Vous êtes sur la page 1sur 16

Aust. Educ. Res.

(2015) 42:117132
DOI 10.1007/s13384-014-0157-9

Turkish Students Science Performance and Related


Factors in PISA 2006 and 2009

Mustafa Sami Topcu Serkan Arkan

Evrim Erbilgin

Received: 24 February 2014 / Accepted: 29 July 2014 / Published online: 7 August 2014
 The Australian Association for Research in Education, Inc. 2014

Abstract The OECDs Programme for International Student Assessment (PISA)


enables participating countries to monitor 15-year old students progress in reading,
mathematics, and science literacy. The present study investigates persistent factors
that contribute to science performance of Turkish students in PISA 2006 and PISA
2009. Additionally, the study investigates whether factors explaining science per-
formance has changed between 2006 and 2009. Multilevel analyses of student level
and school-level variables revealed that the variance in science performance
explained by school-level variables is 13 % for PISA 2006 and 28 % for PISA
2009. Moreover, the variance in science performance explained by student-level
variables is 19 % for PISA 2006 and 20 % for PISA 2009. The six most common
student-level variables in PISA 2006 and 2009 were statistically significant pre-
dictors of science performance. At the school level, location of school was a sig-
nificant predictor of science achievement in both PISA 2006 and 2009.

Keywords Science performance  PISA  Hierarchical Linear Modeling (HLM) 


Student resources  School resources

Introduction

The Programme for International Student Assessment (PISA), a project of the


Organisation for Economic Co-operation and Development (OECD), aims to assess

M. S. Topcu (&)
Department of Elementary Education, Faculty of Education, Yldz Technical University, Istanbul,
Turkey
e-mail: mstopcu@yildiz.edu.tr

S. Arkan  E. Erbilgin
Department of Elementary Education, Faculty of Education, Mugla Stk Kocman University,
Mugla, Turkey

123
118 M. S. Topcu et al.

the extent to which students near the end of compulsory education have acquired
some of the knowledge and skills that are essential for full participation in modern
societies (OECD 2010a, p. 17). Thus, PISA allows the participating countries to
monitor the yields of their compulsory education systems and track the national
progress in accomplishing key learning goals. PISA assesses the performance of
15-year-olds in reading, mathematics, and science in 3-year cycles. In each cycle,
one subject area is designated as the main focus, called the major domain. In PISA
2006 and 2009, scientific literacy and reading literacy were the major domains of
assessment, respectively.
Turkey participated in PISA 2003, 2006, 2009, and 2012; results of PISA 2012
were not released at the time of this study. Among these PISA cycles, PISA 2006
and PISA 2009 provides comparable data for science scores (OECD 2010b). In both
tests, Turkish students performed below the international average for science: 424 in
PISA 2006 (mean score 498) and 454 in PISA 2009 (mean score 501). Poor science
performance of Turkish students in PISA tests generated discussions among Turkish
educational researchers and policy-makers about potential educational policies to
improve the academic outcomes. For example, the science curriculum for students
in grades 18 was revised in 2005. The revised science curriculum highlighted not
only core science contents but also scientific processes such as investigation,
inquiry, problem-solving, and decision-making (Ministry of National Education of
Turkey [MONE] 2005). Another reform initiative was to decrease the grade level
that students start taking science courses from 4 to 3 in 2013. Research studies that
investigate the variance in students science achievement might help to determine
new initiatives to enhance the quality of education in Turkey, and possibly in other
countries.
In addition to assessing students academic performance, PISA also provides a
window into the predictors of student achievement such as student and school-
related resources. Research indicates that students with more resources (e.g., books,
computers, qualified teachers) tend to show higher academic performance (Chiu and
Xihua 2008; Hanushek 2003; Krueger 1999; Sirin 2005; Sousa et al. 2012).
However, research studies are not in agreement about the extent to which student
and school resources influence student performance. A country like Turkey with
limited resources should decide how to distribute resources effectively to improve
its overall student performance. Research investigating student and school resources
will offer new perspectives for making such decisions.
Several studies have been conducted to examine possible causes of Turkish
students science achievement levels in a single PISA test (e.g., Alacac and Erbas
2010; Anl 2009; Gursakal 2012; Yalcin et al. 2012; Ylmaz 2009). Lacking are
studies comparing the students science achievement predictors in more than one
PISA test. The present study contributes to international comparative research
studies by comparing possible determinants of Turkish students science perfor-
mance in PISA 2006 and 2009. The study is significant and unique because it
involves constructing a hierarchical linear model using the same variables from two
different PISA tests to find out the predictors of students performance in science.
Furthermore, the study investigates whether the factors explaining science
performance has changed between 2006 and 2009. The possible persistent common

123
Turkish Students Science Performance and Related Factors 119

factors determined by this study will help evaluate recent initiatives in education as
well as provide data for further research-based education policies. The research
questions that guided this study are:

(1) What is the partitioning variance in Turkish students science performance at


school and student-level in PISA 2006 and PISA 2009?
(2) What common student and school resources are associated with science
performance of Turkish students in PISA 2006 and PISA 2009?

Student Resources Associated with Science Achievement

Students success in school education depends on several factors. Research studies


revealed that student and school resources are among the main determinants of
student achievement (Anderson et al. 2007; Engin-Demir 2009; Sirin 2005; Sousa
et al. 2012). Related research findings about student resources are discussed in this
section and research findings about school resources are presented in the following
section.
Student resources include parental education level, parental occupation, family
income, household wealth, educational materials at home, and cultural possessions
at home (Kalender and Berberoglu 2009; Sirin 2005). Xia (2009) has also
emphasized the importance of family process variables such as parents expecta-
tions and beliefs, and parental involvement. Research studies typically measure
family socioeconomic status (SES) by using one or more student resource variables.
One consistent finding across studies is that most student resources are strongly and
positively associated with student achievement in science (Chiu and Xihua 2008;
Fuchs and Womann 2007; Sirin 2005; Sousa et al. 2012; Xia 2009).
Among the student resources, parental education level and parental occupation
have been found to significantly correlate with student achievement (Barone 2006;
Chevalier and Lanot 2002; Fuchs and Womann 2007; Hanushek and Luque 2003;
Yayan and Berberoglu 2004). For example, in their study of a sample from Great
Britain, Chevalier and Lanot (2002) concluded that parental education level had
more effect than family income on students overall performance in school.
Similarly, Fuchs and Womann (2007) found that student performance increased in
reading, mathematics, and science as the level of parental education increased based
on PISA 2000 data set.
Some student resources are related to familys economic means. Different
research studies used different variables, a narrow set of them, to define this aspect
of student resources. Fathers earning (Chevalier and Lanot 2002), number of items
at home such as books, computers, or dishwasher (Fuchs and Womann 2007;
Hanushek and Luque 2003; Kalender and Berberoglu 2009; Sousa et al. 2012),
index of home possessions (Sousa et al. 2012) are examples of how familys
economic means are represented in studies. Ho (2010a) investigated the effect of
familys economic means on students scientific literacy scores by classifying it into
three categories: cultural resources, educational resources, and material resources.

123
120 M. S. Topcu et al.

She found that students who had more cultural resources at home scored
significantly higher in scientific literacy in PISA 2006 conducted in Hong Kong.
This finding suggests that some forms of familys economic means might be more
influential on student achievement than some other forms do; though more studies
are needed to explore such relationships. In particular, studies that use more
comprehensive measures of student resources will shed light on which resources are
more influential on student achievement (Sousa et al. 2012).

School Resources Associated with Science Achievement

In 1966, sociologist James Coleman reported that student resources were more
predictive of student achievement than differences in school resources based on data
from 4,000 schools in the United States (Coleman et al. 1966). This finding was
interpreted as schools did not matter in making a difference in student achievement
and triggered a tremendous research effort to investigate the effects of school
resources on student achievement (Alacac and Erbas 2010; Hanushek 2003). This
section includes findings of subsequent research that provided further evidence
about the effects of schools on student achievement. In the research studies, school
resources have usually been determined by school facilities, learning time, teachers
education level, teacherstudent ratio, teacher shortage, ability grouping, class size,
school size, location, and school type.
In his review of the studies investigating the effects of school resources on
student achievement, Hanushek (2003) reported that the results of these studies are
inconsistent. For instance, Heyneman and Loxleys (1983) findings suggested that
school resources were more important than family characteristics in developing
countries. On the contrary, in their study of TIMSS 1995 data set, Hanushek and
Luque (2003) argued that there was no clear pattern showing school resources were
more important in developing countries. Hanushek (2003) concluded that resources
devoted to schools did not necessarily promote higher achievement gains. Hanushek
(2003) highlighted that how resources were used by students was more important
than the physical presence of a resource.
The results of research studies about school location, school size, and class size
effects are contradicting. School location (rural, urban, suburban) has not shown a
consistent effect on student achievement; some studies found that schools in rural
communities promote higher student achievement (Boyd 1991; Lippman and Burns
1996), while some other studies showed that there is no relationship between school
location and student achievement (Stewart 2008). In Turkey, it is reasonable to
expect that schools in urban communities would promote higher student achieve-
ment since those schools would have more resources. Regarding the school size,
some studies found a positive effect of school size on student achievement
(Hanushek and Luque 2003), while other studies found a negative effect (Kuziemko
2006). Kuziemko (2006) argued that 50 % decrease in school size would promote
2 % rise in mathematics scores over 2 years. Research on class size affect provides
interesting data with some studies showing a significant positive association
between small class size and student achievement (Krueger 1999; Rivkin et al.
2005), although some other studies did not find such significant relationship

123
Turkish Students Science Performance and Related Factors 121

(Cordero et al. 2010; Fuchs and Womann 2007; Sousa et al. 2012). Fuchs and
Womann (2007) found insignificant association between class size and student
performance, however, they demonstrated evidence that the quality of instructional
materials and teachers, and availability of external and standardized exams
positively correlated with student performance.
One of the school resources that have been found to influence student
achievement is learning time in school. For instance, the weekly number of
mathematics lessons affected students mathematics performance more than
studentteacher ratio did after controlling SES factors (Alacac and Erbas 2010).
Sousa et al. (2012) concurred that learning time in school was positively associated
with students science and mathematics performance, whereas tutoring was
negatively associated with students performance in both subjects. The negative
association of tutoring might be due to cultural factors (Sousa et al. 2012).
Another school resource that attracted researchers interest is ability grouping
since it is related to providing equal educational opportunity to all children. Students
in higher ability grouping tend to have higher family SES and better qualified
teachers. On the other hand, students in lower ability grouping tend to have lower
family SES and teachers with low expectations and standards for their students
(Hallinan 1994; Knipprath 2010). Ho (2010b) informed that Hong Kong schools
grouped students into 5 ability levels before the year 2000 and into 3 ability levels
after the year 2000. Ho (2010b) interpreted that this change resulted in 8 %
reduction in the between-school variance from PISA 2000 to 2006, indicating an
improvement for educational equality in Hong Kong. In Turkey, there is no
structured ability grouping across the country. Some schools implement ability
grouping at their own preference. Investigating the effect of ability grouping in
Turkey is worthwhile for future educational decisions.

Method

Sample

The data analyzed in this study were obtained from the PISA 2006 and 2009 studies.
The sampling procedure for both studies was carried out by PISA project
consortium as described in the published manual (OECD 2005). PISA determined
participants through a two-stage stratified sampling process. First, schools with
15-year-old students were selected based on the following criteria: size, location,
and type. Next, students from a list of all 15-year-old students in the schools were
randomly selected. The sample for Turkey consisted of 4,942 students in 160
schools for PISA 2006, and 4,996 students in 170 schools for PISA 2009. In the
Turkish PISA 2006 sample, there were 2,652 males and 2,290 females attending 7th
(n = 23), 8th (n = 93), 9th (n = 2,007), 10th (n = 2,671), and 11th (n = 148)
grade levels, and in the Turkish PISA 2009 sample, there were 2,551 males and
2,445 females attending 7th (n = 24), 8th (n = 113), 9th (n = 1,258), 10th
(n = 3,393), 11th (n = 196), and 12th (n = 12) grade levels. In multilevel data
analysis, schools with few students could be excluded (Maas and Hox 2005). In the

123
122 M. S. Topcu et al.

current study, two schools with fewer than 5 students in 2006 and 10 schools with
fewer than 5 students in 2009 were excluded from multilevel data analysis. As a
result, the final sample consisted of 4,935 students in 158 schools in PISA 2006 and
4,964 students in 160 schools in PISA 2009.

Measures and Variables

The PISA 2006 and 2009 studies gathered data on students science performance by
means of science literacy test, and data on student and school level characteristics
by means of student and school questionnaires. The present study used the science
literacy test scores and selected common items from the student and school
questionnaires. PISA gives plausible science scores for each participating student.
Therefore, five plausible science scores were used to represent students
performance in science as the dependent variable. Independent variables of the
current study included student and school-level variables. The common items from
both PISA 2006 and 2009 student and school questionnaires were determined. As a
result, 19 student-level variables and 13 school-level variables were used in this
study. The selected items were recoded to carry out the statistical analyses. For the
purpose of the data analysis, group mean centering method at student-level and
grand mean centering method at school-level were used.

The Procedure for Data Analysis

PISA data is embedded with multilevel structures in which students nested within
schools. Therefore, multilevel analysis techniques that take into consideration this
hierarchical structure should be used (Raudenbush and Bryk 2002). Hierarchical
Linear Modeling program (HLM 7) was used in this study because of the
hierarchical structure of PISA data sets with plausible values. This program can
analyze hierarchical PISA data with plausible values as dependent variables
(Raudenbush et al. 2011). In the data set, there were two levels: students at the first
level nested within schools at the second level. By using the multilevel model,
variance in achievement measures were partitioned into student level and school
level variance components. In order to achieve this partitioning, first a simple
multilevel model (Null model) including only dependent variable was developed.
Second, a complex multilevel model (full model) including student and school level
variables was developed. The full model explained variables that are statistically
significant predictors of achievement at the student and school level. For PISA 2006
and 2009 data sets, the same statistical procedures were followed.

Limitation of the Study

The main limitation of this study is that we only used common items from the PISA
2006 and 2009 student and school questionnaires that reflect student and school
resources. Therefore, the findings of this study are only limited to common items
coming from both PISA 2006 and 2009 questionnaires. From 2006 to 2009 there
were changes in many items; therefore we could not use student and school level

123
Turkish Students Science Performance and Related Factors 123

items that were asked only in one administration. As a result, we did not use these
items in our analysis in order to examine whether the same item is effective in
predicting science performance in both PISA 2006 and 2009. Therefore, in the first
phase of the data analysis section, the common items from both PISA 2006 and
2009 student and school questionnaires were determined and used by the
researchers. Similarly, we could not use parent questionnaires because of the
reason that while the items in the parent questionnaire in 2006 were related to
parents perspectives on science activities, items in 2009 focused on the parents
perspectives on reading activities. Therefore, there were no common items in 2006
and 2009, reflecting parents perspectives on science.

Results

All assumptions for HLM 7 analysis reported by Raudenbush and Bryk (2002) were
checked by the researchers. For example, residuals about student variables should
be independent and normally distributed. Histograms of student variables using
residuals were analyzed and concluded that this assumption was not violated.
Similar to this assumption, other assumptions about both student and school
variables (multivariate normality, homogeneity of variance, and multicollinearity)
were checked and not violated. Therefore, these results suggested that we could
achieve the HLM analysis.

Partitioning Variance in Science Performance

In this section, science performance variance components in both PISA 2006 and
PISA 2009 were presented. Variance in science performance was partitioned into
student and school resources (see Table 1). For PISA 2006, approximately a half of
the variance in science performance was explained by school resources (55 % of the
total variance); another half of the variance in science performance was explained
by student resources (45 % of the total variance). For PISA 2009, approximately a
half of the variance in science performance was explained by school resources
(52 % of the total variance); another half of the variance in science performance
was explained by student resources (48 % of the total variance). The PISA 2006 and
PISA 2009 student sample showed the same pattern in science performance about
the variance distribution. In both data sets, students were approximately responsible
for 46 % of the total variance in science performance, whereas schools were
approximately responsible for 54 % of the total variance. Although a higher
percentage of the variance in science achievement was explained by school
resources, the explained variance by student resources was also statistically
significant.

Science Performance and Related Variables in PISA 2006 and 2009

As presented in the previous section, significant variances about both student and
school resources were found to explain students science performance. Therefore,

123
124 M. S. Topcu et al.

Table 1 Partitioning Variance in Science Performance


Between students Between schools

Science performance
PISA 2006 0.45 0.55
PISA 2009 0.48 0.52

student and school resources were used to predict these variances. Table 2 shows
significant predictors (student and school resources) of science performance in PISA
2006. In terms of student resources, 8 of the 19 resources are statistically significant
predictors of science performance. Students having books to help with their school
work scored significantly lower in science performance than those who did not have
books to help with their school work at home. The students having higher number of
computers scored significantly higher in science performance than those who had
fewer numbers of computers at home. The students having classic literature books
scored significantly higher in science performance than those who did not have
classic literature books at home. The students who did not have poetry books scored
significantly higher in science performance than those having poetry books at home.
The students having higher number of books scored significantly higher in science
performance than those who had fewer numbers of books at home. The students who
did not have dishwasher scored significantly higher in science performance than
those having dishwasher at home. The students having higher number of cars scored
significantly lower in science performance than those who had fewer numbers of
cars. The students having higher educational level of father showed significantly
more science performance than those who had fewer educational level of father.
In terms of school resources, 2 of the 13 resources are statistically significant
predictors of science performance in PISA 2006. The students attending schools
having more shortage of science teachers on average scored significantly higher in
science performance than those attending schools with less shortage of science
teachers on average. As the school location changes from village to large city,
students science performances have increased.
The multilevel model explained 19 % of the variance in science performance
among PISA 2006 students. Meanwhile, the model accounted for 13 % of the
variance in science performance among schools attended by these students. These
findings indicated that student resources in the analysis explained science
performance more than school resources in PISA 2006.
Table 3 shows significant predictors (student and school resources) of science
performance among PISA 2009 students. In terms of student resources, 7 of the 19
resources are statistically significant predictors of science performance in PISA 2009.
The students having books to help with their school work scored significantly higher in
science performance than those who did not have books to help with their school work
at home. The students having higher number of computers scored significantly higher
in science performance than those who had fewer numbers of computers at home. The
students having classic literature books scored significantly higher in science
performance than those who did not have classic literature books at home. The students

123
Turkish Students Science Performance and Related Factors 125

Table 2 Explaining Variance in Science Performance among PISA 2006 Students with Student and
School Characteristics
Statistically significant variable Coefficient SE

Student-level effects on science performance


Possessions textbooksa -11.67 (3.08)
How many computersb 9.40 (2.37)
Possessions literaturea 13.59 (2.27)
Possessions poetrya -11.55 (2.15)
How many books at homec 3.41 (0.89)
Possessions dishwashera -6.21 (2.19)
How many carsb -5.23 (1.53)
Educational level of fatherd 2.15 (0.71)
School-level effects on science performance
Location of schoole 24.34 (5.87)
Shortage science teachersf 11.83 (5.52)
Proportion of variance explained
At the student level 19 %
At the school level 13 %

All student-level and school-level variables in the table are statistically significant at the alpha level of
0.05
a
No: 1 and yes: 2
b
None: 1 and three or more: 4
c
0100 books: 1 and more than 500 books: 6
d
None: 1 and ISCED 5A: 6
e
Village: 1 and large city: 5
f
Not at all: 1 and a lot: 4
g
None: 1 and for all subjects: 3
h
No categorization

who did not have poetry books scored significantly higher in science performance than
those having poetry books at home. The students having higher number of books
scored significantly higher in science performance than those who had fewer numbers
of books at home. The students having higher number of cell phones scored
significantly lower in science performance than those who had fewer numbers of cell
phones. The students having higher number of cars scored significantly lower in
science performance than those who had fewer numbers of cars at home.
In terms of school resources, 3 of the 13 resources are statistically significant
predictors of science performance in PISA 2009. The students attending schools
with higher number of students per teacher on average scored significantly lower in
science performance than those attending schools with fewer numbers of students
per teacher on average. The students attending schools having more students
enrolled scored significantly lower in science performance than those attending
schools with less students enrolled. As the school location changes from village to
large city, students science performances have increased.

123
126 M. S. Topcu et al.

Table 3 Explaining Variance in Science Performance among PISA 2009 Students with Student and
School Characteristics
Statistically significant variable Coefficient SE

Student-level effects on science performance


Possessions textbooksa 9.97 (3.06)
How many computersb 6.60 (2.10)
Possessions literaturea 13.71 (2.50)
Possessions poetrya -10.32 (2.43)
How many books at homec 5.17 (0.92)
How many cell phonesb 4.80 (1.76)
b
How many cars -3.00 (1.44)
School-level effects on science performance
Total school enrollmenth -0.02 (0.01)
Teacherstudent ratioh -2.42 (0.48)
Location of schoole 16.42 (4.24)
Proportion of variance explained
At the student level 20 %
At the school level 28 %

All student-level and school-level variables in the table are statistically significant at the alpha level of
0.05
a
No: 1 and yes: 2
b
None: 1 and three or more: 4
c
0100 books: 1 and more than 500 books: 6
d
None: 1 and ISCED 5A: 6
e
Village: 1 and large city: 5
f
Not at all: 1 and a lot: 4
g
None: 1 and for all subjects: 3
h
No categorization

The multilevel model explained 20 % of the variance in science performance


among PISA 2009 students. Meanwhile, this model accounted for 28 % of the
variance in science performance among schools attended by these students. This
finding indicated that statistically significant student-level variables in the analysis
captured important student characteristics that were responsible for explaining
science performance. Moreover, statistically significant school-level variables
captured important school characteristics that were responsible for differences in
science performance among schools attended by PISA 2009 students.

Discussion

Decomposition of Total Variance in Science Performance

In the present study, the first aim was to investigate differences in the mean scores
of science performance among Turkish schools for PISA 2006 and 2009 separately.

123
Turkish Students Science Performance and Related Factors 127

The explained variance in science performance by school resources was 55 % for


PISA 2006, and was 52 % for PISA 2009. Quite similar percentages for school
resources were observed to explain students science performance. Our finding
suggested that for both PISA 2006 and 2009, the variance in science performance
explained by school resources is higher than the variance in science performance
explained by student resources. Aligned with the results of the current study, the
previous studies in the Turkish context (e.g., Delen and Bulut 2011; Ylmaz 2009)
reported that the biggest predictor of students science achievement seems to be
schools, and there are big science achievement gaps between Turkish schools. As a
result, between-school variability is much higher than within-school variability in
terms of Turkish students science achievement (OECD 2004, 2007a, 2010c).

Student Resources and Science Performance

Six out of 19 common resources (possessions textbooks, how many computers,


possessions literature, possessions poetry, how many books at home, how many
cars) from PISA 2006 and 2009 data sets significantly explained students science
performance. Three of these resources, possessions textbooks, how many
computers, and how many books at home are related to family educational
resources. This finding corroborates previous studies reporting that family
educational resources is significantly associated with achievement in science
(Gilleece et al. 2010; Ho 2010a; Ma 2003). When family educational resources
increase, students science performance also increases in general. There is an
exception about one of the family educational resources (possessions textbooks) that
students having books to help with their school work scored significantly lower in
science performance than those who did not have books to help with their school
work at home for the PISA 2006 data set. As opposed to this result, the PISA 2009
data set shows students having books to help with their school work scored
significantly higher in science performance than those who did not have books to
help with their school work at home. This result suggests that the nature of the
relationship between having textbooks and science performance changed from 2006
to 2009 in a positive way for Turkish students. This finding highlights the
importance of educational resources for improving student achievement.
Two of the six significant common resources, possessions literature and
possessions poetry are related to family cultural resources. In line with recent
studies in Hong Kong (Ho 2010a) and in Ireland (Gilleece et al. 2010), the present
study showed that possessions literature is an important socioeconomic resource
explaining students science performance in Turkey. Interestingly, this study found
that possessions poetry inversely influence students science performance in
Turkey. Thus, various forms of family cultural capital might show different effects
in different cultures. One of the six significant common resources, how many cars
is related to family material resources. This study showed that the students having
higher number of cars at home had lower science scores in PISA 2006 and 2009.
The findings of this study corroborate previous research indicating that various
forms of family resources show different effects on students science achievement
(Anl 2009; Gilleece et al. 2010; Ho 2010a). Ho (2010a) found that among the

123
128 M. S. Topcu et al.

familys economic means, family cultural resources had the most effect on students
science achievement in Hong Kong. In the Turkish context, the most important form
of family investment for improving students science achievement seems to be
educational resources. This result provides some insight for educational policies
regarding homeschool cooperation. For instance, schools might inform parents
about making their investments in educational resources rather than material
resources.

School Resources and Science Performance

The current study indicated that only one of the 13 common school resources was
effective in explaining students science performance for PISA 2006 and 2009. This
common resource (location of school) significantly explained students science
performance. Expectantly, for both PISA 2006 and 2009, when the school location
changes from village to large city, students science performances have increased.
This finding reflects that the schools in urban areas of Turkey have more educational
resources and opportunities as opposed to the schools in rural areas. However, in the
literature, there is an inconsistency about the effect of school location (rural, urban,
suburban) on students achievement. While some studies (e.g., Boyd 1991; Lippman
and Burns 1996) reported higher student achievement in the rural area schools,
some other studies (e.g., Stewart 2008) reveal relationships that are not significant
between school location and student achievement.
For PISA 2006, a resource of shortage science teacher showed a significant
relationship with students science performance. Response from the questionnaire
suggests that the students attending schools with more shortage of science teachers
on average scored significantly higher in science performance than those with less
shortage of science teachers on average. Aligned with the results of the current
study, Alacaci and Erbas (2010) reported positive significant relationship between
teacher shortages and mathematics performance of PISA 2006 Turkish students. It is
interesting that when the number of science teachers increases in a Turkish school,
the students in this school inclines to show low science performance. In future
research, it might be useful to examine why there is an interesting relationship
pattern in Turkish school contexts.
Teacherstudent ratio significantly explained students science performance for
PISA 2009. The students attending schools with fewer numbers of students per
teacher on average scored significantly higher in science performance. It is a
reasonable finding because a teacher might have few opportunities to pay special
attention to students individually in crowded classrooms (Ma 2003). Therefore,
school administrators need to be more sensitive when they organize their science
classrooms in their schools because teacherstudent ratio is a critical predictor to
explain students science performance.
For PISA 2009, a resource of total school enrollment showed a significant
negative relationship with students science performance. As it is expected, when
total school enrollment decreases, students achievement increases. However, in the

123
Turkish Students Science Performance and Related Factors 129

literature, while some studies (e.g., Hanushek and Luque 2003) reported a positive
effect of school size on student achievement, other studies (e.g., Kuziemko 2006)
found a negative effect. Therefore, future research is necessary to have deep
information about the nature of this relationship.

Conclusion and Implications

The following conclusions could be derived from the results of the present study.

(1) The findings from both PISA 2006 and 2009 data sets suggest that there are
big science achievement gaps between Turkish schools.
(2) An important contribution of the present study is that the study proposes
significant student and school resources predicting Turkish students science
achievement by using common items in PISA 2006 and 2009.
(3) The findings suggest that six out of the eight significant student resources in
PISA 2006 are also significant in 2009. Similarly, six out of the seven
significant student resources in PISA 2009 are significant in PISA 2006.
Therefore, six significant student resources seem to be persistently effective in
explaining Turkish students science achievement both in PISA 2006 and 2009.
(4) The six significant persistent resources, (possessions textbooks, how many
computers, how many books at home related to family educational resources;
possessions literature, possessions poetry related to family cultural capital;
how many cars related to families economic wealth) for both PISA 2006 and
2009 indicate that these resources are possible causes of students science
performance.
(5) Among the school resources, location of school is the only resource having a
potential to explain students science performance for both PISA 2006 and
2009.

As a result, curriculum developers and policy-makers in Turkish education


system should take into consideration the current research findings that reveal big
science achievement gaps between Turkish schools in light of both PISA 2006 and
2009 data sets. Since the same resources are still effective in explaining Turkish
students low science performance from PISA 2006 to PISA 2009, these resources
should be considered, and necessary precautions should be taken by MONE. For
example, homeschool cooperation should be improved. Parents could be informed
that they should make more investments in educational resources rather than
material resources. Regarding the school resources, teacherstudent ratio in science
lessons might be decreased. As a last word, the research area in Turkey needs a
higher number of comparative studies because with these studies, students science
performance and related resources can be compared within the country and with
other countries to take necessary precautions in order to improve Turkish students
science performance.

123
130 M. S. Topcu et al.

References

Alacac, C., & Erbas, A. K. (2010). Unpacking the inequality among Turkish schools: Findings from
PISA 2006. International Journal of Educational Development, 30(2), 182192.
Anderson, J. O., Lin, H. S., Treagust, D. F., Ross, S. P., & Yore, L. D. (2007). Using large-scale
assessment data sets for research in science and mathematics education: Programme for
International Student Assessment (PISA). International Journal of Science and Mathematics
Education, 5(4), 591614.
Anl, D. (2009). Factors effecting science achievement of science students in Programme for International
Students Achievement (PISA) in Turkey. Education and Science, 34(152), 87100.
Barone, C. (2006). Cultural capital, ambition and the explanation of inequalities in learning outcomes: A
comparative analysis. Sociology, 40(6), 10391058.
Boyd, W. L. (1991). What makes ghetto schools succeed or fail? Teachers College Record, 92, 331362.
Chevalier, A., & Lanot, G. (2002). The relative effect of family characteristics and financial situation on
educational achievement. Education Economics, 10(2), 165181.
Coleman, J. S., Campbell, E. Q., Hobson, C. J., McPartlant, J., Mood, A. M., Weinfall, F. D., et al. (1966).
Equality of educational opportunity. Washington, DC: Department of Health, Education and
Welfare.
Cordero, J. M., Crespo, E., & Santn, D. (2010). Factors affecting educational attainment: Evidence from
Spanish PISA 2006 results. Regional and Sectorial Economic Studies, 10(3), 5576.
Delen, E., & Bulut, O. (2011). The relationship between students exposure to technology and their
achievement in science and math. Turkish Online Journal of Educational Technology, 10(3),
311317.
Engin-Demir, C. (2009). Factors influencing the academic achievement of the Turkish urban poor.
International Journal of Educational Development, 29, 1729.
Fuchs, T., & Womann, L. (2007). What accounts for international differences in student performance? A
re-examination using PISA data. Empirical Economics, 32, 433464.
Gilleece, L., Cosgrove, J., & Sofroniou, N. (2010). Equity in mathematics and science outcomes:
Characteristics associated with high and low achievement on PISA 2006 in Ireland. International
Journal of Science and Mathematics Education, 8, 475496.
Gursakal, S. (2012). An evaluation of PISA 2009 student achievement levels affecting factors. Suleyman
Demirel University the Journal of Faculty of Economics and Administrative Sciences, 17(1),
441452.
Hallinan, M. T. (1994). Tracking: From theory to practice. Sociology of Education, 67(2), 7984.
Hanushek, E. A. (2003). The failure of input based schooling policies. The Economic Journal, 113(485),
6498.
Hanushek, E. A., & Luque, J. (2003). Efficiency and equity in schools around the world. Economics of
Education Review, 22, 481502.
Heyneman, S. P., & Loxley, W. (1983). The effect of primary school quality on academic achievement
across twenty-nine high- and low-income countries. American Journal of Sociology, 88(6),
11621194.
Ho, S. C. (2010a). Family influences on science learning among Hong Kong adolescents: What we
learned from PISA. International Journal of Science and Mathematics Education, 8, 409428.
Ho, S. C. (2010b). Assessing the quality and equality of Hong Kong basic education results from PISA
2000? to PISA 2006. Frontiers of Education in China, 5(2), 238257.
Kalender, I., & Berberoglu, G. (2009). An assessment of factors related to science achievement of Turkish
students. International Journal of Science Education, 31(10), 13791394.
Knipprath, H. (2010). What PISA tells us about the quality and inequality of Japanese education in
mathematics and science. International Journal of Science and Mathematics Education, 8, 389408.
Krueger, A. (1999). Experimental estimates of education production functions. The Quarterly Journal of
Economics, 114, 497532.
Kuziemko, I. (2006). Using shocks to school enrollment to estimate the effect of school size on student
achievement. Economics of Education Review, 25(1), 6375.
Ma, X. (2003). Measuring up: Academic performance of Canadian immigrant children in reading,
mathematics and science. Journal of International Migration and Integration, 4(4), 541576.
Maas, C. J. M., & Hox, J. J. (2005). Sufficient sample sizes for multilevel modeling. Methodology, 1,
8692.

123
Turkish Students Science Performance and Related Factors 131

Ministry of National Education of Turkey (MONE) (2005). Science and technology curriculum of
elementary schools (6th8th grades) [in Turkish]. Board of Education.
Organisation for Economic Co-operation and Development (OECD) (2010b). PISA 2009 results:
Learning trends: Changes in student performance since 2000, vol. V. Paris.
Organization for Economic Co-operation and Development (OECD) (2004). Learning for tomorrows
world. First results from PISA 2003. Paris.
Organization for Economic Co-operation and Development (OECD) (2005). School sampling preparation
manual. [Online] Retrieved May 02, 2013, from http://www.oecd.org/education/school/
programmeforinternationalstudentassessmentpisa/39829698.pdf.
Organization for Economic Co-operation and Development (OECD) (2007a). PISA 2006: Science
competencies for tomorrows world. Paris.
Organization for Economic Co-operation and Development (OECD) (2010a). PISA 2009 Results: What
students know and can doStudent performance in reading, mathematics and science, vol. I. Paris.
Organization for Economic Co-operation and Development (OECD) (2010c). PISA 2009 results:
Overcoming social backgroundEquity in learning opportunities and outcomes, vol. II. Paris.
Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models (2nd ed.). Thousand Oaks, CA: Sage.
Raudenbush, S. W., Bryk, A. S., Cheong, Y. F., Congdon, R. T., & du Toit, M. (2011). HLM 7:
Hierarchical linear and nonlinear modeling. Chicago, IL: Scientific Software International.
Rivkin, S. G., Hanushek, E. A., & Kain, J. F. (2005). Teachers, schools and academic achievement.
Econometrica, 73(2), 417458.
Sirin, S. R. (2005). Socioeconomic status and academic achievement: A meta-analytic review of research.
Review of Educational Research, 75, 417453.
Sousa, S., Park, E. J., & Armor, D. J. (2012). Comparing effects of family and school factors on cross-
national academic achievement using the 2009 and 2006 PISA surveys. Journal of Comparative
Policy Analysis: Research and Practice, 14(5), 449468.
Stewart, E. B. (2008). School structural characteristics, student effort, peer associations, and
parental involvement: The influence of school-and individual-level factors on academic achieve-
ment. Education and Urban Society, 40(2), 179204.
Xia, N. (2009). Family factors and student outcomes. Unpublished doctoral dissertation, Pardee RAND
Graduate School, Santa Monica, CA.
Yalcin, M., Aslan, S., & Usta, E. (2012). Analysis of PISA 2009 exam according to some variables.
Mevlana International Journal of Education, 2(1), 6471.
Yayan, B., & Berberoglu, G. (2004). A re-analysis of the TIMSS-1999 mathematics assessment data of
the Turkish students. Studies in Educational Evaluation, 30, 87104.
Ylmaz, H. B. (2009). Turkish students scientific literacy scores: A multilevel analysis of data from
program for international student assessment. Unpublished doctoral dissertation, The Ohio State
University, Columbus, OH.

Mustafa Sami Topcu is an Associate Professor of Science Education in the Department of Elementary
Education at Yldz Technical University, Istanbul, Turkey. He received a B.S. in science education from
the 19 Mays University, Turkey, an M.Sc. in science education from the 9 Eylul University, Turkey, and
a Ph.D. in science education from the Middle East Technical University, Turkey in 2008. Topcu teaches
courses for an undergraduate science education program, a masters and Ph.D. level elementary science
education programs. His research interests are socioscientific issues, argumentation, teachers beliefs and
educational practices, morality and international student exams such as PISA and TIMSS.

Serkan Arkan is an Assistant Professor of Mathematics Education in the Department of Elementary


Education at Mugla Stk Kocman University, Mugla, Turkey. He received a B.S. in mathematics
education from Bogazici University, Turkey, an M.A. in quantitative research and measurement and
evaluation in education from the Ohio State University, USA, and a Ph.D. in mathematics education from
Middle East Technical University, Turkey. Arkan teaches courses for an undergraduate mathematics
education program related to Statistics. Arkan worked as a researcher for National Testing Organization,
OSYM, and works also as a science committee member for Cito Turkey. His research interests are
measurement and evaluation, test development with IRT, international student exams such as PISA and
TIMSS.

123
132 M. S. Topcu et al.

Evrim Erbilgin is an Assistant Professor of Mathematics Education in the Department of Elementary


Education at Mugla Stk Kocman University, Mugla, Turkey. She completed her masters and Ph.D.
degrees in mathematics education at Florida State University in 2008. Erbilgin teaches mathematics
education courses for the elementary and middle school mathematics education program at Mugla Stk
Kocman University. Her research interests include supervision of mathematics student teachers,
mathematics teacher education, and international student exams such as PISA and TIMSS.

123

Vous aimerez peut-être aussi