Belgian Communities An increasing gap

1
Belgian Communities An increasing gap

• PISA Conference
• Trento, 3-4 April 2008
• Valérie QUITTRE
• A. Baye, A. Fagnant, G. Hindryckx, D. Lafontaine,
  C. Monseur
• French speaking Community
• Université de Liège

2
Student performances PISA 2006
3
TIMSS Science
1971
1985
1995
4
TIMSS Math
1964
1980
1995
5
Context

• Labour and economic indicators are better in the
  Flemish than in the French part

6

• Percentage of unemployed in the three regions
  Eurostat 2006

7

• Economic social and cultural status indice in
  both Communities PISA 2006

8

• Immigration rate is much higher in the French
  part PISA 2006

9

• Flemish and French Communities organize separate
  educational systems ( since 1989 Belgium became
  a Federal State)
• Because of common roots, systems are still now
  similar
• Initial training has the same length

10

• Schools management is either public or private
• Most private schools are sectarian (catholic)
• Freedom of study programmes, but Standards of
  Proficiencies to both public and private school
  management
• Parents and students can freely choose the school
  (no restriction at all)
• Secondary education is organised in tracks in
  both Communities

11
Similar but

• Percentage of students in public and private
  schools are different PISA 2006

12

• Grade repetition is more usual in French part
  PISA 2006

13

• Students are differently distributed in the
  tracks in both Communities PISA 2006

14
BFR
BFL
15
Student performances in science PISA 2006
16

• Percentage of students at each proficiency level
  on the science scale PISA 2006

17
Science achievement differences according to
parents social, economic and cultural status
background - PISA 2006
25 most disadvantaged
25 most advantaged
BFL
BFR
18
39 of the differences of the score between
schools result from the differences of ESCS
41
21
19
Performance in science and the impact of
students socio-economic background
BFL
BFR
20
In summary,

• Differences in outcomes the Flemish students
  perform better in each domain
• Similarities both Communities have inequitable
  system
• Is it possible to answer THE question WHY ?

21
Analysis

• To attempt to understand the differences between
  both Communities, multilevel regression analysis
  have been performed.
• As in both Communities schools strongly differ
  according to the type of tracks they organize,
  the schools in the PISA 2006 sample have been
  separated according to their education offer
• General schools (General only) schools
  organizing only general tracks versus
• Mixed and vocational schools (Others) schools
  organizing either both general and vocational
  tracks, or only vocational tracks
• The regression analysis have then been performed
  separately, by school type.

22
Differences between Flemish and French speaking
Communities PISA 2006
23

• Steps of the analysis
• Empty model (model 0)
• Introduction of the  Community  (model 1)
• Introduction of student level variables (model 2)
• Family characteristics
• Highest parental occupational status (Hisei)
• Books at home (st15q01)
• Immigration status (native vs 1st or 2nd
  generation)
• Academic characteristics
• Grade (st01q01)
• Track (only for  other  schools)
• Introduction of school level variables (model 3)

24

• Steps
• Empty model (model 0)
• Introduction of the  Community  (model 1)
• Introduction of student level variables (model 2)
• Introduction of school level variables (model 3)
• Family characteristics
• School mean Hisei
• School mean number of books at home
• Percentage of immigrants at the school
• Academic characteristics
• School mean grade
• Percentage of students in vocational tracks (only
  for  other  schools)
• School mean homework
• Public / Private school

25
Variance analysis
This refers to the dispersion of dots above and
below the gradient lines (at student and school
level).
26
Variance analysis
27
Variance analysis

• Percentage of variance between schools

28
Variance analysis

• Percentage of variance between schools is
  significantly different from one type of school
  to another
• The high percentage of variance among vocational
  schools shows a difference between high-quality
  vocational schools and low-quality vocational
  schools (ghettos schools)
• The lower percentage of variance among general
  schools shows there are not many difference
  between these schools (against the idea only some
  of them are elite schools)

29
Variance analysis
30

• The introduction of the Community variable
  explains 37 of the between schools variance in
  general schools but only 22 in mixed and
  vocational schools.

31
Variance analysis
32
Fixed effect analysisThis refers to the slope of
the gradient lines. It is an indication of the
change of science score attributable to the
introduced variable
33
(No Transcript)
34
(No Transcript)
35
(No Transcript)
36
General schools

• Once students and schools characteristics taken
  in account
• In French speaking Community, change in science
  score between public and private general schools
  is 6.18 points
• In Flemish Community, this gap is 14.54 points
• Once this interaction is controlled, no
  significant difference between both Communities

37
(No Transcript)
38
(No Transcript)
39
(No Transcript)
40

• Whatever the education offer, individual
  characteristics have a significant impact on the
  performance in science
• In mixed or vocational schools, the grade and
  track have the most powerful influence on
  achievement in science

41
(No Transcript)
42
Mixed and vocational schools

• Once students and schools characteristics taken
  into account
• In French speaking Community, change in science
  score between public and private mixed and
  vocational schools is 1.43 points
• In Flemish Community, the gap is 6.97 points
• Differences between communities remain
  significant in mixed and vocational schools

43

• In general schools, the school variables are not
  significant the quality of education does not
  matter the school once education offer taken
  into account!
• In mixed and vocational schools, the level of
  repeating students in the schools makes the
  difference
• Some professional schools seems to be more likely
  to welcome repeating students

44
PISA Math - 2003

• Similar analysis has been performed on PISA Math
  2003 data
• One step has been added
• Empty model
•  Community  variable (model 1)
• Students variables (model 2)
• Schools variables (model 3)
• Schools  climate and disciplin  variables
  (model 4)
• Student behaviour (STBEHA)
• Teacher behaviour (TEACBEHA)
• Student morale (STMORALE)
• Poor student-teacher relation (MSTREL)
• Student/math teacher ratio (SMRATIO)
• Use of assessment (ASSESS)

45
Student behaviour PISA 2003School principals
perception
46

• Students and schools characteristics (model 3)
  explain less the difference of math performance
  between Communities

47

• Students and schools characteristics (model 3)
  explain less the difference of math performance
  between Communities
• None of the new characteristics introduced (model
  4) has a significant effect on the math
  performance. Model 4 explain only 3 percents of
  the gap between Communities.

48

• Students and schools characteristics (model 3)
  explain very few differences of math performance
  between Communities

49

• Students and schools characteristics (model 3)
  explain very few differences of math performance
  between Communities
• Model 4 only explains 2 supplementary percent but
  students behaviour has a significant impact on
  the math performance

50
Conclusion

• In general schools, the difference between both
  Communities in terms of efficacy seems to be
  linked to higher efficacy of private schools in
  the Flemish Community. Once the interaction
  CommunityPrivate is included, the difference
  in science is no more significant between
  Communities.
• In the other schools, the difference between both
  Communities remain significant, whatever the
  students and schools characteristics included. To
  improve the quality of mixed and vocational
  schools seems to be one of the challenges for the
  French speaking Community.

51
(No Transcript)
52
(No Transcript)