Suppliers of childcare to the market observations and a

							The impact of school choice on students’ university
        entrance rank scores in Australia

                Chris Ryan (ANU) &
        Louise Watson (University of Canberra)
                    October 2009
            Motivation

 Interest on the magnitude of school sector effects in Australia
   • Australian studies point to substantial effects
   • But how should we view these estimates, since they reflect
      non-random choice and, hence, selection
            Motivation

 Interest on the magnitude of school sector effects in Australia
   • Australian studies point to substantial effects
   • But how should we view these estimates, since they reflect
      non-random choice and, hence, selection
 Against a background of:
   • Increasing private student enrolment shares
   • A future of increased ‘information’ on school ‘performance’
            Motivation

 Interest on the magnitude of school sector effects in Australia
   • Australian studies point to substantial effects
   • But how should we view these estimates, since they reflect
      non-random choice and, hence, selection
 Against a background of:
   • Increasing private student enrolment shares
   • A future of increased ‘information’ on school ‘performance’
 Here, study the effects by analyzing the outcomes of students
  forced to make new school choices because of the structure of
  the school system in their jurisdiction
   • We compare their Year 12 outcomes with others from their
      first school (Year 9) who made different later school choices
             Outline from this point

   The Australian education system
   Data
   Characteristics of the new choosers
   Results: Estimated school sector effects
   Conclusions
            The Australian education system

 Schooling as a state government responsibility
            The Australian education system

 Schooling as a state government responsibility
 K-12 grade structure; generally primary (K-6) and secondary (7-
  12) schools
   • In two jurisdictions we have senior colleges for Grade 11&12
            The Australian education system

 Schooling as a state government responsibility
 K-12 grade structure; generally primary (K-6) and secondary (7-
  12) schools
   • In two jurisdictions we have senior colleges for Grade 11&12
 Three sectors: Government (public); Catholic; and Independent
            The Australian education system

 Schooling as a state government responsibility
 K-12 grade structure; generally primary (K-6) and secondary (7-
  12) schools
   • In two jurisdictions we have senior colleges for Grade 11&12
 Three sectors: Government (public); Catholic; and Independent
 Government funding of private schools – 15-80% of school costs
   • Dependent on school ‘resources’ – based on SES of students
            The Australian education system

 Schooling as a state government responsibility
 K-12 grade structure; generally primary (K-6) and secondary (7-
  12) schools
   • In two jurisdictions we have senior colleges for Grade 11&12
 Three sectors: Government (public); Catholic; and Independent
 Government funding of private schools – 15-80% of school costs
   • Dependent on school ‘resources’ – based on SES of students
 ENTER score at the end of Grade 12 as the main basis for entry to
  university
   • High stakes, summary ranking of individuals within their
      cohort
         The Australian education system

Table 2 Key descriptive features of the two cohorts
                                    1995 1998

Government School in Year 9 (%)      67.2 66.4
Catholic school in Year 9 (%)        20.2 20.4
Independent School in Year 9 (%)     12.7 13.2
Obtained a Year 12 certificate (%)   76.1 78.2
Has an ENTER score (%)               53.8 56.2
Attending university (%)             35.2 35.8
ENTER score                          71.8 72.5
            Data

 Longitudinal data based on two cohorts from the Longitudinal
  Surveys of Australian Youth
            Data

 Longitudinal data based on two cohorts from the Longitudinal
  Surveys of Australian Youth
 First surveyed in Year 9, followed to mid-twenties
            Data

 Longitudinal data based on two cohorts from the Longitudinal
  Surveys of Australian Youth
 First surveyed in Year 9, followed to mid-twenties
 Undertook literacy and numeracy tests in Year 9, background
  questionnaire at school, then followed annually by telephone
            Data

 Longitudinal data based on two cohorts from the Longitudinal
  Surveys of Australian Youth
 First surveyed in Year 9, followed to mid-twenties
 Undertook literacy and numeracy tests in Year 9, background
  questionnaire at school, then followed annually by telephone
 Complex survey: stratified by state and school sector
            Data

 Longitudinal data based on two cohorts from the Longitudinal
  Surveys of Australian Youth
 First surveyed in Year 9, followed to mid-twenties
 Undertook literacy and numeracy tests in Year 9, background
  questionnaire at school, then followed annually by telephone
 Complex survey: stratified by state and school sector
 Here analyzing data from the Grade 9 cohorts of 1995 and 1998,
  with their ENTER score outcomes observed in 1999 and 2002
            Data

 Longitudinal data based on two cohorts from the Longitudinal
  Surveys of Australian Youth
 First surveyed in Year 9, followed to mid-twenties
 Undertook literacy and numeracy tests in Year 9, background
  questionnaire at school, then followed annually by telephone
 Complex survey: stratified by state and school sector
 Here analyzing data from the Grade 9 cohorts of 1995 and 1998,
  with their ENTER score outcomes observed in 1999 and 2002
 Around 14000 initial observations in each cohort, 5.5% of Year 9
  cohort
                   Data

The LSAY questionnaire
 
Students who were at school when interviewed in the previous wave were asked:

Are you still going to {school name from the previous wave}? [Yes or No]

       If no, subjects were asked

       In which month and year did you leave {school name}?

              Did {school name} offer Years 11 and 12? [Yes or No]

In all waves, those subjects who had changed school were asked

       Are you now going to a Government school, a Catholic school or some other non-
       Government school?

              1      Government

              2      Catholic

              3      Other
           Characteristics of the new choosers

 The new choosers come from government and Catholic schools
 Most do not change sectors
           Characteristics of the new choosers

 What about normally unobserved characteristics
  • Planned educational attainment, occupational ambition,
    attitudes towards school, homework hours and attitudes
                 Characteristics of the new choosers
    Table 9      Comparison of factors that reflect ‘unobservable’ characteristics
                 across groups – compared with students in originating school

    Standardised comparison with Year 9                          Year 12
    school average                              Government       Catholic     Independent
    In Year 9                                                (standard deviations)
     Planned to undertake Year 12 (%)           -0.01            0.03         -0.02
     Planned to go to university (%)            -0.06            0.07         0.34
     Planned to do an apprenticeship (%)        0.04             -0.05        -0.21
     Planned to do vocational quals (%)         0.03             0.03         -0.42
     SES of future expected job                 -0.02            0.00         0.37
    School satisfaction scale                   -0.01            0.02         -0.01
    Interactions with teachers satisfaction     -0.02            0.01         0.20
    Opportunities or school relevance scale     -0.03            0.07         0.01
    Sense of achievement scale                  -0.02            0.04         0.02
    Positive attitude towards homework (4-12)   -0.04            0.04         0.20
    Hours of weekly homework                    -0.05            0.05         0.23
    Year 9 achievement                          -0.03            0.09         -0.12
            Results: Estimated school sector effects

 Independent school effects – no smaller than when estimated
  over those who do not change school
            Results: Estimated school sector effects

 Independent school effects – no smaller than when estimated
  over those who do not change school
 Catholic school effects are not significant
                    Results: Estimated school sector effects

      There are positive Independent school effects for changers
    Regression-based school sector effects

                                       No other controls         With controls2
                                    Catholic   Independent   Catholic Independent
                                     school       School      school      School

    Group who did not change school
    Regression (standard weights)
           β                         5.73***     10.84***     4.29***     4.84***
           Std error                (1.17)        (1.17)     (0.67)      (0.69)
    Group who changed, no senior program
    Regression (standard weights)
           β                       2.09            6.30       2.13        7.47***
           Std error                (1.99)        (4.03)     (2.26)      (2.19)
            Results: Estimated school sector effects

 Independent school effects – no smaller than when estimated
  over those who do not change school
 Catholic school effects are not significant
 Results seem robust across alternative estimators, including
  matching estimators
                    Results: Estimated school sector effects

      Robust across estimators
    Regression-based school sector effects – group forced to change schools

                                                                  With controls2
                                                              Catholic Independent
                                                               school      School

    Within regression (sample weights)
           β                                                    1.67              6.93**
           Std error                                           (4.55)            (2.83)
    Reweighted regression (private school treatment)
           β                                                    2.01             4.89**
           Std error                                           (1.62)           (2.33)
    Reweighted regression (treatment on untreated)
           β                                                   -2.57             1.70
           Std error                                           (1.92)           (3.36)
    Matching estimator3 (private school treatment)
           β                                                    6.21**           6.67*
           Std error                                           (3.16)           (4.02)
    Matching estimator (treatment on untreated)
           β                                                   -1.92          Too few
           Std error                                           (2.86)         observations
            Concluding remarks

 Independent school effects – no smaller than when estimated
  over those who do not change school
            Concluding remarks

 Independent school effects – no smaller than when estimated
  over those who do not change school
 Why are the Catholic school effects not significant?
            Concluding remarks

 Independent school effects – no smaller than when estimated
  over those who do not change school
 Why are the Catholic school effects not significant?
 How are these effects achieved by schools?
   • Not really so much known about this, but course or subject
     choices differ between the sectors
       Mathematics, languages and science choices differ