DEVELOPMENT OF MATHEMATICAL AND LANGUAGE LITERACY AMONG ROMA STUDENTS

The main goal of the paper is to analyze educational achievements of Roma students in the lower grades of compulsory education in Serbia, as well as to find out what is the influence of the socio-economic status and the quality of education on the educational achievements. The main source of data is the Grade 3 National Assessment Study which is implemented in Serbia during year 2004. The results show that there is big gap regarding to the educational achievements between Roma and non-Roma students (average score of Roma is more than 130 points behind of national average; about 50% of Roma students do not develop even the basic mathematical and language literacy; and the repeating rate is about 11%). Secondary analysis of the results shows that about 40% of this gap can be explained by the lower socio-economical status of Roma, and about 60% is determined by the lower level of education quality delivered to Roma students in Serbian schools. Moreover, based on the analysis, some aspects of the lower education quality are determined: Roma students are in higher extent than non-Roma students into the classes with the lowest education quality, teachers have lower expectations from Roma students and support them in less extent, curriculum is “shortened” and “simplified” for Roma, etc. The conclusion of the study is that without waiting for general improvement of living and social conditions in which Roma population live in Serbia, there is big space for improvement of education quality delivered to Roma students in Serbian schools.

The paper relies on three new important ideas that significantly changed the education in many countries at the beginning of the nineties of the last century, along with some other ideas.The first idea refers to the fact that the right to education is a universal right.At the Conference in Yamatane (Thailand), which took place in 1990, the representatives of 155 countries and 150 organizations set the objective to radically decrease the illiteracy by the end of the decade.Education has been marked as a key factor that may lead to a break of the "vicious circle" and a significant improvement of the economic and social development of the poorest ("Education for all", UNESCO).The second big change is related to the education process itself.It is based on some of the most influential theories of the children's development (Piaget, 1969;Gruber & Vonèche, 1995;Inhelder, Sinclair & Bovet, 1974;Vigotski, 1974) where the student's learning process is placed in the central position, and instruction and everything done by a teacher has been defined as creation of learning opportunities which support and uphold the learning process.Thus, the instruction serves learning and a teacher shapes his/her activities depending on the educational and developmental needs of students.The third change concerns the educational politics that was before focused on the input (curriculum, teachers' salaries, textbooks, etc.), where the supposition was that the outcomes are necessary consequence of the adequate input.However, the new educational policy focuses on the educational achievements of a student, and leaves more space for the autonomy to schools and teachers in relation to the school curriculum, textbooks, instruction and learning organization, etc. (World Bank, 1995;Stewart, 1996;Bakker, 1999;Kellaghan & Greaney, 2001).
In Serbia, like in many other countries (Liegeois & Gheorge, 1995;Pinnock, 2001), Roma population is the most imperiled category of population.All demographic data (length of life, health condition, education, employment, etc.) point out the disastrous status of Roma population in Serbia (Mihajlović, 2004;Burnett et al, 2005).In the "Overall analyses of elementary education in Serbia and Montenegro" (UNICEF, 2001), it is estimated that around 80% of Roma population is functionally illiterate.In the Needs Assessment Study prepared for the Roma Education Fund (REF), which was implemented as a part of preparations for the Roma Decade3 , the following data on Roma population in Serbia have been stated: about 62% of Roma population have not completed elementary education, while about 27% have completed only elementary school.About 60% of Roma population are supported persons, Roma are settled in the poorest settlements, about 50% of Roma do not continue education after the fourth grade of elementary school, etc. (Mihajlović, 2004).Those findings have been confirmed also by other researches dealing with the education of Roma population (Rakočević & Miljević, 2003).
However, these figures express the education status of Roma only on the basis of completion of certain level of education.It is certainly important information, but on its basis it cannot be concluded which knowledge and skills a particular person has.In other words, at this moment, there are no reliable and valid data on the educational achievements of Roma students in respect of the development of the basic mathematical and language literacy.Without those data, as we have seen, it will be very difficult to carry out a modern educational policy directed to an improvement of educational achievements of all children, especially the vulnerable groups like Roma.
Still, in the course of 2003 and 2004, the Institute for Evaluation of the Education Quality carried out the National Assessment Study among the third-grade elementary school students.Having in mind significance of the data on the educational achievements of vulnerable groups of children, an objective of the study was to provide information about achievements of Roma students, as well as the students from refugee and IDP families, regarding to development of the basic mathematical and language literacy.The paper will focus only on the educational achievements of Roma students since it turned out that they are the most imperiled group of children (Baucal et al., 2004).The main objective of the paper is to get the answers to the following questions: To what extent do lower educational achievements of Roma students appear as a result of their lower socioeconomic status (SES), and to what extent are they generated throughout lower quality of education to which the Roma students have been exposed?5 Which aspects of instruction contribute to the lower quality of education to which the Roma students have been exposed?Answers to these questions should be used as a base for the development of measures and activities, which should improve the quality of Roma education.

METHOD Sample
The main sample for the National Assessment includes 4,519 Grade 3 students (212 classes, 113 primary schools) from all 25 districts of the Republic of Serbia (excluding Kosovo and Metohija).From each district (from the list of all schools) two schools were randomly selected from urban areas and two schools from rural areas, except in the case of Belgrade.In the case of Belgrade, 10 schools were randomly selected.From each school, two randomly selected classes were tested.Selected schools are located in 100 places included in 77 different municipalities.
Out of this number, 4178 (92.5%) of students were present in the school on the testing day.
The data obtained in this way were weighted in order to get representative sample of Grade 3 students in Serbia.

Instruments
In the main study, two tests (Serbian language, and mathematics) were used, designed and standardized on the basis of the pilot study realized during 2003.As usual in the Assessment studies, incomplete design was used (Johnson, 1992).The tests included total of 157 items in the Serbian language and 162 items in mathematics organized into 12 booklets per subject.These items covered 5 domains from Serbian language subject (reading, writing, grammar, semantics, literature), and 6 domains from Mathematics (the understanding of numbers up to 1000, calculation operations, geometry, measuring and measures, fractions, the application of mathematics -rules and graphs, and money operations).
Based on results in the tests of the Serbian language and Mathematics, 5 hierarchically ordered levels of achievements were defined (from the lowest E, through D, C, B up to the highest, A).These levels were defined qualitatively based on the knowledge and skills the students from the level developed (Table 1 and 2).The fact that these levels are hierarchically ordered means that students from certain level can demonstrate also the knowledge and skills from all lower levels.
Besides the tests in Serbian language and Mathematics, questionnaire for students, questionnaire for teachers, and questionnaire for school principal were also used in order to collect data on relevant contextual factors that could provide explanation of students' achievements.

Assessment Procedure
Each student was assessed using one test per subject.The tests in the Serbian language and mathematics were distributed according to the spiral method (Allen, Carlson & Zelenak, 1999).Students had 45 minutes to solve Serbian language test, and 45 minutes for Mathematics test.A half of students were first tested in the Serbian language, followed by the test in mathematics, while another half of students was tested in different sequence.After the tests in the Serbian language and mathematics, students completed student questionnaire.At the same time, teachers and principal completed the questionnaire for teachers and questionnaire for principal, respectively.

Data Processing
Since the students were tested using 12 different tests in the Serbian language and 12 different tests in mathematics, students' achievement was calculated on the basis of single-parameter IRT using Winsteps program.
Students' achievements at the National Assessment Study were expressed at the scale where national average is 500 points, and standard deviation is 100 points (which means that around 2/3 of students have the score between 400 and 600 points).
Based on the score, each student was allocated to relevant achievement level (from E to A), and those students who failed to demonstrate the knowledge and skills defining the lowest level E were allocated to the category "below E".

School Marks of Roma and non-Roma Students
The first indicators of Roma education achievements are school marks which are result of teacher assessment.Within the National Assessment, data on school marks of students (Roma and non-Roma) were collected for the first three grades of elementary school (Table 3).
The data on school marks indicate a very significant difference in education achievements of Roma and non-Roma students regarding to two subjects.It may be seen that the majority of Roma pupils (45-55%) have the lowest passing grade (2), while over 40% of non-Roma pupils in all three grades in both subjects have the highest mark (5).Great difference in education achievements expressed through school marks is also perceived in terms of the number of pupils who had mark (1) at the end of school year (failed).Among Roma students, this percentage ranges from 7% to 11% for the Serbian language, and from 10% to 14% for the Mathematics, while less than 1% of non-Roma children got the mark 1 at the end of school year.On the other hand, when the highest school mark is analyzed (5), it is seen that 40-55% of non-Roma students get the highest mark at the end of school year, while this percentage among Roma children ranges between 5% and 10%.Thus, when school marks of Roma and non-Roma children are compared, a very high and significant difference is perceived.

School Marks of Roma and non-Roma Boys and Girls
When discussing school marks of the Roma and non-Roma, it is important to pay attention to school marks of boys and girls (Table 4, Figure 1, and Figure 2).As it may be seen, average mark of Roma students ranges between 2.5 and 3, while the average mark of non-Roma students ranges between 4 and 4.5.When the marks of boys and girls are compared, it may be perceived that, among non-Roma students, girls are more successful than boys both in the mathematics and the Serbian language.With Roma children, boys are to some extent more successful than girls in terms of the mathematics in the first grade, but this difference decreases until the third grade when Roma girls have somewhat higher mark in the mathematics than boys have.As far as the Serbian language is concerned, it is perceived that Roma girls and Roma boys have the same average mark in the first grade, but the difference increases in favor of girls until the third grade.Basic reason of difference increase in average mark is in the fact that the average mark of boys in the Serbian language decreases, while the average mark of girls from the first to the third grade remains at the same level.

The Achievement of Roma and non-Roma Students at the National Assessment Study
The average achievement of Roma students in mathematics test amounts to 366 points, which is by 134 points below national average, while average achievement in the Serbian language test amounts to 356 points, i.e. 154 points below the national average.There is no statistically significant difference in terms of the result of Roma boys and girls, but it is perceived that educational achievements among girls are significantly more heterogeneous (table 5).In some international studies (TIMSS, PISA) it was determined that one year of education impacted, on average, that students achieve the progress of around 60 points at the scale used for the National Assessment Study in Serbia.If it is assumed that it is the same case in Serbian education, i.e. that one school year results in the progress of 60 points, this would mean that Roma students lag behind other students for 2.2 school years in the mathematics, i.e. 2.6 school years in the Serbian language!Besides, those are the students who have spent only 3 years at school!In other words, Roma students, after the third grade of elementary school, have the achievement level in the mathematics non-Roma students have had at the end of the first grade, and in the Serbian language, they have the achievement level non-Roma students have had in the middle of the first grade.These data may be interpreted in another way.If existing quality of education of Roma students was maintained, they should have from 2.2 to 2.6 times more classes (or time spent with teacher) to reach national average.
Besides, it is perceived that the variability of the achievements of Roma students is higher than the variability at national level indicating that individual differences among Roma students in terms of educational achievements are higher than among non-Roma students.When two examined subjects are compared, it is perceived that the variability of the achievements of Roma students is higher in the Serbian language test then the variability of the achievement in the mathematics test which is most probably the consequence of the fact that the Serbian language is not mother tongue to Roma students.
As it is seemed, based on the achievement tests used in the National Assessment Study the five levels of achievements were singled out.Each level was defined on the basis of what students, at given achievement level, know or can do in the mathematics and the Serbian language (see the table 1 and 2).The data presented in the table 6 indicate that 50% of Roma students after 3 years of education have failed to acquire even very basic level of mathematical literacy, or failed to be capable of applying very basic mathematical competencies in simple situation (Below E level).On the other side, at the national level, only 11% of students failed to rich level E.Besides those 50% of Roma students, additional 26% of Roma students, after 3 years of education, managed to acquire only knowledge and skills from the level E. It means that, at national level, around 29% of students are at E level or below it, while among Roma students, there are 76% of students who are at E level or below it.When the achievements of Roma boys and girls are compared, it is shown that the difference is not statistically significant.
It is similar situation with achievements of Roma students at the Serbian language assessment (table 7).Around 56% of Roma students, after three years of education, failed to master even basic level of language competencies, while, at the national level, this is the case with only 14% of students.It means that four times more Roma students than others rested below the lowest level of educational achievements.With respect to the gender difference, it is shown the difference between Roma girls and Roma boys is not statistically significant.Although these findings about the educational achievements of Roma students are already disturbing, it should be stressed that the real situation is somewhat worst.Namely, when the students who were not present at school on testing date are taken into account, it may be perceived that both among non-Roma and Roma students there were more absent students who have lower marks, but this proportion with Roma students is higher (table 8).Thus, if all students took part in the assessment study, it could be expected that obtained picture on educational achievements of Roma students would be even more unfavorable.

The Repeating Rate of Roma and non-Roma Students
The data on repeating rate in the first three grades of elementary school were also gathered within the National Assessment Study of Grade 3 students of elementary school.The data indicate that, at the national level, repetition rate is 1%, while the repetition rate among Roma students in the first three grades of elementary school is 11%.In other words, out of 100 students of non-Roma nationality, only 1 child will repeat grade during the first three years of education, while there will be 11 students among Roma children who will repeat some of the first three grades.On the other hand, when the representation of Roma students is analyzed within the group of the students who have repeated a grade during the first three grades of education, it is perceived that 48% of those students are of Roma nationality, which is far higher than the representation of Roma students within the population of Grade 3 students (3-4%).
***** On the basis of above presented data on educational achievements of Roma students in Serbia (school marks, the result at the National Assessment Study, and the repeating rate), it may be concluded that Roma students have significantly lower education achievements in relation to the national average.Perhaps, the perceived gap in educational achievements of Roma and non-Roma students could be indicated as dramatic.Perhaps, regardless of the fact whether this gap in the achievements would be indicated with more or less dramatic attributes, it is actually of such a level that it points out the necessity and urgency to design and implement the measures impacting the decrease of this gap.Besides, it should not be ignored that the majority of the data used herein have been provided in the third grade of the elementary school.Thus, already after three years of education, the gap between Roma and non-Roma students is very high.If similar data for older students would be available, the picture would probably be even more dramatic.

To What Extent Does School Contribute to Lower Achievements of Roma Students?
One interpretation of the finding that Roma students have significantly lower educational achievements than non-Roma students may be related to the fact that Roma students, in higher number that non-Roma students, originate from the families having very low socio-economical status (SES).In other words, in accordance with this interpretation, it would not be realistic to expect that Roma students have the same achievements as non-Roma students since there have already been significant differences in their family background.This interpretation can be tested by comparison of achievements of Roma students with the achievement of the group of non-Roma students having the same SES as Roma students have.If only unfavorable conditions in which Roma students live are responsible for their lower achievements, then non-Roma students having the same SES as Roma students, would have the same achievements as Roma students have.In this case, it could be concluded that school provides equal quality of education both to Roma and non-Roma students, that is, it could be concluded that the education system and schools do not contribute to the gap between Roma and non-Roma students.Of course, even in this case, it would be quite justifiable to require from education to provide even higher quality of education to Roma students since the education is not only to provide equal education quality to everyone, but that children get the education quality which will meet their developmental and educational needs.
To check this interpretation, additional analyses of the data on students' achievements were made at the National Assessment Study in mathematics test.Taking into account the similarity of the results of Roma students in the Serbian language and mathematics, it may be assumed that the same findings would be obtained if the achievements in the Serbian language were additionally analyzed.Through additional analysis (ANCOVA with socioeconomic status as the covariate variable) the data on what would be the achievements of Roma and non-Roma students in the mathematics would be unless the students from these two groups would be different in terms of their family socioeconomic status (SES) 4 .
The results of this additional analysis indicate that, in case there is no difference in terms of the SES between Roma and non-Roma families, there still would be the significant difference in the achievement of Roma and non-Roma students.Namely, the ANCOVA shows that there is a statistically significant effect of the SES (F=792.7;df=1,4125; p<0.000), but the difference between Roma and non Roma is also statistically significant (F=147; df=1,4125; p<0.000).
Table 9 shows the data on average achievement of Roma and non-Roma students in mathematics test before and after the introduction of the SES in the Analysis of Covariance (ANCOVA).As it may be seen in the table 9, the fact that Roma students come from the families with low SES may explain only 54 points of observed difference between Roma and non-Roma students.Yet, the difference of 80 points remains, which may not be explained with low SES of Roma families.In consideration of the fact that socioeconomic family status is highly related to the level of preparation when starting school, as well as the duration of attending pre-school institutions, support level provided to children by parents during their education, and other so called "family" factors of educational achievement of students, it may be said that the majority of remaining difference is determined with the quality of education to which Roma students are exposed in Serbian schools.Thus, it may be concluded that around 40% of the difference in terms of the achievements between Roma and non-Roma students may be explained with lower SES of Roma students, but around 60% of this gap may not be explained using family and socioeconomic factors.In other words, the substantial part of determined difference in terms of educational achievements may be solved in such a way that the quality of the education of Roma students gets improved.

Teachers' Approach to Roma and non-Roma Students
However, the question of how it is possible that Roma students do not get the same education quality as non-Roma students can be raised taking into consideration that Roma students who participated in the National Assessment Study were in the classes with the majority of non-Roma students.Thus, the same teachers work with both Roma and non-Roma students.However, it seems that the same teachers have different relation and approach to Roma and non-Roma students.
One of important differences in view of this may include different expectations of teachers from Roma and non-Roma students.Namely, it often assumed that teachers have lower educational expectations from Roma students, the consequence of which may be the fact that Roma students will be less supported to develop.All this may result in the fact that even those Roma students who could achieve much better educational results will not achieve them for they are not supported appropriately by teachers.The data obtained within the National Assessment Study provide to determine whether teachers have the same educational expectations from Roma and non-Roma students.To determine this, the achievement at the National Assessment Study of both Roma and non-Roma students who got the same school marks from their teachers was analyzed.Namely, if teachers have the same expectations from Roma and non-Roma students, then criteria for particular school mark will be the same for Roma and non-Roma students.In other words, if the criteria of marking and expectations of teachers are the same, Roma and non-Roma students having the same school mark will have the same average achievement at the National Assessment. Figure 3 shows the data on average score in mathematics test of Roma and non-Roma students who got passing school marks in the mathematics at the end of Grade 3 by their teachers.On the basis of the data shown in the Figure 3, it is seen that teachers have lower criteria, i.e. lower expectations from Roma students.Besides, it should be emphasized that this comparison was made only within those classes including both Roma and non-Roma students.These data suggest that Roma students should demonstrate lower achievement level in the mathematics to get the same mark as non-Roma students by teachers.The difference in terms of marking criteria amounts to around 45 points on average, which is almost a whole mark.In other words, if Roma students would be marked according to the same criteria as non-Roma students, each Roma student would have one mark lower in the mathematics (those who got the mark 5 at the end of the third grade would get the mark 4, those who got the mark 4 would get the mark 3, etc.).
Besides lower expectations, teachers demonstrate some other differences in their relation to Roma and non-Roma students, as well as in their work with these students.Within the National Assessment Study, students completed the questionnaire that also included questions on the relation of students and teachers, and the teaching/learning process.The comparison of answers by Roma and non-Roma students regarding these questions provides the opportunity to determine how these two groups of students experience their teachers.On the basis of data analysis, the following is determined: • Teachers more often mark the homework of non-Roma students.
• Teachers more often provide explanation of what is good and what is bad in what student has answered regarding non-Roma students.• Roma students estimate that they have better relation with teachers than non-Roma students.• Non-Roma students think that teachers are more frequently stricter to them than this is the case with Roma students.• Non-Roma students think that teachers warning them more frequently when they do not know something than this is the case with Roma students.• Non-Roma students think, to higher extent than Roma students, that teachers expect them to know the mathematics well.These findings indicate that the same teachers, in some important aspects of their work with students, do not behave in the same way to Roma and non-Roma students.What should be emphasized as positive fact is that Roma students often have the feeling that they have good relation with teachers than this is the case with non-Roma students.Yet, this may be the consequence of the fact that teachers have lower expectations from Roma students, they require less work from them, stimulate them less to learn, etc., which may be experienced by Roma students as positive relation to them.
Within the context of the issue whether education quality exposed to Roma students is the same as education quality exposed to non-Roma students, the issue of distribution of Roma students in classes in the course of enrollment in school may be opened.Namely, when a Roma child is enrolled in an elementary school, there is the possibility that it is placed at the class where there is more quality teaching/learning or at the class with less quality teaching/learning.If Roma students would be placed at the classes of less education quality, this would only result in the creation of higher gap in education achievements with Roma children.In this case, Roma children would be "victims" of both lower socioeconomic status and lower education quality.On the other hand, if Roma children would be placed at the classes with more education quality, this would have a compensatory effect and impact the decrease of the gap in the achievement of Roma students.
To find out what is the practice in Serbian schools in terms of the distribution of Roma students in classes, it is first necessary to get the measure of education quality at the class level.The data collected within the National Assessment Study provided the opportunity to obtain this information.Starting assumption is that students' achievement at the National Assessment Study depends both on socioeconomic status of family (which is not impacted by teacher and school) and education quality a child is exposed to at school (which may be impacted by teacher and school).On the basis of the Analysis of Covariance (ANCOVA) in which SES is covariate variable, the data on what would be average achievement of students in a class would be obtained when students from various classes would not be different in terms of average SES.In other words, on the basis of such an analysis, the estimate of the achievement of students at a class would be obtained when all teachers, at the beginning of the first grade, would get the same group of students, i.e. when "starting point" of students from various classes would be the same.In this case, the class in which students' progress to higher extent in relation to the same starting point would be the class with higher quality of education.Thus, the measure of the education quality is the actual measure indicating the extent to which the students of a class have progressed, while it is assumed that this progress is the result of what happens within the teaching/learning process.On the basis of such obtained measure of quality instruction at class, all classes participating in the National Assessment are dividing in 5 categories according to the education quality.
Table 10 presents data on Roma and non-Roma students and their enrollment into classes with various education qualities.As it may be seen above, there is a significant difference in terms of education quality at the classes with Roma and non-Roma students.Namely, as seen from the data presented in the table 10, 40% of Roma students are in the classes with the lowest education quality, while there are twice less (around 20%) of non-Roma students enrolled in such classes.The situation is opposite when attention is paid to the classes with the highest education quality (category 1 and 2 from the table above).At such classes, there are around 39% of non-Roma students and around 23-24% of Roma students.

CONCLUSIONS
The main aim of this paper was to, on the basis of collected data within the Grade 3 National Assessment Study implemented by Institute for the Evaluation of Education Quality in 2004 determine the level to which Roma students manage to develop some basic mathematical and language competencies.Besides, one of the objectives of the paper was to determine the extent to which the differences in the achievement of Roma and non-Roma students could be determined with the fact that Roma students originate from families that have lower SES in relation to non-Roma families, and to what extent they may be explained with the lower education quality to which Roma students are exposed during their education.Finally, this paper had an objective to indicate at least some aspects of teaching/learning process impacting poorer educational achievements of Roma students.
Obtained data indicate that there is a high and significant gap in terms of the level of the development of basic mathematical and language literacy between Roma and non-Roma students after three years of education -the average achievement of Roma students in the mathematics was by 134 scores lower than the national average, and by whole 154 scores lower than the national average in the Serbian language.On the basis of these data, it may be said that Roma students, only three years after their education, lag behind other students by 2.2-2.6 of school years!Moreover, after three years of education every second Roma student failed to develop the knowledge and skills defining lowest achievement level of Grade 3 students.Based on this information it may be concluded that Roma students "failed to keep pace" already at the start of their education, therefore, it is no wonder that almost 50% of Roma students do not continue their education after the forth grade of elementary school (Mihajlović, 2004).
Besides, presented data also indicate that Roma children are triple "victims".First, they are the "victims" of poorer development conditions during their early childhood, above all due to poverty of their families, lower educational status of parents, and other family factors impacting that Roma children do not have an appropriate opportunities to develop all those skills developed with other children before starting education.The first opportunity for Roma children to get exposed to stimulating environment is already missed in pre-school period for a minor number of Roma children are included in pre-school institutions (Mihajlović, 2004).However, on the basis of the data presented in this paper, it may be perceived that Roma children become the "victims" of lower education quality obtained during the first grades of elementary school.It may be said that Roma children in this way simply do not get their chance for the development of their potentials with which one of their basic rights is affected (Convention on the Rights of the Child5 ).Besides, Roma children become the "victims" of many stereotypes based on the fact that Roma children lag behind non-Roma children in terms of basic literacy, and this lagging behind fails to be attributed to utterly disastrous conditions in which these children are developed, but are attributed to inborn development capacities of Roma children.In such a situation, it should be no wonder that "vicious circle of Roma poverty" passes from generation to generation.In such a situation, the explanation why some Roma children fail to develop basic literacy should not be sought, but one should wonder how some Roma children, despite those multiple "burden" manage to develop some important competencies during their education.
Obtained data indicate some aspects of teaching/learning process that could point out the way in which school and teachers additionally contribute that Roma children fail to develop during their education as much as they could.It was noted that Roma students are arranged, to higher extent than non-Roma students, into the classes with the lowest education quality, that teachers have lower expectations from Roma students, that they are not sufficiently stimulated, that curriculum is "shortened" and "simplified" for them, etc.The question can be raised whether Roma students are less favored in relation to non-Roma students also in terms of other important aspects of education quality.In any case, it is seen that existing high difference in the achievements of Roma students in relation to non-Roma students may only partially (around 40%) be explained with less favorable socioeconomic status of Roma families and that there is a significant space to improve the achievements of Roma students if education quality to which Roma students are exposed at schools is significantly improved.This does not mean that instruction quality to which Roma students will be exposed should be equal to the quality of non-Roma students, but it should be even much better to compensate unfavorable conditions in which Roma children are developed before starting education.Educational needs of Roma students are higher than educational needs of non-Roma students.The teaching activities need to be adjusted to educational needs of students rather than vice versa.In this way, school and teachers may provide adequate education to all children, which will result in the fact that not any child will be the victim of its social background it could not choose.

Figure 3 .
Figure 3. Average score in mathematics test of Roma and non-Roma students who got passing mark in the mathematics at the end of Grade 3 of elementary school