Multilevel analysis of the relationship between school-level variables and student achievement

Abstract

The purpose of this study was to investigate the relationship between student achievement and a set of school-level variables, including distributed leadership, academic optimism, teacher collaboration and enabling school structure. The study was designed as correlational research. A Hierarchical Linear Modeling (HLM) analysis was conducted with a data set collected from 23,053 students and 426 teachers from 40 randomly selected public schools in Turkey. The data were collected using previously developed scales and student achievement data from the Ministry of National Education. HLM results revealed that two dimensions of academic optimism – namely collective efficacy and trust in clients – and hindering bureaucracy significantly predicted between-school differences in student achievement. The tested HLM model explained 60% of the variation in student achievement across schools. The results revealed that student achievement is shaped by school-level variables that are tied to the structural and functional characteristics of schools in Turkey. However, these school characteristics are rooted in the societal structures and cultural characteristics of the country. Hence, it is concluded that a reinterpretation of common school-level variables used to predict student achievement in the contexts of different countries is necessary.

Keywords

Academic optimism distributed leadership enabling school structure teacher collaboration school effectiveness

Introduction

The role of schools in student outcomes has always been a top concern of policy makers and educational researchers. Since the publication of the seminal Coleman Report (1966), educational researchers in different countries have tried to document the role of schools in affecting student outcomes (Reynolds et al., 2014). Despite the numerous studies conducted on school effects on student achievement, the core concerns of ‘What makes a “good” school?’ and ‘How do we make more schools “good”?’ are still not evident, especially in differing educational contexts. The literature still largely reflects the ‘Western’ contexts, although this has started to change lately (Bush, 2018; Hallinger, 2018).

The global trends around student achievement measurement programmes such as the Programme for International Student Assessment (PISA) and Trends in International Mathematics and Science Study (TIMSS) have raised concerns over the role of school-level variables in student achievement. In general, the tendency was to imitate the high-performing countries in the way they design their schooling systems. The rush to change school systems for better student outcomes resulted in borrowing policies and practices without adequately assessing the contextual relevance of these variables. For example, several scholars suggest a clear effect of leadership on student outcomes (Leithwood et al., 2008; Sammons et al., 2014), while several others suggest a very limited effect of leadership on student achievement (e.g., Kyriakides et al., 2010). The case for leadership can be extended to other school-level variables. Likewise, the role of formal structures on student outcomes has not been evidenced enough. The literature provides controversial evidence on the role of formalization regarding student outcomes. Various studies argue about the negative impact of processes and humans in organizations while other sources report on the role of formalization on positive work behaviours (Michaels et al., 1988). New conceptualisation on formalisation in schools gives new perspectives, which provide an opportunity to test the new perspective on student outcomes. Finally, academic optimism, as a construct depicting an important set of school-level variables related to school climate (collective efficacy, collective trust and academic emphasis), is a variable completing a holistic picture on school-level variables on student outcomes. These controversial findings across the contexts of different countries suggest that the literature may not consider the unique structural characteristics of educational organizations in different country contexts, and the political, social and cultural realities surrounding them. In that sense, there is limited research on the effectiveness of schools operating under the unique context of Turkey (Kondakci and Sivri, 2014). The following two sections provide information on the Turkish Education System and the need for school effectiveness research in Turkey.

The context: Turkish education system (TES)

Improving the quality of schools has been the key concern governing the reform movements in Turkey. Turkey has been experiencing a high demand for the schooling of its young population. Currently there are 16.4 million students in the primary and secondary schools of Turkey (Turkish Statistical Institute, 2017). Over the last two decades, Turkey has managed to improve its education system quantitatively (author). Consequently, key quantitative indicators such as schooling rate, number of teachers and number of classrooms have steadily increased (Turkish Statistical Institute, 2017).

Despite these quantitative gains, the TES has been suffering from quality and effectiveness problems (World Bank, 2011); particularly the effectiveness issue pushes governments to conceive and implement further reforms to the system. Over the previous 12 years, Turkish government institutions have initiated several major reforms to the system to deal with the effectiveness issue. Adopting the constructivist programme policy in 2005, the technology integration project in 2012, restructuring the levels of education in 2012, altering the selection system for principals in 2013, redefining the transitions between levels of education, and pursuing ongoing change in the content of instructional programmes are just some of the large-scale reform initiatives, which were motivated by the desire to improve the effectiveness of the educational system at large.

However, although Turkey has invested extensive monetary and human resources in large and small-scale reform initiatives, these reforms have not proven to be widespread solutions to the problem of low performance in student outcomes. Indeed, the TES has been identified as one of the world’s underperforming educational systems (Kondakci and Sivri, 2014), and international assessment programmes such as PISA and TIMSS have consistently shown the underperformance of Turkish students since their establishment.

Hence, the question of why increased investments have not led to quality improvement makes Turkey an interesting case for documenting the significance of school effectiveness in education. Typical to a developing country context, educational reforms in Turkey target major structural issues (i.e. changing levels of education) and ignore issues of individual schools. According to Aksit (2006), over-concentration on system while ignoring school-level implementation is one reason contributing to low school effectiveness. As a result, identifying the role of the school in student outcomes and defining school-level variables that contribute to positive student outcomes is a core concern of the TES (Kondakci and Sivri, 2014).

School effectiveness research (SER) in Turkey

The School Effectiveness Research (SER) movement has explored different variables in educational contexts and looked at their effect on various outcomes such as students’ academic outcomes or social output variables. These educational context variables included, for example, several factors such as leadership, climate, formal structures, social collaboration, curriculum or the differential levels in educational contexts such as districts and schools (Reynolds, 2010; Teddlie and Reynolds, 2000).

Although SER started to popularize around the world in the 1960s, studies within this paradigm have begun to be published in Turkey only since 1990s and gained momentum in 2000s (Polatcan and Cansoy, 2018). The work of Balcı (1993) and Şişman (1996) are considered to be one of the first scholarly studies on school effectiveness in Turkey. The foci of SER conducted in Turkey are affected by the contemporary issues of the period in which the studies are conducted. Accordingly, while early studies focused on family socioeconomic status, subsequent studies shifted their focus on topics such as leadership, achievement, school culture and climate (Polatcan and Cansoy, 2018).

A recently published review on SER in Turkey reported that these studies are mostly quantitative and associational in nature (Polatcan and Cansoy, 2018). The review maintains that most of the SER conducted in Turkey include data collected from teachers and administrators but tend to overlook inputs from students. More specifically, the authors contend that these studies have rarely incorporated student achievement data. Further, Polatcan and Cansoy (2018) assert that studies conducted in Turkey within the SER paradigm looked at the school effectiveness phenomenon with a limited number of variables. Additionally, it is very hard to spot SER conducted in Turkey that used a multilevel methodology. Using a multilevel methodology is important for educational research, especially for those studies that look at school-level effects on student achievement, as schools have stakeholders both inside and outside of the organization and it is crucial to distinguish between their effects (Hox et al., 2018).

Considering the need for SER for TES and the gaps in the research literature, especially among those conducted in the Turkish context, this study aimed at investigating the contribution of a set of purposefully selected school-level variables to student achievement within the context of Turkey. The specific objectives of this study are to investigate the student achievement variation between schools and how this variation is related to the distributed leadership, academic optimism teacher collaboration and enabling school structure in public schools in Turkey. Specific hypotheses based on these objectives are stated in each section covering our discussions on the variables of the study.

Theoretical framework

To advance our knowledge regarding the contributions of school-level variables to student achievement in the Turkish context, this study adopted the SER paradigm as a guide for its design and interpretation of its findings. Studies within this paradigm have explored various variables in educational contexts and investigated their effect on learning outcomes, which may be academic output variables or various social output variables. These educational context variables included, for example, various factors such as leadership, climate, formal structures, social collaboration, curriculum or the differential levels in educational contexts such as districts and schools (Reynolds, 2010; Teddlie and Reynolds, 2000).

Parallel with this understanding, this study aimed at capturing the relationship between key organizational aspects of schools and student achievement. Building on the literature, the study included distributed leadership as a concept to capture the leadership aspect, teacher collaboration as a measure to capture the informal collaboration, enabling school structure as a concept to capture the formal/bureaucratic structure and academic optimism as a measure to capture the climate in schools as organizations. The main proposition of the study is that these variables are key to establishing a conducive school atmosphere, which in turn impacts student outcomes. Based on this broad aim, this study answered the following research question: What is the relationship between a set of school-level variables (i.e. distributed leadership, academic optimism, teacher collaboration and enabling school structure) and student achievement among primary and middle school students in Turkey? Considering the research question, we provided below a detailed discussion on why we included each of these variables and their relationship with the student outcomes. At the end of the discussion on each variable, we developed a specific research hypothesis to test in the study.

Distributed leadership

Leadership has been shown to be the second most powerful factor explaining the differences between schools in student learning (Sammons et al., 2014). It is proposed that this relationship between student achievement and leadership is indirect in nature (Hallinger, 2011). This indirect relationship could work through mechanisms such as culture, work processes and people (Hallinger, 2011), professional climate and focused instruction in the US context (Seashore Louis et al., 2010), and academic climate (De Maeyer et al., 2007) in the Belgian context. Conversely, some studies reveal a weak or non-existent relationship between leadership and student outcomes, such as the study conducted by Krüger et al. (2007) in the Netherlands, and the meta-analysis by Kyriakides et al. (2010).

Furthermore, most of the studies on leadership have focused exclusively on people in leadership positions and have tended to overlook those who exercise leadership without a traditional leadership position (Diamond and Spillane, 2016). Accordingly, we adopted a distributed perspective on leadership as proposed by Spillane (2006) for this study. According to Spillane’s (2006) understanding, leadership in schools can be understood best by seeing it as a distributed phenomenon; it is a practice that is stretched over the school as an organization. The literature reveals that distributed leadership may have strong relations with school performance and outcomes (Hallinger and Heck, 2010; Harris, 2004). However, it is possible to locate skeptical positions regarding this relationship in the literature too (e.g. Lumby, 2013; Tian et al., 2016). Considering the general position in the literature regarding leadership’s impact on student learning, the following is hypothesized:

H₁: Distributed leadership contributes to between-school variation in student achievement.

Academic optimism

Academic optimism has consistently been shown to have an impact on student achievement in various contexts (Hoy, 2012). Furthermore, it was shown to work together with distributed leadership (Malloy, 2012) and enabling school structure (ESS) (McGuigan and Hoy, 2006; Wu et al., 2013) in contributing to student achievement. Academic optimism from Hoy’s (2012) perspective is a latent variable with three dimensions: academic emphasis, trust in clients and collective efficacy. Academic emphasis is the extent to which a school emphasizes student academic achievement. Trust in clients, on the other hand, is related to the level of the faculty’s trust in parents and students. Moreover, collective efficacy is the faculty’s belief that they as a whole can make a positive difference in student achievement (Hoy, 2012).

Research has shown that academic optimism has an impact on student learning regardless of student socioeconomic background (e.g. Bevel and Mitchell, 2012; Smith and Hoy, 2007). Furthermore, academic optimism was found to be related to other variables, including ESS in both the US and Taiwan (McGuigan and Hoy, 2006; Wu et al., 2013), collective responsibility in Taiwan (Wu et al., 2013), community engagement in the US (Kirby and DiPaola, 2011) and distributed leadership in Canada (Malloy, 2012).

A set of studies has been conducted on academic optimism in the Turkish context. Çağlar (2014) found that academic optimism predicts openness to change. Çoban and Demirtaş (2011) revealed a positive and moderate relationship between schools’ academic optimism and organizational commitment. Also, Kılınç (2013) observed a statistically significant positive relationship between teacher academic optimism and supportive, directive and intimate school climates. Moreover, Özdemir and Kılınç’s (2014) study revealed a significant and positive association between ESS and teacher academic optimism in the schools of Turkey. Having said that, no study investigating the relation between optimism and student achievement in the Turkish context could be located. Considering the literature outlined here, the following hypotheses are proposed:

H₂: Academic emphasis contributes to the between-school variation in student achievement.

H₃: Trust in clients contributes to the between-school variation in student achievement.

H₄: Collective efficacy contributes to the between-school variation in student achievement.

Teacher collaboration

Teacher collaboration was chosen to measure the level of informal collaboration among the faculty for the purposes of improving instruction processes, and thus improving student academic achievement. This argument is based on the understanding proposed in the influential work of Goddard et al. (2007), who state that ‘when teachers have opportunities to engage in professional discourse, they can build upon their unique content, pedagogical, and experiential knowledge to improve instruction’ (p. 880). Also, teacher collaboration and distributed leadership are interconnected variables that can be asserted to work together (Bush, 2011).

Studies exploring the relationship between student achievement and teacher collaboration are not abundant in the literature, although a few can be spotted. To start with, teacher collaboration seems to work well in the US context. Goddard et al. (2007) found that teacher collaboration significantly predicted differences in student achievement among schools. Additionally, Tschannen-Moran et al. (2000) concluded that collaboration in schools fosters organizational learning. Nevertheless, teacher collaboration has been shown to have its own challenges. To illustrate, it has been pointed out that egalitarian assumptions between teachers may hinder teachers from sharing their expertise, and ‘top-down’ approaches to leadership may hinder teacher collaboration as they limit teachers from taking initiatives (Katzenmayer and Moller, 2001).

Few studies have been conducted about teacher social interaction in the Turkish context. Beycioğlu and Aslan (2012) found that administrators and faculty members believed that there is collaboration in the primary schools of Turkey, but they also believed that the existing collaboration is not adequate. Furthermore, Aslan’s (2015) comparative study on the teaching profession in Turkey and South Korea using TALIS 2008 data indicated that teachers in South Korea were far better than teachers in Turkey in terms of professional collaboration. That being said, no study investigating the relationship between teacher collaboration and a learning outcome variable could be found in the literature on Turkey. Given the unique features of the Turkish schooling system, this can be accepted as a gap that awaits fulfilling. Based on this discussion, the following is hypothesized:

H₅: Teacher collaboration contributes to the between-school variation in student achievement.

Enabling school structure (ESS)

ESS was included in the study as there has been evidence showing its relation to student achievement in a few country contexts (e.g. McGuigan and Hoy, 2006; Wu et al., 2013). Also, it was specifically chosen to be studied along with teacher collaboration since the literature does not adequately cover studies that take both informal and bureaucratic aspects of schools into account when explaining student achievement (Freiberg, 1999).

Schools are bureaucratic structures; they have hierarchy of authority, division of labour, and rules and regulations (Hoy and Sweetland, 2001). While the bureaucratic structure of schools is generally seen as a negative phenomenon that affects school improvement (Murphy, 2013), there are some authors that take a more neutral stance on this issue. Hoy and Sweetland (2001), for example, do not ignore this hindering side of bureaucratic structures; however, they assert that school bureaucracies may also be enabling in nature. Accordingly, they proposed a framework for bureaucracy in their article arguing that it can be enabling or hindering in nature.

Enabling bureaucracy from this perspective was found to be positively related to trust in principals and negatively related to truth spinning and role conflict (Hoy and Sweetland, 2001), predicting organizational citizenship (Messick, 2012), predicting student achievement in a direct manner (Tarter and Hoy, 2004) and having an indirect relationship with student achievement through academic optimism (McGuigan and Hoy, 2006) in the US context. Also, its relation to student achievement through academic optimism was replicated in Taiwan (Wu et al., 2013). A handful of studies have been conducted on ESS in Turkey. To illustrate, enabling bureaucracy was shown to significantly predict transformational leadership (Buluç, 2009) and academic optimism (Özdemir and Kılınç, 2014). In addition, according to Cerit (2012), the bureaucratic structure of schools significantly predicts the professionalism of classroom teachers. Briefly, ESS has been found to be related to variables shown to be influential on student achievement. However, studies investigating its relation to student achievement directly are hard to come across in the literature and, to our knowledge, are non-existent among studies conducted in the Turkish context. Hence, the following hypotheses are proposed:

H₆: Enabling bureaucracy positively contributes to the between-school variation in student achievement.

H₇: Hindering bureaucracy negatively contributes to the between-school variation in student achievement.

Methods

Research design

This study was designed as correlational research as it investigated the relationship between two sets of quantitative variables (Fraenkel et al., 2012). It is designed and implemented on the basis of quantitative research traditions. Accordingly, in terms of epistemology, the study strives to ensure the objectivity of the researchers while acknowledges that dualism is not possible (Guba and Lincoln, 2004).

Hierarchical Linear Modeling (HLM) was chosen as the analytical technique as it provides several advantages in educational research. The most important advantages include having the ability to account for the nested structure of educational data, and the ability to distinguish between- and within-school variation (Raudenbush and Bryk, 2002).

Sampling and data collection

The data for the study was collected from Adana, Turkey. A list of all 1642 schools in Adana Province was obtained from the official website of the Provincial Directorate of National Education. Primary and middle schools were randomly selected from the districts of Çukurova, Seyhan and Sarıçam in Adana. High schools were not included in the sample to control for school structure. High schools in Turkey are differentiated based on the education they provide, including science, religious vocational and sports high schools. Thus, the grade-point averages (GPA) obtained from them may have differing implications. Eventually, the sample consisted of 23,053 students and 426 teachers from 40 public schools.

Utilizing HLM as the analytical technique, this study had two participant levels. Student participants were included at the first level of the study. The anonymous student achievement data were collected with the help of principals through the centralized computer system of the Ministry of National Education. This way, the achievement data for all the students enroled in participating schools were obtained. Furthermore, schools were the unit of analysis for the second level of the study. The school-level data were obtained through questionnaires completed by teachers working at the participating schools. The researchers visited the teachers’ lounge in each participating school and asked the available teachers to participate in the study by filling in the questionnaires. These questionnaires were grouped according to the school they were collected from, and the aggregate values obtained for each variable constituted the school-level variables. The teachers who filled in these questionnaires were mostly female (63.8%). Teacher participants’ mean age was 39 years (SD = 8.03) and their mean teaching experience was 15.72 years (SD = 7.93).

Measures

All the independent variables of the study were Level 2 variables; that is, the school was taken as the unit of analysis for these variables. Four previously developed scales were utilized to collect the data for the independent variables.

Distributed leadership scale

The scale was developed by Özer and Beycioğlu (2013) with ten 5-point Likert-type items and a single dimension. Özer and Beycioğlu (2013) reported the test-retest reliability score of the scale to be .82. Moreover, the scale’s Cronbach’s alpha reliability coefficient was calculated as .94 for the present study.

Schools’ academic optimism scale

Adapted to Turkish by Çoban and Demirtaş (2011) from the original scale developed by Hoy et al. (2006), this scale was used to measure the academic optimism perceptions of teachers. Eleven items were excluded from the Turkish version of the scale since they either had low loading values or their content was explained by other items. The end result was a 19-item scale composed of 5-point Likert-type items. Cronbach reliability scores were calculated for the present study to be .71, .85 and .89 for collective efficacy, faculty trust in clients and academic emphasis, respectively.

Teacher collaboration sub-scale of teacher leadership culture scale

Developed by Demir (2014), this scale has three dimensions: teacher collaboration, managerial support and work environment. For the purposes of this study, however, only the teacher collaboration sub-scale was used to measure teacher perceptions of collaboration among teachers in their school. The Cronbach’s alpha value of this sub-scale was calculated to be .91 for the present study.

Enabling school structure scale (ESSS)

This scale was originally constructed by Hoy and Sweetland (2001) and was adapted to Turkish by Özer and Dönmez (2013). In contrast to its original form, the Turkish form of the ESSS is a two-factor scale: enabling bureaucracy and coercive bureaucracy. The adaptation of the ESSS comprised 5-point Likert-type scales similar to its original form. Cronbach’s alpha values of the scale were calculated to be .92 and .89 for enabling and hindering bureaucracy, respectively.

Grade-point average (GPA) scores

This was the dependent variable of the study. It was a Level 1 variable; that is, it was not aggregated to any level. It is the individual scores that fourth graders from primary schools – since grades 1–3 are not scored – and all the middle school students had accumulated over the semester from all their school courses. Both primary school and middle school courses included mathematics, Turkish language, English as foreign language, sciences, religion and ethics education, music, physical education and visual arts. Both sectors can be asserted to be comparable since the difference in compulsory courses was in the inclusion of a technology course and the omission of two courses – namely, traffic security, and human rights and citizenship – once students pass to middle school from primary school. GPA scores range from 0 to 100 and are calculated in such a way that the courses with most hours in the weekly programme impact the weighted score most. These student GPA scores were automatically linked to their schools by the HLM7 software via the school IDs they were encoded with.

Data analysis

This study included two levels of data, Level 1 being individual student GPA and Level 2 being derived from teacher-completed questionnaires that are aggregated to school level. Because of this, an HLM analysis was conducted using HLM7 software (Raudenbush and Bryk, 2002). Also, to check the health of the data, bivariate correlations were calculated, and missing value and measurement model analyses were conducted before the HLM analyses.

A totally unconditional model was run using HLM7 software before the main analysis, as suggested by Raudenbush and Bryk (2002). It is a model with no predictors at either level. This step was crucial to distinguish within and between variation components, and to come up with preliminary ideas about where the main variation occurs.

Then, the main analysis was conducted using HLM7 software. In the main analysis, student GPA values were set as the outcome variable in Level 1. The school-level independent variables were distributed leadership, dimensions of academic optimism, teacher collaboration, dimensions of ESS. No independent variable was added to the student level as the study aimed at explaining variation across schools.

Results

Descriptive statistics

The descriptive statistics of the variables included in the main analysis are provided in Table 1. As can be seen, 40 schools participated in the main analysis. All the variables under the school-level section denote school means. All the scales used for school level were Likert-type scales that ranged from 1 to 5. Further, there were 23,053 student GPA scores at Level 1, and they ranged from 0 to 100. Also, in order to check the correlations between variables and examine multicollinearity, bivariate correlations were calculated. The results indicated that no correlation value exceeded .70, which is well below the critical value of .90 suggested by Field (2005). Bivariate correlations are summarized in Table 1.

Table 1.

Descriptive statistics and bivariate correlations.

Variable	α	M	SD	MIN	MAX	1	2	3	4	5	6	7
School Level (N = 40)
1.Distributed leadership	.94	4.02	0.66	2.10	4.80	1	.294**	.196**	.133**	.305**	.684**	−.445**
2.Academic emphasis	.89	3.47	0.51	2.42	4.89		1	.683**	.136**	.365**	.329**	−.165**
3.Trust in clients	.85	3.22	0.45	2.20	4.22			1	.086	.350**	.217**	−.112^*
4.Collective efficacy	.71	4.44	0.28	3.60	5.00				1	.267**	.140**	−.390**
5.Teacher collaboration	.91	4.13	0.29	3.47	4.68					1	.309**	−.111^*
6.Enabling bureaucracy	.92	3.66	0.48	2.32	4.46						1	−.472**
7.Hindering bureaucracy	.89	4.00	0.48	2.67	4.71							1
p < .05. *p < .01.
Student Level (N = 23053)
Student Achievement		73.7	16.04	0	100

Missing value analysis

Little’s MCAR test results revealed that all the variables except two yielded insignificant results. Subsequently, to check if a certain group of participants refused to answer these two scales, t-tests were run by using all the categorical variables in the dataset as grouping variables, namely gender and schooling level. The t-tests yielded insignificant results. Thus, it was concluded that the missingness in these variables was not the result of a bias by a certain group. Further, there were no missing data at the student level. Thus, no such analysis was conducted for the student level.

Measurement model results

To investigate the health of the relationship among the latent variables and their indicators, a seven-factor measurement model analysis was conducted (Byrne, 2010). AMOS 22 was used to conduct the analysis. The initial results indicated close values for acceptable fit, but some improvements were necessary. Accordingly, with modification indices and estimates provided by AMOS, two items of the distributed leadership scale (items number 6 and 9) were eliminated with the permission of the scale developers. The final results of the measurement model analysis indicated improved and acceptable fit values, x ²(1013) = 2299,631, p < .05, x²/df = 2.27, RMSEA = .051 (90% CI = .048 -.054, pclose = .30), CFI = .91, TLI = .90.

HLM analysis results and hypothesis testing

Before the main HLM analysis, a fully unconditional model was produced. Results indicated a reliability score of .98 for the intercept values of the unconditional model. As expected, student achievement was found to vary both between and within schools significantly. The great proportion of the variation in student achievement occurred within schools. This result is compatible with what was expected, and with the related literature (e.g. Walker et al., 2014). The intraclass correlation was 16.75% and, more importantly, the chi-square test results showed a significant non-zero score for between-school variation (x ²(39, N = 40) = 3788.27, p < .001). Hence, we moved on to the proposed model to explain this significant non-zero score and test the proposed hypotheses. Further details about the unconditional model are summarized in Table 2.

Table 2.

HLM unconditional model results.

Variable	Variation
Intercept	73.09 (1.05)
Between-school parameter variance	44.57^a
Within-school parameter variance	221.53
HLM reliability estimate	0.98
Proportion of variance between schools	0,17

N = 23053 students from 40 schools. Standard Error is displayed in parentheses.

^a x ²(39, N = 40) = 3788.27, p < .001.

The proposed hypotheses were tested through HLM analysis. H₁ and H₂ proposed that distributed leadership and academic emphasis would significantly contribute to student achievement variation between schools, respectively. However, the HLM results did not support these hypotheses (p = .44 for H₁; p = .09 for H₂). Furthermore, H₃ and H₄ proposed that trust in clients and collective efficacy would significantly contribute to student achievement differences between schools. The HLM analysis supported these hypotheses in that a one-unit increase in trust in clients and collective efficacy would predict 5.44- and 8.16-point increases in student GPA scores, respectively (p < .05 for H₃; p < .01 for H₄).

Nevertheless, H₅ and H₆, which proposed positive relationships between across-school variation in student achievement, and teacher collaboration and enabling bureaucracy, were not supported by the HLM results (p = .95 for H₅; p = .40 for H₆). Contrary to what H₇ proposed, the HLM results showed that a one-unit increase in hindering bureaucracy would predict a 5.99-point increase in student achievement (p < .05). The original H₇ proposed a negative relationship between the two. More details on the relationship between these variables are provided in Table 3.

Table 3.

Proposed variables as the predictors of variation in student achievement between schools (with robust standard errors).

Variable	Coefficient	SE	T ratio	p
Intercept	73.06	0.61	119.45	<.001
Distributed leadership (H1)	−2.02	2.58	−0.78	0.44
Academic emphasis (H2)	4.58	2.64	1.73	0.09
Trust in clients (H3)	5.44	2.20	2.48	<.05
Collective efficacy (H₄)	8.16	2.73	2.99	<.01
Teacher collaboration (H5)	−0.16	2.36	−0.07	0.95
Enabling bureaucracy (H6)	3.45	4.08	0.85	0.40
Hindering bureaucracy (H₇)	5.99	2.77	−2.17	<.05

N = 23053 students from 40 schools.

Additionally, the HLM results revealed that the tested model explained 60% of between-school variation, and thus decreased the proportion of variation in student achievement between schools from 16.75% to 7.44%. In addition, as expected, no change in the within-school variation of student achievement was observed. As a side note, the remaining between-school variation for student achievement was still statistically non-zero, x ²(32, N = 40) =1698.59, p < .001. In other words, although the proposed model explained the majority of between-school variance in student achievement, there are still some other variables that may explain across-school differences in student achievement. More detail on student achievement variation is provided in Table 4.

Table 4.

HLM proposed model results: Variation between schools in student achievement.

Variable	Variation
Intercept	73.06 (061)
Between-school parameter variance	17.80^a
Within-school parameter variance	221.53
HLM reliability estimate	0.95
Proportion of variance between schools	0.07
Proportion of between-school variability explained by the model^b	0,60^b

N = 23053 students from 40 schools. Standard Error is displayed in parentheses.

^a x ²(32, N = 40) = 1698.59, p < .001.

^bCalculated as the reduction in between-school parameter variance reported in Table 1.

Discussion

This study analyzed the effects of key school-level factors on student achievement. Building on the previous scholarly work, this study tested a holistic model incorporating distributed leadership as a concept to capture the leadership aspect, teacher collaboration as a measure to capture the informal collaboration, ESS as a concept to capture the formal/bureaucratic structure, and academic optimism as a measure to capture the climate in schools as organizations. The main findings of study estimations are the following.

Firstly, analysis results justified this study’s focus on the role of school-level factors on individual student achievement scores as significant differences among schools was observed. According to the analysis results, 16.75% of the student achievement variation occurred between schools. This finding shows that a fundamental assertion of highly centralized schooling systems such as Turkey’s that every student benefits from schooling services in an equal manner (Kondakci et al., 2016) may not altogether hold true, at least not for the Turkish context. Indeed, several previous analyses depict the huge difference in student achievement from different school types. For example, several analyses on PISA results suggest that although the overall performance of the country is very low, a tiny group of students in Turkey are able to perform even higher than the best performing countries. This performance difference is associated with school type or school-level variables. In other words, Turkey possesses the means to cultivate high student achievement but simply cannot distribute this achievement throughout the system (Berberoğlu, nd.). On the other hand, the tested HLM model explained 60% of this between-school variation in student achievement. As a result, among the total variation, the unexplained variation of between-school differences in student achievement decreased to a value of 7.44%. Thus, the findings of this study have important implications about ‘where to look’ to decrease school-level differences in student achievement, and to promote quality education for all in schooling systems such as Turkey’s.

Moreover, the findings for the dimensions of academic optimism were mainly congruent with the literature from different cultural contexts. All three dimensions had positive associations with between-school differences in student achievement, with two of them being significant. Collective efficacy was the strongest predictor of student achievement (H₄). A one-score increase in collective efficacy meant an 8.16-point increase in student GPA scores (p < .01). Furthermore, trust in clients played a significant role in achievement attainments (H₃). A one-score increase in trust in clients meant a 5.44-point increase in student GPA scores (p < .05).

The finding regarding collective efficacy can be related to the concept of the collectivistic society (Hofstede et al., 2010). People in collectivistic societies emphasize interpersonal relationships and define their self-image more in terms of ‘we’ than ‘I.’ In that sense, in a collectivist culture, key constituencies of the schools voluntarily share their capacity with other constituencies. Regarding trust in clients, it can be asserted that school-family relationship might be an important factor for improved student achievement results in the public schools of Turkey. Besides, Turkey’s hierarchical societal structures (Esmer, 1997) and the roles attributed to school in society may influence these findings. Historically, the citizenship development function of the schools has been prominent in Turkish society, facilitating close ties between families and schools. These close ties breed a positive image of school in Turkish society (Demir, 2007).

Nevertheless, although positive, the relationship between academic emphasis and student achievement differences across schools was not significant (p = .09) (H₂). One explanation for this could be the inclusion of fourth graders from primary schools in the analysis. It is possible that fourth graders may not be as academically oriented as middle-school students. We decided to include fourth graders from primary schools in the analysis since the GPAs obtained from them do not have differing implications like that of high schools, which we left out; this is further explained under methods section.

The concept of academic optimism was first developed in the US and it might not have the same effect on student achievement in different contexts. Accordingly, the finding that H₂ was not supported could be accepted as an indication of a possible need for refinements in the concept of academic optimism, especially for differing contexts such as Turkey. Also, to our knowledge, this study was the first to explore the relationship between the dimensions of academic optimism and student achievement in Turkey. Hence, more study is necessary to refine the results.

Another finding was that distributed leadership did not have a significant direct effect on student achievement (p = .44) (H₁). There could be several reasons for this, including the nature of leadership practice. Hallinger (2011) pointed out that leadership has an indirect effect on student achievement through different organizational features. In this sense, the results of this study can be considered as compatible with the literature in that the leadership variable was not found to be directly related to student achievement. Also, these findings support the skeptical positions regarding distributed leadership’s relationship with school performance and outcomes in the literature (e.g. Lumby, 2013; Tian et al., 2016).

Another explanation could be a preoccupation with the student performance indicators that are generally used in the performance tables. To illustrate, a leader may promote deep learning; however, this may not be reflected in the short-term exam results. Thus, the actual effectiveness of leadership may not be seen on a radar that detects only the student achievement variables that are used on performance tables (Barker, 2007).

Looking through a comparative perspective, distributed leadership in Turkey seems to show similar features with studies conducted in differing Western contexts such as Canada and Belgium (e.g. Hulpia et al., 2011) in terms of having important relationships with various organizational variables in schools, as it had significant bivariate correlations with all the other school-level variables included in the study in expected directions. However, the findings do not support the assertions about distributed leadership’s direct relationship with school outcomes (e.g. Harris and Spillane, 2008) in the Turkish context.

As for bureaucratic structure, the study revealed that hindering bureaucracy was significantly (H₇), and enabling bureaucracy was insignificantly (H₆) related to student achievement. Based on this finding, it can be argued that the existence of hindering bureaucracy is more effective on student achievement than the existence of enabling bureaucracy. This finding is in line with Baumeister, Bratslavsky, Finkenauer, and Vohs’s (2001) argument that negativity is stronger than positivity. According to the authors, even if they have the same intensity, negative situations/emotions are stronger than positive ones.

An interesting finding about hindering bureaucracy was that it had a positive relationship with student achievement differences between schools (H₇). Every point increase in hindering bureaucracy meant an increase in student GPA scores by 5.99 (p < .05). An important explanation for this finding could be the role of cultural differences. It would not be reasonable to expect the findings to be consistent all over the globe, especially when Hoy and Sweetland (2001) themselves accepted the culturally bound nature of ESS. Thus, the results should be considered as a mirror reflecting the cultural context of Turkey. Moreover, the findings of the present study are the result of a relationship analysis. Although the results claim a predictive relationship between hindering bureaucracy and student achievement, no claims are made about causation. Moreover, to our knowledge, this study is the first to examine the direct relationship between ESS and student achievement. Clearly, more research is necessary to gain further insights.

No significant direct link between teacher collaboration and student achievement differences was observed between schools (H₅). This finding signals that teacher collaboration may not be as significant to student achievement in the Turkish context as it is shown to be in the US (e.g. Goddard et al., 2007). One explanation for why teacher collaboration was not found to significantly relate to student achievement could be that collaboration may be the opposite of being effective if the staff does not have the necessary skills to communicate in an effective way, or if it creates conflict. This, in turn, may lead to draining teachers’ energy for non-instructional purposes (Marks and Louis, 1997).

Conclusion and implications

This study investigated the relationship between a set of potentially effective school-level variables and student achievement differences between schools. Based on its scope and methodological approach, this study presents an incremental advance in the state of the art of research on SER in the Turkish context and thus, has important implications.

Firstly, the findings revealed that teacher beliefs play a role in improving student academic achievement. This is a crucial finding with respect to centralized education systems in that teachers need more decision latitude over their own practices to cultivate positive beliefs over their roles in student achievement. In this regard, more attention to teacher beliefs – such as collective efficacy and trust in clients – is necessary.

Secondly, findings indicated that schools’ relationship with students and parents plays a significant role in improving student achievement and has the potential to decrease the student achievement inequalities among public schools in Turkey. As also supported by Colemans’ seminal work (1988) and recent SER conducted in Turkish context (Polatcan and Cansoy, 2018), schools should focus on promoting closer ties with their surroundings, especially with key stakeholders such as parents and students. Organizing activities that involve the surrounding community, especially parents and students, may play a role in helping with this endeavour (Çubukçu and Girmen, 2006).

Study results revealed that teacher collaboration does not have positive effects on student achievement within the current context. This could be related to Marks and Louis’ (1997) proposition that lack of communication skills or differing goals among teachers may cause the efforts of collaboration to have opposite effects on student achievement. Based on this, communication skills of both teachers and school leadership should be emphasized more in on-the-job training activities, master’s education opportunities and professional development seminars so that collaboration efforts can lead to improved student achievement results, as also suggested by Polatcan and Cansoy (2018). School leadership could play a more active role in supporting healthy ways of communication among faculty. In addition, differing goals or conflicts among faculty could be alleviated by a clear stance and continued reminding by principals on the mission and vision of the organization.

Further, the findings draw attention to the contrasting effects of school-level variables such as ESS, distributed leadership and teacher collaboration on student achievement within different contexts. This finding can be interpreted as an invitation to revisit the validity of these concepts in differing country contexts such as Turkey. Given that the effects of school-level variables on student achievement have not been adequately studied in the context of Turkey (Polatcan and Cansoy, 2018), future research can benefit from this study as a starting contribution to the revision of these concepts.

Finally, as results of the study revealed, although a significant amount of variation occurred between schools, most of the variation in student achievement occurred within schools. This study focused on school-level factors that were shown to be effective on the between-school variation of learning outcomes in Turkey and it provided important implications. However, a great research potential lies in within-school variation of student achievement. Further studies should consider including other variables that may potentially explain within-school variation in learning outcomes.

Footnotes

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

ORCID iD

Yusuf Ikbal Oldac

References

Aksit

(2006) A comparative case study on school effectiveness characteristics of two private high schools in Ankara. PhD Thesis, Middle East Technical University, Turkey.

Aslan

(2015) A comparative study on the teaching profession in Turkey and South Korea: Secondary analysis of TALIS 2008 data in relation to teacher self-efficacy. Eurasian Journal of Educational Research 61: 1–22.

Balcı

(1993) Etkili okul ve okul geliştirme: Kuram, uygulama ve araştırma [School effectiveness and School improvement: Theory, practice and research]. Ankara: Yavuz Dağıtım.

Barker

(2007) The leadership paradox: Can school leaders transform student outcomes? School Effectiveness and School Improvement 18(1): 21–43.

Baumeister

Bratslavsky

Finkenauer

Vohs

(2001) Bad is stronger than good. Review of General Psychology 5(4): 323–370.

Berberoğlu

(undated) Türk bakış açısından PISA araştırma sonuçları [PISA research results from Turkish perspective]. Available at: http://www.konrad.org.tr/Egitimturk/07girayberberoglu.pdf (accessed 15 May 2017).

Bevel

Mitchell

(2012) The effects of academic optimism on elementary reading achievement. Journal of Educational Administration 50(6): 773–787.

Beycioğlu

Aslan

(2012) Öğretmen ve yöneticilerin öğretmen liderliğine ilişkin görüşleri: bir karma yöntem çalışması [Teachers’ and administrators’ views on teacher leadership: A mixed methods study]. Kuram ve Uygulamada Eğitim Yönetimi Dergisi 18(2): 191–223.

Buluç

(2009) İlköğretim okullarında bürokratik okul yapısı ile okul müdürlerinin liderlik stilleri arasındaki ilişki [The relationship between bureaucratic school structure in primary schools and school principals’ leadership style]. Eğitim ve Bilim 34(152): 71–86.

10.

Bush

(2011) Theories of Educational Leadership and Management, 4th ed. London: Sage.

11.

Bush

(2018) Research on educational leadership and management. Educational Management Administration & Leadership 46(3): 359–361.

12.

Byrne

(2010) Structural Equation Modeling with AMOS: Basic Concepts, Applications, and Programming, 2nd ed. New York: Routledge.

13.

Cerit

(2012) Okulun bürokratik yapısı ile sınıf öğretmenlerinin profesyonel davranışları arasındaki ilişki [The relationship between schools’ bureaucratic structure and their teachers’ professional behaviors]. Kuram ve Uygulamada Eğitim Yönetimi 18(4): 497–521.

14.

Coleman

(1988) Social capital in the creation of human capital. American Journal of Society 94: 95–120.

15.

Coleman

Campbell

Hobson

McPartland

Mood

Weinfeld

York

(1966) Equality of Educational Opportunity. Washington, DC: US Government Printing Office.

16.

Çağlar

(2014) The relationship between academic optimism levels of schools and their levels of openness to change. Kuramsal Eğitimbilim Dergisi 7(1): 94–113.

17.

Çoban

Demirtaş

(2011) Okulların akademik iyimserlik düzeyi ile öğretmenlerin örgütsel bağlılığı arasındaki ilişki [The relationship between the level of schools’ academic optimism and teachers’ organizational commitment]. Kuram ve Uygulamada Eğitim Yönetimi Dergisi 17(3): 317–348.

18.

Çubukçu

Girmen

(2006) Ortaöğretim kurumlarının etkili okul özelliklerine sahip olma düzeyleri [Middle school institutions’ levels of having effective school characteristics]. Sosyal Bilimler Dergisi 16: 121–136.

19.

De Maeyer

Rymenans

Van Petegem

van den Bergh

Rijlaarsdam

(2007) Educational leadership and pupil achievement: The choice of a valid conceptual model to test effects in school effectiveness research. School Effectiveness and School Improvement 18(2): 125–145.

20.

Demir

(2007) Metaphors as a reflection of middle school students’ perceptions of school: A cross-cultural analysis. Educational Research and Evaluation 13(2): 89–107.

21.

Demir

(2014) Öğretmen Liderliği Kültürü Ölçeği: Geçerlik ve güvenilirlik çalışması [Teacher Leadership Culture Scale: A study of validity and reliability]. Elementary Education Online 13(2): 334–344.

22.

Diamond

Spillane

(2016) School leadership and management from a distributed perspective: A 2016 retrospective and prospective. Management in Education 30(4): 147–154.

23.

Esmer

(1997) Devrim, evrim, statüko: Türkiye’de ekonomik, siyasal ve kültürel değerler [Revolution, Evolution, Status Quo: Economic, Political and Cultural Values in Turkey]. Istanbul: TESEV.

24.

Field

(2005) Discovering Statistics Using SPSS, 2nd ed. London: Sage Publications.

25.

Fraenkel

Wallen

Hyun

(2012) How to Design and Evaluate Research in Education, 8th ed. New York, NY: McGraw-Hill.

26.

Freiberg

(1999) School Climate: Measuring, Improving and Sustaining Healthy Learning Environments. Philadelphia: Psychology Press.

27.

Goddard

Tschannen-Moran

(2007) A Theoretical and empirical investigation of teacher collaboration for school improvement and student achievement in public elementary schools. Teachers College Record 109(4): 877–896.

28.

Guba

Lincoln

(2004) Competing paradigms in qualitative research: Theories and issues. In: Hesse-Biber

Leavy

(eds), Approaches to Qualitative Research: A Reader on Theory and Practice. New York, NY: Oxford University Press, pp. 17–38.

29.

Hallinger

(2011) Leadership for learning: Lessons from 40 years of empirical research. Journal of Educational Administration 49(2): 125–142.

30.

Hallinger

(2018) Surfacing a hidden literature: A systematic review of research on educational leadership and management in Africa. Educational Management Administration & Leadership 46(3): 362–384.

31.

Hallinger

Heck

(2010) Leadership for learning: Does collaborative leadership make a difference in school improvement? Educational Management Administration and Leadership 38(6): 654–678.

32.

Harris

(2004) Distributed leadership and school improvement: Leading or misleading? Educational Management Administration & Leadership 32(1), 11–24.

33.

Harris

Spillane

(2008) Distributed leadership through the looking glass. Management in Education 22(1): 31–34.

34.

Hofstede

Minkov

(2010) Cultures and Organizations: Software of the Mind, 3rd ed. London: McGraw-Hill.

35.

Hox

Moerbeek

van de Schoot

(2018) Multilevel Analysis: Techniques and Applications, 3rd ed. New York: Routledge.

36.

Hoy

(2012) School characteristics that make a difference for the achievement of all students: A 40-year odyssey. Journal of Educational Administration 50(1): 76–97.

37.

Hoy

Sweetland

(2001) Designing better schools: The meaning and measure of enabling school structures. Educational Administration Quarterly 37(3): 296–321.

38.

Hoy

Tarter

Hoy

(2006) Academic optimism of schools: A force for student achievement. American Educational Research Journal 43(3): 425–446.

39.

Hulpia

Devos

Van Keer

(2011) The relation between school leadership from a distributed perspective and teachers’ organizational commitment: Examining the source of the leadership function. Educational Administration Quarterly 47(5): 728–771.

40.

Katzenmayer

Moller

(2001) Awakening the Sleeping Giant. Helping Teachers Develop as Leaders, 2nd ed. Thousand Oaks, CA: Corwin Press.

41.

Kılınç

AÇ

(2013) The relationship between individual teacher academic optimism and school climate. International Online Journal of Educational Sciences 5(3): 621–634.

42.

Kirby

DiPaola

(2011) Academic optimism and community engagement in urban schools. Journal of Educational Administration 49(5): 542–562.

43.

Kondakci

Sivri

(2014) Salient characteristics of high-performing Turkish elementary schools. Journal of Educational Administration 52(2): 254–272.

44.

Kondakci

Zayim

Oldac

Senay

(2016, May) Türkiye’de okul müdürlerinin sosyal adalet rolleri [School Principals’ social justice roles in Turkey]. In: Paper presented at the 11th Ulusal Eğitim Yönetimi Kongresi, Izmir.

45.

Krüger

Witziers

Sleegers

(2007) The impact of school leadership on school level factors: Validation of a causal model. School Effectiveness & School Improvement 18(1): 1–20.

46.

Kyriakides

Creemers

BPM

Antoniou

Demetriou

(2010) A synthesis of studies searching for school factors: implications for theory and research. British Educational Research Journal 36(5): 807–830.

47.

Leithwood

Harris

Hopkins

(2008) Seven strong claims about successful school leadership. School Leadership & Management 28(1): 27–42.

48.

Lumby

(2013) Distributed Leadership. Educational Management Administration & Leadership 41(5): 581–597.

49.

Malloy

(2012) Effects of distributed leadership on teachers’ academic optimism and student achievement. PhD Thesis, University of Toronto, Canada.

50.

Marks

Louis

(1997) Does teacher empowerment affect the classroom? The implications of teacher empowerment for instructional practice and student academic performance. Educational Evaluation and Policy Analysis 19(3): 245–275.

51.

McGuigan

Hoy

(2006) Principal leadership: creating a culture of academic optimism to improve achievement for all students. Leadership and Policy in Schools 5(3): 203–229.

52.

Messick

(2012) Examining relationships among enabling school structures, academic optimism and organizational citizenship behaviors. PhD Thesis, Auburn University, USA.

53.

Michaels

Cron

Dubinsky

Joachimsthaler

(1988) Influence of formalization on the organizational commitment. Journal of Marketing Research 25(4): 376–383.

54.

Murphy

(2013) The architecture of school improvement. Journal of Educational Administration 51(3): 252–263.

55.

Özdemir

Kılınç

AÇ

(2014) The relationship between bureaucratic school structure and teachers’ level of academic optimism. Journal of Theory & Practice in Education 10(1): 1–23.

56.

Özer

Beycioğlu

(2013) Paylaşılan Liderlik Ölçeğinin geliştirilmesi: Geçerlik ve güvenirlik çalışmaları [The development of Distributed Leadership Scale: A study of validity and reliabilitiy]. İlköğretim Online 12(1): 77–87.

57.

Özer

Dönmez

(2013) Kolaylaştırıcı Okul Yapısı Ölçeğinin Türkçe formunun psikometrik özelliklerinin yeniden değerlendirilmesi [The reinvestigation of the psychometric properties of the Turkish form of Enabling School Structure Scale]. Pegem Eğitim ve Öğretim Dergisi 3(4): 57–68.

58.

Polatcan

Cansoy

(2018) Türkiye’de okul etkililiği araştırmaları: Ampirik araştırmaların analizi [School effectiveness research in Turkey: The analysis of empirical studies]. Sakarya University Journal of Education 8(3): 8–24.

59.

Raudenbush

Bryk

(2002) Hierarchical linear models: Applications and data analysis methods. Newbury Park, CA: Sage.

60.

Reynolds

(2010) Failure-Free Education? The Past, Present and Future of School Effectiveness and School Improvement. New York: Routledge.

61.

Reynolds

Sammons

De Fraine

Van Damme

Townsend

Teddlie

Stringfield

(2014) Educational effectiveness research (EER): A state-of-the-art review. School Effectiveness and School Improvement 25(2): 197–230.

62.

Sammons

Davis

Day

(2014) Using mixed methods to investigate school improvement and the role of leadership. Journal of Educational Administration 52(5): 565–589.

63.

Seashore Louis

Dretzke

Wahlstrom

(2010) How does leadership affect student achievement? Results from a national US survey. School Effectiveness and School Improvement 21(3): 315–336.

64.

Smith

Hoy

(2007) Academic optimism and student achievement in urban elementary schools. Journal of Educational Administration 45(5): 556–568.

65.

Spillane

(2006) Distributed Leadership. San Francisco: Jossey-Bass.

66.

Şişman

(1996) Etkili okul yönetimi [Effective school management]. Unpublished research report. Eskişehir.

67.

Tarter

Hoy

(2004) A systems approach to quality in elementary schools: A theoretical and empirical analysis. Journal of Educational Administration 42(5): 539–554.

68.

Teddlie

Reynolds

(2000) The International Handbook of School Effectiveness Research. London: Falmer Press.

69.

Tian

Risku

Collin

(2016) A meta-analysis of distributed leadership from 2002 to 2013. Educational Management Administration & Leadership 44(1): 146–164.

70.

Tschannen-Moran

Uline

Hoy

Mackley

(2000) Creating smarter schools through collaboration. Journal of Educational Administration 38(3): 247–272.

71.

Turkish Statistical Institute (2017) Basic statistics: Educational statistics. Available at: http://tuik.gov.tr/PreTablo.do?alt_id=1018 (accessed 26 May 2007).

72.

Walker

Lee

Bryant

(2014) How much of a difference do principals make? An analysis of between-schools variation in academic achievement in Hong Kong public secondary schools. School Effectiveness and School Improvement 25(4): 602–628.

73.

Hoy

Tarter

(2013) Enabling school structure, collective responsibility, and a culture of academic optimism. Journal of Educational Administration 51(2): 176–193.

74.

World Bank (2011) Improving the Quality and Equity of Basic Education in Turkey: Challenges and Options. Report no. 54131-TR, World Bank, Washington, DC.