Evidence from prePIRLS 2011 Faculty of Education Social Interaction Determinants of South African Reading Literacy Achievement: Evidence from prePIRLS 2011 Surette van Staden, University of Pretoria Annika Bergbauer, Ifo Institute, University of Munich Quantitative Methods Conference, Stellenbosch, 20 September 2016

Structure of the presentation Theoretical background Method and analyses Descriptive results OLS results Discussion Faculty of Education

Theoretical Background: Vygotsky’s Social Interaction Theory Cultural Context Social Interactions Cognitive Development Language Faculty of Education

Theoretical Background Vygotsky’s view: children’s sociocultural environment as a variety of tasks and demands. In completing these, the child is dependent on other people, in many cases the parents, who introduce them to what to do, how to do it and what not to do. Parents act as a conduit through which the cultural conditions pass to the child and actualise their instructions through language. Thus, as stated by Mahn (1999), the child co-constructs meaning through social interaction while language plays a role in the development of thinking. A central principle in sociocultural theory: individual and social processes are interdependent in the co-construction of knowledge. Faculty of Education

Theoretical Background The role of language is further elaborated by Mercer and Howe (2012) citing Vygotsky (1962, 1978); that the acquisition and use of language transforms a child’s thinking. Language becomes a cultural tool (in terms of the ability the child now has to share knowledge among members of the community) and as a psychological tool that structures the processes and content of individual thought. Mercer and Howe (2012) argue that language acquisition therefore has a profound effect on collective and individual thinking. Faculty of Education

Method and Data Source prePIRLS 2011: A nationally representative sample of 15 744 Grade 4 learners from 342 schools that provided achievement data. Learners were assessed across all 11 official languages and were assessed in the Language of Learning and Teaching (LoLT) to which they were exposed in Foundation Phase. Background questionnaires to Grade 4 learners, teachers, principals and parents of Grade 4 learners. Faculty of Education

Selected variables for the current analyses: prePIRLS 2011 learner and teacher questionnaire data Parents check homework Grade 4 learners were asked the frequency with which their parents checked their homework (variable ASBG07D). Response options: every day or almost every day, once or twice a week, once or twice a month and never or almost never. Parent talk about school Grade 4 learners were asked the frequency with which their parents talked to them about school (variable ASBG07B). Teachers’ emphasis on the curriculum Teachers were asked the extent to which they characterised the degree of success in implementing the school’s curriculum (variable ATBG06C). Response options: very high, high, medium, low and very low. Learners know what their teachers expect of them Learners were asked their agreement in terms of knowing what their teachers expected of them (variable ASBR05C). Response options: agree a lot, agree a little, disagree a little and disagree a lot.

Method and Data Source The International Database Analyser (IDB) was used to generate descriptive statistics. To control for learner characteristics, a learner asset scale and a school asset scale, learner age, sex and test language were used as controls for the regression analysis using Stata version 13.0 software. Faculty of Education

Social interactions as measured by: Cognitive development Populating Vygotsky’s Social Interaction Theory with prePIRLS 2011 variables Social interactions as measured by: Learners talking to parents about school Parents assisting learners with homework Teachers’ degree of success in implementing the curriculum The extent to which learners know what their teachers expect of them Cognitive development Knowing the language of testing before school Having books read to them before school Early literacy skills Language As evidenced by overall prePIRLS 2011 reading literacy achievement scores

How did we limit possible bias? Descriptive Results: How did we limit possible bias? Learners of higher socio-economic status are suspected to self-select into higher-quality schools in search for higher learning outcomes, while more apt learners may opt for higher-quality schools. Both endogeneities violate OLS assumptions and may lead to biased coefficients of school quality. We control for school quality by including average school test scores as explanatory variable. While there could still be unobserved factors that are not mentioned here, the inclusion of these characteristics limits the risk of bias substantially. Non-random school choice is more likely in rural areas, where fewer schools are situated and learners often have to attend the closest institution. For robustness checks, we limit the learner demographic to remote rural location totaling, resulting in similar estimates as for the whole sample. Faculty of Education

Descriptive Results We observe differences across the school quality spectrum when applying frequency weights. Literacy test scores expose a relatively large gap between the two test languages (English and African languages as a group) and school quality tiers. English-tested learners scored on average 531 points. African-tested learners score on average 425.73 points. This literacy score gap across test languages may be attributable to socio-economic status intertwined with school quality. Faculty of Education

Descriptive Results Conventional controls: Learner sex appear relatively similar across test languages. Learner age hints at slightly older African-tested students (10.61 years) opposed to English-tested students (10.33 years). Variables of socio-economic status appear higher for the English-tested tier. On average, English-tested learners have an asset index of 0.84, opposed to African language-tested learners with an asset index of -0.49. Mean school assets reflect this pattern too. Faculty of Education

Descriptive Results Parental attributes: Parents who have employment as a professional, English-tested learners expose an average score of 0.26, opposed to African-tested learners with an average of 0.03. Parental university attendance reaches an average of 0.35 for English-tested learners, opposed to 0.08 for African language-tested learners. Key teacher variables, such as educational qualification, are relatively similar across the tiers. Conventional indicators of school quality, class size, suggests a gap: on average, 36 learners attended English-tested classes, 43 learners attended African-tested classes. Faculty of Education

Descriptive Results We observe differences across the higher cognitive skills when starting school when applying frequency weights. Knowledge of the test language before starting school = 25 points higher. More frequent book reading before school = 15 points higher. Early literacy skills (not formal pre-school) seems to translate into higher Grade 4 literacy skills. Faculty of Education

Regression Results Faculty of Education (1) (2) (3) (4) VARIABLES   (1) (2) (3) (4) VARIABLES testscore 2.Learners know what teacher expects: Agree a little -11.68*** -12.82*** (2.717) (2.698) 3.LEarners know what teachers expect: Disagree a little -21.20*** -23.94*** (3.501) (3.507) 4. Learners know what teachers expect: Disagree a lot -6.955* -8.403** (3.804) (3.773) 2.Teacher implementation of curriculum: Medium 26.20*** -25.30*** (7.260) (7.161) 3.Teacher implementation of curriculum: Low -11.83 -10.76 (11.41) (13.05) 4.Teacher implementation of curriculum: Very low 111.7*** -96.41*** (22.44) (23.64) 1.Parents talk about school 13.63*** 14.02*** (3.696) (3.303) 2.Parents check homework: Every day or almost every day 18.13*** 21.62*** (2.825) (2.409) Constant 487.9*** 498.2*** 471.3*** 462.3*** (52.89) (52.91) (60.95) (46.05) Observations 7,803 8,070 8,944 9,258 R-squared 0.651 0.645 0.643 0.648 Faculty of Education

Discussion Findings from the current study provides evidence for the significant social dimension of parents and teachers (when controlling for background factors) in predicting increased reading literacy achievement scores. Results of large-scale assessment programmes could be interpreted applying the lens of social interaction theory in terms of consideration to the local context, the purpose of the test and the policy context in which the testing took place. Currently, no policy guides parental involvement or parent-teacher interactions, yet growing evidence of projects aimed at involving parents: Nalibali Tshwane Municipality project Matthew Goniwe School of Leadership Family mathematics and family science project in the Free State Faculty of Education