## TWENTY YEARS OF PISA STUDY IN SERBIA: (UN)LEARNED LESSONS

**Marina Videnović**

University of Belgrade, Faculty of Philosophy, Institute of Psychology | *mvidenov@f.bg.ac.rs*

PISA is one of the most comprehensive international comparative studies of educational systems. The study estimates 15-year-old students’ readiness to solve real-life challenges by using reading, mathematics and science knowledge. Accordingly, mathematical, reading and science literacy have been the main variables in every research cycle. Research includes background variables at the school, family and student-level factors that could influence achievement. Investigating system equity and equality is one of the major focuses of the PISA study. Through every cycle, it has been investigated how students’ achievement is predicted by their economic, cultural and social status (ESCS). The PISA research framework also covers current educational issues. The study from 2022 opens questions related to the sustainability of education during the pandemic. Results show that around 50 percent of countries have significantly declined in mathematical literacy compared to the previous cycles. Average scores in the OECD countries are the lowest obtained from 2020: 472 points in mathematical literacy, 476 points in reading literacy, and 485 points in science literacy. The difference in mathematical literacy scores between advantaged (highest ESCS status) and disadvantaged (lowest ESCS status) students is 91 points. Almost every third of students (31%) from the OECD countries didn’t meet the basic level of mathematical literacy. The Serbian educational system has been part of this assessment since 2003 or for five cycles (2003, 2006, 2009, 2012, 2018, 2022). During the twenty years of the PISA study in Serbia, 15-year-old students’ achievement was relatively stable. Achieved scores have been around 40 points below the OECD average. In this cycle, the scores are 447 in mathematical literacy, 440 in reading and 447 in science. The difference between advantage and disadvantage students’ scores on the mathematical literacy scale is a significant 81 points. The estimation is that around 40% of students would face difficulties answering contemporary world demands due to a lack of literacy skills. The symposium aims to discuss open questions, main findings and policy implications.

**Keywords: **PISA study, literacy, equity, education, pandemic

Acknowledgment: This work was supported by the Ministry of Education, Science and Technological Development of the Republic of Serbia (No. 451-03-68/2022-14/200163).

______________________________________________________________________________

**THE POWER OF STUDENT SOCIO-ECONOMIC STATUS IN SHAPING PERFORMANCE IN MATHEMATICS**

**Dragica Pavlović Babić**

Faculty of Philosophy, University of Belgrade| dpavlovic@f.bg.ac.rs

**Aleksandar Baucal**

Faculty of Philosophy, University of Belgrade

The main aim of this paper is to determine the impact and change in impact over time of students’ socio-economic status on its performance in mathematics, estimated in two consecutive cycles of OECD Programme for International Student Assessment (PISA). Based on additional secondary analysis of PISA 2018 and 2022 data, we have compared national performance, as well as performance of a number of selected countries (Croatia, Montenegro, Estonia, North Macedonia, Finland), in order to shed light on potential reasons for lower equity in Serbia. As for all previous cycles, socio-economic status of student has been the strongest predictor of performance in mathematics in Serbia. This is the same for all other domains of performance. It is also similar in the most PISA participating countries. For PISA cycle 2022, SES accounted for 13% of the variation in mathematics performance in PISA 2022 in Serbia, which is slightly in favor of the education system in Serbia compared to 15% on average across OECD countries. However, compared with the previous cycle, equity was higher in the previous PISA 2018 study than in PISA 2022 (In 2018: 8%, in 2022: 13%). Moreover, a comparison of math achievements in PISA 2018 and PISA 2022 suggests that gap between the highest-scoring students (10% with the highest scores) and the weakest students (10% with the lowest scores) narrowed. Socio-economically advantaged students (the top quarter in terms of socio-economic status) outperformed disadvantaged students (the bottom quarter) by 81 score points in mathematics. This is similar to the average difference between the two groups (93 score points) across OECD countries. Taking together these findings, it is an open question why the equity in domain of math achievements is lower than it was in previous cycles. The possible factors influencing the significantly lower equity of the education system of Serbia compared to the previous period will be additionally discussed.

**Keywords**: socio-economic status of students, performance in mathematics, OECD/PISA

______________________________________________________________________________

**DISTANCE LEARNING IN SERBIA DURING THE COVID PANDEMIC**

**Marina Videnović**

University of Belgrade, Faculty of Philosophy, Institute of Psychology | *mvidenov@f.bg.ac.rs*

The pandemic of COVID has caused a shock to the educational system all around the globe. An estimation was that more than 90 per cent of the students were affected by some kind of school closure. PISA study 2022 addressed this issue. This study aims to distinguish different student profiles or groups based on their experience with self-directed learning during the COVID-19 pandemic in Serbia using a PISA sample. Different questions were designed as part of the PISA study to investigate students’ experience with distance learning. Students’ self-efficacy in self-directed learning was examined with eight statements. Students reported how often (“never”, “a few times”, “about once or twice a week”, “every day or almost every day”) they experienced the defined eight problems during distance learning. School and family support to maintain learning and well-being while school buildings were closed were also assessed with two groups of statements. Students’ responses were combined into four indexes (students’ self-efficacy with self-directed learning, problems with self-directed learning, school actions/activities to maintain learning and well-being and family support) with average scores around zero and standard deviations are one across OECD countries. Four groups were identified with hierarchy cluster analysis (Ward method, Squared Euclidean Distance). Three statistical significance discriminative canonical functions were used to describe groups’ characteristics (λ = .189, *R*= .765, *p*<.001; λ = .456, *R *= .686, *p*<.001;* λ *= .860, *R *= .374, *p*<.001). The first group met fewer problems with online learning than the rest while perceived self-efficacy as higher. The second group, the largest one, dealt with problems during self-directed learning and had lower self-efficacy than other groups. The third group is characterised by lower family support, while the third group experienced the highest family support. A one-way ANOVA demonstrated that the effect of group membership was significant for students’ economic social and cultural status (*F*(3, 32284) = 51.24, *p* < .001) and mathematical literacy achievement (*F*(3, 32284) = 184.13, *p* < .001). The results indicate that the educational system’s actions in times of crisis should be individualised taking into account differences in the students’ needs.

**Keywords:** COVID-19, PISA study, self-efficacy, family support, educational system

Acknowledgment: This work was supported by the Ministry of Education, Science and Technological Development of the Republic of Serbia (No. 451-03-68/2022-14/200163).

_____________________________________________________________________________

**STUDENTS’ MATH ACHIEVEMENT AND PERCEIVED CHARACTERISTICS OF MATH CLASS: PISA 2022 STUDY**

**Gordana Čaprić**

Institute for Education Quality and Evaluation, Belgrade, Serbia| *gcapric@ceo.gov.rs*

This study aims to investigate the association between students’ perception of math class (perceived mathematics instruction quality, discipline climate and mathematics teacher’s support) and students’ mathematical literacy according to the PISA 2022 study. Students estimate the overall perceived mathematics instruction quality on a scale from 1 to 10. Discipline climate was assessed through the seven statements (scales from 1 happened at every class to 4 never). Teacher support was evaluated using the same scales (1 to 4) with the four statements. PISA 2022 students’ sample was used to obtain the results. The average perceived mathematics instruction quality is 6.1. The analysis of discipline climate statements reveals that a significant number of students report noise and disorder (28.6%), distractions caused by digital devices used by other students (28.1%), and a lack of listening to the teacher by students (40.4%), in all or most math classes. In terms of teacher support during math classes, around two-thirds of students report that their teacher shows interest in each student’s learning (60.5%) and provides help when needed (66.5%), while roughly half report that their teacher helps them with learning (58.8%) and continues to teach until they understand the lesson (56.4%). Also, disadvantaged students report support from teachers more frequently than advantaged students. Linear regression was used to investigate the relation between mathematical literacy (dependent variables) and perceived mathematical instruction quality, discipline climate in math class and teacher support. The model is statistically significant, explaining the relatively low percentage (3%) of mathematical literacy variance (*R² = *0.03, *F*(2, 32284) = 68.33, *p *< .001). The analysis of regression coefficients shows that students with higher achievement perceived mathematical instruction and discipline climate as better and teacher support as lower. However, the association is not strong, indicating that teachers have similar approaches to providing math classes for higher and lower achievers. The results suggest that it is necessary to find effective solutions to increase students’ motivation to work in the classroom and establish clear rules regarding using personal electronic devices on a policy level. Furthermore, it is essential to review teaching instructions and enhance the capacities of teachers to provide support to all students in their classes.

**Keywords:** mathematical literacy, quality of mathematical instruction, teacher support, discipline climate, PISA study

______________________________________________________________________________

**CULTURAL CORRELATES OF GENDER GAP TRENDS IN PISA MATHEMATICS PERFORMANCE ACROSS COUNTRIES AND TIME**

**Katarina Mićić **

Department of Psychology, Faculty of Philosophy, University of Belgrade | katarina607@gmail.com

**Barbara Blažanin **

Faculty of Special Education and Rehabilitation, University of Belgrade | barbara.blazanin@gmail.com

Gender difference in mathematics performance favouring males shows geographical and temporal variations indicating its dependence upon cultural influence. Societies differ in the width and direction of the gender gap in mathematics achievement. Also, there is an overall trend of improvement of this issue – the gap has been getting more and more narrow over the last couple of decades. However, the pace of the improvement varies across cultural contexts and is not detected in all countries. Studies show that cultures that stress ambition and competitiveness tend to have a relatively higher achievement disadvantage of girls compared with boys. The goal of this study is to investigate cultural dimensions connected to the changes in gender gap in mathematics performance over the last decade.

To approximate cross-countries’ trends in gender difference in mathematics performance, we used data from PISA from 2012-2022. From this data we used the average decennial trend in gender difference in mathematics performance across the last ten years of PISA cycles. The measure was expressed for 69 countries using linear regression showing an average change in the boys-girls difference in the last four cycles. Negative value of the measure indicates a trend of closing the boys-girls gap across the four cycles, and vice versa. Six dimensions based on Hofstede cultural dimensions theory were used to predict the decennial trend. Analyses were performed on the dataset consisting of 51 countries that had both PISA data and data on the cultural dimensions.

As in other studies, the data depict a general trend of improvement of gender equality in the last 10 years – on average and in most countries. However, results from the last cycle indicate a slight but significant regress in comparison to the 2018 cycle. Narrowing of the decennial gender gap favouring boys was significant in 10 countries (out of 33 where it was detected), whilewidening of the gap was significant in three countries out of 18 where it was detected. A regression analysis shows that Hofstede’s cultural model explains 35% of the variance in the average decennial trend in gender difference (*F*(6,43)=3.282, *p*=.011). The trend is best predicted by individualism (β=.116, p=0.018) and indulgence (*β*=-.107, *p*=0.045), while another two dimensions were marginally significant predictors – masculinity (*β*=-.067, *p*=0.073) and uncertainty avoidance (*β*=-.072, *p*=0.077). Collectivist unrestrained societies showed the greatest improvement in gender equality over the last ten years. Identification of relevant dimensions is informative for intervention programs to stress the contributing processes and values and to counter the undesired ones.

**Keywords**: PISA, mathematics, gender, Hofstede cultural dimensions, individualism

Acknowledgment: This work was supported by the Ministry of Education, Science and Technological Development of the Republic of Serbia (No. 451-03-68/2022-14/200163).

______________________________________________________________________________

**EDUCATIONAL ASPIRATIONS OF STUDENTS AND THEIR RELATIONSHIP WITH SOCIO-EMOTIONAL AND DEMOGRAPHIC CHARACTERISTICS**

**Barbara Blažanin **

Faculty of Special Education and Rehabilitation, University of Belgrade | barbara.blazanin@gmail.com

**Katarina Mićić **

Department of Psychology, Faculty of Philosophy, University of Belgrade | katarina607@gmail.com

One of developmental tasks for adolescents is to choose an educational path that is the right fit for them. Sometimes students lack sufficient knowledge or confidence regarding their competences, which can lead to mismatch between educational aspirations and competences, followed by potential unfavourable outcomes. Aims of this study are to test how many students are realistic about their future education, and to describe extracted profiles based on their demographics and socio-emotional characteristics (SEC). We used data from PISA 2022 which consisted of 6256 Serbian students. First, we crossed educational aspirations with test performance. Descriptive analysis showed that 24% of students are low performers (did not attain minimum proficiency in all three tested domains). Among low performers, 22% expected to complete high school (Realistic low performers (RLP)), 36% expected to complete tertiary education (Unrealistic low performers (ULP)), while 42% didn’t respond. 19% of students stood out as high performers (attained at least minimum proficiency in three domains, and high proficiency in at least one domain). Among them, 8% aspired to complete high school (Unrealistic high performers (UHP)), 80% expected to complete tertiary education (Realistic high performers (RHP)) and 12% didn’t respond. Chi-square, MANOVA, and Bonferroni post hoc tests were conducted to further understand the difference between RLP, ULP, UHP and RHP based on demographics and SEC. RLP are more often from vocational schools (*χ*2 = 676.08, *p*< .01), boys (*χ*2 = 34.48, *p*< .01) and from lower economic, social and cultural status (ESCS) (*χ*2 = 343.35, *p*< .01). They have the lowest scores on perseverance (*F*(3, 1668) = 16.55; *p*< .001, partial *η*2 = .029), curiosity (*F*(3, 1668) = 25.41; *p*< .001, partial *η*2 = .043), empathy (*F*(3, 1668) = 21.34; *p*< .001, partial *η*2 = .037) and assertiveness (*F*(3, 1668) = 18.49; *p*< .001, partial *η*2 = .032). ULP are more often from vocational schools and from lower ESCS. They have low scores on all SEC domains, but are significantly higher on curiosity and assertiveness compared to RLP. RHP are more often from general schools, girls and from higher ESCS. They have the highest scores on perseverance, curiosity, empathy and assertiveness. UHP are more often boys, with mid-level of perseverance, empathy and assertiveness. Compared to RHP they have lower levels of curiosity. Better understanding of unrealistic students is important for designing a strategy to help them set high, yet realistic expectations for future education, which is important for both social mobility and economic prosperity.

**Keywords**: PISA, educational aspiration, academic performance, socio-emotional characteristics