Performance tables - what are they good for (absolutely nothing?)

1
Performance tables - what are they good for
(absolutely nothing?)

• Professor Steve Strand
• Institute of Education
• University of Warwick

Lambeth Raising Achievement Making Use of Data
and Good Practice Annual Conference,
International House, 3 November 2011
2
Objectives of the session

• Review the development of school performance
  data in England What are their objectives, do
  they meet them?
• How are performance tables (PTs) changing with
  the coalition government (White Paper, Nov 2010)?
  What is planned and what effects might the plans
  have?
• What are the conditions that can maximise the
  effective use of data for school improvement?

3
25 years of school performance data

• What data was routinely available in the mid
  1980s to evaluate school performance?
• All secondary schools required to publish their
  examination results from 1982, but no format
  specified, wide inconsistencies
• Widespread testing but by LA or individual school
  choice, rarely public
• LEA had option for inspection service but more
  often advisory services, no reports published,
  HMI typically restricted to thematic reports

4
The 1980s

• 1982 All secondary schools required to publish
  their examination results, although the
  Regulations didnt specify the precise form.
• 1987 Nationwide extension of TVEI for 14-18 year
  olds. Central curricula innovation was
  introduction of work experience. Split into local
  projects, carefully monitored to establish good
  practice many performance indicators.
• 1988 Education Reform Act, Introduction of LMS,
  National testing 7-14, Governors detailed annual
  reports to parents, CIPFA DES aide memoire,
  list of 100 PIs in schools

5
Sample of TVEI / CIPFA / DES PIs

• 5th year with 4 GCE O levels / CSE Grade 1
  passes
• attendance
• Total days exclusion
• Y11 who transfer to FTE post 16
• PTR / class size by year group
• Staff qualified to degree level in their
  subject
• staff attendance
• school periods not taught by designated
  teachers
• staff involved in significant CPD
• Number of formal parental complaints
• pupils whose parents attend parent consultation
  sessions
• Capitation per pupil on books / Computers IT /
  resources
• Are incidents of internal vandalism increasing?
• Are the schools objectives for community links
  being achieved?

6
Making sense of PIs

• Mere inspection of a list of indicators will
  not typically reveal all the intricacies of their
  inter-relationships without a detailed
  knowledge of the trade off between inputs and
  outputs policy making may simply be confused by
  these additional data.
• Mayston Jesson (1988).

7
Timelines (1990s)

• May-1991 First Statutory KS1 tests published
  LA level (not intended)
• Nov-1992 First secondary performance tables
  published in England
• Government loose interest in PIs
• May-1993 First statutory KS3 tests
• Sep-1993 Start of secondary OFSTEDs (reports on
  web, summary to
  
  
  parents). Primary
  school OFSTEDs from 2004.
• May-1995 First Statutory KS2 tests
• Nov-1996 TIMSS International study (age 13)
  England 16/25 in Maths (
  (widely cited) 6/25 in Science
  (not widely cited!)
• Mar- 1997 First primary performance tables
  using 1996 results (England)
• Sep-1998 NLS introduced
• Sep-1998 First Statutory Baseline Assessment in
  Reception (age 4)
• Jan- 1999 First OFSTED LEA Inspections
• Sep-1999 NNS introduced
• Nov-1999 First Autumn Package published

8
Crucial thread of VA/CVA

• "Without a value-added dimension, the obvious
  basis for judgement is that 'higher' scores
  represent better practice and 'lower' scores
  worse. This could lead to unwarranted complacency
  on the part of some schools whose pupil
  population comprise more able pupils and,
  conversely, to despair on the part of others,
  who, however hard they try can never expect to
  raise the absolute level of their pupils scores
  to those obtained in schools with more able
  pupils." (SEAC 1993 Dearing interim report,
  Annex. 5, par. 3).

9
Willms (1992) analysis for 20 Scottish secondary
schools (p60)
10
Above/Below Floor Target 2010
Demographics Group 1 Group 2 Odds-ratio
FSM (primary) 30 16 2.0
FSM (secondary) 27 14 2.3
primary in top third deprived (IDACI) 2009 77 lt33 1.7
secondary in top third deprived (IDACI) 2009 70 lt33 1.8
SEN KS2 28 21 1.5
SEN KS4 33 21 1.9
Looked after KS2 KS4 0.7 0.4 1.7
Persistent absence secondary (not FSM) 6 3 2.1
Mobility - with gt4 joining during Y6 42 27 2,0
BME secondary 21 22 1.0
CVA ? ? ?
Source DFE (2011). Underperforming schools and
deprivation (RR141). London DFE.
11
Timelines (2000s)

• Sep-2001 KS3 Strategy Introduced - (Pilot of
  science TLF)
• First national targets for KS3 in 2004 (En,
  Ma, Sc, ICT)
• Jan- 2002 First full PLASC
• Nov-2002 First KS3 published tables
• First VA reported in
  secondary KS4 tables
• Sep-2003 Interactive AP becomes the PAT
• Jan -2004 David Milliband North England speech
  announces NRwS
• May-2005 Election of Labour for Third Term
• Sep-2005 NRwS (SEF, School Profile, SIA, single
  conversation)
• FFT exceptions reports
• Nov-2006 First CVA reported in Secondary AA
  tables
• Jun- 2009 School Report Card proposed
• May-2010 Coalition Government formed
• Nov-2010 Schools White paper
• Nov-2011 CVA to be removed from tables

12
Winning the argument, but..

• Value-added included in secondary performance
  tables from 2002, but
• Median line methodology ceiling effect means
  level 3-gt5 cannot show VA (similar for 5-gt7 KS3),
  introduced systematic bias against high baseline
  schools (Tymms, 2004)
• Prior attainment only no other pupil / school
  context factors
• Narrow focus, still primarily on a single
  threshold measure (5A-C)
• Absence of confidence intervals around school
  value-added estimates

13
However CVA arrived in 2006

• Differentiated PA, regression methodology
• KS1 average points score and divergence
• KS2 fine grades and divergence
• Levels 345 48 chance 5A-C (1.4 of cohort)
  Levels 543 75 chance 5A-C (0.3 of cohort)
• Pupil factors FSM, Deprivation (IDACI), SEN,
  Gender, Age within year, mobility, in-care,
  ethnic group, EAL
• School composition
• School mean and SD of prior attainment
• With 95 confidence intervals

14
The aims of Performance tables

• Accountability as publically funded institutions
  to government public
• To support parental choice of schools (market
  driven) aligned with open enrolment
• To raise standards / support school improvement

15
1. Accountability

• Strong parental support
• Parents should be able to compare one schools
  performance against another (86)
• Tests and exam results are one important measure
  of a schools performance (87)
• The performance of each school in tests and exams
  should be published and publically available
  (87)
• Nationally representative sample of 1,624 adults
  (including 550 parents of child 0-18) in England,
  November 2008. DCSF (2009). School accountability
  and school report card omnibus survey.
  DCSF-RR107. London DCSF.

16
And among the media

• All the major newspapers publish school (and
  university) league tables (Times, Guardian,
  Telegraph etc).
• Education is a perfect media topic. It has heat
  and light. Heat because everybody cares about it
  ,and light because they all think that they
  understand it. (Journalist, quoted in Earl 2001,
  p6).
• With such an alliance of parents and media, any
  Government is going to listen.

17
2. Parental Choice
Telephone survey of 3,005 parents in summer 2004.
Wiseman et al (2005). London Challenge Second
survey of parents carers 2004 (RR624). London
DfES.
18
Social gradient in choice

• Academic outcomes was most common reason (43)
  offered by parents for wanting a place in their
  favoured school. But likelihood of citing
  academic outcomes was significantly higher for
• Owner Occupiers (21 relative to parents renting)
• Mothers in Social Class I II (1.71 against
  manual)
• BME mothers (1.71 against White)
• Parents residing in London boroughs (2.5 times
  more likely not to apply to their nearest schools
  than parent in Shire LEAs)
• Increases social segregation by indirectly
  informing parents which schools have high
  concentrations of high SES students or
  democratises the information that high SES
  families are already aware of though their social
  networks?

2,170 parents of Y6 children in 2000. Flately,
J., et al.(2001). Parents' experiences of the
process of choosing a secondary school (RR 278).
19
3. School Improvement

• Meta-studies from the US using independent
  measures of attainment (NAEP, NELS) suggest a
  modest positive impact on average (ES0.24), but
  studies provide mixed findings and tend to
  polarise between the extremes (Lee, 2008)
• Difficulty of disentangling school performance
  tables from other policies adopted at the same
  time (teacher certification rewards/sanctions
  for schools e.g. teachers performance pay, school
  vouchers, school takeover threats etc.)
• But recent study from UK comparing Wales-England
  has had a substantial impact

20
The Wales experiment

• Broad context is a substantial decline in Wales
  in Programme for International Student Assessment
  (PISA) scores for maths, reading and science

21
Burgess, Wilson Worth (2010)

• Welsh Assembly announced it would cease
  publishing secondary school performance tables in
  July 2001 (after exams taken). Natural experiment
  otherwise similar to England in inspection
  regimes, exams, etc.
• Any change in school results in Wales vs. England
  from 2002 onwards?
• Controls for
• prior attainment at KS3, entitled to FSM, Cohort
  size
• Pupil funding, population density,
  Church/Maintained
• Local competition (N schools with 5Km)

22
Results
23
Burgess conclusions

• Decline for Wales seen for APS as well as 5AC
  congruent with PISA results not gaming.
• Top 25 Welsh schools (by high PA or low poverty)
  not affected. Effect concentrated particularly
  for lowest performing schools falling furthest
  behind.
• Public scrutiny through PTs puts low performing
  schools in England under great pressure to
  improve, similar schools in Wales maybe coasted
• Removal of tables did not reduced school
  segregation and sorting (by FSM or by KS2 score)
  in Wales

24
Unintended consequences

• Excellent review by Smith (1995) on eight
  unintended consequences of PIs
• In school performance tables includes
• Gaming C/D boundary, GNVQ equivalents, switching
  exam boards, SEN school action, etc.
• Depressing baseline scores
• Teaching to the test
• Selective student admissions / removing
  difficult students
• Always happens in the school up the road! Little
  information on whether these work in long term.

25
Should they stay or should they go?

• One of the least evidence-based areas of school
  policy, but what evidence there is suggests PTs
  may be a driver/energiser of improvement
• Both political left and right support, but for
  different reasons (Markets, FOI/Openess/Empowermen
  t)
• Better than the alternatives? (remember the
  1980s)
• Have achieved a level of consensus among school
  leaders based on moral argument around CVA and
  fairness Every school, regardless of the SES
  circumstances in which it operates, must have a
  fair opportunity of achieving a good score DFE
  (2009) A school report card, par.47.
• Tamper with this at peril

26
How are the PTs changing?

• Addressing some of the gaming issues In 2004,
  when recognised as equivalent to GCSE, 1,882
  gained level 2 passes, risen to 462,182 students
  in 2010 (see Wolf review). Also SEN SAP or above
  only.
• Ebacc- will it drive out vocational? 22 eligible
  in 2011, 33 in 2012 and 47 in 2013 (DFE survey
  of 692 schools)
• VA for low/mid/high prior attainment and for each
  Ebacc subject (though are PTs the place for
  this? see later)
• Widen the range of indicators (average PA, FSM,
  EAL etc). Broadly good, but remember fate of PI
  schemes. Some (e.g. university destinations)
  largely outside of schools control?

27
New website (GoCompare)
http//www.education.gov.uk/researchandstatistics/
stats-search
28
Removal of CVA

• Took from 1992 2006 to get CVA, now discarded,
  why?
• The CVA measure is difficult for the public to
  understand... It is morally wrong to have an
  attainment measure which entrenches low
  aspirations for children because of their
  background. For example, we do not think it right
  to expect pupils eligible for FSM to make less
  progress from the same starting point as pupils
  who are not eligible for FSM. (DFE, 2010,
  Schools White Paper, p68).

29
Fundamental misunderstanding

• Confuses student level expectations from school
  level accountability. Pupils on FSM not only have
  lower attainment but DO make less progress at
  school.
• However this does not mean lower expectations for
  students, target setting and progress measures
  are not adjusted (and never were) for pupil
  background.
• The point is factors that are outside a schools
  control (gender, deprivation, EAL, ethnicity,
  SEN, mobility) need to be included in a school
  accountability indicator.

30
Suggested replacements

• Two levels of progress but highly subject to
  threshold effects
• Alternative approach to contextualisation through
  families of 10 to 15 schools with similar
  intakes for all regions of the country (DFE,
  2010, p76), but
• CVA, because it is based on individual pupil
  characteristics and their attainment, is not
  prone to the biases that can be created by
  comparisons based on school-level similarities.
  We therefore believe that some form of CVA is the
  best means of contextualising pupil progress.
  (DFE 2009, School Report Cards, par.77).

31
Options for schools?

• Does data speak for itself, like dials on the
  dashboard of a car? Does feedback always lead to
  improvement? Is providing data in PTs the best
  way of securing school improvement?
• There are a wide range of School Performance Data
  Services PIPs/MidYIS/ALIS, FFT, NFER-PASS, RAISE
  online, LA services like Lambeth
• Data is inevitably more detailed than that needed
  for performance tables (see following examples)
  but they also offer the training support to use
  the data effectively

32
(No Transcript)
33
Differential effectiveness by subject VA
against CAT
34
Value added for different groups of pupils

• Are pupils who make significantly more, or
  significantly less, progress
• boys/girls, SEN, EAL?
• Different ability levels?
• joined school recently?
• Alternatively, are they pupils
• Who missed particular classes for long periods?
• With a particular teacher?
• In a particular set?
• Who had extra support / intervention?
• Who followed a particular scheme of work?
• Whose teachers used different teaching practices?

35
(No Transcript)
36
Conclusions

• Accountability and PTs are here to stay
• Wales is reintroducing Last week all secondary
  schools were told which of 5 bands they had
  been placed into as part of a new accountability
  regime (based on raw scores, VA, attendance).
  Parents and the media will get the information in
  December 2011. A primary school model is being
  developed and will follow next year. (TES,
  23/09/11).
• We will see more and more data published and
  publically available (on the web, GoCompare
  style) but it will be harder and harder to make
  sense of it

37
Conclusions (Continued)

• The credibility of the data is key to users
  (Saunders, 2000), removal of CVA breaks the
  trust, should publically oppose this change
• Resist the shift of the entire burden from the
  State to schools, keep a focus on the policy
  issues (like EMA and equity in University entry)
• PTs can provide the incentive (if seen as fair)
  for both low attaining and (if CVA included) high
  attaining schools (avoiding complacency)
• but for SI need much more detailed data, training
  support It should be a bright future for
  School Performance Data services!

38
End of Session Thank you

• Professor Steve Strand
• Institute of Education
• University of Warwick
• steve.strand_at_warwick.ac.uk
• Tel. [phone]