Estonian Education; The significant influencers of the 2015 PISA results

Leiden University - Campus: The Hague

Faculty of Governance and Global Affairs - Institute of Public Administration

Masters Thesis

Estonian Education; The significant influencers of the 2015 PISA results

Supervisor: Prof. M.C. Berg

Acknowledgements

Firstly, I would like to thank my supervisor, Professor Maarten Berg and the Leiden

University community for their flexibility, patience and intellectual support.

Secondly, my dear friends, Mea, Elza, Virág, Ruairidh, Laura and Jaanus who

supported me emotionally throughout this project. I would like to thank Ms. Tire and

Ms. Henno from the Estonian Ministry of Education, who have always answered my

questions about the Estonian PISA results and therefore contributed greatly to this

paper. Last but not least, I would like to thank my mom and my sister who have

always supported my education and constantly encouraged me to learn.

Abstract

The PISA test results are eagerly awaited by scientists, policy makers and the public alike every three years upon their release. The perfect recipe for good education is an extremely sought after and highly esteemed formula the world over. While Estonia has performed outstandingly on the last two tests conducted, it would be of great interest to investigate what exactly influences local education in the small Baltic state. There are numerous general concepts such as socio-economic status, ethnicity and social mobility that can enhance one’s cognitive and professional development or prevent one to participate in the educational system. This study is concerned with how socioeconomic status, ethnicity and IT facilities affect Estonian high school students' PISA score. Considering that Estonia is a very rich and multiethnic society, with a quarter of the citizens being Russian, their entry into modern educational systems and later to the labour market should be of high priority in integration policies. Additionally, Estonia is famous for its digitalization and Information Technology advancement. There is strong evidence that supports the theory that technology facilitates learning but has this translated to the classrooms yet and if so, does it have an effect on the PISA scores of local children? The research concluded that while Estonian classrooms and homes are very well equipped with IT devices it does not necessarily have a direct effect on a student’s results. However, socio-economic status, parental involvement, ethnicity and teacher student relationship are found to be of high significance in terms of Estonian students' PISA test scores.

Table of Contents

Limitations of the study ... 47

Conclusion ... 48

5

Estonian Education; The significant influencers of the 2015 PISA results

Introduction

In recent years, Estonia has become a very interesting country to examine given that from immense economic and political turmoil, the small state became one of the Nordics according to UN classification within just 20-some years of its formation. (Estonian World, 2017, web; OECD, 2001, p. 197). After being suppressed for almost a thousand years, the Estonian nation finally took matters into their own hands. A period of economic growth followed, democratic institutions were set up and living standards around the country have risen since the period of independence (OECD, 2001, p. 12). With help from the West, Estonians established one of the world’s most outstanding educational systems via government policies and rigorous public service labour (Garcia, 2011, p. 1-4; Toots, 2009, p. 66-67). According to the widely acknowledged Programme for International Student Assessment, the PISA test, Estonian students scored amongst the highest in the world while ranking highest overall in Europe (Estonian Ministry of Education and Research). Policy makers all around the world use the PISA results and the results of the highest-performing nations to justify regulations and policies which are expected to enhance education and thus the economy (Wolf, 2004, p. 316). This paper will examine Estonian education and within that try to pinpoint which characteristics of Estonian education could shape students’ performance. There are a multitude of reasons a student can perform well or poorly; from very random facts, such as book ownership to teacher-student relationship, many factors affect performance (Mikk, 2015, p. 333; Clark & Poulton, 2011, p. 5-6; Evans et al, 2010, p.189; Muller, 2015, p. 140). The question posed by this paper is; what are the significant influencers of performance in Estonia and why are they influencing results? This research aims to point out if the Estonian educational system is particularly outstanding because of the policies within it, such as the Tiger Leap, which installed a lot of equipment and digitalized education, or due to other factors, such as home based IT equipment, or extracurriculars paid for by the parents. For this I will use variables from the 2015 PISA test results.

Most believe that academic achievement can be measured and numerified, thus there is a lot of attention on standardized international test results (Biesta, 2009, p. 37; Toots, 2009, p. 59).

6

Contemporarily, the PISA test is considered to be one of the best indicators of education, therefore, it is safe to say that the Estonian results are valid. The tests, done by the OECD, are conducted every 3 years in order to provide information on comparative performance data on schooling systems along with other national education policy developments (Sellar & Lingard, 2014, p. 917). This repetition allows to capture the alteration in each country’s accomplishments (Lenkeit & Caro, 2014, p. 147-148). With the PISA test, the OECD became a center of calculation and later, a global provider of technological know-how in the measurement of educational performance in both member and non-member states (Sellar & Lingard, 2014, p. 918). The Organization for Economic Cooperation and Development collects data from all over the world, not merely on education but also educational resources available, including students’ socioeconomic statuses via measuring their possessions, and considers discrimination by measuring girl-boy and immigrant performance (Lenkeit & Caro, 2014, p. 146-147 and 155-156). The PISA test and the OECD findings serve as premier sources for educational statistics and are referred to in order to legitimize the creation of new educational policies (Sellar & Lingard, 2014, p. 919; 923). However, there are shortcomings of the PISA tests; including the fact that there are many schooling systems with different criteria and length which were not taken into consideration when measuring performance (Lenkeit & Caro, 2014, p. 148).

Education policies are an important area of public administration; knowing how education works enables policy makers to install strategies that change communities (Biesta, 2009, p. 40.; Banach 1995). The majority of states offer education for free in order to invest into human capital and indirectly into the job market. The process of transmitting accumulated knowledge - education - affects one tremendously. It raises awareness, teaches skills, customs and values from generation to generation (Ismaili & Latifi, 2012, p. 4731). Education affects growth and in addition to creating awareness of being part of a higher structure it also creates a surface for effective functioning of public administration (Ismaili & Latifi, 2012, p. 4731). The public are administered smoothly if they feel connected to the system of which they are part of and, schools are the perfect tool. Educational institutions are one of the greatest influencers of a child’s knowledge, values and life (Banach, 1995; Ülavere & Veisson 2015 p. 109; Muller, 2015, p. 140; Smith, 1995, p. 216). Both Public Administration and Education function to introduce the public to (modern) culture, civilization and also to transmit social norms and traditions while creating social circles (Yeaky et al, 2008, p. 12; Biesta, 2009, p. 40; Ülavere & Veisson, 2015, p. 116;119).

7

It is the key to the future generation’s knowledge, well-being and job market positions and therefore it intertwines with economics and politics as well (Smith, 1995 p. 215; Gylfason, 2001, p. 858; Toots, 2009, p. 58; Biesta, 2009, p. 40-41). Education is closely linked to a country’s prosperity which is often measured in GDP; usually countries with a high GDP have better education (Wolf, 2004, p. 316-317; Pastor et al 2018, p. 1632-1633). Moreover, it fosters Research and Development because it is an indirect investment into human capital (Wolf, 2004, p. 316). However, studies find education fosters development and economic growth only if the curriculum is set out to do so, which is a necessary condition but not a sufficient one (De Meulemeester & Rochat, 1995 p. 385).

As mentioned previously, there are numerous reasons why we measure the development of knowledge and, at least democratic societies, if not all societies, should have a constant discourse about their public education (Biesta, 2009 p. 37; Yeaky et al, 2008, p. 12). Governments are working to excel; their purpose and legitimacy stem from the need to make their countries better (Raadschelders, 1999, p. 290). To achieve this goal Public Administration is key; it is the management of the government and other public activities, citizens’ participation and also provides them with information and structure. It is the tool to govern society; public administration exists to realize the solutions to citizen’s needs (Ismaili & Latifi, 2012, p. 4732; Raadschelders, 1992, p. 288). To govern education is a societal need in a fully functioning 21st century country. Academic and vocational knowledge are only useful if, by possessing them, one gains access to a better position in society; thus, education brings about better labour market options, social mobility, and even social equity (Biesta, 2009, p. 37; Muller, 2015, p. 146; Lenkeit & Caro, 2014, p. 149-150; Wolf, 2004, p. 316).

In our globalized world it is no longer enough to cater to the local needs of the job market; the international labour market also proposes a challenge for citizens and it is the government’s job to solve this. The demands of the international labour market should be met for a small country to achieve prosperity. The openness of the educational system is a key element in attaining this goal (Toots, 2009, p. 59). De Meulemeester and Rochat also state that it is crucial that what students learn in schools is valuable in the job market, and in the social, political and economic structures of their country (De Meulemeester & Rochat, 1995, p. 385). Furthermore, the OECD even made a statement that skills are the global currency of twenty-first century economies (Sellar

8

and Lingard, 2014, p. 921-923). In addition to that, the level of technological development within a society should be high enough for students to make use of their knowledge (De Meulemeester & Rochat, 1995, p. 385; Toots, 2009. P. 59; Wolf, 2004, p. 316). Sellar and Lingard call this phenomenon the ‘economization of education’ which entails the growing significance of certain skills and human capital; therefore, ‘good’ education is crucial when it comes to development (Sellar and Lingard, 2014, p. 921-923).

Education also fosters integration into bigger international units; such as the European Union, because education directly affects the overall development of society and the direction of its development (Ismaili & Latifi, 2012, p. 4735). Essentially, it could bring about democracy, transparency and the elimination of social disadvantages (Witty, 2002, p. 12; Ismaili & Latifi, 2012, p. 4735).

Historically speaking, Estonia did not have the chance to work on independent education and define Estonian education until the 1980s (OECD, 2001, p. 53). This is due to the fact that this relatively small Baltic country has been occupied by various forces throughout history and only managed to gain its independence briefly during the 20th century. During Soviet times, Estonian education was fully under central control, meaning that the entire curriculum taught was supervised by the Soviet government along with strong Russification; beginning from kindergarten, all education was made bilingual (Russian - Estonian) (Garcia, 2011, p. 6; Krull & Trasberg, 2006, p. 3-4). However, the nation early on realized that in order to integrate into the European Union or the global market economy, post-soviet Estonia needed to catch up and improve. While still part of the Soviet Union, public and local politicians pushed for more independence from the Soviet educational institutions which entailed having their own curriculum, a larger voice in political decision making and participation in international forums on education, trying to transition from the strictly engrained communist ideologies and soviet principles (OECD, 2001, p. 53; Krull & Trasberg, 2006, p. 3-4). The years 1990-1992 were spent establishing a constitution and the legal framework for education; this is where public administration played a crucial role because the Estonians set up an entirely different system for education. The main focus of this was freedom from the ideological controls of the past, opportunities for private institutions and autonomy for educational units (Toots, 2009, p. 63; Krull & Trasberg, 2006, p. 6; OECD, 2001, p. 14). With the assistance of Finland and other western donors, Estonia sought to Europeanize and raise the level of education using policy transfer from their northern neighbour

9

(Toots, 2009, p. 63-64; OECD, 2001, p. 14-15). The period until EU consolidation is characterized by dynamic and profound educational innovations in Estonia. There was an excessive computerization of schools, integration of Russian schools and the reorganization of higher education (Krull & Trasberg, 2006, p. 9-10; OECD, 2001, p. 14-15, 55, Kankaanrinta & Marandi 2002). All of these changes, were welcomed by the public because they were made in prospects of joining NATO and the European Union which thus became a rapprochement towards the Western world (Krull & Trasberg, 2006, p. 9-10). Estonia then realized that computers were to be the future and in turn established the Tiger Leap program, Tiigrihüppe, which focused on innovation, computerization and the digitalization of schools. A visible indicator of this change in educational strategy could be seen as early as 2001, when most Estonian schools already had access to computers and internet connection within the school. In addition to that, thanks to this specific government measure, the faculty has received computers, and dial up internet was established for students in need throughout the country (Kankaanrinta & Marandi, 2002, web). The introduction of practical learning enhanced participation within the internal labour market. Later, with the EU succession it enhanced Estonian representation on the international labour market which influenced Estonian educational policies (Toots, 2009, p. 59; Biesta, 2009, p. 40; OECD, 2001, p. 55). After becoming a member state of the European Union, Estonia further developed its education following EU protocol (Toots, 2009, p. 61). The Estonian government reformed school childcare institutions as well; a key element here was the shift from a knowledge-based pre-school to a value-based one, which fosters the child’s creative and analytical skills better than a knowledge-based approach. Furthermore, children become ambitious and more resourceful citizens (Ülavere & Veisson, 2015, p. 109). As mentioned above, this is a key element to including people in public administration processes (Ismaili & Latifi, 2012, p. 4735) Since the year 2010 until the present day, the objective of educational progress was curriculum reform, revision of the evaluation regarding both students and schools and to provide equal access to education (Garcia, 2011, p. 9; Toots, 2009, p. 62). The most recent goal is to integrate more women into IT and to cater to the demands of the IT labour market (HITSA, web). All in all, there was a lot of help coming from the North, mostly from Finland, and from the West, Canada and the USA, thus, there was a clear intention to integrate Estonia (OECD, 2001, p. 14-5).

10

Theoretical Framework

Public administration is an interdisciplinary field; it merges knowledge together from economics, history, social sciences, political science and organizational theories (Van der Waldt, 2017, p. 1; Raadschelders, 1999, p. 295). Systems theory entails that society is a complex arrangement of elements - it is a system that relates to a whole, just as Education is part of Public Administration or the global market economy. The concept of Systems theory was developed by Herbert Spencer, an English Philosopher from the 19th century. His views were highly affected by the evolution theory. He suggests that we all thrive for perfection even if it is more complex than the previous system we established (Gibson, 2018).

Some other versions of Systems theory argue that society is not working to approach perfection but evolving into an increasingly complex entity. This sub-theory is called Structural Differentiation, which proposes that society is constantly adapting to its environment through changes in its internal complexity. (Gibson, 2018). Social Differentiation has two important aspects; firstly, the approach - the direction by which social change occurs, and secondly, how changes in the structure of the system relate to the processes of the system (Gibson, 2018). Both versions of Systems theory could be applied here because Estonian education is genuinely and evidently improving (not necessarily into perfection) very steadily. Furthermore, in recent years Estonia has seen many signs of Social Differentiation; there was indeed, a very strong social change; in fact, there were multiple changes; Estonia became a democracy, an independent state, and a significant Information Technology provider. There was EU, NATO and OECD succession Economic, public administrational and societal change all took place. From a Systems theorist point of view, Estonia has adapted to its environment through adjustments in its structure, processes and output (Miles & Huberman, 1994, p. 18; Van der Waldt, 2017, p. 6). The government of the small Baltic state realized early on that in order to stand out in a world full of larger economic powers they would need to specialize and excel in one field. However, in the case of Estonia that is not only Information and Communication Technologies but education as well using the resources from ICT (Kattel & Kalvet, 2006, p. 15-17).

Just like public administration, education and learning are also multidimensional processes. Learning is multidimensional because it starts with motivation, is followed by training and finishes with acquiring knowledge which could be cognitive, affective or skill capacities (Kraiger et al,

11

1993, p. 311). We have tests like the PISA, measuring systems, in order to examine if the learning process has been successful or not (Kraiger et al, 1993, p. 312). From a Systems theory point of view education evolves day by day in order to provide a better future for the next generation. There is a need for educational and cognitive functional theories for this change to optimize the learning process for the next era and minimize wasted resources. Failed approaches and will also help to improve already established ones (Gottlieb et al, 2017, p. 293).

Cognitive learning theory or the Constructivist learning theory entails that the learner is

principally a respondent to the events in their surrounding environment; learning is an internal action formed by the learner as a response to what they receive, recall and code (Gottlieb et al, 2017, p. 294). Learners seek to understand the way that knowledge is built up (Torre et al, 2006 p. 904). The teacher has tremendous responsibility in order to guarantee that students receive the information in a way that they can process and construct conceptual networks, categories and memories from it (Sullivan, 2009, p. 83; Gottlieb et al, 2017, p. 294). Torre et al states that the teacher’s role is to teach ‘how to learn,’ thus, to expand the students’ capacities for self-directed learning (Torre et al, 2006, p. 904). Scholars believe that the key to Cognitive learning theory is that knowledge is based on logic, memory and the structures we understand in order to link them together (Sullivan 209, p. 83; Torre et al, 2006, p. 904). Students also need to be active because they have to constantly recall this schema in order to retain knowledge (Gottlieb et al, 2017, p. 294). All learning theories emphasize that a student always need an expert to learn from, however this expert shall deliver the knowledge in a way that is understandable for the student. Albeit, it is crucial in Cognitive Learning Theory particularly because it states that knowledge is made up of structures, therefore the student might not be able to build upon that if the concepts that they are about to learn are too intricate and do not ‘fit’ their ‘underdeveloped’ structures (Gottlieb et al, 2017, p. 298; Torre et al, 2006 p. 905).

When reflecting Cognitive Learning Theory on the case of Estonia, one should observe that a student in this country is constantly accumulating knowledge from their surroundings, they have been able to appropriately process and practice it, and eventually have constructed structures from it which will later go on to serve as building blocks for future knowledge. Cognitive Learning Theory also adds that teachers are adequately able to pass on knowledge and create an environment suitable for learning. This could be tied in with the variable Student-Teacher relations, where the

12

teacher is not hostile, but helpful. In addition to this, the variable regarding how many extra hours are students spending at home studying should be connected too; if learning is done properly and to a satisfactory level at school, then the student may not need to spend time studying at home anymore, because they have understood the material sufficiently in class.

The IT facilities are of significant importance as well, because they offer a lot of easily digestible ways for students to learn, and for Cognitive Learning Theory it is often important to present knowledge in a comprehensible way. However filling a classroom with IT facilities is not enough for good education. The student might not necessarily be able to use them, or the tasks may possibly become difficult. The facilities schools are equipped with are not capable of assessing the students’ level of education as well as a faculty member. Hence, the student might not learn anything at all; the effective use of technology in the classroom is only possible if it is properly integrated (Plowman et al, 2013 p. 30; Dincer, 2018, p. 2702).

One of the most interesting questions raised by this research concerns integration; if non-Estonian born immigrants are able to participate properly in the educational system. The variables for immigration are seen to be connected to Cognitive Learning Theory as both Sullivan and Torre et al. state that language and learned symbols are an eminent part of one’s knowledge structure which they build on (Sullivan, 2008, p. 83; Torre et al, 2006, p. 298). Therefore, each individual is the architect of the information pile in their mind, and so this may well vary from person to person, however, if someone from an alternative cultural background arrives to Estonia, they will not necessarily have a similar construction in their head on which to build on.

Hypotheses

1.) At home resources - may that be digital or regular objects and a room - are crucial to a child’s education thus PISA score.

2.) Good student teacher relationship creates a favourable climate for education

3.) Foreign children are not integrated well into the Estonian school system, their scores will be affected by this.

Expectations from the hypotheses are the following: the Estonian government created a

13

classrooms, installing equipment and integrating children. In addition to that they have trained the staff to use the equipment. This will be observed by the values IT equipment in home and at class positively affecting PISA results.

The first hypothesis is supported by measurements from the OECD. The research states that Information and Communication Technology (IT or ICT) is crucial to the development of twenty-first century economies in order to stay efficient and modern (OECD, 2006, p. 8). The theory of Cognitive absorption states that learning is best with technologies that are visually appealing and have a user friendly interface, thus it is key to give students something that they know how to use, and moreover, can use independently in order to boost creativity and make learning interesting and fun (Agarwal & Karahanna, 2000, p. 688). A regular classroom setting did not change much in the past decade, there is a board with chalk or pens, tables all facing towards that, however technology and the amount of knowledge a high schooler is expected to possess have been exponentially growing (Roschelle et al, 2000, p. 76-77). Yet we do not really see how much more technology children study with in school. Finkelstein et al pose the question; Why should a student do something digitally, such as chemistry experiments or repetitive mathematical exercises if they can do it in real life with two beakers or on paper, and generations before have done so and understood it just fine (Finkelstein et al, 2005, p. 1). Scholars point out that it can be immensely useful for a child given that it enhances studying, creativity, communication skills and makes learning a ‘fun and interesting’ experience (Finkelstein et al, 2005 p. 1-2; OECD, 2006, p. 8;34; Roschelle et al, 2000, p. 76; Dincer, 2018, p. 2701; Olivares & Castillo, 2018, p. 2320). In addition to that basic ICT literacy is needed just to navigate in our complex world (Finkelstein et al, 2005 p. 1-2; OECD, 2006, p. 8;34; Roschelle et al, 2000, p. 76; Dincer, 2018, p. 2701; Olivares & Castillo, 2018, p. 2320). Certain studies also show that especially computers helped to open up and facilitated the process of learning for children who are perceived as less intelligent or slower and children who are less social (Roschelle et al, 2000, p. 81). Most studies concentrate on ICT being employed in science and math classes, such as in physics experiments, where the computer does not only give more detailed results and information on the measurements but students who worked on computerized experiments were more capable at reconstructing the knowledge they have gained and were also more precise and fast when asked to reproduce the analysis (Finkelstein et al, 2005, p. 6). Researchers also argue that testing on computer could be safer because there are no actual harmful substances or electricity being involved. However while it does not require

14

safety goggles and fire extinguishers in a lab, it is still resource intensive for the school when it comes to providing computers for all students (Roschelle et al, 2000, p. 77; Finkelstein et al, 2005, p. 6). In addition to this, many jobs are done via computers in the present day and therefore, it would be very useful for students to learn and socialize in a digital environment, perhaps even crucial to mastering the usage of such equipment as a preparation for the job market (OECD, 2006, p. 8). Researchers at the OECD state that IT could make significant positive effects in classroom teaching and other school environments such as libraries. It has the potential to create a more dynamic interaction within the classroom, increase collaboration and teamwork, stimulate learning and help both learners and the faculty control and oversee the learning process. Moreover the ‘joy’ of using technology could easily translate into productivity (OECD, 2006, p. 8; Agarwal & Karahanna, 2000, p. 688).

Student teacher relationship have been proven to be important in terms of student achievement in class. The linear regression conducted will demonstrate if this significance stands in terms of Estonian PISA scores as well. The Teacher plays an important role in the establishment of a favourable school climate, as their positive attitude is ‘catchy’ and captivates students better (Hagenauer et al, 2015, p. 398). Although, if the student perceives their environment and faculty as scary and hostile they will be stressed and respond with rebellion or conforming unproductively (Ruus et al, 2007, p. 921). In the long run a negative educational milieu could impede the child’s cognitive, social and personality growth (Ruus et al, 2007, p. 923). This ties in perfectly with Cognitive learning theory. If a student is stressed, and scared, they will not be able to build properly on the blocks of knowledge and connect various already learnt concepts as well as other students, who are happy and at ease.

While Estonia has various integration policies their biggest minority, the Russians, are still quite separated socially, culturally, economically and academically speaking from their local counterparts (Garcia, 2011, p. 3;18; Vetik & Helemäe, 2011, p. 15). Albeit local Russians have their political representation, separate cultural spheres and private universities, the integration of native Russians into Estonian culture up until nowadays is one of Estonia’s greatest internal issues because research has shown that both parties are scared, or at least unwilling, to interact with each other (Vetik & Helemäe, 2011, p. 14-17). This comes from the fact that Estonians only recently gained independence from the Soviet Union, which local Estonians perceive as Russian

15

suppression, and Russians have now became a minority or even immigrants in a place where they were formally a majority (Vetik & Helemäe, 2011, p. 14-17; Nimmerfeldt et al, 2011, p. 86-88; Garcia, 2011, p. 3). Furthermore, research has shown that minorities and migrants tend to underperform on cognitive tests (Cobb-Clark et al, 2012, p. 19; Mikk, 2015, p. 326). If Estonians perceive Russians as a threat, Russian children will not receive the same treatment as Estonian ones. Innove, an Estonian independent research center, conducted an assessment of Russian schools in Estonia and did indeed conclude that the standard of education in such schools was worse than that of the average school within the country (Innove, 2009). Garcia adds that educational reform lags behind when it comes to improving the institutions in the Russian populated regions of the small Baltic state (Garcia, 2011, p. 9).

Research Design

Operationalization of Variables

For this research the dependent variable is the PISA results in 2015 and from that the independent variables which will influence the score shall be determined. I have chosen the PISA test because as mentioned above, it is internationally acknowledged as an indicator of the quality of education in a given country and of the level of education of an individual; how well-taught students are (Mikk, 2015. P 326., Lenkeit & Caro, 2014, p. 146-147.). The test measures students’ performance in math, science and reading, thus the focal points of education, while also recording data on students’ socioeconomic states (OECD). Estonia has performed dynamically well throughout the recent years, showing an improvement of accomplishment in all fields (Estonian Ministry of Education - PISA). That is why it is important to look into the changes of individual students’ scores in Estonia. What are the factors that make a student stand out in Estonia and contrarily, what can possibly make them underachieve?

There are numerous factors that could be considered as independent variables which have potential to influence the dependent variable, which in this case is the results of the international standardized test, the PISA. The OECD provides researchers with vast amount of data available in

16

the various sections of the PISA testing but the scope of this research is limited thus only some of these will be taken as variables.

The first category, which has 3 variables, will be called Student teacher relationship. It builds on classroom discipline and students’ perception of how adequate their teacher is, which have all been proven to influence the level of a student positively (Mikk, 2015, p.325-326; Lenkeit & Caro, 2014, p. 151). This variable measures how the student perceives their teacher’s attitude towards them. Although the student's experiences with and opinions on their teachers are certainly valid and important, we must not forget that in the end they are perceptions that are subjective thus do not necessarily reflect reality. The questions which will make up my variable are respectively:

ST039Q03NA Teachers gave me the impression that they think I am less smart than I really am. ST039Q04NA Teachers disciplined me more harshly than other students.

ST039Q05NA Teachers ridiculed me in front of others.

These three variables are coded into one in order to examine the significance of the Student-Teacher relationship. The final variable that measures frequency in the SPSS is named as Student Teacher relationship and has the following increasing values: 1 (never), 2, 3, 4 (once a week). It is also found that students who receive a longer education (in terms of time) perform better, even though Banach questions this, arguing that length does not necessarily equal better quality, moreover, the curriculum can be packed or very spread out (Banach, W. J. (1995). Furthermore, Smith asserts that quantity and quality of education should be separate characteristics as one does not mean the other, moreover quality depends on more than just the length (Smith, 1995. p. 215). Therefore, I will select the variables that show how much education they receive in terms of time. These will be:

ST071Q01NA This school year, approximately how many hours per week do you spend learning in addition? <School Science>

ST071Q02NA This school year, approximately how many hours per week do you spend learning in addition? <Mathematics>

ST071Q03NA This school year, approximately how many hours per week do you spend learning in addition? <Test language>

ST071Q04NA This school year, approximately how many hours per week do you spend learning in addition? <Foreign language>

17

ST071Q05NA This school year, approximately how many hours per week do you spend learning in addition?

These variables are coded into one category; Avg learning add in SPSS in order to measure how significant additional - outside of school or extracurricular - learning is in order to obtain a good PISA result in Estonia in 2015.

According to Mikk and Muller, available equipment does have significance when it comes to countries with a low income but is not significant when it comes to countries with a high GDP/capita (Mikk, 2015, p. 333; Muller, 2015 p. 140). Estonia is in the top richest quarter in the world according to the CIA database (Central Intelligence Agency - The World Factbook 2017), however, it might matter on the individual level. Therefore, I will test this by filtering the possessions of the students and the equipment at school. Book ownership and reading are also important because a recent study shows how reading and ownership of books actually correlates with how smart an individual is and how easily they can process the material (Mikk, 2015, p. 333; Clark & Poulton, 2011, p. 5-6; Evans et al, 2010, p.189; Muller, 2015, p. 140).

These will be categorical variables, taking up either 1 or 2 as a value, which will mainly consist of equipment at school and at home.

These variables are the following: Do you possess: ST011Q01TA In your home: A desk to study at ST011Q02TA In your home: A room of your own ST011Q03TA In your home: A quiet place to study

ST011Q04TAIn your home: A computer you can use for school work ST011Q05TA In your home: Educational software

ST011Q06TA In your home: A link to the Internet

ST011Q07TA In your home: Classic literature (e.g. <Shakespeare>) ST011Q08TA In your home: Books of poetry

ST011Q09TA In your home: Works of art (e.g. paintings)

ST011Q10TA In your home: Books to help with your school work

All these variables mentioned above are grouped into a variable called Home Equipment in SPSS. In addition to that, are coded to be more fluently translated into significant or not significant. The original database gave students the option to choose between 1- ‘yes, I own it and use it’, 2- ‘yes,

18

I own it but don’t use it’ and 3 - ‘I do not possess one’. I coded these into 1- ‘yes, I use it’ (first option) and 2-3 into 2 - ‘I do not use one’.

IC001Q01TA Available for you to use at home: Desktop computer

IC001Q02TA Available for you to use at home: Portable laptop, or notebook IC001Q03TA Available for you to use at home: <Tablet computer>

IC001Q04TA Available for you to use at home: Internet connection IC001Q05TA Available for you to use at home: <Video games console> IC001Q06TA Available for you to use at home: <Cell phone>

IC001Q07TA Available for you to use at home: <Cell phone> (with Internet access) IC001Q08TA Available for you to use at home: Portable music player

IC001Q09TA Available for you to use at home: Printer

IC001Q10TA Available for you to use at home: USB (memory) stick IC001Q11TA Available for you to use at home: <ebook reader>

Available at school:

IC009Q01TA Digital devices available at school: Desktop computer

IC009Q02TA Digital devices available at school: Portable laptop or notebook IC009Q03TA Digital devices available at school: <Tablet computer>

IC009Q05NA Digital devices available at school: Internet connected school computers

IC009Q06NA Digital devices available at school: Internet connection via wireless network IC009Q07NA Digital devices available at school: Storage space for school-related data, e.g. a folder for own documents

IC009Q08TA Digital devices available at school: USB (memory) stick IC009Q09TA Digital devices available at school: <ebook reader>

IC009Q10NA Digital devices available at school: Data projector, e.g. for slide presentations IC009Q11NA Digital devices available at school: Interactive Whiteboard,

These will be grouped and called Avg IT equipment, coded similarly as the previous one, 1 - ‘yes, I own one and I use it’, 2 - ‘I do not use one’.

Students in Estonia had the option to fill out the PISA test in their native language, which meant that in 2009, 78% filled it out in Estonian and 22% filled it out in Russian. According to a recent study, only around 40% of Russian-Estonians are proficient in the Estonian language (Nimmerfeldt et al, 2011, p. 86). This is important because ‘immigrant’ children tend to perform worse in school than natives to the country according to certain studies (Mikk, 2015, p. 326), while other papers claim that it actually does not affect their performance (Lenkeit & Caro, 2014, p. 154-155). Estonia

19

offers two distinct type of high school in terms of language. Estonian ones operate entirely in the local language and Russian high schools. At high school the Russian school transitions to a 60/40 system, 60% of subjects taught in Russian and 40% in Estonian; to begin with, children get rigorous Estonian language education so that later they can choose subjects to take in Estonian. (Estonian Ministry of Education & Research, 2015; Innove, 2009). While Russian people are not entirely willing to integrate they often chose Estonian or 60/40 primary and middle schools instead of Russian ones because they are aware of how Estonian language is the key to a better life, so their child can start learning in the local language earlier (Innove, 2009). The PISA results of foreign children could be a bit distorted; certain 15 years olds study some subjects in Estonian and other subjects in Russian, however, can only fill out one version of the PISA test, the Russian or the Estonian language one.

Variables to test this - immigrant students perform worse - are the following: ST019AQ01T Country of Birth International - Self

ST019BQ01T Country of Birth International - Mother ST019CQ01T Country of Birth International - Father ST022Q01TA International Language at Home

For these variables PISA used the following 1 - local 2 - else, thus 1 for all answers meaning a student and his family being born in Estonia, speaking Estonian at home. After being grouped into one variable - Avg Foreign I will examine how significant it is to be a ‘foreigner’ in Estonia, in a local school regarding PISA results.

Case Selection & Method of Data Collection

I will examine the results and answers of the Estonian students who filled out the PISA 2015 questionnaire from the 205/206 high-schools which participated. I will acquire the data from the national database and I have gotten the PISA results from the OECD website where every student has an individual score moreover one can filter for Highschool by looking at the School ID. Given that data is confidential one can not figure which highschool is which or who exactly is the student who filled the test out.

20

In this cross-sectional analysis everything is already quantified in order to create a multi regression - logistic regression model. I will put the numbers in SPSS and test for their significance. Dummy variables are either 0 or 1, student teacher ratio is going to be x/y, number of classes a week will be a single number for math, science and language + grammar, etc. according to the given value in the database.

Once finding the least significant (independent) variable I will leave it out and test again, continuing until I find what is/are the truly significant variable(s), taking one insignificant variable out from the regression at a time. In addition to this, I will calculate the mean values for the answers in order to find out if the variable is relevant. There is a possibility that no estonian child does any additional learning, or that only a small section of them have ever been disciplined harshly by their teachers. These tables are found in the Analyis section. In the next chapter, The Empirical Findings section, the exact PISA calculations will be elaborated on.

Reflection on Validity, Reliability

It might happen that there is a very important aspect of Estonian education that I simply miss and do not include in my research yet it does contribute to the great PISA results.

Additionally, validity could fail if I quantify my variables inappropriately.

I trust in the reliability of my test given that I examine all students who filled out the questionnaire, thus I check all individuals concerned. However, I have to state that I am only looking at the students who participated in 2015, thus with other years one might get different results about why the good performing students are performing well, and why the poor ones are performing poorly.

Consent:

When using public data from the internet consent is not needed. I will ask consent from all other parties who may contribute to my research and all who is concerned.

21

Empirical Findings

Calculations

The data the OECD gathers via the PISA test is both assessed and calculated in a rather complicated manner. In order for the reader to understand why the scores of each student are not simply put in an excel sheet and then averaged to count country or individual scores, the following section will elaborate on the calculation of results and how to use them for further analyses.

There are a number of difficulties that arise from the nature of the PISA information, the first one being fatigue. Students tend to tire out after hours of testing and consequently will not perform to their normal standard at the end of a long test as they would at the beginning. Principals would not allow them to have the tests on multiple days because that would decrease school participation. In order to avoid overwhelming a student, each of them is assigned into a pool, thus only filling out part of the PISA test. The information gathered is anticipated by a large group of researchers, policy makers and even the public, therefore, there is a need to summarise detailed item-level information for communicating the outcomes (OECD, 2009, p. 78-79).

Additionally, the OECD PISA data manual points out that the results are highly dependent on the difficulty level of the test. Therefore, one could misinterpret the size of the difference in results between two countries (OECD, 2009, p. 78). Furthermore, counting the mean will not allow researchers to access the statistical dispersion of results; some countries might have regions/schools or individual students who may, for a certain reason, score very low while others may score very high (OECD, 2009, p. 78). This dispersion could become apparent, for example, in the case of a country where schools around the capital city usually perform better and schools in poorer regions perform worse, which in turn could help their government allocate educational funds more appropriately. To make sure students get the same difficulty level of tests, these tests are trialed and the raw score is checked for mean and variance. The obtained raw score is later standardized; however, standard deviation and average can still show a slight difference (OECD, 2009, p. 79). In order to avoid the above illustrated mistakes and biases, PISA data is scaled with the Rasch model. This will help researchers interpret the data and the performance of the students adequately.

The Rasch model is a complex way to model data and so further information on this and the calculations can be found in the PISA Data Analysis Manual, pages 79-90 published in 2012.

22

To put it concisely, the Rasch model can show a concession between two variables “in order to assess the extent to which a set of persons has a certain level of an attribute of interest (e.g., mathematical proficiency or level of anxiety) and the extent to which a positive answer to the questions or statements demands a certain level of that attribute” (Frey, 2018, p. 1370-1371). This attribute regarding the PISA test is difficulty. The model constructs a scale where both student performance and item difficulty are located and a probabilistic function links these components together. Low ability students and easy items are on the left of the continuum while high ability students and difficult items are on the right. (OECD, 2009, p. 79-83). Taking this into account, the model calculates the probability of a student being able to solve the exercise or not by analysing the distance between difficulty and student capability (OECD, 2009, p. 83). All of these computations will demonstrate to the investigators how difficult one question was and how reasonable it was that the student obtained a good score. Subsequently, the performance of the student is named as Plausible Values, which are the ones analysts could work with when evaluating anything about the PISA scores (Monseur, 2009, p. 117)

While in the SPSS file, the Plausible Values seem somewhat easy and manageable to work with; however, there is still a sufficient amount of alterations that need to be applied before the model produces a wholly satisfactory result. In the manual, the variable which aims to compute the average of the ten plausible values is labelled alarmingly as ‘Fatal Error.’ Because, in this case, one gets a score very similar to how the student was expected to do, and not how they actually performed. This value could “underestimate the standard deviation, overestimate the correlation between the student performance and some background variables, and underestimates the within-school variance” and would therefore only give us biased results (Monseur, 2009, p. 127). Running an analysis with only one plausible value at the time gives us unbiased scores (Monseur, 2009, p. 129). In the data analysis manual it is listed that one should do the exact following steps when computing regression;

1. computing, using one of the five plausible values, the statistical estimate and its sampling variance by using the final student weight as well as the 80 replicate weights;

2. computing the statistical estimate by using the final student weight on the four other plausible values;

3. computing the final statistical estimate by averaging the plausible value statistical estimates;

23

5. combining the imputation variance and the sampling variance, as previously described.

As this analysis worked with a regression model, I looked at certain aspects of a student’s life and how the PISA score was determined. When all of the calculations are completed, the function Regression will show the significance of the results. If the p-value is less than or equal to 0.05, the independent variable is significant in the determination of the dependent variable. After following the steps above, a coefficient out of the 30 plausible values (10 for reading, 10 for science and 10 for the mathematics test) was calculated. This coefficient determines that for every 1 point of increase in the value of the independent variable, the dependent variable, PISA score, increases by the value of the aforementioned coefficient. All of the values are an increase or decrease on a linear function if all other independent variables are held constant.

24 MATH B Average avg_leaning_add -4.319 -4.342 -4.399 -4.614 -4.633 -4.211 -4.347 -4.394 -4.345 -4.268 -4.387 avg_studteacher_rela -14.813 -18.059 -16.577 -15.825 -16.542 -16.065 -18.237 -16.525 -17.057 -15.096 -16.480 avg_home_equip -105.258 -101.001 -88.354 -99.670 -104.164 -96.204 -106.873 -103.809 -102.048 -97.564 -100.495 avg_it_equip 39.043 28.271 25.890 41.005 34.803 24.483 30.137 23.235 24.531 20.612 29.201 avg_foreign -35.581 -38.290 -37.167 -38.021 -43.264 -44.081 -38.317 -33.443 -41.441 -42.318 -39.192 avg_school_equip 17.366 13.722 19.699 12.088 13.056 18.507 17.090 21.657 20.685 19.164 17.303 sig avg_leaning_add 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 avg_studteacher_rela 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 avg_home_equip 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 avg_it_equip 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.001 0.001 0.003 0.001 avg_foreign 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 avg_school_equip 0.000 0.005 0.000 0.014 0.008 0.000 0.000 0.000 0.000 0.000 0.003 R^2 0.134 0.136 0.128 0.138 0.143 0.132 0.145 0.136 0.137 0.128 0.136 sig 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 Constant 650.998 673.928 648.759 653.966 675.659 667.330 674.432 665.209 671.644 671.501 665.343

25

Mathematics

The first independent variable is Additional Learning - labelled as Avg_learning_add in the table. One can observe that this independent variable is statistically highly significant, given that the value of p takes up 0.000 for all ten Plausible Values. The figure for the coefficient is -4.387 which means that for each additional unit of learning (in this particular case, each added study hour after school) the student’s PISA score is expected to decrease by 4.387 points if all other variables are held constant. If the student studies for 2 extra hours, the PISA score will expect to alter by -8.774 points (2x -4.387), and if the student studies 3 extra hours it will be -13.161 (3x -4.387) points off the test score and so it continues given that this function is linear. Therefore, it can be concluded that additional study hours did not help Estonian students achieve a higher score on the PISA test in 2015.

Average Student Teacher relationship - labelled as Avg_studteacher_rela - has also been found to be of statistical significance. The p-value here also takes the value of 0.000. The coefficient equals -16.480, meaning that if the student answered all three questions with a yes, their score on the PISA decreased by 16.48 points. This means 3x = 16.48, therefore for each question answered yes the PISA score of the pupil decreased by 5.493. This value is obtained by dividing 16.48/3. Conclusively, if the student perceived that they are mistreated by their teacher they were expected to perform worse on the PISA math test in 2015 in Estonia than the average person.

The third independent variable is Home Equipment ownership, Avg_home_equip in the table. This variable reports whether ot not the student has their own quiet place to study and a home that is intellectually inspiring for them. This was found to be significant as well, p equals 0.000 throughout and the coefficient is evidently very high. Given that if the student answered all no then their score would decrease by 100.49 points, which is an abnormally large decrease. No to all questions signifies that the student is from a very underprivileged household, not having any books, nor even a desk or a place to study, therefore, they have a very underwhelming environment for learning, not having much privacy or being able to do homework comfortably at home with helping material such as the internet or books. For each item not possessed, out of the 10, on the list the student lost 10.0495 points from their PISA test. As a result, for a child to achieve good points on the PISA 2015 test it can be concluded that it is of very high importance to have certain

26

qualities of living such as a private room, a desk, internet and books. However, there is a very high possibility that there are indeed no students in Estonia who are deprived of every single one of these things ;thus, there is a high possibility that no one has -100.49 points on their test compared to the average because of this.

Informational Technology devices also play an important role in the PISA score in Estonia. This fourth variable showed that if a student possesses and uses IT equipment or does not use it or does not have one at home. Access to equipment at school was analysed in a separate category. As previously mentioned, this variable could take up 3 values, however, for the analysis it has been coded to 2 only. The p-value equals 0.001, even though there are some slight differences at the last three plausible values, they are all below 0.05 therefore are statistically significant. If the student said no to each equipment, they scored 29.20 points better if all other independent variables are constant.

The fifth variable was concerned with a pupil being somewhat foreign in order to see if immigrant children, or Russian children, receive the same quality of education as their Estonian counterparts. The value was coded as either 1 - Estonian, 2 - foreign. In terms of the PISA math result, this independent variable is significant, the p-value = 0.000 all over. If the student answered with 2 to all four of the questions; neither of their parents nor themselves were born in the Baltic state, neither did they speak Estonian at home, then they scored 39.192 points less on the PISA test in 2015. If they only answered 2 to one then it is only a 9.798 decrease, (39.192/4), because this function is linear.

To measure school IT resources, the final variable, Avg_school_equip was used. Similar to all of the other independent variables, this was also found to be significant; p=0.003 throughout, with little variation in the second, fourth and fifth Plausible Value. If a school did not have any of the devices listed at the variable section, a student performed 17.303 points better if all else is held constant.

27 READ B Average avg_leaning_add -4.578 -4.594 -4.692 -4.555 -4.605 -4.611 -4.717 -4.693 -4.840 -4.668 -4.655 avg_studteacher_rela -20.803 -21.218 -21.562 -20.253 -21.519 -20.866 -21.265 -19.345 -21.124 -21.213 -20.917 avg_home_equip -96.330 -91.528 -90.491 -105.296 -102.194 -105.035 -100.315 -96.010 -98.701 -94.430 -98.033 avg_it_equip 64.346 54.177 56.457 60.996 54.544 50.459 46.671 54.635 53.123 56.408 55.182 avg_foreign -54.061 -51.973 -54.200 -48.160 -49.871 -50.454 -48.828 -56.135 -55.201 -49.203 -51.809 avg_school_equip 25.797 26.170 24.584 28.692 29.320 26.415 25.745 26.573 23.884 23.937 26.112 sig avg_leaning_add 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 avg_studteacher_rela 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 avg_home_equip 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 avg_it_equip 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 avg_foreign 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 avg_school_equip 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 R^2 0.155 0.147 0.150 0.151 0.156 0.149 0.149 0.146 0.156 0.150 0.151 sig 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 Constant 628.500 633.258 635.622 633.099 640.139 651.184 650.015 638.011 652.357 634.804 639.699

28

Reading

The reading scores were found to be very similarly affected by the independent variables as the math scores were. However, by looking more closely, it can be seen that the same difference in the independent variables tend to create a larger difference to the final PISA score than they did with their math counterpart, except the variable of avg_home_equipment. All independent variables were found to be of statistical significance here too.

The coefficient for average additional learning - avg_learning_add - is -4.655. This signifies that a student who studies one extra hour will perform -4.655 points below their fellow counterparts if all else - independent variable - is constant. This variable is significant as p = 0.000 with all 10 Plausible Values.

Student-Teacher relationship is significant in determining the PISA scores in the field of reading for Estonian students as well given that p = 0.000. If the student perceived that the teacher misjudged them, or made fun of them often, then this student scored -20.917 worse than others in general.

Home equipment is crucial for a child’s high score in the PISA test as illustrated above. However, it relates to a slightly smaller change in PISA score in terms of reading than in mathematics. Significance is evident because p takes the value of 0.000 for all PVs. If a child does not possess any of the 10 listed things in the group ‘home equipment’, then their score will be expected to be 98.033 points worse if all other independent variables are held constant.

IT equipment possession again proved to be of significance to the model p equals 0.000 for all PVs . If one possessed all of the equipment listed, they would be expected to score 55.182 points less than the average student. However, it must be noted that this is shown as a gain in the table, because, as a two categorical variable ‘yes’ was 1, and ‘no’ was 2, thus, the more ‘no’s the student gaveas answers, the more their score increased. Therefore, owning a lot of ICT devices will affect a student’s score on the PISA test in Estonia negatively.

29

Being non-Estonian was also found to decrease the PISA results. The p-value was also found to be 0.000, thus it proves to be of statistical significance. If a student answers yes to the four questions relating to nationality (parents’ nationality, born inside/outside Estonia and whether or not they spoke Estonian in their household) then they will be expected to lose 51.809 points on their test score if the other independent variables are held constant. For example, if only one of their parents is foreign, the situation is already better, they only lose 12.952 points, this is calculated by dividing the coefficient by four given that there are four subvariables to this category and a linear regression model was conducted.

School equipment was found to cause the student’s PISA score to decrease; it is statistically significant because p = 0.000. If the school has none of the 10 items listed above the student’s PISA score is expected to increase by 26.112 points. For each item available at school this increment will decrease by 2.611 points, therefore if all digital devices are available the student will score 26.112 points less than the mean Estonian value and if 6 of the 10 items were available to the student at school then their score would be expected to decrease by 15.666 points (2.611 x 6) in case all the other independent variables are held constant.

30 SCIENCE B Average avg_leaning_add -5.172 -5.085 -5.191 -5.262 -5.235 -5.215 -5.209 -5.377 -5.272 -5.228 -5.225 avg_studteacher_rela -19.797 -17.678 -19.437 -19.631 -19.398 -19.652 -19.275 -16.810 -20.006 -18.705 -19.039 avg_home_equip -95.459 -98.212 -95.475 -100.061 -97.961 -95.480 -97.732 -99.543 -101.166 -99.861 -98.095 avg_it_equip 40.941 42.231 41.271 44.921 49.823 43.784 35.709 45.348 40.078 40.428 42.453 avg_foreign -61.267 -56.287 -50.315 -57.737 -55.646 -60.348 -58.865 -54.396 -57.681 -58.486 -57.103 avg_school_equip 25.717 25.474 21.568 27.003 20.105 27.539 28.969 25.368 26.489 25.049 25.328 sig avg_leaning_add 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 avg_studteacher_rela 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 avg_home_equip 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 avg_it_equip 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 avg_foreign 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 avg_school_equip 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 R^2 0.156 0.143 0.146 0.159 0.152 0.156 0.150 0.147 0.158 0.154 0.152 sig 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 Constant 681.792 674.726 672.985 675.647 673.948 673.679 682.184 669.629 684.059 682.871 677.152

31

Science

The coefficients were also found to be altogether quite similar in the students’ Science results as they were in Mathematics and Reading.

Once again, it was found that Additional learning - avg_learning_add - does indeed not increase a student’s PISA score, while it is significant, as p takes up the value of 0.000, it does in fact, decrease a person’s score by 5.225 for every extra hour spent studying outside of school.

Relationship with the teacher, thus the independent variable labelled as avg_studteacher_rela, is of statistical significance p equals 0.000 throughout. If the student perceived that the teacher is fully discriminative towards them or the student feels like the teacher considers them less smart than they actually are, then the student’s score was expected to decrease by 19.093 points 2015.

The variable for home equipment is labelled as avg_home_equip in the table. This entails rooms, books, places to study et cetera. p was 0.000 for all ten Plausible Values; therefore, a statistical significance was observed. If a student lacks all of the 10 items listed as home equipment then they performed 98.095 points less if the other variables are held constant. If they only own half of the objects listed they will have a decrease of 49.047 given that we can divide the coefficient because it is a linear function.

The fourth valuable, avg_it_equipment, measures how the possession of IT equipment at home affects a student’s score. It was proven to be significant, p =0.000, and for every item possessed by the children their score would drop by 4.245 points compared to the average Estonian student’s score. Therefore, if they own all 10 devices their score is expected to be 42.453 points worse. This number is calculated by a dummy variable, students could answer with either 1, yes I own this device and I use it or 2, yes I own it but don’t use it or I do not possess it. The more often they gave the answer no, the higher the coefficient is, therefore if they said no to all questions, they are expected to obtain a PISA score which will be 42.453 higher.

The analysis showed that, when it came to scientific proficiency, the most valuable trait that a student could have was in fact, to not be foreign. Significance can be observed by the fact that p=

32

0.000 for the Plausible Values. If a person is as foreign as they can be, meaning them and both of the parents are born outside of Estonia and the language spoken at home is not Estonian then they are expected to obtain 57.103 points less.

Finally, the sixth independent value, school IT equipment - avg_school_equip in the table - also was found to be significant, p = 0.000 for all ten PVs. This was again a dummy variable, students could answer with 1: yes, our school has this particular item or 2: no, we do not have access to this at school. The more that they answered no, the higher the coefficient is. Accordingly, if there is not a single device available, then they were expected to achieve 25.328 points higher in 2015 in the PISA test if the other independent variables are held constant. If all of the 10 devices listed above are available, the score of the pupil was expected to decrease by 25.328 points.

The R^2 values - r squared - are the coefficients of determination. Notice that there is only one for each section; thus three in total. They explain how close the values are to the linear function line, hence demonstrating if a model is capable of explaining the data (Enders, 2018, web). An R^2 value will show how much percentage of the variation in the outcome could be explained just by using the covariates included in the model (Enders, 2018, web). The R^2 values for the models are the following: 13% for mathematics, 15% for science and 15% again for reading. This means that the model explains the results to an extent of 13%, 15% and 15% respectively

In conclusion, the analysis of the data found that all independent variables were of statistical significance However, some of the previously set up hypotheses are now refuted. In all three cases the highest predictive power of a student’s PISA result was the presence or absence of home equipment; without exception the coefficient was near 100 points. This signifies that the most important basis for a good PISA score is an intellectually conducive home or a calm place to study. It is confirmed that being foreign does negatively affect a child’s performance on the PISA test. In both Math and Science, those who are entirely foreign, they and both of their parents are born outside Estonia, they would be expected to score particularly lower than their ‘local’ Estonian counterparts.

33

Analysis

Explanation for the values obtained

Additional learning

Possibly students who lack the necessary skills or knowledge to attain the top grades are understandably the same students who spend more time on additional learning at home, and conversely, those who are more advanced and able to comprehend the material taught at school with more ease do not feel the need to conduct more (or any at all) extracurricular learning within their own household. Smith asserts that quantity and quality does not equal thus studying longer does not mean studying more (Smith, 1995, p. 215). Estonian children spend around 3-4 hours extra per week additionally to study at home. They spend the most time on math and the least on the language of the test that would be Estonian or Russian including grammar and literature.

Additional learning was a variable selected to check if Estonian students perform well because of extracurricular classes, tutoring or having supplementary practice at home. These activities are mostly not school controlled thus are not related to the actions of the municipality or the government. Usually it is a decision in the family to hire a tutor for their child, accordingly government policies and regulations hardly reach this segment of education. Therefore it is of little importance to Public administration if it does not affect a child’s PISA score significantly. While this variable has proven to be significant, it negatively affects the PISA results thus the more the student studies at home the worse their score is if all other variables are held constant. We can conclude that Estonian teenagers do not perform better if they take a lot of time at home to recite school material. This probably stems from the fact that those learners who spend extra time at home being busy with school material are below average in their cognitive abilities. Therefore they need help at home to keep up with the school.