1The 1997 foreign exchange crisis widened the income gap in Korea, which in turn lead to dramatic increases in the poverty ratio (Yoon, 2014). This widening gap, coupled with the growing number of households living in poverty weighed heavily on school education (Sung et alii, 2015). The type of poverty that arose in the aftermath of the foreign exchange crisis was not simply one of economic shortages; it had more to do with other forms of shortage and lack of access, including to enriching cultural experiences, social relationships, and peer networking opportunities, and increasing levels of psychological anxiety (Kim, 2012; Peng, 2012).
2Research has also shown that children from poor households are at a higher risk of being neglected at home than are those from higher-income households (Coleman, 1988). Low-income students are less engaged in class and have a weaker understanding of what is being taught (Jensen, 2013; Lee et alii, 2014). They are also more exposed to being alienated from their peers, experiencing violence, and developing behavioural problems (Conger et alii, 1992; Kim & Cha, 2007; Lee & Gang, 2004; McLoyd, 1990; Park et alii, 2006).
3The South Korean government acknowledged this issue, and in coordination with the local community, it began to take action by nudging schools to assume a more proactive role in addressing the adverse conditions faced by children from poor households. As one of its policy measures, the government launched the Education Welfare Priority Zone Program (ewpzp) in 2003 by benchmarking the zep (Zone d’Éducation Prioritaire) from France and the eaz (Education Action Zone) from the uk, both of which were education-priority investment policies targeting children living in poverty-stricken urban areas. They also both emphasised building networks between schools and social welfare agencies to enhance quality of life and academic achievement among socio-economically disadvantaged students (Kim, 2012; Sung et alii, 2015; Yoon, 2014).
4The aim of establishing the ewpzp was to identify educationally disadvantaged student groups in South Korea and provide support to meet their needs. In particular, it classified the needs of the target zone population into three categories — learning support, compensation for cultural deprivation, and psychological and emotional support — with the objective of delivering these forms of support through the joint efforts of schools and local communities (Ryu et alii, 2011b; Kim, 2012; Sung et alii, 2015). In terms of learning support, the ewpzp provides various programmes (such as after-school tutors) that have been designed to provide academic support to under-achieving students so that they can become interested in learning and eventually become self-regulated learners. In terms of compensation for cultural deprivation, the programme provides free food and medical assistance to students with financial difficulties and also helps them develop cultural capacity (including by free access to arts and music performances). In terms of psychological and emotional support, the ewpzp provides psychological tests and interventions (for example, counselling services) to students with emotional and behavioural problems.
5The policy differs from traditional welfare policies in South Korea, which are principally aimed at improving academic achievement. Specifically, it attempts to set up a solid system in which civil groups in the local community can provide resources and support to children after school so that they will feel secure and cared-for. For instance, it introduces small-scale after-school programmes for children from poor households aimed at developing these children’s trust in adults in the local community and helping them feel a sense of emotional security and support.
6It also seeks to provide intensive support to certain groups and simultaneously improve the overall quality of the surrounding environment. Schools working with the programme select students who have been given Priority Support (ps) based on their socio-economic circumstances and review their needs in order to provide customised support to individual students. It acknowledges that teachers’ enthusiasm and passion can make or break the programme, and provides them with training to help them better understand the characteristics of children from poor households (moehrd, 2005). Coordinators are assigned to schools to inform teachers about their lives outside school, communicate with each other, and help them with appropriate attention and interventions.
7Third, ewpzp represents an attempt to provide the necessary support mechanisms to children in an integrated manner to as great an extent as possible. Although the special after-school programmes are divided for the sake of convenience into learning, cultural experiences, and emotional/psychological support, some take a mixed approach, tackling multiple areas, while others focus on a single particular area in accordance with the students’ characteristics.
8Finally, it has set itself the objective of creating an education community in which schools and the local community can work together to solve the educational issues caused by the growing number of households living in poverty. Specifically, it seeks to help students who are receiving intensive support from the programme to better adapt to school education by improving their attitudes and social skills and mitigating behavioural problems in order to ultimately boost their academic abilities (Kim, 2012; Sung et alii, 2015).
9The pilot ewpzp programme was launched in 2003, targeting 45 schools (29 elementary schools and 16 middle schools) in eight districts in Seoul and Busan, Korea’s two largest cities, and gradually expanded its scope each year until by 2011 the programme had grown to involve 1,300 schools nationwide (670 elementary schools and 681 middle schools). The Ministry of Education then handed over management of the ewpzp to municipal education departments so that the programme would be able to respond to local needs.
10Previously, policies for disadvantaged students had principally focused on paying for education and meals, whereas the ewpzp placed its emphasis on providing the various kinds of support required for students from poor households. It first selected an area encompassing 3 or 4 elementary schools and 1 or 2 junior high schools with a high proportion of children from low-income households. School coordinators were then assigned to schools in these zones with the task of understanding the circumstances of individual students through interviews with target students, selecting students to whom ps should be offered, and preparing the necessary measures with teachers.
11Teachers use the information about the ps students provided by the school coordinator so that they will be better able to understand the situation and interact appropriately (moehrd, 2005). For example, the coordinators help teachers understand the difficulties faced by students who are often late or leave early, who miss breakfast, who come from grandparents-grandchildren families, or who are silent in the classroom. With this shared understanding, the teachers and coordinators jointly devise a response plan.
12Based on the needs of the ps students, schools run activities in four main categories: supplementary learning, experiential activities to compensate for cultural deprivation, emotional and psychological support, and care after school and during school holidays. Since full provision of this support using an individual school’s resources alone is not possible, the coordinators play a role in cooperative networking with local youth institutions, welfare centres, psychological counselling centres, hospitals, and sports facilities.
13The school coordinators and teachers cooperate to arrange suitable programmes according to the needs of the ps students to ensure that the various types of supports for the development of poor students are integrated. A number of after-school programmes for ps students attempt to reduce feelings of stigma by allowing non-poor children to participate. Some schools open programmes for parents or grandparents of students from poor households.
14Where necessary to resolve the difficulties or problems faced by ps students, schools seek an integrated solution with teachers and local professional counsellors, welfare centre managers, doctors, police, and volunteers. As a result, communications between schools and regional institutions that can be used by the children are encouraged, and activities for children and adolescents in the region are promoted. Studies analysing the effects of the uk’s eaz have also shown that it helped raise social capital in the relevant area (Franklin, 2005; Gewirtz et alii, 2005).
15The Ministry of Education provides an additional budget of more than 100,000 usd for each school in the early stages of this policy to be used in particular for various after-school programmes and for experts such as psychological counsellors, if necessary. It is not allowed to use this budget to repair or extend school facilities (improvements to outdated school facilities were handled through a separate budget). As an exception, however, it is permissible to use the budget to create a space where the coordinator communicates with students and to remodel school libraries. This is because these facilities are viewed as spaces where poor students can openly discuss their problems at school at any time, or that provide quality cultural activities related to learning.
16Coordinators are also assigned to the regional education department to manage mutual cooperation, information sharing, and joint programmes between schools at an area level. Through a variety of activities at a school level and collaborative support with the community, the aim of the ewpzp is that students in poverty and their parents will have a friendly awareness of and trust in the school and the regional community.
17The steady expansion of the ewpzp was not merely driven by a need to deal with the newly-emerging social issue of the increasing income gap. Some argued that it might potentially satisfy the concrete needs of schools on the education front line and actually help the school lives of students in every conceivable way. However, regardless of what changes or benefits for schools and students were noted for those participating in the programme, they might be written off as subjective or coincidental unless they are described in an objective manner (Lee, 2008).
18This study argues for the necessity and importance of variable selection when designing a programme effectiveness analysis model. The selected variables should reflect the emphasis placed on the actual processes at work within the programme in terms of the way it measures adaptation to the school environment. It goes without saying that verifying a programme’s effectiveness with variables that are irrelevant to the factors and aims of the programme will mean it will not be implemented adequately.
19Previous studies examining the effectiveness of the ewpzp either selected outcome variables such as teacher expectations, teacher efficiency, or the leadership of school principals solely on a theoretical basis (Ryu et alii, 2011a), or simply focused on the school’s ewpzp status as a treatment measure, and ignored the actual mechanism of ps selection within ewpzp schools (Baek et alii, 2013). Therefore, as noted by Lee et alii (2005), in order to evaluate whether the ewpzp was properly designed, how these factors work to exert their effects through various pathways needs to be examined further, starting from the beginning of the policy implementation process.
20In this study, we specifically focused on analysing the programme’s impact from the perspective of students’ adjustment to school, since the ultimate goal of the programme is to revitalise activities in the areas of learning, emotional support, and cultural experience so that students from households living in poverty can receive more exposure to positive educational experiences and adapt better to their school lives (Widdowson et alii, 2015). Two studies by Ladd (1990) and Ladd and Price (1987) have treated school adjustment as a psychological phenomenon, viewing it in terms of anxiety and aversion, and reported that peer relations can be a key variable for predicting maladjustment to school. Likewise, Dubois et alii (1994) found a correlation between school adjustment and frequency of school absence, self-reported scholastic self-concept, and grades. Many studies on school adjustment carried out in Korea have attempted to measure school adjustment through academic ability, social relations, and problematic behaviours (Ryu et alii, 2007; Yoon et alii, 2011).
21A path model analysis enabled us to examine the underlying mechanism of the ewpzp policy. This helped us gain a better understanding of the effects of the programme on the degree to which students adjust to life at school. Some research suggests that the level of teacher interest toward students, participation in school activities, and social capital through networks within the local community, inter alia, are confirmed to have positive relationships with the improvement of school adjustment (Lee et alii, 2013; Kim et alii, 2007). The hypothesised path model therefore included the level of teacher interest and devotion, the presence of adults who were concerned about the local community and had an active network within it (such as social capital), and school activities that were stimulated in part by the ewpzp project, including class activities and club activities.
22The following research questions were formulated to examine the effectiveness of the ewpzp in South Korea:
RQ1: What is the effect of ewpzp/ps on students’ school adjustment compared to non-ewpzp school students?
RQ2: What is the effect of ewpzp/non-ps on students’ school adjustment compared to non-ewpzp school students?
RQ3: Do teacher interest and support, school activities, and community social capital mediate the effect of the ewpzp on students’ school adjustment?
23Data from the 2012 Korean Educational Development Institute Longitudinal Effectiveness Study of the ewpzp (Ryu et alii, 2012) were analysed in the course of this study. Initially, using stratified multi-stage cluster sampling, 50 nationally representative ewpzp schools were selected for the first stage in 2009. In addition, for each of the ewpzp schools, a non-ewpzp school with similar conditions was selected. The key criterion for the non-ewpzp control group selection was the number of students receiving basic livelihood security within the same local school district. At the second stage, one class of fourth graders from each of the 50 ewpzp and 50 non-ewpzp elementary schools was randomly sampled. The sampled students were surveyed annually until 2011. However, owing to the expansion of the ewpzp programme, some initially non-ewpzp schools changed their status to become ewpzp schools over the course of the period. Moreover, among the initial 50 ewpzp schools, only 15 began the ewpzp in 2009. Therefore, to make a more precise comparison in terms of ewpzp effect, we selected the 15 ewpzp schools that started the ewpzp in 2009 and kept up the programme until 2011 and the 28 non-ewpzp schools that did not change their ewp status during the study period (2009-2011). The final sample we compiled in this way was made up of 1,237 students from 43 schools (501 students in 15 ewpzp schools and 736 students in 28 non-ewpzp schools), as shown in Table 1. It should be noted that 736 students in non-ewpzp schools formed our control pool, from which 501 treated students (193 non-ps and 308 ps) were 1:1 matched based on the propensity score.
24The outcome variables in our study, which are the indicators of school adjustment, included students’ class attitudes, social skills, and behavioural problems. These variables were collected in 2011. From a temporal standpoint, therefore, we can assume that the outcome variables (measured in 2011) reflect the effectiveness of our treatment (ewpzp and ps, administered for 3 years from 2009 to 2011). The class attitudes variable was composed of six 5-point Likert scale items that asked the extent to which students concentrate on class activities, answer questions, study diligently, prepare class materials, and do homework (α= .859). The social skills variable was also measured by means of four 5-point Likert scale items that asked the extent to which students work well with friends, participate in school activities with friends, are able to resolve conflicts with friends, and get along with their friends (α= .849). The behavioural problems variable was composed of twelve items that asked about maladaptive behaviours, including absence; late attendance; lack of class participation; disobedience; bullying; playing violent video games; watching or looking at adult rated films or magazines; running away from home; engaging in extortion, gang activities, or fighting; smoking; and drinking alcohol (α= .814). All these questions were posed directly to the students in 2011. Although the questionnaire was submitted annually, it was partly modified in 2010. It was therefore not possible to construct a longitudinal measure of the outcome. Instead, we adopted the propensity-based matching method to achieve inter-group pre-treatment balance on key background variables. Before calculating the class attitudes and social skills composite scores, the dimensionality of each scale was checked using principal component analysis (pca). pca with extraction criterion of eigenvalue >1 indicated that both class attitudes and social skills were unidimensional, with eigenvalue of 3.204 (64.08% of total variance) and 2.768 (69.20% of total variance) respectively.
25It was hypothesised that the process variables would mediate programme participation and school adjustment. These variables included teacher interest, school activity participation, and community social capital. Teacher interest consisted of four items that asked the extent to which teachers understood students, treated students warmly, listened to students’ concerns, and provided comfort when students found themselves in difficulty (α= .940). The school activity participation was composed of three binary items that asked whether students participated in classroom, club, and career experience activities. Finally, the community social capital was composed of two questions that asked whether students knew many adults in their communities, and whether they had many adults they could ask for help when needed (α= .757). With the exception of the school activity items, all the variables were measured using 5-point scale items. The applicable items were averaged to represent each constant. The pca result with eigenvalue >1 criterion indicated that both teacher interest and community social capital scales were unidimensional, with eigenvalues of 3.395 and 1.609 respectively, and the proportion of the total variance was explained by each component as 84.87% and 80.46% respectively.
26Finally, a student background variable was included, comprising family income (measured by krw), parental occupation (measured by the Ganzeboom scale [Ganzeboom & Treiman, 2003]), level of education in years, whether the student received free school lunches, whether the student received basic livelihood security, and whether the student was from a single- or no-parent family. These variables were used to statistically adjust out the pre-treatment difference in estimating the effect of ewpzp and ps. The descriptive statistics are presented in Table 1 below.
Table 1. Descriptive statistics for dependent, process, and background variables
27Following the work of Frangakis and Rubin (2002), it was possible to define two groups of students in such a way that the first group was composed of those who would receive ps if assigned to ewpzp schools (Principal Stratum 1) and the second was made up of those who would not receive ps even if they were assigned to an ewpzp school (Principal Stratum 2). Therefore, the two types of treatment effect that were of interest in this study corresponded to (1) comparing ewpzp/ps students with their comparable non-ewpzp counterparts and (2) comparing ewpzp/non-ps students with their non-ewpzp counterparts. These two treatment effects correspond to our Research Questions 1 and 2 respectively. The two effects may be described differently as follows: the effect of ewpzp assignment for (1) those who received ps and (2) those who did not.
28The effectiveness of the ewpzp can be viewed from two perspectives. On one hand, it can change an individual school’s overall educational environment and the attitudes of both students and teachers. Improvement in school adjustment due to this type of change can be expected from students who do not receive ps support but simply attend ewpzp schools — which we refer to as the participation school effect. On the other hand, ps can have its own unique effect. The ps participants are provided with various after-school programmes and activities. If the ps participants’ school adjustment is changed due to ps participation, only the actual ps participants would be expected to change. We will refer to this effect as the ps-specific effect. Therefore, the total effect of the ewpzp is the sum of the two types of effects:
Total ewpzp effect= Participation school effect + ps specific effect
29Because ps participants enjoy both participation school effect and ps specific effect, our Treatment Effect 1 will actually estimate the “Total ewpzp effect.” In addition, because non-ps students will only be subject to the participation school effect, our Treatment Effect 2 will estimate the “participation school effect.” Figure 1 visualises how the two treatment effects are formulated with the ewpzp data.
Figure 1. Defining treatment effects for each principal stratum
30To construct a comparable matched sample, the propensity score method (Rosenbaum & Rubin, 1983) was used as a standard application. Adopting this approach would normally have required us to estimate the probability of ps assignment for each and every student (the propensity score) and either (1) match ps students in ewpzp schools with non-ewpzp school students (for Research Question 1) or (2) match non-ps students in ewpzp schools with non-ewpzp school students (for Research Question 2) based on the estimated propensity score. In this study, however, because the actual ps status could not be observed for the students in non-ewpzp schools, it was not possible to estimate their propensity scores using a conventional logistic or probit regression approach. To address this problem, we adopted the propensity score matching method via the out-of-sample extrapolation used in Hong and Raudenbush (2005), Kim (2006), and Wang (2006). First, the propensity score for receiving ps was estimated for students in ewpzp schools. Second, the regression equation from the first step was applied to the students in non-ewpzp schools to produce a projected propensity score. In this way, a propensity score was assigned to every student.
31The resulting propensity score was used to compose a matched control for each principal stratum: that is, for the first treatment effect, ps students in ewpzp schools were matched on a one-to-one basis with non-ewpzp school students based on their propensity scores, while for the second treatment effect, non-ps students in ewpzp schools were matched on a one-to-one basis with non-ewpzp school students based on their propensity scores. Propensity-based nearest-available matching with caliper size 0.1 was used to construct each matched control (D’Agostino, 1998).
32After constructing a matched sample for the two principal strata, path analysis was applied to the two matched samples in order to estimate the direct and indirect total ewpzp effect (Principal Stratum 1) and participation school effect (Principal Stratum 2). Figure 2 indicates our hypothesis regarding the ways in which this programme was able to promote levels of school adjustment. For each component of school adjustment (class attitudes, social skills, and problem behaviours), the first mediator between programme participation and school adjustment was teacher interest. We hypothesised that the programme first boosted teacher interest. Increases in teacher interest either directly or indirectly affected school adjustment. As can be seen, the indirect path was mediated by school activity and community social capital.
Figure 2. Path model from programme participation to school adjustment
33The probability of ps assignment was modelled using 308 ps and 193 non-ps students in ewpzp schools. Because family economic status and disadvantaged circumstances were the main criteria applied for ps assignment, we placed 6 SES- and family structure-related background variables in Table 1 in the logistic propensity model with binary ps status as a dependent variable. The results indicate that low income and the provision of free school meals are significantly associated with a higher probability of ps assignment. The directions of coefficients were consistent with the general ps selection criteria: that is, students from low-income and socially disadvantaged families. For the non-ewpzp school students who did not have ps status, ps assignment probabilities were calculated using the coefficients from Table 2.
Table 2. Logistic regression coefficients for PS selection probability
34The estimated mean propensity scores were .782 for ewpzp/ps, .343 for ewpzp/non-ps and .437 for non-ewpzp. Since the likelihood of ps participation was significantly different among the three groups, ewpzp/ps students were matched with non-EWP students to construct a matched control for a “total ewpzp effect.” Similarly, ewpzp/non-ps students were matched with non-ewp students to construct a matched control for “participation school effect”, as defined in Figure 1.
35Table 3 summarises the post-matching balance in background variables along with the propensity score. Note that acceptable matched pairs were found for 249 out of 308 ewpzp/ps students within ±0.1 calliper in logit scale. ewpzp/ps and matched non-ewpzp showed no significant difference in all the background variables. Also, ewpzp/non-ps and matched non-ewpzp had no significant difference in the distribution of background variables. We can therefore conclude that posterior balance on observables is achieved and analysis using the matched samples would approximate the causal effects of programme participation.
Table 3. Post-matching balance in propensity score and background variables
36The path model described in Figure 2 was fitted on the two matched datasets (ewpzp/ps v. matched non-ewpzp and ewpzp/non-ps v. matched non-ewpzp). The fit indices are presented in Table 4. Chi-square tests showed that there is no significant difference between observed and model-based expected variance-covariance structure. Also, the rmsea (.00 and .02), cfi (1.00 and .99) and srmr (.01 and .02) indicated a decent fit according to the general recommendations of cfi >.9, rmsea<.05 and srmr <.1 (Kline, 2016). It should be noted that the rmsea and cfi for the ewpzp/ps v. non-ewp dataset are set at 0 and 1 respectively due to the chi square value being smaller than the degree of freedom (Muthén & Muthén, 2015).
37Figure 3 presents standardised path coefficients for ps v. non-ewp path analysis. It should be observed that non-significant paths (at a significance level of α=.05) are represented by a dashed line. The test of indirect effects with the delta method (Hayes, 2013) for individual pathways is presented in Table 5. These results indicate that receiving ps in ewpzp schools (ewpzp/ps) significantly increased students’ perception of their teacher’s interest and simultaneously boosted student participation in school activities compared to similar students in non-ewpzp schools. Increased teacher interest directly affected class attitudes and social skills. The effect of increased levels of teacher interest on class attitudes is also mediated by school activity participation and community social capital. In terms of social skills, community social capital works as a major mediator. However, for problem behaviours, no direct or indirect effect was found.
Table 4. Fit indices for two-path model
|
χ²
|
df
|
p
|
RMSEA
|
CFI
|
SRMR
|
PS v. non-EWPZP
|
2.93
|
4
|
0.57
|
0.00
|
1.00
|
0.01
|
Non-PS v. non-EWPZP
|
4.34
|
4
|
0.36
|
0.02
|
0.99
|
0.02
|
Table 5. Test of indirect effects from PS to the three outcomes (PS v. Non-EWPZP m odel)
Path
|
Standardised
Coefficient
|
s.e.
|
p-value
|
PS -> Community social capital -> Class attitudes
|
0.030
|
0.018
|
.096
|
PS -> School activity participation -> Class attitudes
|
0.046
|
0.019
|
.017
|
PS -> Teacher interest -> Class attitudes
|
0.041
|
0.019
|
.017
|
PS -> Teacher interest -> Community social capital -> Class attitudes
|
0.014
|
0.006
|
.022
|
PS -> Teacher interest -> School activity participation -> Class attitudes
|
0.010
|
0.005
|
.035
|
Total indirect effect from PS to class attitudes
|
0.141
|
0.035
|
.000
|
PS -> Community social capital -> Social skills
|
0.045
|
0.026
|
.080
|
PS -> School activity participation -> Social skills
|
0.010
|
0.014
|
.450
|
PS -> Teacher interest -> Social skills
|
0.041
|
0.019
|
.028
|
PS -> Teacher interest -> Community social capital -> Social skills
|
0.021
|
0.008
|
.011
|
PS -> Teacher interest -> School activity participation -> Social skills
|
0.002
|
0.003
|
.456
|
Total indirect effect from PS to social skills
|
0.120
|
0.037
|
.001
|
PS -> Community social capital -> Behavioural problems
|
0.009
|
0.009
|
.350
|
PS -> School activity participation -> Behavioural problems
|
-0.014
|
0.015
|
.342
|
PS -> Teacher interest -> Behavioural problems
|
-0.023
|
0.016
|
.147
|
PS -> Teacher interest -> Community social capital -> Behavioural problems
|
0.004
|
0.004
|
.311
|
PS -> Teacher interest -> School activity participation -> Behavioural problems
|
-0.003
|
0.003
|
.352
|
Total indirect effect from PS to Behavioural problems
|
-0.028
|
0.023
|
.219
|
Figure 3. Standardised path coefficients for total EWPZP effect
(PS v. non-EWPZP) model (*** :p<.001, ** :p<.01, * :p<.05)
Figure 4. Standardised path coefficients for the participation school effect
(non-PS v. non-EWPZP) model (*** :p<.001, ** :p<.01, * :p<.05)
Ized path coefficients for the participation school effect (non-ps vs. non-ewp) model (***:p<.001, **:p<.01, *:p<.05)
38Figure 4 shows the results for the non-ps v. non-ewpzp path analysis. For those who were unlikely to receive ps, simply attending ewpzp schools did not increase levels of teacher interest. Therefore, the participation school effect as depicted in figure 1 was not detected in our analysis. However, for those students who were not eligible to receive ps, levels of teacher interest had a significant impact on all the school-adjustment variables and process variables. For the two process variables, community social capital was significantly related to all three outcome variables and school activity was positively associated with class attitudes. These results imply that increasing levels of teacher interest for the non-ps students in ewp schools is a key factor that may help to realise a participation school effect.
39This study focused on the total effect and participation school effect of the ewpzp and ps in ewpzp elementary schools in Korea. The results are summarised below.
40First, for those students who were eligible to receive ps, being in ewpzp schools and receiving ps can result in increased teacher interest. This increased level of teacher interest can directly and indirectly improve these students’ class attitudes and social skills. This implies that ewpzp is effective in promoting students’ school adjustment via increased levels of teacher interest. However, no empirical evidence was found to suggest or prove that more serious forms of maladjustment or behavioural problems are mitigated by increased levels of teacher interest.
41Second, for the student group that was not eligible for receipt of ps, simply attending ewpzp schools but not receiving ps did not improve their school adjustment. However, regardless of ewpzp participation, it was found that teacher interest positively affected students’ behavioural problems as well as their class attitudes and social skills. This finding underscores the role of teachers in creating a school environment that is conducive to student growth.
42Third, one of the main objectives of the ewpzp programme is to increase the level of community social capital for disadvantaged students. In this study, it was not possible to see any clear evidence that ewpzp (including ps) participation directly increases the levels of community social capital for the students who were the subjects of this research. However, there was an indirect effect for ps recipients. This reveals that ps increases levels of teacher interest and support, which has the effect of increasing the level of the students’ community social capital. This may be due to teachers’ efforts to connect the ps students to community facilities and resources. Increased community social capital positively changes levels of student school adjustment.
43This study has sought to confirm the pathways of effect that occurred during the ewpzp programme as well as examining its effectiveness of the programme itself.
44Something that seems to be of far greater importance than the statistical issues described above is steering the ewpzp in a fruitful direction. This would appear to be a daunting task in real terms due to the apparent incompatibility between enhancing overall school education and achieving social justice aims through the ps for specifically targeted groups. It would seem that in the case of social justice, if this principle were to be applied rigidly in a way that only gave preferential support to disadvantaged groups, the notion of overall school improvement might possibly be regarded as a secondary concern. It is critical to share the vision of the ewpzp and gain the co-operation of teachers in order to improve overall school education for the future.
45In the early stages of the ewpzp, construction of an educational community through a partnership between schools and local communities is encouraged. The purpose of the ewpzp is to provide integrated support in accordance with the needs of disadvantaged learners. However, its aim is not to provide a form of service that is simply available to be purchased. This study focused on the variable of social capital within a local community because it seemed that the local community needed to work for the growth of children living in poverty as much as the schools themselves do. However, the result of our analysis confirms the position that ewpzp participation does not fully mobilise the social capital within a local community, as had been anticipated. This finding suggests that creating the basis for social capital for disadvantaged learners between schools and the local community should not be ignored while we are absorbed with providing a customised service to end users (Kim, 2012). In the long term, social capital means building meaningful social networks around children. Therefore, its influence may be ultimately more significant than merely providing services that can be purchased. In order to establish such networks, a great deal of effort to develop communication and co-operation between schools and various private agencies in local communities is required.
46The findings from this study suggest that the ewpzp does indeed stimulate interest among teachers regarding students living in poverty, and that interest plays a key role in positive changes for these students. The interest and effort dedicated to understanding a disadvantaged group of students on the part of these teachers is significant for the children’s development. Considering that most teachers increasingly come from middle class socio-economic backgrounds, intentional endeavours to understand students living in poverty are required. The effectiveness of the ewpzp was confirmed during the process of tracking its progress in terms of sparking teacher interest in students living in poverty. However, more work in areas such as teacher training is needed in order to increase its effectiveness. Showing concern and a deeper understanding of students are important for school adjustment; however, they are also highly significant in terms of their relevance to designing educational activities, developing teaching materials, and being in a stronger position to relate to the experiences of students during the process of learning.