Organizational Health Scale: A Scale Development Study

The purpose of this study is to develop a valid and reliable measurement tool for measuring organizational health of schools. The study group consists of 429 teachers working in secondary schools in the central districts of Eskişehir/Turkey in the 2015-2016 academic year. The construct validity of the scale was examined by exploratory factor analysis. According to the results of the analysis, the scale measures a four-factor structure. The four-factor structure of the scale was confirmed by confirmatory factor analysis. The total variance ratio explained by the scale was determined as 71.101%. In order to determine the reliability of the scale, the internal consistency coefficient of Cronbach’s alpha was evaluated and this value was determined as .915. The results of all validity and reliability analyzes show that the Organizational Health Scale can be used as a valid and reliable measurement tool in the studies that teachers will be taken as a working group.


Introduction
In order for organizations to develop continuously, to catch up with change and to be successful, all their units must be functional. When the organization is viewed from this perspective, one of the prominent concepts is organizational health.
Organizational health is a concept that addresses the harmony of the organization with all its internal components and environment, its ability to achieve its goals and objectives. At the same time, organizational health addresses the state of the organization's physical environment and the tools appropriate to realize the organization's purpose, the state of communication between the organization and the senior management, the potential for problem solving, development, growth and innovation. Organizational health also deals with the quality of inputs (competence status) and raw material safety, management and decision-making activities of the organization, the moral, psychological and physical health of its employees, as well as employees' welfare, performance and positions within the organization (Akbaba-Altun, 2001;Argyris, 1958;Argyris, 1959;Miles, 1965;Ardıç & Polatcı, 2007, Uras, 2000Xenidis & Theocharous, 2014). Miles (1965) stated that it would not be enough for an organization to be considered healthy to exist only within its own boundaries for a certain period of time; and he defined the healthy organization as an organization that has been in existence for a long time continuously developing and extends its abilities of survive and overcome. It is seen that the first studies on organizational health were made at the beginning of the second half of the 20th century. It is thought that Argyris used the concept for the first time.
result, the loyalty to the organization increases.
3) Morale: It refers to the welfare or satisfaction of employees. Instead of feelings of discomfort, reluctance, tension and discontent, taking the center of feelings of happiness, satisfaction and pleasure and addressing these feelings raises the morale of the organization. In this way, employees look at life positively and their dominant personal reaction to events becomes "goodness".

Dimensions Related to Growth and Variability
In the third group, there are dimensions of innovation, autonomy, mutual harmony with the environment and problem-solving competence, which are related to growth and changefulness. 1) Innovativeness: A healthy system tends to find new methods, move towards new goals, produce new types of products, diversify itself and differentiate over time. In a sense, it can be said that a system with this tendency has grown, developed and changed rather than continuing in the normal course.
2) Autonomy: The healthy person moves "from his or her center towards to out" In other words, it establishes a relationship with the environment by taking its own self into consideration. He does not obey authority figures unconditionally. He does not take what he says to others as instruction and does not see it as a predictor of his own behavior. Similarly, the healthy organization will not respond to demands from outside in a destructive or unruly manner, but will not take a passive stance, feeling itself as part of the environment. As a result, it will tend to have some kind of independence from the environment.
3) Adaptation: There must be a certain harmony between what the organization presents to the environment that surrounds it and the expectations of the environment. When environmental demands and organizational outcomes do not match, a separate problem-solving and restructuring process begins, both within the environment and within the organization. Thus, the organization and the environment undergo an adaptation process. Healthy individuals, groups or organizations have a realistic and effective relationship with their environment. 4) Problem-solving adequacy: Even in healthy organisms, there can always be problems, difficulties, tensions. What matters is not the existence or absence of problems, but how the person, group or organization deals with these problems. Argyris (1964) stated that in an effective system, problems are solved with minimal energy, problems solved cease to be problems permanently, problem-solving mechanisms used are not weakened, on the contrary maintained or strengthened (as cited in Miles, 1965).

Dimensions of Organizational Health Scale
The dimensions of Organizational Health Scale developed with this study are briefly explained below.

Academic Emphasis
This dimension is concerned with factors aimed at revealing the academic status of the school. Determining the academic status of the students and the school from the perspective of the teachers working in the school, in other words, from the perspective of the employees working in the organization is considered within the scope of this dimension. Academic emphasis has been conceptualized as a key feature of effective schools that link productive teacher and student interactions. The guidance services and equipment support provided to the students are also handled within the scope of this dimension.

Morale
It is aimed to make sense of the relationship of teachers with each other and organizational climate. It is also within this dimension that the co-worker supports and the impact that the headmaster has on the teachers.

Supportive Leadership
This dimension deals with participatory school management. The indicators of the supportive school management are that the school administrator gives importance to the communication with the teachers, respect of the rights of the teachers and also includes the teachers in the decisions making. Preventing the emergence of unethical behavior in the school by the principal, and is respect teachers' rights, in terms of understanding the support given to teachers is important.

Environmental Factors
This dimension is related to organizational health indicators other than academic activities of the school. In this context, the school's safety and discipline system, physical conditions, supports that received from field experts and student parents are within this dimension.
The literature survey concluded that there are not enough valid and reliable scales related to the organizational ies.ccsenet.org International Education Studies Vol. 13, No. 7;2020 health developed based on teacher perceptions in Turkey and other countries. Therefore, it can be said that there is a need to develop a multidimensional scale that can be used to detect organizational health of schools. The aim of this research is to develop a valid and reliable measurement tool that can measure the organizational health of schools based on teacher perceptions.

Goal of the Research
The goal of this study is to develop a valid and reliable measurement tool for measuring the organizational health of schools based on the perceptions of teachers working in public secondary schools.

Method of the Research
The research is a scale development study and includes the development process of Organizational Health Scale developed by the researchers.

Participants
The participants of the study consists of 429 teachers working in secondary schools located in Odunpazarı and

Measurement Tool
In the first phase of the process of developing the "Organizational Health Scale", a literature review was conducted. In the literature review, the books written about organizational health (Akbaba-Altun, 2001;Hoy, Tarter, & Kottkamp 1991), theses (Akbaba, 1997;Ayduğ, 2014;Çiftçi, 2014;Özdemir, 2006;Gürkan, 2006;Karakuş, 2008;Kurum, 2013;Tacar, 2013;Taneri, 2011;Yıldırım, 2006;Yıldız, 2014) and articles (Argyris, 1959;Cemaloğlu, 2007;Güçlü, Recepoğlu, & Kılınç, 2014;Hoy & Feldman, 1987;Karaman & Akıl, 2005;Korkmaz, 2005;Korkmaz, 2006;Miles, 1965;Ordu & Tanrıöğen, 2013;Polatcı, Ardıç, & Kaya, 2008;Uras 2000) has been examined in this study. As a result of literature reviews, dimensions that can best measure the health of the organization have been tried to determine and in this direction, a pool of 61 items has been established based on institutional basis (Tezbaşaran, 1997). In the second stage, the draft scale was reduced to 43 items as a result of the elimination of the items that were not related to the dimensions that were based on the research, or which were estimated to have a low relationship and screening between similar expressions. In the third stage, the opinion of the field experts (Balcı, 2001) was taken for the scope and appearance validity of the measurement tool. In this context, 8 expert faculty members in the field of Educational Sciences (3 Prof. Dr., 5 Dr.) have been consulted for opinion. After the examination of the experts, the number of items was reduced to 42 with the regulation of the expressions on the said scale. In the fourth stage, in order to ensure the comprehensibility of the measurement tool in terms of language, the opinion of the Turkish language and literature expert was consulted. In line with the opinions on spelling rules and the use of punctuation marks, the scale items have been reviewed. Finally, without any dimensioning on the draft scale, the items were listed directly and 4 items were added to the scale to obtain personal information about the participants with a directive explaining the purpose of the study and the answers expected from the participants.
The draft scale prepared for implementation is called the Organizational Health Scale (OHS). The scale, which was designed to determine the responses of the participants with the Likert-type five-point rating consisted of I do not agree at all (1), I agree very little (2), I agree at medium level (3), I agree substantially (4), and I fully agree (5) options. To determine whether questions on the draft scale were understood by teachers, a pilot study was conducted into a group of 20 teachers. In this application, there was no criticism by the teachers of the pre-application of the items on the scale, and it was stated that the items were quite understandable. Thus the final form of the draft scale is given.

Data Analysis
In order to determine the validity and reliability of the measurement tool, questionnaire was applied to 448 teachers employed in secondary schools located in Odunpazarı and Tepebaşı districts of Eskişehir province by the researchers in the 2015-2016 academic year. In order to collect data in practice, necessary written permissions were obtained from the relevant authorities and the scales distributed were then collected by the researchers. When the filled scales are examined, it is found that 19 measuring tools were missing or incorrect (more than one option was checked). The number of scales to be analyzed was determined as 429 as a result of subtracting the incorrectly ies.ccsenet.org International Education Studies Vol. 13, No. 7; 2020 132 filled scales. After the data was obtained, statistical analyses were carried out to reveal the psychometric properties of the measurements. It is important that the sample group represents the universe. Therefore, the appropriate sample group and number should be determined.
In the literature, there are also opinions that suggest that the number of people to be applied in relation to the number of items in the scale, which can also be expressed as sample size, be determined. Some authors consider it sufficient for the number of people to be applied to be at least five times the number of items on the scale (Bryman & Cramer, 2001), but there are also authors who suggest that it should be 10 times (Nunually, 1978) or even 15 times (Gorusch, 1983, as cited in Delice & Ergene, 2015). According to the specified criteria, it can be said that the number of participants in the research group is sufficient for factor analysis.
The appropriateness of the correlation matrix for factor analysis was established using the Kaiser-Meyer-Oklin (KMO) test of sampling adequacy and the Bartlett test of Sphericity (BS). To determine the reliability status of the scale, Cronbach's alpha value, which determines the internal consistency measure, and item total correlations were examined. The structure validity of the OHS was examined by exploratory factor analysis (EFA). Exploratory factor analysis was used to determine whether OHS which is composed of 43 items, is single or multi-factor. Accordingly, principal components analysis and varimax rotation method were used.
Researchers often make use of confirmatory factor analysis (CFA), especially when the tests are supposed to be multidimensional. For this, a covariance matrix is calculated over the scores of a number of subjects and CFA is then applied to test whether a presumed factor structure or pattern is not contradicted by this matrix. CFA is executed by means of structural equation modeling (SEM), a very sophisticated statistical procedure for testing complex theoretical models on data. Since a computer program became available for SEM (LISREL), and this method has gained much in popularity. LISREL has been updated several times, and there are several similar programs available now, e.g., EQS, and Mplus (Prudon, 2015). In this study, data analysis and the suitability of the factor structure in the measurement tool was tested through the SPSS and LISREL. There are a lot of adaptation statistics in SEM literature and new ones are constantly being developed. Compliance is called the ability of a model to reproduce data, i.e. the variance covariance matrix. Because of the diversity in compliance statistics, it is important for the researcher to decide which compliance statistics are appropriate to use in the studies. In this study, Chi-Square, CFI, NFI, NNFI, IFI, GFI, PNFI, PGFI, AGFI, RMSEA and SRMR fit indices were used to test the suitability of the model.

Construct Validity
In order to test the construct validity of the measurements obtained from the OHS, explanatory factor analysis and confirmatory factor analysis were applied.

Explanatory Factor Analysis
Analysis studies were started by examining BS and KMO test results. The KMO test determines whether the selected sample data is appropriate to infer a factor, or the degree of conformity with a different expression. High test result value ranging from 0 to 1.0 means that each variable on the scale is perfectly predictable by the other variables on the scale (Giesen, 2004(Giesen, , as cited in Şencan, 2005. In this study, KMO value of the collected data .936 and BS test results were significant (p < .05, df = 861). These values indicate that the data are suitable for factor analysis. Eigenvalues were first examined to determine the number of factors. According to Köklü (2002), factors with eigenvalues above 1 and 1 should be considered stable.
In the first factor analysis based on the data within the scope of the research, emerged 6 factors with a factor eigenvalue greater than 1. Although factors with an eigenvalue above 1 are considered stable, Thompson (2004) stated that the scree plot reduces (determines) the number of factors more successfully than the eigenvalue (as cited in Çokluk, Şekercioğlu, & Büyüköztürk, 2014). When the scree plot was examined (Figure 1), it was observed that the slope started to flatten from point 4. Therefore, it was decided that the number of factors on the scale should be 4.

Figure 1. Scree plot (SPSS output)
In the second analysis, the rotation process was performed to better determine the factor loads of the items. "Varimax" was chosen as the rotation method. As a result of the second analysis, 11 items on the scale consisting of 4 factors (8,10,11,12,13,14,21,23,24,29 and 32) were found to be the confluent. Items with less than 0.1 differences between the load values of two or more factors are called confluent items. In exploratory factor analysis, component is undesirable because it is desirable that an item measure only one property (as cited in Çokluk, Şekercioğlu, & Büyüköztürk, 2014). For this reason, 11 items, which are confluents, were removed from the scale and the analysis was renewed.
The third analysis showed that there was no confluent item left on the scale. By examining the correlation between the items of the correlation matrix, the number of items with an acceptable relationship (r > 30) was quite high. In addition, the fact that the determinant of the matrix is 1.190 indicates that factor analysis is possible according to Can (2014) (Determinant>0.0001).
The existence of strong relationships between independent variables is called connection or multiple linear connections and shows the undesirable situation in regression analysis (Orhunbilge, as cited in Albayrak, 2005, p. 109). When the correlation matrix table is examined, 40 to 1 (0.804); 38 to 3 (0.839); 19 to 5 (0.829) 27 to 4 (0.807) and 36 to 2 (0.899), in items multiple co-linearities (r>0.8) between the items were observed. Çokluk, Şekercioğlu, and Büyüköztürk (2014) also stated that if items were decided to be removed in exploratory factor analysis-on the grounds that the removal of one item may result in changes in the factor load values of other items-the items should be excluded from the analysis one by one. Therefore, the communalities table was examined and the lower factor load was eliminated. When the communalities table is examined, it is seen that factor loads of items 1, 2, 3, 5 and 27 are lower than factor loads of items 4, 19, 36, 38 and 40. Therefore, these items were excluded from the analysis and then the analysis was renewed.
In the communalities table are basically given the rates of together explanation the variance in a common factor of each item. Among the items in this table, it should be carefully evaluated whether the items that are found to be of lower value than others are functioning (Çokluk, Şekercioğlu, & Büyüköztürk, 2014). When the table of Communalities was examined, the factor load of item 39 was observed to be quite low (0.226) compared to other substances. This item was excluded from the scale because it significantly reduced the mean of common variances (Can, 2014).
ies.ccsenet.org Vol. 13, No. 7;2020 In the analysis and evaluations, items 7 and 20, which were found to have the highest load value in different dimensions than theoretically expected dimensions, were removed from the scale in accordance with expert opinion.
The communalities of the items in the scale, the factor load values after the rotation process and the item total correlations obtained from the subsequent analysis are given in Table 1.The Factors which that forming the scale were named by making use of the studies Hart, Wearing, Conn, Carter, and Dingle (2000), Ardıç and Polatcı (2007), Polatcı, Ardıç, and Kaya (2008), Tutar (2010), Güçlü, Recepoğlu, and Kılınç (2014) and Guidetti, Converso, and Viotti (2015). In this context, the first dimension is called "academic emphasis"; the second dimension is called "supportive leadership"; the third dimension is called "morale" and the fourth dimension is called "environmental health". Communality values of the data obtained within the scope of the research were also examined. Communalities vary between .507 and .823 as shown in Table 1.According to Field (2005), in studies where the number of samples exceeds 250, the means communalities averages above 0.6 may increase the reliability of the criteria used in the research (as cited in Can, 2014). In this study, the community average was calculated as 16434/23 = 714. In accordance with this determination, there was no need to remove any item from the scale according to the communalities of the substances.
After rotation with Varimax method, it is observed that the factor loads of items in "academic emphasis" dimension changed between .607 and .851 factor loads of items in "supportive leadership" dimension changed between .632 and .895 factor loads of items in "morale" dimension changed between .729 and .865 and factor loads of items in "environmental factors" dimension changed between .476 and .812 It is generally desirable in researches that the factor loads of the items should be at least.45 (Büyüköztürk, 2010;Seçer, 2013). From these data, it is understood that factor load values of Organizational Health Scale are sufficient.
When the item total correlations of the Organizational Health Scale were examined (Table 2), the values were positive and generally high; however, it is seen that item 4 has a lower value (0.226) than other items. Since it was evaluated by the researchers that item 4 was necessary for the scale, it was decided to remain in the scale in accordance with the expert opinion. ies.ccsenet.org International Education Studies Vol. 13, No. 7;2020 In its final form, the Organizational Health Scale consists of 23 items (Appendix A). 7 of the 23 items in Appendix A (1, 2, 3, 4, 5, 6, 7) constitutes the Academic Emphasis factor; 6 (8, 9, 10, 11, 12, 13) are constitutes the Supportive Leadership factor; 5 (14,15,16,17,18) are constitutes the Morale factor and 5 are constitutes the Environmental Factors (19,20,21,22,23). In addition, the items 5, 16, 18 in scale are intended as reverse items. Thus, the scoring of these items was done in reverse order. The eigenvalues and variance ratios of the factors of the Organizational Health Scale (final version) are presented in Table 2. When Table 2 is examined, it is seen that the Organizational Health Scale, which took the final form before the confirmatory factor analysis with the exclusion of 19 items from the scale as a result of exploratory factor analysis, shows a four-factor structure. It is understood that the variance ratio explained by each factor is 20.681% in the first factor; 18.939% in the second factor, 18.774% in the third factor and 12,756% in the fourth factor. According to Table 2, the first three factors are stronger than the fourth factor in terms of the explained variance. Total variance explained by four factors was determined as 71.101%. "The variance ratio explained by a measurement tool must be higher than the unexplained variance ratio (Seçer, 2013)". In this case, the total variance explained is sufficient.
In order to allow the total score to be obtained from the whole organizational health scale or from each dimension separately, the ranges of scores and their meanings were determined. For this purpose, firstly the range coefficient was calculated. Likert-type five-point scoring technique is used in the scale, grading items "1 point", "2 points", "3 points", "3 points", "5 points" options. Based on the (Maximum measurement -smallest measurement)/desired number of groups rule, the range coefficient is determined as 5-1 = 4 and 4/5 = 0.80 and the option ranges are arranged in this way. The organizational health scale's score ranges, rating options and their meanings are given in Table 3. In the literature, many fit indices are used to determine the fit adequacy of the model tested in CFA. According to the results of goodness of fit tests, the model is accepted or rejected (Ayyıldız, Cengiz, & Ustasüleyman, 2006). In this study, Chi-Square, CFI, NFI, NNFI, IFI, GFI, PNFI, PGFI, AGFI, RMSEA and SRMR fit indices were used to test the model's suitability. It should be clear that these rule of thumb cutoff criteria are quite arbitrary and should not be taken too seriously. Fit indices may be affected by model misspecification, small-sample bias, effects of violation of normality and independence, and estimation method effects (Hu & Bentler, 1998as cited in Schermelleh-Engel & Moosbrugger, 2003 Therefore it is always possible that a model may fit the data although one or more fit measures may suggest bad fit (Schermelleh-Engel & Moosbrugger, 2003). The results of confirmatory factor analysis of Organizational Health Scale are given in Table 4 and Table 5. For the factor structure tested in DFA, modifications were made between items 41 and 35, 38 and 37 and 31 and 30 in line with the modifications recommendations for improvement. After modification, the Chi-square value of the scale (χ2 = 626.46, N = 429, df = 221, p = 0.00) significant; the ratio of χ2-sd was found to be χ2/sd = 2.83.  References: Schermelleh-Engel, Moosbrugger, and Müller, (2003), Hu and Bentler (1999), Kline (1998) cited in Wakslak, Jost, Tyler, andChen (2007), Marsh, Balla, andMcDonald (1988) cited in Ergül, Baydık, and Demir (2013), Meyers, Gamst, and Guarino (2006). Table 4 shows that NFI, NNFI, CFI and IFI fit index values are within the range of "good fit" while other fit indexes are within the range of "acceptable fit". The results of the confirmatory factor analysis of the Organizational Health Scale were also shown in Table 5. References: Mulaik et al. (1989), James, Mulaik, and Brett (1982), Schermelleh-Engel, Moosbrugger, and Müller (2003), Meyers, Gamst, and Guarino (2006).
As can be seen from Table 5 and Table 6, the findings show that the fit index values of the model are within the desired range. Considering these results, it can be argued that the four-factor structure obtained as a result of confirmatory factor analysis is an acceptable model. The diagrams of the model are shown in Figure 2 and 3. In the confirmatory factor analysis, t values of the model were examined. Parameter estimates are significant at 0.05 level if t-values exceed 1.96; If it exceeds 2.56, it is significant at 0.01 level. In the analysis carried out within the framework of the structural equation model, non-significant t-values should be excluded from the analysis (Çokluk, Şekercioğlu, & Büyüköztürk, 2014, p. 304). When the t-values of the Organizational Health Scale are examined (Figure 2), it was seen that the values ranged between 10.84 and 24.12, and the t-values of the scale were significant at 0.01 level.  It is useful to check the error variances in confirmatory factor analyzes. The error variance represents the unexplained portion of the variance for the data set. Error variances should not be too high (very close to 1.00) (Büyüköztürk, 2002;Çokluk, Şekercioğlu, & Büyüköztürk, 2014;Çepni, 2010, p. 50). When the error variances of the scale are examined (Figure 3), the value of the 4th item is slightly higher than the others (0.75); values of other items ranged from 0.18 to 0.56; however, it is seen that there is no item with very high error variance in the scale.  Vol. 13, No. 7;2020 separate concepts (Çepni, 2010). When the factor correlations of Organizational Health Scale are examined, it is seen that there is no value exceeding 0.85 ( Figure 3, Table 6). This shows that model-data fit can be achieved by 4 factors and 4 factors are separate concepts and that the results of exploratory factor analysis are validated.

Reliability
Cronbach Alpha is the most widely used objective reliability criterion. George and Mallery (2003) provided the following rules of thumb: "_ > .9 -Excellent, _ > .8 -Good, _ > .7 -Acceptable, _ > .6 -Questionable, _ > .5 -Poor, and _ < .5 -Unacceptable" (p. 231). While increasing the value of alpha is partially dependent upon the number of items in the scale, it should be noted that this has diminishing returns. It should also be noted that an alpha of .8 is probably a reasonable goal (Gliem & Gliem, 2003). The Cronbach's alpha values obtained of the organizational health scale are given in Table 7. As seen in Table 7, Cronbach alpha reliability coefficient of the organizational health scale was determined as 0.915.The Cronbach's alpha coefficients for the dimensions of the scale have values between .919 and .870. As shown in Table 8, the Cronbach alpha coefficient calculated for Academic Emphasis factor was .910, for Morale factor .919, for Supportive Leadership factor .918 and for Environmental Factors factor .870.

Discussion, Conclusion and Suggestions
In this study, it is aimed to develop a valid and reliable measurement tool for determining the organizational health of schools. The scale was prepared as a five-point likert type. The scale, which consisted of 42 items as a draft, was applied to a total of 448 teachers; however, it was found that 19 scales were answered inadequately or incorrectly (more than one choice was marked). Therefore, the total number of scales analyzed was determined as 429.
The results of the KMO and BS tests were examined before EFA was performed and it was concluded that the data was suitable for EFA as a result of the KMO value being less than .936 and BS value being less than p<.05.In order to determine the construct validity of the organizational health scale, 19 items were excluded from the scale and it was determined that the scale consisted of 4 factors. The scale was dimensioned according to these factors. While the first dimension of the scale (Academic) explained 20.681% of the total variance; second dimension (Supportive Leadership) 18.939%; third dimension (Morale) 18.774% and fourth dimension is explain 12,756% of the total variance. The total variance explained by the four dimensions (Environmental Factors) were determined as 71.101%. As a result of the DFA after EFA, the factor structure of the scale revealed by the EFA was confirmed and it has been seen that the scale is 4-dimensional. In the DFA, the fit indices of the scale were examined and The Chi-Square value (χ2 = 626.46, N = 429, DF = 221, p = 0.00) has been seen to be significant. When the studies and analyzes are evaluated together, it is possible to say that the organizational health scale is a valid and reliable measurement tool that can be used to determine the opinions of teachers working in schools on organizational health.