UNIVERSITY OF JYVÄSKYLÄ Characteristics of low and top performers in reading and mathematics. Exploratory analysis of 4th grade PIRLS and TIMSS data in Nordic countries Sari Sulkunen, Department of Languages Kari Nissinen, Finnish Institute for Educational Research Pekka Kupari, Finnish Institute for Educational Research University of Jyväskylä UNIVERSITY OF JYVÄSKYLÄ Introduction Consider three student groups: – Top performers – Low performers – ’In-betweens’, intermediate performers Vast majority of students are neither low or top performers – ’In-between’, intermediate pupils – Define these pupils as the reference group of analyses UNIVERSITY OF JYVÄSKYLÄ Introduction How to minimize the number of low performers and maximize the number of top performers? Look for characteristics which distinguish the low performers and top performers from the reference group – Student – Home – Teaching – School Are there key characteristics that could be managed by school and educational system? UNIVERSITY OF JYVÄSKYLÄ Previous findings Analyses of earlier large-scale data sets (TIMSS, PIRLS, PISA, national surveys etc.) => Certain characteristics repeatedly appear important 1. Socio-economic background – Parents’ education and occupation – Resources at home – Appreciation of education and culture at home – Often these explain school differences also UNIVERSITY OF JYVÄSKYLÄ Previous findings 2. Student’s attitudes and engagement – Motivation to school work – Free-time reading vs other activities – Self-concept 3. Student’s basic skills – E.g. reading accuracy and fluency • Lack of basic skills makes all learning difficult 4. Student’s language background – Command of instruction language highly important • ’Everyday’ vocabulary not enough? UNIVERSITY OF JYVÄSKYLÄ Previous findings Role of gender? – Girls outperform boys in reading – Gender differences usually smaller in mathematics – Gender itself not always significant in multivariate analyses => question of attitudes and engagement more than question of sex? UNIVERSITY OF JYVÄSKYLÄ Previous findings School and class-level factors play minor role (in Nordic countries) – Student characteristics explain much of (already small) school and class differences Why is this? Are teachers’ and schools’ efforts relevant at all? – Overall level of teachers’ and schools’ efforts and resources relatively standard (high)? – Question of teacher-student interaction; student’s contribution (engagement) crucial – Difficult to catch with questionnaire data! UNIVERSITY OF JYVÄSKYLÄ Data We utilize Nordic PIRLS 2011(reading) and TIMSS 2011 (mathematics) data sets – 4th grade pupils (except 5th grade for Norway) – Several sources of data: • student questionnaire • parent questionnaire (PIRLS) • teacher questionnaire • school questionnaire UNIVERSITY OF JYVÄSKYLÄ Data Finland, Norway and Sweden implemented TIMSS and PIRLS on the same schools and students In Denmark PIRLS and TIMSS had separate student samples Norway assessed 5th grade students also (with smaller sample!) – Comparable with other countries’ 4th grade FinIand, Norway and Sweden implemented an additional test of student’s basic reading skills UNIVERSITY OF JYVÄSKYLÄ Data Number of Students schools assessed in participating PIRLS in PIRLS Denmark 4th Finland 4th Norway 5th Sweden 4th 232 145 53 152 4594 4640 1258 4622 Number of Students schools assessed in participating TIMSS in TIMSS 216 145 54 152 3987 4638 1270 4482 UNIVERSITY OF JYVÄSKYLÄ Overall results In reading (PIRLS), all four Nordic countries performed internationally relatively high – Finland (568) and Denmark (554) in Top 10 – Norway 5th grade (549) and Sweden (542) in Top 20 • Scale midpoint = 500 In mathematics (TIMSS), Norway 5th, Finland and Denmark performed high while Sweden was mediocre – Norway 5th grade (548) and Finland (545) in Top 10 – Denmark (537) in Top 20 – Sweden (504) close to scale midpoint (500) UNIVERSITY OF JYVÄSKYLÄ Performance groups Define low performers as students below 475 points 475 = intermediate international TIMSS and PIRLS benchmark Define top performers as students over 625 points 625 = advanced international TIMSS and PIRLS benchmark UNIVERSITY OF JYVÄSKYLÄ Group percentages / reading % low performers in reading: girls Denmark 4th Finland 4th Norway 5th Sweden 4th 10 6 9 13 % low performers in reading: boys 14 10 13 17 % top performers in reading: girls 14 23 12 11 % top performers in reading: boys 11 14 8 7 UNIVERSITY OF JYVÄSKYLÄ Group percentages / mathematics % low performers in maths: girls Denmark 4th Finland 4th Norway 5th Sweden 4th 19 17 14 34 % low performers in maths: boys 17 17 13 32 % top performers in maths: girls 9 10 9 2 % top performers in maths: boys 12 13 16 4 UNIVERSITY OF JYVÄSKYLÄ Statistical modelling Three-level logistic regression models for each country A number of explanatory variables were introduced in the multivariate regression – Suggestions from previous research – Interesting variables from the TIMSS and PIRLS background questionnaires UNIVERSITY OF JYVÄSKYLÄ Statistical modelling After preliminary steps a reduced variable list was used in the final models – Variables appearing significant in at least one country – Two-way interactions were originally considered but left out from the final regression models Recall: multivariate model => effects are adjusted for all other variables in the model UNIVERSITY OF JYVÄSKYLÄ Results Student’s basic skills – Important predictor of performance in every case – Only minor variation between countries – Strong relation between skills at school start and 4th grade performance (stability) • Early literacy activities with child predict top reading performance Student’s attitudes – Confidence <=> performance – Motivation and engagement in learning, no show??? – Liking to read related to reading performance • Not found with mathematics! UNIVERSITY OF JYVÄSKYLÄ Results Student’s activities – Reading activities: • No role in distinguishing low performers from intermediate ones • Instead, top performers stand out from the rest – Free-time computer use negatively associated with performance Home resources (SES) – Important in every country, with minor variations – Educational resources, books, parent’s education,… – Wealth of neighborhood (DEN, SWE) UNIVERSITY OF JYVÄSKYLÄ Results Language at home / language difficulties in class – Some role found in every country Parental and teacher support – Highly significant in every country (in various forms) – But: more support <=> lower performance • Top performers’ need for support not recognized? • Top performers do not express need for support? UNIVERSITY OF JYVÄSKYLÄ Results Male gender – Low performance in reading: DEN – Top performance in maths: FIN, NOR, SWE Female gender – Low performance in maths: SWE – Top performance in reading: FIN UNIVERSITY OF JYVÄSKYLÄ Results School-related variables largely non-significant (given the student, home and teaching related variables) – Discipline and safety at school – School emphasis on academic success – School size and location – School resources – Only occasional exceptions: • Sweden: low performance associated with urban schools • Denmark: top reading performance associated with urban schools and good resources • Denmark: low maths performance associated with small schools UNIVERSITY OF JYVÄSKYLÄ Results Teaching-related variables completely non- significant except for supporting low performers (given the student, home and school related variables) – Teacher’s experience and education – Student ability grouping – Activities in engaging students – Teacher’s confidence in teaching maths – Memorization or elaboration in teaching maths – Calculator or computer use in class – Amount of given homework in maths – Characteristics of reading materials UNIVERSITY OF JYVÄSKYLÄ Conclusions Gap between low and top performers seems stable – School cannot even out the impact of background and differences in basic skills at school start support for struggling students still insufficient? In reading: – Early help with problems in linguistic development – Support in reading engagement, also for (disadvantaged) families • Bring school’s and student’s textual landscapes closer (text materials) • Family literacy programs UNIVERSITY OF JYVÄSKYLÄ Conclusions In reading: – Immigrants: need for intensified support both in instruction language and in mother tongue – Development of reading and learning strategies • Special benefit for struggling readers! In mathematics: – Support in reading! – Self-concept / confidence / positive attitudes crucial • Influence of teacher actions and attitudes • Appropriately tuned tasks and homework • Bring school’s and student’s ’mathematical landscapes’ closer (tasks, learning materials) UNIVERSITY OF JYVÄSKYLÄ Conclusions What to do with top performers? – Teachers and parents may overlook them since everything goes fine => implications e.g. for school enjoyment? – Mild interest from adults’ side and lack of challenges may hinder reaching full potential Need for individualized pedagogical solutions, how about resources? Need for continuing professional development of teachers, how about resources? UNIVERSITY OF JYVÄSKYLÄ Tak! Takk! Tack! Kiitos!