M6 4or8 fits into the ideas a+ove +ecause 4e are tr6in" to
determine 4hich varia+les have a statisticall6 si"nificant effect on %ir"inia S51 scores.
A&A
3+id.# 22.
A&2
3+id.# 2&.
2e can also anal6@e the effects over time for several test 6ears. M6 4or8 4ill sho4
4hich varia+les are si"nificant. 3t can ans4er the :uestion of 4hich varia+les 4e should
+e focusin" on 4ith education reform polic6. 7o4ever# this stud6 is not fla4less. For
starters# the data 4e use is districtlevel data# not schoollevel data# so it 4on/t +e as
accurate as it could +e. 3t 4ill miss some of the variation 4hich could +e captured if 4e
had access to school level or student level data. Secondl6# if the main varia+les that sho4
up as si"nificant are varia+les that polic6 cannot counteract# the recommendations for
com+atin" such pro+lems 4ill not +e eas6 to enact. For e.ample# if povert6 is
statisticall6 si"nificant# the o+vious recommendation is to eliminate povert6# +ut this is a
ver6 unrealistic solution. 3n that case# su""estions 4ill +e "iven for minor and "radual
improvements.
1II. .! Research
M6 research 4ill primaril6 focus upon updatin" and replicatin" the 4or8 of
others. Specificall6# 3 4ill follo4 the 4or8 of Brat and *fit@ner <200A=.
A&&
3n Brat and
*fit@ner/s 4or8# there is a discussion of the findin"s of 7anushe8 and 9rue"er. 3n short#
A&&
David Brat and !. Barr6 *fit@ner# CE.plainin" S51 %ariation $cross %ir"inia School Districts? $
*reliminar6 $nal6sis#D http?MMfacult6.rmc.eduMd+ratMresearchpapersM200A%$EWS51varWF'to00.doc.
7anushe8 found Cno convincin" effect on student performance for teacherMpupil ratios#
teacher education# teacher salar6# or e.penditures per pupil.D
A&(
Braph ) displa6s the
modest "ains in test scores over the last decade despite lar"e increases in spendin"# and
Braph ' sho4s U.S. performance and spendin" compared to other countries.
7o4ever# 9rue"er advises us to +e cautions a+out trustin" these results. 2e must
remem+er that this data is at the district level and not the school level# so there is most
certainl6 some measurement error. 7o4ever# since the district level data are more
readil6 availa+le# that is the data 3 have chosen to utili@e for this research.
-his portion of the paper 4ill e.amine m6 re"ression results. 3 4ill +e"in +6
discussin" the varia+les 4hich 4ere not si"nificant in the re"ressions. Secondl6# 3 4ill
discuss the re"ressions 4hich include a stri8in" varia+le 4hich captures first "rade
aptitude First$pt. 1astl6# 3 4ill e.amine the re"ressions e.cludin" this varia+le. 3 4ill
also discuss the "eneral conclusions of this 4or8.
1II.1 The Non"Significant 1ariales
Before 4e "et to the main stor6# it is important to note that the main stor6 has
emer"ed throu"h a lot of hard 4or8. -he construction of the data set 4as itself the most
demandin" tas8 in this pro;ect. -he most important part of this 4or8 is ma8in" sure that
4e have properl6 specified our model. 3f one varia+le is omitted# then our findin"s 4ould
+e in :uestion. 2e have carefull6 follo4ed the literature in this respect and 4e have
+een careful to run all possi+le independent varia+les in our re"ressions# not ;ust those
found to +e si"nificant in the 4or8 of others. 3n reportin" our results# therefore# it is
important that 4e leave others in the field# the results of all research 4or8 done usin" this
model# not ;ust the punchline. -his also ensures that there is inte"rit6 to the re"ressions
A&(
3+id.# 5.
4e do provide in the end. 3 4as not fishin" for a stor6# +ut instead 3 let the data do the
tal8in" and 3 am reportin" 4hat the data revealed. 3n addition# 3 ran several tests to ensure
that our eventual model specification is ro+ust.
-here are five t6pes of varia+les that 4ere found to +e statisticall6 insi"nificant in
our model# 4hich is 4h6 the6 4ere e.cluded from m6 final re"ressions. -hese five t6pes
of varia+les are al"' <a pro.6 for ri"or=# income and povert6 varia+les# a measure of the
percenta"e of rural housin" units# a cit6dumm6 varia+le# and a varia+le that accounts for
teachers 4ith master/s de"rees.
$l"' 4as a varia+le used +6 Brat and *fit@ner to represent the percenta"e of
students ta8in" $l"e+ra +efore the F
th
"rade. 3t 4as a pro.6 for academic ri"or. Schools
4hich encoura"ed and offered $l"e+ra in the '
th
"rade 4ere considered to +e pro"ressive
in terms of ri"or. Brat and *fit@ner found that this varia+le Cpla6ed a si"nificant role in
all of the re"ressions at the ei"hth "rade level# and in none of the re"ressions at other
"rade levels for AFF'.D
A&5
3 tested this varia+le on a variet6 of the 2002 S51 scores# and
it 4as not si"nificant. Since this data 4as not updated for the 2002 school 6ear# this
could have s8e4ed m6 results.
$s far as income and povert6 are concerned# there are man6 varia+les that can +e
chosen as pro.ies. 3 tested man6 such varia+les on a variet6 of different S51 testin"
"rades +efore determinin" 4hich 4as the most si"nificant and most ro+ust across all
e:uations. 3 found that per capita income# median famil6 income# median household
income# median famil6 income# and the percenta"e of children +et4een a"e 5 and A) in
families considered +elo4 povert6 level 4ere all inferior to the varia+le that measures the
percenta"e of students 4ho received free lunch in the 200(2005 school 6ear. -herefore#
A&5
3+id.# '.
the school lunch measure 4as used in m6 final re"ressions. 3t e.plained the most
variation in student test scores.
$ measure of the percenta"e of rural housin" units 4as also found to +e
insi"nificant. 3 speculated that havin" more ur+an or more rural housin" units could
affect S51 scores# +ut the re"ressions don/t +ac8 up this theor6# so this varia+le 4as also
e.cluded from m6 final re"ressions.
!losel6 related to this varia+le is the cit6dumm6 varia+le# 4hich simpl6
accounted for a %ir"inia school district +ein" classified as either a cit6 or a count6.
Richmond is coded as !it6. 7enrico or 7anover are coded as !ount6 data. Does this
matter> 1i8e the rural housin" measure# this varia+le 4as also insi"nificant in most all
e:uations.
$nother varia+le that 4as insi"nificant in m6 re"ressions 4as a varia+le that
measures the percenta"e of facult6 4ith post"raduate de"rees. -his 4as another varia+le
used in Brat and *fit@ner/s 4or8.
A&I
-his varia+le is from AFFA and AFF2# thou"h# so the
a"e of the data could affect the results. Brat and *fit@ner e.perienced similar results 4ith
this varia+le. -he6 found that Cthe percenta"e of teachers 4ith master/s de"rees 4as
statisticall6 si"nificant in onl6 four of the thirt6si. re"ressions.D
A&)
-hou"h man6
citi@ens +elieve this varia+le is important# the re"ression results don/t full6 support this
notion.
A&I
3+id.# ).
A&)
3+id.# '.
1II.2 Regressions Including +irstA*t
-he re"ression results for each of the ten test scores for 2002 are reported in -a+le A.
-he e.planator6 varia+le set includes?
8iloma percenta"e of females 25 6ears and over 4ith a Bachelor/s# Master/s#
*rofessional School# or Doctorate De"ree <from 2000 census=
Lunch percent of students "ettin" free lunch <school 6ear 200(2005=
<from http?MM444.pen.8A2.va.usM%D5E=
Truancy percent of students 4ith 4hom a conference 4as scheduled after the
student had accumulated si. a+sences durin" the school 6ear
<200&200(= <from http?MM444.pen.8A2.va.usM%D5E=
Sending per pupil S total current spendin" for support services <200&=
<from http?MMftp2.census."ovM"ovsMschoolMelsec0&t..ls=
S,T students per teacher <2002= <from http?MM444.schoolmatters.com=
Black percent of +lac8s in the population <from 2000 census=
Tchsal actual avera"e salar6 for FJ 200(
<from http?MM444.pen.8A2.va.usM%D5E=
+irstAt results of co"nitive a+ilities test administered to first "raders
<o+tained from Dr. David Brat=
-he results in these re"ressions aren/t ver6 promisin"# 4ith one R
2
+ein" as lo4 as
0.2&. -his indicates that these re"ressions did not do a "ood ;o+ of e.plainin" the
variation in the dependent varia+le. 9eepin" this in mind# 4e vie4 the results cautiousl6
+ecause these varia+les are not e.plainin" the entire stor6 of the variation in S51 test
scores. Further research 4ill attempt to +etter specif6 the model.
Third %rade Results
For the third"rade level# across all su+;ects# the Truancy and Sending
coefficients 4ere +oth ne"ativel6 si"ned and statisticall6 si"nificant. -his means that the
"reater the percenta"e of students 4ith more than I a+sences# the lo4er the test scores. 3t
is important to note at this point that all results reported are at the count6 or cit6 level.
2hen 4e sa6# for e.ample# students 4ith more than I a+sences# this means on avera"e
for the count6 as a 4hole. -his result for the Truancy varia+le is ver6 lo"ical and 4as
e.pected. 3f students aren/t present to learn the material# it is clear that this 4ill
ne"ativel6 affect their test scores. $nd districts 4ith hi"her truanc6 rates 4ould +e
e.pected to score lo4er on avera"e. Fortunatel6# this is a polic6 varia+le# 4hich means
that schools can 4or8 on reducin" a+sences and encoura"in" attendance to help raise the
test scores of their students. Brat and *fit@ner/s attendance varia+le 4as also si"nificant
in man6 of their re"ressions# so it is note4orth6 that data from t4o different sources#
applied to test scores for different 6ears# 6ields similar results.
-he Sending varia+le does not# ho4ever# follo4 e.pected intuition. 3nstead# the
re"ressions results indicate that the more mone6 per pupil that is spent on support
services# the lo4er the test scores. *erhaps "reater support e.penditures means less
mone6 or time for instruction. -his could +e the reason for the ne"ative result and lo4er
test scores. $"ain# this result holds after controllin" for all of the other varia+les.
+ifth %rade Results
$t the fifth"rade level# Lunch and Sending 4ere +oth statisticall6 si"nificant#
+ut a"ain Sending 4as not si"ned as 4e mi"ht have e.pected. 8iloma# Black# and
Tchsal 4ere each si"nificant in one re"ression# 4hich indicates that the findin" ma6 not
+e ro+ust. -he fact that Lunch 4as si"nificant means that the "reater the percenta"e of
students 4ho receive free lunch# the lo4er the test scores +6 count6. Since Lunch is a
pro.6 for income# school polic6# unfortunatel6# can/t com+at this issue directl6.
2igh school Results
$t the hi"h school level# +irstAt 4as the onl6 consistentl6 si"nificant varia+le.
3t 4as si"nificant in three of the four re"ressions. Black 4as si"nificant in half of the
re"ressions. Lunch and Sending 4ere each si"nificant in one re"ression. Since +irstAt
4as si"nificant# this indicates that the lo4er the students score on the first "rade aptitude
test# the lo4er their test score performance in later 6ears. 3t is important to understand
4h6 4e have put another test score on the ri"ht hand side of our e:uation. -he reason 4e
did so is +ecause it is an aptitude test and# if constructed properl6# it should control for
the incomin" aptitude of students at the cit6Mcount6 level. -he later S51 test scores
4ould then +e considered true 8no4led"e accumulation over and a+ove the incomin"
aptitude. 3f it trul6 captures aptitude# one could not as8 for a +etter control varia+le. But
as there is some de+ate on this :uestion# 4e have run all re"ressions 4ith and 4ithout this
ver6 important varia+le.
3n addition# this is 6et another varia+le that school polic6 cannot directl6 affect. 3f
schools are supplied 4ith students 4ho score hi"her on the aptitude tests# those same
students 4ill perform +etter on the S51s than their peers. -he +irstAt varia+le indicates
that not all variation in S51 test scores can +e attri+uted to +etter schoolin" +6 district.
-he school/s added value ma6 not +e as "reat as 4e thin8.
$ccordin" to these results# some polic6 varia+les can have a si"nificant effect on
test score results. 3t is interestin" to note that the num+er of students per teacher 4asn/t
si"nificant in an6 of the re"ressions# thou"h this is a varia+le that man6 citi@ens deem to
+e important for :ualit6 education. 3t is also interestin" to note that Blac8 came up as
positivel6 si"nificant in t4o of the re"ressions. -his "oes a"ainst the literature 4hich
states that +lac8 students tend to perform 4orse than their 4hite peers on avera"e. Upon
further inspection# the data all seem to +e in "ood form and the summar6 statistics seem
to match our e.pectations# so 4e loo8 for4ard to future research 4hich can further
e.amine this findin".
1II.# Regressions E7cluding +irstA*t
-he re"ression results for each of the ten test scores for 2002 are reported in -a+le
2. -he e.planator6 varia+le set includes all of the varia+les from the previous
re"ressions# 4ith the e.ception of +irstAt. +irstAt 4as e.cluded from these
re"ressions +ecause some ma6 dou+t the relia+ilit6 of such an aptitude test. !an such
tests in fact reall6 capture the aptitude of a student# and not the stoc8 of 8no4led"e the6
have alread6 o+tained> -his is the :uestion. 2e report +oth sets of results so that 6ou
can ma8e up 6our o4n mind. 3n addition# the +irstAt varia+le 4e used is dated +6 a+out
one decade. -his ma6 not +e a pro+lem. -he data 4ould have had to have chan"ed
s6stematicall6 from count6 to count6 for a real chan"e to occur. Differences in levels
over time 4ould not affect the re"ression results unless some counties chan"ed
proportionatel6 more than other counties. 2e do not see 4h6 this 4ould +e the case.
Further research 4ill attempt to update this varia+le and to +etter understand its
properties.
2ith the e.clusion of this varia+le# the results chan"ed some4hat. 8iloma 4as
statisticall6 si"nificant in three out of the ten re"ressions# 4hich indicates a lac8 of
ro+ustness. Lunch remained statisticall6 si"nificant in si. of the ten re"ressions as
e.pected. 3t 4as also ne"ativel6 si"ned in all +ut one of the ten re"ressions. Sending
4as statisticall6 si"nificant in si. of the ten re"ressions# as opposed to five in the prior
set. $"ain# ho4ever# hi"her spendin" 4as associated 4ith lo4er student test scores# and
this 4as not anticipated. Even in most of the re"ressions in 4hich it 4as not statisticall6
si"nificant# it 4as ver6 close to +ein" so.
Black also +ecame more statisticall6 si"nificant# sho4in" up as si"nificant in
seven of the re"ressions# as compared to five in the first set. 3nterestin"l6# it 4as
positivel6 si"ned and statisticall6 si"nificant in onl6 one "radeGMath '. -his is an
interestin" result 4hich "oes a"ainst the +od6 of literature on S51 test results. 7o4ever#
this varia+le 4as ne"ativel6 si"ned in all of the other re"ressions.
$s for our main polic6 varia+les# Tchsal 4as statisticall6 si"nificant in t4o of the
re"ressions. 3t 4as positivel6 si"ned in all re"ressions# as 4e 4ould e.pect. 3t 4as also
relativel6 close to +ein" statisticall6 si"nificant in several other re"ressions. S,T# on the
other hand# remained insi"nificant in all re"ressions. !onversel6# Truancy 4as
statisticall6 si"nificant in five of the re"ressions# as compared to three in the first set. 3t
4as also ne"ativel6 si"ned in all re"ressions. -his could indicate that perhaps school
polic6 does have an impact on student test performance.
1II., Summar! of Regression +indings
$fter all of the re"ressions have +een run# several conclusions can +e dra4n. -he
first is that Lunch# our pro.6 for income# appears to have a note4orth6 impact on S51
test scores across districts. -his is in line 4ith the e.istin" literature. 3n addition# 4e find
interestin" results 4ith the sending varia+le# as it is ne"ativel6 si"ned. Further research
4ill +e necessar6 to full6 investi"ate this findin". Black also appears to +e moderatel6
important for test scores# 4hich is also in line 4ith the literature.
$s for the polic6 varia+les# their effects aren/t :uite as stron" in most cases#
thou"h Truancy had some impressive# thou"h e.pected# results. Since income is a
leadin" varia+le in the e.planation of variation in test scores# this paints a +lea8 picture
for improvin" scores amon" students. Short of economic "ro4th and redistri+ution of
4ealth# it seems that there is little that can +e done to help less affluent children.
Fortunatel6# some of the polic6 varia+le# such as Truancy# can +e used to improve scores
some4hat.
1III. Conclusions
$fter revie4in" the education literature# it is m6 personal +elief that 0!1B is a
"ood idea for this nation. 1i8e an6 le"islation# it has shortcomin"s. -he ma;or
4ea8nesses seem to +e the a+solute dependence on test scores as the measurin" stic8 as
4ell as the states/ a+ilit6 to "ame the s6stem. *erhaps too much 4ei"ht is placed on test
scores. 3f several alternate measures could +e used in con;unction 4ith test scores to
determine conse:uences for schools# this 4ould +e a step in the ri"ht direction. 2hat
these measures should +e is a :uestion that can +e ans4ered +6 future research. $s for
the incentives attached to "amin"# these could +e some4hat controlled if the "overnment
placed more specific re"ulations a+out the development and scorin" of state tests. 5f
course# if there are too man6 re"ulations# the tests 4ill +ecome national instead of state
tests. 3t/s a tou"h +alance to reach.
Despite the man6 criticisms of 0!1B# 3 +elieve it is a positive step in the ri"ht
direction +ecause it forces schools to +e accounta+le for the education of their students.
3t/s an effective mechanism for identif6in" schools 4here students are clearl6 not "ettin"
the education the6 need to +ecome productive mem+ers of societ6. 0!1B is certainl6 a
ma;or force pushin" schools to ma8e the necessar6 chan"es so that the6 can provide their
students 4ith the +asic s8ills needed to succeed. -hese are chan"es that ma6 not have
+een made if 0!1B hadn/t +een created.
$s for the %ir"inia S51s# 4e can/t simpl6 "et rid of them +ecause some don/t
li8e them. -he 0!1B re:uires each state to test its students. 2hat 4e can do is ma8e
su""estions for improvement. -he ma;or area for these su""estions is in relation to
"amin". -he state can certainl6 monitor schools more carefull6 to ma8e sure that the6
are not "amin" the s6stem. -hese should +e monitored +ecause the6 harm the children
instead of helpin" them. 5n the 4hole# accounta+ilit6 is a positive thin" 4hen 4e
monitor for "amin" to protect the nation/s 6outh from harm.
M6 o4n research and re"ressions have helped me to +etter understand ho4
difficult it is to ma8e real pro"ress in education. Man6 of the varia+les that 3 4ould have
chosen to fund are not statisticall6 related to student performance. -his 4as an e6e
opener. -he ma;or varia+les that do affect scores often relate to famil6 education and
income status. -hese are o+viousl6 not polic6 varia+les 4e can 4or8 4ith. 3t appears that
4e can 4or8 on discipline and truanc6 and academic ri"or.
3n the future# the choice aspects +uilt into the 0!1B and S51 s6stems 4ill come
into pla6. Some schools ma6 actuall6 lose fundin" and students ma6 +e a+le to choose
+etter schools. 3 loo8 for4ard to investi"atin" those issues and to see 4hether student
performance "oes up as choice e.pands. 3 +elieve that 3 have developed the tools to do
that research on m6 o4n. $nd that is the most important part of m6 summer SURF
research pro;ect.
%ra*h 1
%ra*h 2
%ra*h #
%ra*h ,
%ra*h 4
%ra*h 6
%ra*h 8 9from htt*:;;ed*ro.stanford.edu;eah;*ulished;E<.=ournal>2?*ulication.*df
%ra*h @ 9from htt*:;;ed*ro.stanford.edu;eah;*ulished;E<.=ournal>2?*ulication.*df
Tale 1
Diploma 1unch -ruanc6 Spendin" S*- Blac8 -chsal First$pt 3ntercept R
2
En"l & 2'.FF
<A.F&=
0.A5F
<A.)5=
"#6.6?
<2.I0=
"?.??6
<2.(F=
0.&0(
<0.5F=
A0.2F
<A.&0=
0.000
<0.0F=
0.AA'
<0.'2=
'2.5'
<5.)A=
0.50
Math & '.0)'
<0.I&=
"?.1A?
<2.(I=
"26.6@
<2.2&=
"?.??6
<&.A&=
0.)2A
<A.I5=
"2#.22
<&.(I=
0.000
<A.((=
0.0I2
<0.5A=
A0&.'
<'.((=
0.5)
Diploma 1unch -ruanc6 Spendin" S*- Blac8 -chsal First$pt 3ntercept R
2
En"l 5 22.8#
<2.A'=
"?.2?,
<&.2&=
AI.I5
<A.)0=
"?.??6
<&.FA=
0.5&&
<A.(F=
"1#.??
<2.&)=
0.000
<A.50=
0.0FF
<A.00=
'I.0F
<'.5)=
0.)0
Math 5 20.(&
<A.2F=
"?.2#@
<2.()=
20.&0
<A.&I=
"?.??@
<&.50=
0.(A)
<0.)I=
AA.A2
<A.&&=
?.??1
<2.25=
0.AA(
<0.)5=
IF.()
<(.5(=
0.5'
Diploma 1unch -ruanc6 Spendin" S*- Blac8 -chsal First$pt 3ntercept R
2
En"l ' 2A.0I
<A.I0=
"?.248
<&.22=
A).I)
<A.(&=
0.00(
<A.)'=
0.025
<0.0I=
I.A50
<0.'F=
0.000
<0.2(=
?.28?
<2.A5=
I2.0A
<(.''=
0.5'
Math ' A.(55
<0.0'=
"?.,64
<(.AF=
"#A.?,
<2.2)=
0.002
<0.')=
0.IF)
<A.AA=
2#.#1
<2.(2=
0.000
<0.I'=
?.,@#
<2.)I=
5I.5I
<&.20=
0.5'
Diploma 1unch -ruanc6 Spendin" S*- Blac8 -chsal First$pt 3ntercept R
2
En"lA2 A.&)A
<0.AA=
0.0()
<0.I&=
A(.02
<A.22=
0.00&
<A.'2=
0.(FF
<A.A)=
16.81
<2.5I=
0.000
<0.()=
?.,@,
<(.05=
I5.A2
<5.('=
0.&0
$l" A 5.&'5
<0.2I=
0.0)&
<0.5F=
A(.FA
<0.))=
0.005
<A.(5=
A.02&
<A.(I=
(.0A)
<0.&)=
0.000
<A.2I=
0.&&0
<A.I'=
I'.0)
<&.((=
0.&2
Beom F.FIA
<0.5)=
0.0))
<0.)5=
0.)&'
<0.05=
"?.??6
<2.&I=
0.5F0
<A.00=
"1@.6A
<2.0'=
0.00A
<A.)5=
?.414
<&.A&=
(I.')
<2.'I=
0.5I
$l" 2 (2.0F
<A.5'=
?.#2@
<2.A0=
A).2&
<0.)2=
0.002
<A.&A=
0.'FA
<A.00=
A'.')
<A.&'=
0.00A
<A.&A=
?.61A
<2.()=
).A2A
<0.2F=
0.2&
Tale 2
Diploma 1unch -ruanc6 Spendin" S*- Blac8 -chsal 3ntercept R
2
En"l & #2.42
<2.2I=
"?.2?1
<2.(0=
"#8.61
<2.)5=
"?.??6
<2.5)=
0.&AF
<0.I&=
"1#.24
<2.A2=
0.000
<0.A&=
'F.'2
<'.(0=
0.(F
Math & I.AFA
<0.5A=
"?.181
<2.(&=
"24.82
<2.2(=
"?.??6
<&.A)=
0.IFF
<A.I(=
"21.,,
<(.0'=
0.000
<A.(&=
FF.'5
<AA.A2=
0.5)
Diploma 1unch -ruanc6 Spendin" S*- Blac8 -chsal 3ntercept R
2
En"l 5 24.@1
<2.I0=
"?.2#?
<&.F'=
"1@.6A
<A.FF=
"?.??6
<(.A'=
0.5'5
<A.I)=
"16.14
<&.)5=
0.000
<A.IA=
F2.)A
<A2.5F=
0.)0
Math 5 2(.0I
<A.5F=
"?.246
<2.F0=
2(.A0
<A.I'=
"?.??A
<&.'&=
0.52I
<0.FF=
"14.4#
<2.&I=
?.??1
<2.(0=
)).IF
<I.F2=
0.5)
Diploma 1unch -ruanc6 Spendin" S*- Blac8 -chsal 3ntercept R
2
En"l ' 2A.1?
<2.0)=
"?.#62
<(.(2=
A'.F)
<A.(2=
0.00(
<A.I2=
0.02(
<0.05=
0.0)(
<0.0A=
0.000
<0.2)=
)'.20
<).(F=
0.5A
Math ' AI.IF
<0.F2=
"?.46#
<5.2F=
"42.84
<&.05=
0.00(
<A.5A=
A.0)F
<A.I'=
4.A16
<0.)5=
0.000
<A.02=
F0.((
<I.IF=
0.5(
Diploma 1unch -ruanc6 Spendin" S*- Blac8 -chsal 3ntercept R
2
En"lA2 A&.&5
<A.00=
0.05F
<0.'A=
"24.AA
<2.AI=
"?.??4
<2.&'=
0.))(
<A.)0=
0.2'&
<0.05=
0.000
<A.0I=
FI.22
<F.)2=
0.A'
$l" A A5.IA
<0.)F=
0.A5'
<A.&'=
2A.FA
<A.A)=
0.005
<A.)I=
A.20F
<A.)(=
A(.5F
<A.)0=
0.000
<A.(&=
F0.A)
<I.AI=
0.&0
Beom 22.IA
<A.2)=
0.0(2
<0.(&=
A2.')
<0.'A=
"?.??8
<2.)F=
0.'IA
<A.(2=
"#6.,6
<(.FF=
?.??1
<2.AF=
)F.IF
<I.05=
0.52
$l" 2 2).&F
<A.0&=
0.AI&
<A.AA=
2'.F)
<A.2A=
0.00A
<0.&(=
0.II5
<0.)&=
"#@.81
<&.5(=
0.00A
<A.I(=
(5.AA
<2.2F=
0.A)
References
Bloc8# $ndre4. C$ttention to S51s $ppears to Be 2idenin" the Braduation Bap.D
5ichmond Times'8isatch, Nune '# 2005# sec. $.
Brat# David# and !. Barr6 *fit@ner. CE.plainin" S51 %ariation $cross %ir"inia School
Districts? $ *reliminar6 $nal6sis.D
http?MMfacult6.rmc.eduMd+ratMresearchpapersM200A%$EWS51varWF'to00.doc.
Fi"lio# David 0.# and Noshua 2inic8i. CFood for -hou"ht? -he Effects of School $
ccounta+ilit6 *lans on School 0utrition.D 0ational Bureau of Economic Research
<!am+rid"e# M$= 2or8in" *aper 0o. F&AF# 0ovem+er 2002.
Fi"lio# David 0.# and Maurice E. 1ucas. C2hat/s 3n a Brade> School Report !ards and
7ouse *rices.D 0ational Bureau of Economic Research <!am+rid"e# M$=
2or8in" *aper 0o. '0AF# 0ovem+er 2000.
Fi"lio# David 0. C0ames# E.pectations and the Blac82hite -est Score Bap.D 0ational
Bureau of Economic Research <!am+rid"e# M$= 2or8in" *aper 0o. AAAF5#
March 2005.
Fi"lio# David 0. C-estin"# !rime and *unishment.D 0ational Bureau of Economic
Research <!am+rid"e# M$= 2or8in" *aper 0o. AAAF(# March 2005.
Fi"lio# David 0. 3hat )ight School Accountability 8o9,
http?MM444.n+er.or"MreporterMfall0AMfi"lio.html.
7anushe8# Eric $.# and Mar"aret E. Ra6mond. C1essons a+out the Desi"n of State
$ccounta+ilit6 S6stems.D Taking Account of Accountability2 Assessing ,olicy
and ,olitics, :arvard !niversity# FAA Nune 2002.
7o.+6# !aroline M. E2here Should Federal Education 3nitiatives Be
Directed> 9A2 Education %ersus 7i"her Education.D in M. 9osters# ed.
+inancing *ollege Tuition. 2ashin"ton# D!? $E3 *ress# 200A.
Naco+# Brian $. C$ccounta+ilit6# 3ncentives and Behavior? -he 3mpact of 7i"hSta8es
-estin" in the !hica"o *u+lic Schools.D 0ational Bureau of Economic Research
<!am+rid"e# M$= 2or8in" *aper 0o. 'FI'# Ma6 2002.
1ehr# David# and Melanie Mar8s. C*erformance %ariance on %ir"inia/s Standards of
1earnin"? $ *reliminar6 School1evel Empirical $nal6sis.D Su+mitted to
;irginia $conomic <ournal, <200A=.
Meier# De+orah# and Beor"e 2ood. )any *hildren Left Behind. Boston? Beacon *ress#
200(.
*opham# 2. Names. America-s 6+ailing7 Schools. 0e4 Jor8? Routled"eFalmer# 200(.
,resident Bush-s +. /001 $ducation Budget2 Sending )ore, and Sending it )ore
3isely.
http?MMed4or8force.house."ovMissuesMA0'thMeducationMfundin"M+ud"etfactsh
eet020(0&.htm.
5eview of +actors and ,ractices Associated with School ,erformance in ;irginia"
http?MM;larc.state.va.usMSummar6MRpt&05MSch*erfm.htm.
CRevie4 of Factors and *ractices $ssociated 4ith School *erformance#D Senate Noint
Resolution &(F O200& SessionP.
R6an# Names E. C-he *erverse 3ncentives of the 0o !hild 1eft Behind $ct.D 4ew .ork
!niversity Law 5eview# Sprin" 200(.
Statewide $ducation #nsights" http?MM444.schoolmatters.com.
!nderstanding ;irginia-s 5eort *ard"
http?MM444.thomas;effersoninst.or"MpdfMarticlesMe.ecutiv.pdf.

You might also like