Educational Strategic Leadership Practices among Leaders of Selected Malaysian Risky Schools

This paper explored the practice of educational strategic leadership (ESLP) among selected leaders of Malaysian risky schools, with the aim of developing, validating, and confirming a hypothesised structural model for leaders’ ESLP. Data were collected from 472 school leaders of 141 risky secondary schools across Malaysia using a set of five-point rating scales as the research instrument. For data analysis, the study used descriptive statistics, confirmatory factor analysis, and full-fledged SEM. The findings revealed that Malaysian risky schools’ leaders had a high proclivity and inclination to practice educational strategic leadership (ESLP) at work and reported high levels of ESLP. Strategic orientation, strategic translation, strategic intervention, strategic alignment, strategic competencies, restlessness, absorptive capacity, adaptive capacity, and wisdom were among the nine practices of educational strategic leadership identified in the survey responses. SEM procedures were used to confirm that the hypothesised model of ESLP for Malaysian risky schools’ leaders was empirically valid and reliable. The findings emphasised the importance of planning and developing a specific-context training program in strategic educational leadership for Malaysian risky schools’ leaders. The training was deemed essential in the pursuit of effective leadership and positive school outcomes in Malaysian schools. The study was able to develop, test, and validate an ESLP model for the strategic management of risky schools and it is regarded as one of the few studies on the factors influencing risky schools in Malaysia.


INTRODUCTION
A "risky school" is one in which students face circumstances that may jeopardise their ability to complete their education, such as teenage pregnancy, violence, disciplinary issues, grade retention, or other learning-related factors that may harm their educational performance and achievement. Being in such a school is alarming as it threatens the safety and welfare of school children and adolescents. The same threat exists for all students attending risky schools in Malaysia. Despite significant national efforts to improve the school environment and the learning and teaching culture, negative and physically alarming incidents on school grounds are still frequently reported (UKEssays, November 2018).
School safety is every society's biggest concern and one of the gravest issues currently affecting Malaysian public schools-both at the primary and secondary levels-as the number and severity of alarming incidences involving violence, gangsterism, theft, vandalism, poor discipline, and student immorality continue to escalate (Tie, 2014). The presence and recurrence of these physically and morally unsafe situations leave a negative impact on students' physical well-being, emotional health, and intellectual development, sometimes with lasting consequences that persist into adulthood (David-Ferdon et al., 2021;Darling-Hammond & Cook-Harvey, 2018). Such adverse situations have made it extraordinarily challenging for schools to maintain and ensure the safety of all students, forcing them to employ proper and more effective administrative practices to combat the threat on students' well-being. Despite the ongoing efforts to protect the school environment, Malaysia has still not achieved its aim of creating a really conducive and safe surrounding for students in all of its schools (Sabu, 2005;David-Ferdon et al., 2021;Darling-Hammond & Cook-Harvey, 2018). Given the criticality of the situation, Malaysian school leaders need to understand and embrace effective leadership practices and must immediately employ them to mitigate the increasing lack of students' safety on school grounds.
Much debate and discussion has taken place on positive school characteristics and ways of enhancing them, with the broader aim of advancing equitable education for students of all backgrounds (Ololube et al., 2012). But to specifically improve the physical and psychological safety of schools and the quality of their social environment, researchers have suggested integrating educational strategic leadership practices (ESLP) into principals' school management. The belief is that the employment of ESLP, especially in the principal leadership of risky schools, will turn the condition and performance of these schools around. Essentially, leadership experts contend that ESLP is a vital component in the effective management and development of schools (Davies, 2006;Davies & Davies, 2009;Eacott, 2008) and could well be the solution to the safety crisis occurring in risky schools. Premised on this belief, this study aimed to explore the levels of ESLP among selected school leaders of high-risk Malaysian schools and validate a proposed model of ESLP relevant to them. In so doing, the research will thereby address the gap in the literature on school leadership practices as they pertain to the management of risky schools and contribute to a better understanding of ESLP as practiced in these schools.

EDUCATIONAL STRATEGIC LEADERSHIP PRACTICES (ESLP): AN OVERVIEW
Leadership competency with strategic elements expressed in a systematically validated ESLP model must be developed if schools are to maintain their academic performance and move into deep learning. The ESLP model is appropriate for strategic leaders as it theoretically develops the individual characteristics (i.e., restlessness, absorptive capability, adaptive capability, and wisdom) and organisational capability (i.e., strategic orientation, strategic translation, strategic intervention, strategic alignment, and strategic competencies) of school leaders, which in turn contribute to school effectiveness and safety. The development of this model is important as it shows school leaders how to guide their subordinates towards realizing the organisation's vision and mission (Davies & Davies, 2004, 2006Hairuddin, 2012Hairuddin, , 2016. In addition, the model also assists school leaders in recognizing their organisations' ability and intelligence, understanding problem solving, and finding a strategic relationship among staff (Davies & Davies, 2004, 2006. In short, such a model explains the great importance of organisational ability and individual characteristics in achieving the vision or aims of the organisation. This study's conceptual framework was based on the review of relevant literature, investigating the relationship between these variables ( Figure 1).

Figure 1
The Proposed ESLP Model for Risky Schools (Davies & Davies, 2006Hairuddin, 2012Hairuddin, , 2016Hairuddin & Inas, 2017, 2018 Leithwood and Sun (2012), in their meta-analysis of unpublished work, raised the point that effective leadership models include many of the same practices that shape schools into good learning institutions and concluded that researchers and practitioners ought to give more attention to the impact of specific leadership practices on school improvement, especially, and less on leadership models.
This study used a nine-factor model of ESLP, which integrates five organisational capabilities and four individual characteristics of strategic leaders (Davies, 2004;Davies & Davies, 2006). There are a few strong arguments for using this model in this study.
First, Davies is one of the most well-known experts when it comes to ESLP in educational management and leadership (Davies, 2004;Davies & Davies, 2006. Second, ESLP is rather a new concept in the Malaysian educational system. Third, the model appears to be applicable in the current state of Malaysian education, where the government is focusing on the outcomes outlined in the Malaysia Education Blue Print 2013-2025. Finally, the leaders of Malaysian risky schools are expected to acquire and demonstrate all nine ESLP characteristics. Possessing the ESLP characteristics is important as it facilitates and drives the strategic implementation of a strategically focused school (Davies, 2004;Davies & Davies, 2004, 2006Eacott, 2008). Therefore, based on the examination and synthesis of all nine constructs proposed by Davies and Davies (2006), Hairuddin (2012, and Inas (2017, 2018) as discussed above, the nine-factor model of ESLP was proposed for the better management of Malaysian risky schools. The proposed model is depicted in Figure 2.

Statement of the Problem
The prevailing trend of research on leadership is to focus mainly on leader characteristics or leadership styles. Our review of previous research pointed to school leaders as the pivot of school management and leadership. Thus, there is an urgent need to investigate the nature of their strategic leadership practices in risky schools, as well as how those practices can save schools that are on the verge of academic and moral failure. Based on the previous body of research, studies into strategic leadership have primarily and largely addressed the practices of educational leaders in vocational colleges (Hairuddin & Inas, 2018), administrators and teachers in primary education (Prasertcharoensuk & Tang, 2017), Kenyan national secondary school principals (Gakenia et al., 2017), and Malaysian secondary school principals (Jones et al., 2015). There has been very little research done on the school leaders of risky schools, particularly in Malaysia. Thus, this study provides a new inquiry into national secondary school leaders in Malaysia, particularly, the principals and assistant principals of high-risk schools.
Although some previous research had used quantitative methods and statistical procedures, the vast majority of studies did not attempt to empirically evaluate strategic leadership, particularly in the field of school effectiveness or performance as seen through the eyes of principals and assistant principals. Furthermore, Malaysia appears to be sceptical or hesitant about ensuring school improvement in terms of strategic leadership practices (Ahmad Masrizal, 2013;Kamaruzaman, 2012;Muhamad Bustaman, 2015;Siti et al., 2018).

Research Objectives and Questions
The objectives of this study were to investigate the levels of strategic leadership practices among leaders of selected Malaysian risky schools and to empirically validate a proposed model of ESLP for the management of schools with physical and psychological risks for students. The intention was to validate a set of leadership competencies and leader characteristics that need to be present in the effective management of schools with known risks. Subsequently, the following research questions were asked: 1. What are the levels of ESLP practiced by leaders of selected Malaysian risky schools? 2. Is the measurement model of ESLP of Malaysian risky schools' leaders valid and reliable? 3. Is the proposed ESLP model of Malaysian risky schools' leaders valid and reliable?

Research Hypotheses
H1: The measurement model of Malaysian risky school leaders' ESLP is valid and reliable. H2: The proposed ESLP model of Malaysian risky school leaders is valid and reliable.

Research Design
As per Muijs' (2004) contextualisation, this study adopted a quantitative, non-experimental research design, which included the cross-sectional survey method, to examine the impact of ESLP on risky schools in Malaysia. Since this study was confirmatory in nature, this design ensured high levels of reliability, validity, and generalisability of the gathered data in testing the proposed ESLP model.

Population and Sampling
This study's population consisted of the topmost leaders of the 384 selected (HIGH-RISK) government secondary schools throughout Malaysia. The topmost leaders were the principals and all three deputy principals in charge of the 384 schools. In deciding the sample size, three important parameters were considered, namely the population size (N = 384), margin of error (5%) and level of confidence (95%). Using the Raosoft online calculator for calculating minimum sample sizes, the study came up with N = 193 as the minimum number of HIGH-RISK secondary schools that should be surveyed for the study. The study had the confidence that the sampling process used had provided a strong statistical basis for a representative sample that should be generalizable to the entire population of Malaysian risky schools (Creswell, 2014;Edwards et al., 1997;Fink, 1995, as cited in Kotrlik & Higgind, 2001.

Instrumentation and Data Collection
The survey instrument measuring school leaders' ELSP was adapted from a number of questionnaires used in previous research on strategic leadership. A five-point Likert scale was developed consisting of 49 items that measured the respondents' perceived practices of strategic leadership in managing their respective hotspot schools, in addition to the demographic questions. The items on educational strategic leadership practices were partly adapted from Davies (2004), Davies and Davies (2004), Hitt, Ireland and Hoskisson (2007) Face and content validation were carried out prior to the pilot study. Three experts validated the face and content validity of the items for ambiguities, unclear wordings, leading (and misleading) questions and biases. The pilot test sample was drawn from the hotspot schools in Selangor and Kuala Lumpur. There were 48 hotspot schools in Selangor and Kuala Lumpur, out of a total of 193 in Malaysia. Twenty of these 48 schools were piloted (i.e., 16 schools from Selangor and four schools from Kuala Lumpur). The sample size determination for the pilot test was guided by Treece and Treece (1982), Baker (1994), and Connelly (2008). The Cronbach's alpha test was used to check the internal consistency of the subscales within the primary questionnaire using the pilot data. All schools in Malaysia that were involved in the study, with the exception of those in Kuala Lumpur, Selangor, and Putrajaya, were required to mail the completed survey questionnaires back to the researchers using the envelopes provided.

Data Analysis
The analysis procedures began with coding and integrating the data into SPSS version 25, followed by the cleaning and screening process to detect errors, missing values, and outliers in the data set. Two statistical procedures were run on the cleaned data after the data screening process. The study decided to apply a descriptive analysis to address RQ1 and an SEM analysis to address RQ2 and RQ3, hence H1 and H2, respectively. The two analyses were preferred as they were considered the most appropriate statistical techniques.

Demographic Profile of the Respondents
The final sample comprised 472 respondents, of whom almost 52% were male (n = 245) while 48% were female (n = 227). The respondents' ages ranged from 41 to 56 years. Among them, 78.4% had a bachelor's degree as their highest academic qualification, while those with a master's and doctoral degree constituted 21% and 0.4%, respectively. In terms of professional qualification, a majority were Diploma in Education/KPLI holders (57.6%), while 26.9% and 15.5%, respectively, were NPQH/NPQEL and other professional qualification holders.
More than 70% of the respondents had served as government teachers for more than 26 years (71.6%). Slightly more than one-fifth had taught in public schools for 21 to 25 years (24.4%), while the remaining 4% of the respondents had between 16 to 20 years of government service.
In terms of current leadership position, 24.2% were principals, 25.8% were deputy principals of Academic and Administration, 25.6% were deputy principals of Student Affairs, and 24.4% were deputy principals of Co-Curricular Affairs. Most respondents had served in their leadership role for less than three years (51.5%), while just about 18% had been school leaders for four to five years. The rest of the sample (30.5%) had more than six years of experience as school leaders.

Perceived Levels of ELSP among Malaysian Risky Schools' Leaders
The levels of ESLP of Malaysian risky school leaders were assessed mainly using means (M) and standard deviation (SD). The highest score for the levels of ESLP (organizational capabilities) was for strategic orientation ( Table 1 provides the details of the ESLP levels among the selected Malaysian risky schools' leaders.

Measurement Model of ESLP for Malaysian Risky Schools' Leaders
ESLP comprised two underlying components, namely Organisational Capabilities (or OC, which was represented by five constructs) and Personal Characteristics (or PC, which was represented by four constructs). The application of SEM, particularly Confirmatory Factor Analysis (CFA), to determine the measurement models of ESLP is explained separately in the subsequent sections.

Organisational Capabilities (OC)
OC consists of five latent constructs, which are strategic orientation (Orientation), strategic translation (Translation), strategic alignment (Alignment), strategic intervention (Intervention), and strategic competencies (Competencies). The study had expected all 24 items to load perfectly under organizational capabilities, but the preliminary results fell short of the expectation. This finding is statistically insignificant, specifying an inadequate fit between the observed data's covariance matrix and the model's implied covariance matrix. Other fit indices were also found to hold deficient values like the GFI = .875, AGFI = .845, TLI = .869, NFI = .843, and CFI = .885, which fell below the threshold value of .90. As a result, the proposed measurement model did not fit the data, thereby necessitating a revision. Figure 3 below depicts the revised 17 items from the five-factor measurement model of OC.

Figure 3 Re-specified Multi-dimensional Constructs of Organisational Capabilities (OC)
The above re-specified measurement model showed an improved GFI, which was consistent with the data. The (χ 2 ) 105= 296.910, while the p value = .000 remained the same. Similarly, the cmin/df = 2.828 and RMSEA = .062 fell between the ideal range of < 3.0 and < .08. Other fit indices also show improvement and hold acceptable values. The GFI = .932, AGFI = .901, TLI = .928, NFI = .918, and CFI = .945 fell above the threshold value of .90. Additionally, the parameter estimates were also examined and were found to be statistically significant as all the loadings were more than .60. Besides that, the values of the squared multiple correlation (R 2 ) are more than the threshold value of .25, which means no items are candidates for removal. In short, the re-specified model is now significant and fit the sample data as it fell in the range of accepted fit indices values. The results of the GFI of the generated model and the re-specified model are shown in Table 2. Sources: Awang et al. (2018), Kline (2011);Hair et al. (2010); Bentler (1995, 1999); Schumacker and Lomax (1996); and Bentler (1995) The convergent validity of the OC measurement models was investigated further. Table 3 summarises the factor loadings, the average variance extracted (AVE), and the critical ratio (CR) for the revised hypothesised model of OC. Based on Table 3, it can be concluded that all the five latent constructs have achieved the CR value of > .7, while two constructs, i.e., Translation and Alignment, reached the AVE value of > .5. Even though the AVE's for Orientation, Intervention, and Competencies are slightly below .5, but according to Fornell and Larcker (1981), if the AVE is less than .5, but the CR is higher than .6, the convergent validity of the construct is still adequate. This means that if the AVE < .5, on average, the item loadings for this study is less than .70 (Hair et al., 2010). Thus, the re-specified multidimensional constructs of organisational capabilities have evidence of convergent validity. Based on the fit indices and convergent validity of the re-specified constructs, the constructs were found to be valid and reliable.

Personal Characteristics (PC)
PC consists of four latent constructs, restlessness (Restlessness), absorptive capacity (Absorptive), adaptive capacity (Adaptive), and wisdom (Wisdom). The interrelationship with all the 25 items was statistically significant but the four latent constructs, Restlessness, Absorptive, Adaptive, and Wisdom, show poor fit model, and therefore the measurement model did not fit the sample data. To improve the model's fit, seven items were removed and five others were co-variated (Figure 3).

Figure 4 Re-specified Multi-dimensional Constructs of Personal Characteristics (PC)
The re-specified PC measurement model showed an improved GFI with χ 2 (126) = 346.361, while the p = .000 remains unchanged. Likewise, cmin/df = 2.749 and RMSEA = .061 fell between the recommended good range of < 3.0 and < .08. Other fit indices also show improvement and hold acceptable values. GFI = .921, TLI = .924, NFI = .906, and CFI = .938 were higher than the threshold value of .90, while AGFI = .893 were slightly lower than .90, but is still in the acceptable range. In addition, the parameter estimates were also examined and were found to be statistically significant as 99.8% of the loadings exceeded .60. Besides, the R 2 values are greater than the threshold value of .25, which means that no items are to be removed. The re-specified model is now significant and fits the sample data as it fell within the range of accepted value of the fit indices. Table 4 shows the GFI results of the generated model and the re-specified model of PC. Source: Awang et al. (2018), Kline (2011);Hair et al. (2010); Bentler (1995, 1999); Schumacker and Lomax (1996); and Bentler (1995) The PC measurement models, convergent validity was investigated further. The factor loadings, AVE, and CR for the revised measurement model of PC is exhibited in Table 5. Based on the Table 5 above, the four latent constructs have achieved the CR value of > .70. Even though the AVE values are lower than the recommended value of .50, but the CR values are greater than .60, therefore the convergent validity of this construct is still acceptable (Fornell and Larcker, 1981). This implies that the item loadings for this study are on average less than .70 (Hair et al., 2010). Consequently, the re-specified multi-dimensional PC constructs have proof of convergent validity. The analysis reveals that the constructs were valid, reliable, and deemed acceptable based on the fit indices and convergent validity of the re-specified constructs. Therefore, the RQ2 was addressed by the study and H1 was supported.

Proposed ESLP Model for Leaders of Malaysian Risky Schools
The causal relationship between OC and PC were examined via the structural model. The proposed structural model of this study comprised of two latent variables measured by a total of nine indicators. Statistically, the interrelationship between these indicators were significant. Figure 4 depicts the proposed model of the study.
The cmin/df = 3.399 and RMSEA = .071 fell between the good range. The other fit indices also reflected good fit values, like the GFI = .969, AGFI = .930, TLI = .973, and CFI = .980, which fell beyond the threshold value of .90. The above results prove that the proposed model of the study is now valid and reliable and therefore represents the study's model. SEM is also a statistical tool used for testing the hypotheses of a multivariate analysis. The hypotheses could be tested at the 0.05 significant level if the revised measurement models and the proposed model were both valid. After extensive modifications, the study's measurement model is supported and considered valid and reliable. The proposed study model closely resembles the study model with determined re-specification. As shown in Table 5, H1 and H2 were supported.

DISCUSSION ESLP Levels: The Perception of Malaysian Risky Schools' Leaders
Based on the descriptive statistics, the finding reveals that the scores of every construct for ESLP were high. The respondents obtained high levels of OC and PC of ESLP and therefore RQ1 was fully addressed. The findings were consistent with the findings reported by Adelakun (2015), Ali (2013), Kitonga (2017), Hairuddin (2012), Hairuddin and Inas (2018), Inas (2016), Lear (2012), Nasruddeen (2015), Nazifah (2012), Thanomwan and Tang (2017), William and Johnson (2013), as well as Deeboonmee and Ariratana (2014) who also identified a high level of ESLP among educational leaders or administrators.

ESLP Model for Malaysian Risky Schools' Leaders: The Validity and Reliability.
CFA was also utilised to confirm if there were presence of multi-dimensional constructs of ESLP. The validity and reliability of the multi-dimensional constructs of OC and PC were analysed separately. Interestingly, both OC and PC models exhibited convergent validity and internal reliability. Therefore, RQ2 was fully addressed, and the findings had supported H1. The results of the revised measurement model of OC and PC concluded that the Malaysian risky schools' leaders possessed and practiced all nine attributes as suggested by Davies and Davies (2006;), Hairuddin (2012, Hairuddin and Inas (2018) and Inas (2016). Further analysis was conducted to test the validity and reliability of proposed model of ESLP and the findings had successfully addressed RQ3, and supported H2. ESLP was treated as a nine-factor model, which is in line with Adelakun (2015), Ali (2013), and Nazifah (2012), Hairuddin (2012), Hairuddin and Inas (2018) and Inas (2016).

Theoretical and Practical Implications
For the management of risky schools in Malaysia, a model of ESLP with nine underlying factors is proposed and verified in the study. The significant findings have enriched the literature on strategic leadership, particularly in regard to the Malaysian context of school management. This would inspire further research by more scholars in different contexts, thereby generating immense benefit to the sustainability of the Malaysian educational system.
The findings of this study will indeed facilitate school leaders and stakeholders in designing, developing, and implementing required courses and workshops on ESLP to help school leaders establish and maintain risk-free schools. This is because a safe school requires continuous enhancement and an integrated approach of ESLP. It will act as a pillar of support for school leaders in boosting the quality and ensuring the impact of ESLP in schools. This could result in having competent school leaders that will, in turn, generate and boost the efficiency and quality of the schools under their management.

CONCLUSION
This study has affirmed the importance of nine factors in the effective management of risky schools, hence contributing to the body of knowledge on strategic school management and improving the research literature on ESLP. The results have come in full support of the efforts of stakeholders, practitioners, and future researchers, both directly and indirectly, in improving the quality of Malaysian schools, particularly those with a high degree of at-riskness. The insights are beneficial for guiding school leaders in managing and navigating the at-risk schools under their leadership towards greater success by reducing their degrees of at-riskness via the principals' effective organisational capabilities and positive personal characteristics.
This research was conducted in a cross-sectional manner. The information was gathered at a single point in time. There was no pre-and post-event testing, and no longitudinal processes were assessed. The temporal limitation of cross-sectional studies is that the constructs under investigation may fluctuate over time or in reaction to external real-life circumstances. So, future research should consider replicating the study using longitudinal or mixed-methods procedures to identify the factors influencing school leaders' SLPs with risky schools. Besides, because this research involved a specific group of people, future researchers should look at a different group of respondents, such as school administrators, senior teachers, and counsellors of various ages and educational levels from all Malaysian government schools and examine the structural differences in the model using the same structure of the ELSP model. While this research had used a robust statistical method of SEM to explore school leaders' ESLP, the results are predicted to vary if qualitative methods are used for an in-depth inquiry into the applicability of this model in the future. Even though there have been many other studies conducted throughout the world on the variables of risky schools based on their respective school contexts, none of the studies had covered the variables encompassing ESLP examined in this research in the context of Malaysian at-risk schools. This study has successfully identified the fundamental determinants of ESLP (SLP) through the application of CFA on the at-risk school data. Indeed, the ESLP body of knowledge is enriched to another level through the findings of this study.