The current issue and full text archive of this journal is available at www.emeraldinsight.com/0951-354X.

htm

Examining culture and performance at different middle school level structures

Martin Omar Gomez
John OConnell High School, San Francisco, California, USA

Examining culture and performance 205

Received 25 February 2011 Accepted 7 April 2011

George A. Marcoulides
University of California, Riverside, Riverside, California, USA, and

Ronald H. Heck
Department of Educational Administration, University of Hawaii, Honolulu, Hawaii, USA
Abstract
Purpose The purpose of this study is to propose and test a model of school culture and examine data from schools in Southern California to identify educationally important aspects of teacher-perceived cultural variables and how these perceptions differentially impact school performance in K-8 and middle school structures. Design/methodology/approach Data were collected using a sample of 628 teachers from 59 schools (17 K-8 schools and 42 middle schools) in ve different schools districts in Southern California. The proposed model was tested using structural equation modeling techniques. Findings The proposed model was determined to t the data well. The theoretical and practical implications of the model concerning culture and school performance within the framework of educational management and school conguration are discussed. Originality/value This paper identies educationally important aspects of teacher-perceived cultural variables and how they impact school performance, and also it discusses the theoretical and practical implications of the proposed model. Keywords Teachers, Schools, Cultural techniques, Educational administration, United States of America Paper type Research paper

Introduction A number of different models for school structure have been used over the past century in the American educational school system to address issues with helping adolescents transition from elementary school settings to high school settings. At the turn of the twentieth century, completing a K-8 education was standard (kindergarten through eighth grade levels) for most individuals, with fewer than 10 percent of students graduating from high school (Tyack and Cuban, 1995). Although enrollment rates continued to increase overall for ve to 19-year-olds during the rst part of the century (i.e. from 51 percent in 1900 to 75 percent in 1940), even by the beginning of World War II, over 50 percent population had completed no more than an eighth grade education (see Snyder, 1993).
This research is based on the dissertation of the rst author submitted in partial satisfaction of the doctoral degree requirements at the University of California, Riverside.

International Journal of Educational Management Vol. 26 No. 2, 2012 pp. 205-222 q Emerald Group Publishing Limited 0951-354X DOI 10.1108/09513541211202004

IJEM 26,2

206

In an effort to assist students in transitioning to and graduating from high school, educators tinkered with various school structures that might better address the social and academic needs of 12-14 year old adolescents. During the early decades, the junior high school conguration (seventh through ninth grade levels) appeared. This conguration mirrored the departmentalized curricular structure of the high school. By the 1960s and 1970s, the middle school (sixth through eighth grade levels) approach became increasingly popular, with students beginning with one teacher during sixth grade and gradually transitioning to a departmentalized structure across core and elective courses in the seventh and eighth grades. This model was seen as more directly addressing the social and academic needs of adolescents as they transitioned between the elementary setting, typically with one teacher at one grade level, to the more complex departmentalized instruction typical of the high school setting. Recently, interest in the K-8 conguration has returned, mainly because some of the current research ndings concerning school outcomes for this particular model have been positive (Arcia, 2007; Byrnes and Ruby, 2007; Connolly et al., 2002; Juvonen et al., 2004; Offenberg, 2001; Weiss, 2008; Weiss and Kipnes, 2006; Yecke, 2006). One of the obvious advantages to this conguration is that students do not have to transfer to another school, so there is no break in their instruction. Previous results have noted that student achievement tends to dip after students transfer from elementary to middle or junior high school (Alspaugh, 1998). Positive results for K-8 schools compared to other school models include overall higher student test scores, increased involvement of parents and staff members, higher participation rates in extracurricular activities, greater levels of student leadership and self-esteem, student attendance in more prestigious high schools, increased safety and well being, fewer discipline issues, and more positive views of teachers (Arcia, 2007; Byrnes and Ruby, 2007; Connolly et al., 2002; Juvonen et al., 2004; Offenberg, 2001; Weiss, 2008; Weiss and Kipnes, 2006; Yecke, 2006). Other recent research results, however, has shown that K-8 schools do not always outperform middle schools, once a number of other educational process variables are considered in addition to more obvious differences in grade structure. This work suggests that the major factors in determining student achievement include a motivated staff and strong leadership, and not simply the conguration of the school or a change in the educational program (Balfanz et al., 2002; Erb, 2006; Viadero, 2008; Weiss, 2008; Whitley et al., 2007; Yecke, 2006). Research focus This study attempted to expand on this latter line of research and compare various dimensions of school culture in K-8 schools in contrast to middle schools, in order to explore whether or not the positive effects associated with the K-8 setting highlighted in the extant literature actually occur and, if so, to provide an understanding of the effects. We drew on research establishing a linkage between educational processes (values and beliefs, expectations, academic and organization, teacher practices) and school outcomes. These include processes related to the instructional environment surrounding classrooms, teaching practices, and school-level strategies aimed at academic improvement (Creemers, 1994; Kyriakides et al., 2010; Opdenakker and Van Damme, 2007). These various beliefs, attitudes, and processes are often referred to as a schools academic culture or capacity for producing quality school instructional practices and learning (Mulford and Silins, 2003).

More specically, we investigated the following: . the generalizability of a previously validated school culture model originally proposed by Heck and Marcoulides (1993) concerning how teacher perceptions of school culture explain student achievement in a sample of K-8 and middle schools in California; and, more specically; and . the hypothesis that K-8 schools exert stronger positive cultural inuences over school performance than middle schools do. The remainder of the paper is comprised of three sections. We discuss the studys participants and the constructs comprising our proposed theoretical model in the method section. We then provide the ndings of our model tests in the results section. Finally, we draw conclusions and suggest some implications of our research. Method Data A total of 59 schools from ve different school districts participated in the study. Out of the 59 schools, 17 were identied as K-8 schools and the remaining 42 were middle schools. Several criteria were used to select school districts for participation. Selected school districts had: . K-8 and/or middle schools; . higher than 60 percent free and reduced lunch participation; . higher than 20 percent of English language learners; and . parent Education Level greater than 1.90 (high school graduate) but less than 2.90 (college graduate). All data conrming the district and schools SES measured by percentage on free and reduced lunch, population, and CST scale scores were taken from the California Department of Education (2007) (www.cde.ca.gov) web site in the DataQuest feature. All schools with similar student SES levels (measured by percent of free and reduced lunch) and similar percentages of ELL (English Language Learner) students were surveyed in order to reduce the difference in student performance that is explained by those two variables. All teachers from each school site were contacted to participate in the study. The sample consisted of 628 teachers within the 59 schools who returned usable surveys (i.e. 154 K8 teachers and 474 middle school teachers). The response rates were similar across the two types of schools in the study; that is, 25.4 percent for K-8 teachers in the sample and 26.3 percent for middle school teachers (with approximately 11 teachers per school). Although return rates in such surveys of school processes are typically less than 50 percent, previous research has determined that stable structural equation modeling solutions can be obtained with samples ranging from ve to 15 teachers per school site. School performance was measured using mean score values on ELA, Algebra 1, Science and History in the eighth grade for the 2008-2009 school year. An arithmetic average value of the reported CST scores was calculated for each school and then each school was ranked from lowest score (one) to highest score (60). Sixth-grade CST scores for the same schools were also used to compute mean score value using data from the 2006-2007 school year. These data were examined in order to determine if there was a

Examining culture and performance 207

IJEM 26,2

208

drop in student achievement due to the transition to middle school (as has been noted in previous research). All CST information was collected from the California Department of Education (CDE) web site. Additionally, school demographic variables were also obtained from the CDE web site. These variables included percentage of non-white students, percentage of students in free and reduced lunch, percentage of students in the English Language Learners (ELL) program, percentage of students with disabilities, and average parent education level. Proposed model Figure 1 depicts the originally proposed theoretical model by Heck and Marcoulides (1996b) utilized in this study. As a group, the factors or latent variables (represented by ellipses) are viewed as loosely comprising students perceptions of the three subsystems of school culture (i.e. socio-cultural, organizational process, and individual beliefs). We consider organizational structure and organizational values to be givens in the context of the proposed model; that is, they are considered exogenous variables since their variability is determined by other variables outside of the proposed model. Of course, we recognize that organizational structure and values are themselves dynamic processes; however, since the data are cross-sectional, we reasoned that the schools structure and its values would be reective of wider sets of cultural values in the environment and, thus, relatively stable at any one point in time. In contrast, we consider organizational climate, managerial processes, and teacher attitudes to be endogenous variables, in that other variables in the proposed model determine their variability. The exogenous variables, therefore, indirectly affect organizational performance through the endogenous variables in the model. In order to describe school achievement through organizational culture, Heck and Marcoulides (1996b) developed the The Organization of the School and Teacher Satisfaction with Their Work Environment: A Survey of Secondary School Teachers in Singapore (OSTSWE). The OSTSWE was created based on research by Marcoulides and Heck (1993a), which is a study about how organizational culture can make a difference in for-prot organizational productivity. The instrument was designed to measure a variety of strategic interactions between principals and teachers, focusing on how the school is structured and governed, how it is organized instructionally, and how teachers perceive

Figure 1. Proposed model of organizational culture and school performance

elements of its culture and climate. The OSTSWE has already been shown to possess good psychometric properties (e.g. Heck and Marcoulides, 1996b). Moreover, these previous examinations using the instrument suggest that 75-80 percent of the variance in perceptions about culture lies within schools. Given that we were interested in examining differences in teacher perceptions about two types of school structures, we conducted our examination of organizational culture at the individual teacher level. Constructs comprising the model In the OSTSWE, the rst important cultural subsystem (the sociocultural subsystem) is measured by the following two factors: organizational structure and managerial processes. Organizational structure is intended to measure the attitudes and beliefs related to the organizations structure and operational processes implemented to achieve desired results. Organizational structure is measured by one scale: level of bureaucracy and this scale is measured utilizing ve questions on the survey instrument. Low scores for organizational structure are associated with less rigidity and more teacher autonomy. Managerial processes is intended to measure how the organization functions over time as a result of its particular structure, purposes, value and belief systems, and is measured by three questions on the OSTSWE: availability of resources (measured by three questions on the survey instrument), administrative responsiveness (measured by four questions on the survey instrument), and principals leadership (measured by six questions on the survey instrument). The second important component of culture, the organizational value subsystem, is measured by the following factors: organizational values and organizational climate. The organizational values factor is intended to measure principles, ideologies, and activities representing values thought to be important in achieving school productivity. On the OSTSWE, organizational values are measured by three scales: time for teacher collaboration and meetings among teachers (measured by three questions on the survey); support for innovation (measured by three questions on the survey); and encouragement for teacher participation in decision making (measured by three questions on the survey instrument). The second factor, organizational climate, is intended to measure teachers perceptions of how things are on a day-to-day basis regarding a variety of topics. Organizational climate is measured by three questions: willingness to socialize with staff (measured by six questions on the survey), open staff communication (measured by three questions on the survey), and teacher collegiality and the availability of help (measured by four questions on the survey). The third important component of culture, the individual belief subsystem, is composed of teacher attitudes. The teacher attitudes factor is measured within the areas of teacher classroom instruction, student academic ability, and student background. On the OSTSWE, teacher attitudes is measured by two scales: perceptions about students capabilities to learn academic material and the attitudes they bring from home (measured by three question on the survey), and teacher perceptions of the parents support of their child, the teacher, and the school (measured by three questions on the survey). Finally, school performance (which is the dependent variable in the proposed model), describes the level of academic performance within the school, and was measured in this study by the CST 8th grade mean scale scores of students obtained in English Language Arts (ELA), Algebra 1, Science and History for the 2009 school year.

Examining culture and performance 209

IJEM 26,2

210

Results The proposed model was tested using structural equation modeling (SEM) techniques reskog and So rbom, 2006). The analysis implemented through LISREL 8.80 ( Jo attempted to determine whether or not the K-8 and middle school data t the originally proposed Heck and Marcoulides (1996b) model. The assessment of model t was conducted by examining the Comparative Fit Index (CFI), Goodness of Fit Index (GFI), Normed Fit Index (NFI), Chi-Square Test, the Root Mean Square Error of Approximation (RMSEA), and the 90 percent Condence Interval of the RMSEA t criteria. These t indices were selected due to their widespread use and their usefulness in comparing samples of unequal sizes (Marcoulides and Heck, 1993b). It is commonly accepted that individually examined values for the CFI, GFI, and NFI approaching or above 0.95 are expected for good model t. In contrast, RMSEA values close to 0.05 and small condence intervals with the left tail end including the value zero are indicative of good model t. Additionally, a non-signicant value obtained for the chi-square (x 2 ) test and evaluated through its corresponding p-value is indicative of good model t. However, because it is well known that the chi-square test is notoriously sensitive to sample size and has a tendency to reject models that are only marginally inconsistent with the data, more emphasis is placed on the other t criteria (Raykov and Marcoulides, 2008). Since the study utilized an a priori dened model to be tested, a rst objective lies in determining the adequacy of model t to the MS and K-8 data. Once model t is determined, then the signicance of the various parameter estimates can be ascertained for each school structure type. If the model does not show adequate t to either the considered MS or K-8 data, then the paths and even the variables included in the model might potentially have to be re-conceptualized. On ascertaining good model t separately to the MS and K-8 data, a subsequent test of model invariance will be used in order to compare if the parameter estimates are the same for the two school structure types. Tests of model invariance allow researchers to study potential group differences and expand the literature as it attempts to add insights into the construct validity of measures (Marcoulides and Heck, 1993b). In particular, these tests enable a discussion with regards to the potential similarities and differences within the proposed model across the K-8 and middle school data when attempting to explain how culture variables impact school achievement. The value of a proposed theoretical model is said to be greatly-enhanced, if the same model can be replicated in samples from the same or from different populations (Heck and Marcoulides, 1990). On determining overall model invariance, parameter estimates can then also be compared in order to examine the contribution of each observed and latent variable to the overall model across the K-8 and middle school types. Assessing the magnitudes of the parameter estimates can provide information with regards to the strongest and weakest path estimate of culture variables (direct and indirect) on school achievement in the proposed model (Koufteros and Marcoulides, 2006; Marcoulides and Heck, 1993b). Additionally, assessing the magnitude of the parameter estimates of the paths in the model provides insight concerning the predictive power of any one variable on another. In order to compare the achievement differences between MS and K-8 schools included in this study, tests of group difference using t-tests will also be conducted. These t-tests compare the mean scaled CST scores of the K-8 schools against those of the

middle schools at both the sixth (mean CST scores on Math and ELA) and the eighth grade level (mean CST scores on Algebra 1, ELA, Science, and Social Studies). Examining the results of these t-tests provides insight regarding which type of middle-level structure may be signicantly outperforming the other by comparing the averages on mean CST scores at the sixth grade level and then at the eighth grade level. Table I provides descriptive information about the observed variable means and standard deviations across both teacher groups. The observed variable means, standard deviations, and mean comparison t-statistic with corresponding p-value are presented for each parceled variable included in the model. The observed variables are listed according to which particular latent variable (italicized) they measure. They focus on teacher perceptions regarding: . their schools level of bureaucracy; . how much innovation is encouraged by school administrators; . the principals responsiveness to teacher issues; . their social relationships with their peers; and . their collegiality with their peers were similar across both school groups. More specically, MS teachers, more than K-8 teachers, believe they are provided more: . time for collaboration; . situations to participate in educational decision making; . availability to resources; . effective principal leadership practices; and . open communication with their peers.
K-8 Mean (n 154) 13.93 9.62 11.00 10.12 8.86 13.18 21.74 23.42 10.01 13.99 10.43 MS Mean (n 474) 14.10 10.60 11.40 10.99 9.89 13.63 23.11 24.10 10.81 14.20 9.19

Examining culture and performance 211

Variables Organizational structure Degree of bureaucracy (x5) Organizational values Time for collaboration (x11) Encourage innovation (x12) Participate in decisions (x13) Managerial processes Availability of resources (x8) Principal responsiveness (x9) Principal leadership (x10) Organizational climate Social relationships (x2) Open communication (x3) Collegiality of teachers (x4) Teacher attitudes Perceptions of parents (x7)

SD 3.03 2.97 2.31 3.10 3.01 3.01 6.54 4.06 3.57 2.16 3.21

SD 3.41 2.75 2.28 2.97 2.87 2.75 5.45 4.04 3.14 2.09 3.02

t-value 2 0.54 2 3.70 2 1.82 2 3.11 2 3.79 2 1.70 2 2.57 2 1.80 2 2.63 2 1.03 4.32

p-value 0.59 0.00 0.70 0.00 0.00 0.90 0.01 0.72 0.01 0.30 0.00

Table I. Descriptive statistics of variables in both groups

IJEM 26,2

212

In contrast, K-8 teachers tended to have higher perceptions of parental support than teachers in the MS group. Table II presents the obtained criteria t for the proposed model examined across three separate snapshots of the collected data: (1) The entire sample of teachers regardless of school structure. (2) Teachers from the k-8 schools alone. (3) Teachers from the MS structure alone. As can be seen by examining these t indexes, with the expected exception of the signicant chi-square value, all the measures of model t indicated good t of the model to each of the three snapshots of the collected data. Thus, it is clear that the a priori proposed model ts the obtained data at the total sample level, the K-8 level, and the MS level. The parameter estimates, standard errors, and t-values for the endogenous and exogenous latent variables obtained for the total sample and the separate K-8 and MS data sets are summarized in the Appendix 1 (see Tables AI and AII). All factor loadings for the latent variables included in the model were found to be statistically signicant as determined by examining their t-statistic values and ensuring that the t-value was greater than 1.96 (the critical value at a 0:05; Raykov and Marcoulides, 2008). The obtained factor loadings provide evidence that the variables included in the proposed model are important measures of the considered latent variables. All parameter estimates exhibited small standard errors conrming that the observed variables are stable estimates of the latent variables (see Appendix 2 (Table AIII) for detailed information concerning the parceled variables and survey questions that were used to measure each latent variable). The obtained parameters estimates for the path coefcients examining the impact of the exogenous latent variables (e.g. Organizational Values) on the endogenous latent variables (e.g. Teacher Attitudes) are presented in Figure 2. The rst provided value corresponds to the parameter estimate, the value below in parenthesis to the corresponding standard error, and the third the corresponding t-test value. Generally speaking, an effect size of 0.2 to 0.3 is considered to be a small or weak effect, around 0.5 a medium effect, and 0.8 to 1.0, a large or strong effect. We contrast the effects across K-8 and middle schools in Figure 2. The direct effect of organizational values on managerial processes was found to be statistically signicant across the both the K-8 and MS groups. The obtained values for K-8 [0:65; t 628 17:71; p , 0:01] and MS [0:64; t628 26:62; p , 0:01], respectively,
Total sample (n 628) Chi-Square (x 2 ) p-value Goodness of t index (GFI) Normed t index (NFI) Comparative t index (CFI) Root mean square error of approximation (RMSEA) 90 percent condence interval for RMSEA 332.17 0.00 0.97 1.00 1.00 0.10 (0.09; 0.11) K-8 (n 154) 141.09 0.00 0.96 1.00 1.00 0.12 (0.09; 0.14) MS (n 474) 267.67 0.00 0.96 1.00 1.00 0.10 (0.08; 0.11)

Table II. Measures of model t

Examining culture and performance 213

Figure 2. Examining standardized structural coefcients in the proposed model between K-8 and middle schools

indicate that teachers in both school structures reported that the latent variable, organizational values, had similar direct effects and is considered to be a strong predictor of managerial processes. As specied previously, this parameter estimate can be interpreted just like a regression coefcient and indicates that for every one unit change in the measure of organization values there is respectively in each school structural type a corresponding 0.64 or 0.65 increase in the value of the measure of managerial processes. This nding essentially implies that teachers that are given time for collaboration, are encouraged to be innovative, and participate in school wide decisions also had similar access to the resources they needed and a principal that was responsive with effective leadership practices. The direct effect of organizational values on teacher attitudes was also found to be statistically signicant across the K-8 and MS considered groups. Nevertheless, for K-8 teachers, the latent variable organizational value (0.83) was more important in predicting teacher attitudes than for the MS teachers (0.30). K-8 schools with more innovation, and more teachers participating in decision-making were much more likely to be associated with higher teacher perceptions of the parental support that their students have. The direct effect of organizational values on organizational climate was also found to be signicant across the K-8 and MS teacher groups. For both groups the latent variable organizational values appeared to be equally important in predicting organizational climate (0.76 and 0.72). Thus, K-8 and MS teachers from schools with high organizational values were similarly associated with teachers in schools that exhibited high teacher relationships, communication, and collegiality. All other paths considered in the model were found to be non-signicant. For example, the magnitude of the path between organizational structure and managerial processes was found to be 2 0.01 for the K-8 teachers and 2 0.09 for the MS teachers. We noted similar non-signicant results for the paths between organizational structure and organizational climate. The parameter estimates for the K-8, and MS groups between the endogenous variable paths are also presented in Figure 2. As can be seen by examining the results

IJEM 26,2

214

displayed in Table III, the magnitude of the path between organizational climate and teacher attitudes was similar for MS teachers [0:60; t628 7:50; p , 0:01] and K-8 teachers [0:37; t628 4:62; p , 0:01]. This suggests that, the impact of organizational climate on teacher attitudes is perceived as being similar by MS and K-8 teachers. Thus in both groups, teacher perceptions of their relationships, communication, and collegiality with their colleagues were positively associated with perceptions of parental support that their students have. The parameter estimate between managerial processes and teacher attitudes was found to be signicant for teachers from both types of school structures. Nevertheless, for teachers from K-8 schools [0:26; t154 2:31; p , 0:01] it appears to be somewhat less important than for teachers in the MS conguration [0:64; t474 2:67; p , 0:01]. MS teachers and K-8 teachers associate higher levels of principal leadership methods to higher levels of teacher perceptions of their students parental support. The magnitude of the path between managerial processes and organizational climate was also found to be signicant in both the K-8 and MS teacher groups (K-8 0.42, MS 0.46). Thus both teachers from both K-8 and MS congurations perceive principal actions to be similarly important in explaining the schools collegiality and communication. The direct path between organizational climate and school performance was determined to be signicant only in the K-8 teacher group [0:33; t 154 2:40; p , 0:01]. It appears that organizational climate is not perceived by MS teachers as impacting school performance. Similar differences between the perceptions of teachers from K-8 and MS school congurations were also determined for the path between teacher attitudes and school performance [0:17; t154 3:93; p , 0:01]. This result indicates that K-8 teachers were more likely to feel that higher teacher perceptions of students parental support were predictive of school performance. Thus, attitudes of teachers in K-8 school congurations were a better predictor of school performance than attitudes of teachers in middle schools. All other paths considered in the model were found to be non-signicant at p , 0:05. In addition to examining direct effects, the magnitude of indirect effects in the model can also be investigated. Based on a review of the literature, a number of indirect effects on school performance were examined in this study (see Table III for a complete list of the indirect effects examined in this study). It is noted that the magnitude of the indirect effect can be readily computed using the so-called tracing rule (which in most cases is essentially the product of the corresponding path coefcients between various direct effects). For example, to determine the indirect effect of organizational values on school performance (0.25) for K-8 teachers, the product of the direct effects between organizational values on organizational climate (0.76) and between organizational climate and school performance (0.33) is taken. As can be seen by examining the indirect values displayed in Table III, the magnitude of the indirect effects of organizational values on school performance via organizational climate and on school performance via managerial processes were sizeable for the K-8 teachers. In the MS teacher sample the indirect effect for organizational values on school performance via managerial process was found to be sizeable. For the total sample, the magnitudes of indirect effects of organizational values on school performance via organizational climate and on school performance via teacher attitudes, as well as the indirect effect of organizational climate to school

Total sample (n 628) (0.74) (0.42) 0.31 (0.80) (0.62) 0.50 (0.65) (0.19) 0.12 (2 0.03) (0.42) 2 0.01 (0.81) (0.62) 0.50 (0.83) (0.17) 0.14 (0.65) (0.24) 0.16 (2 0.03) (0.33) 2 0.01 (0.37) (0.17) 0.06 (0.76) (0.33) 0.25

K-8 (n 154)

MS (n 474) (0.72) (0.05) 0.04 (0.30) (0.24) 0.07 (0.64) (0.54) 0.35 (2 0.06) (0.05) 2 0.003 (0.60) (0.24) 0.14

Organization values ! Organizational climate ! School performance Organizational values ! Teacher attitudes ! School performance Organizational values ! Managerial processes ! School performance Organizational structure ! Organizational climate ! School performance Organizational climate ! Teacher attitudes ! School performance

Examining culture and performance 215

Table III. Indirect effects to school performance

IJEM 26,2

216

performance via teacher attitudes were sizeable. All remaining indirect effects were noted to be of little importance in predicting the various outcome variables considered. A t-test comparing the combined mean scaled CST scores of ELA and math achievement between K-8 and middle schools at the sixth grade level indicated that the sixth grade students at K-8 schools (m 335:58; SD 27:18) did not signicantly outperform their middle school sixth grade peer schools [(m 332:72; SD 20:36); t 626 1:39; p 0:165]. However, a second independent sample t-test comparing the mean scaled scores between achievement (average of ELA, Algebra 1, Science, and Social Studies CST scores) of K-8 and middle schools at the eighth grade level indicated that eighth grade students in K-8s ( m 344:24; SD 33:22) signicantly outperformed eighth grade students in middle schools [m 338:15; SD 23:72); t 626 2:49; p 0:013]. It is noted that the sixth grade CST scaled scores were taken during the 2006-2007 school year and the eighth grade scores were taken from the 2008-2009 school year in order to theoretically study the same cohort of students. Discussion The purposes of this study were to: . examine the generalizability of a proposed model of organizational culture and its relationship to student outcomes; and . determine whether organizational processes that might explain differences in achievement outcomes in different school structures serving 12- to 14-year-old students. These types of school settings have not been as frequently studied as either elementary or high school settings. Our results contribute to an understanding of educational processes and outcomes at middle grade levels in at least two important ways. First, the study provides support concerning the construct validity of the originally proposed Heck and Marcoulides model (1996b) regarding the effects of organizational processes comprising school culture on school outcomes across teachers from both K-8 and MS school structures. More specically, it indicates that the component dimensions of a schools academic and social culture are positively related to each other and to student outcomes in schools serving young adolescents. This provides additional evidence of the proposed models construct validity in describing educational processes and outcomes in these types of schools. This was particularly evident in terms of the positive indirect effects of organizational values and school climate to explaining school outcomes, and the direct effect of teacher attitudes in being positively related to outcomes regardless of school type (see also Heck et al., 1990; Leithwood and Mascall, 2008; Leithwood et al., 2010). Overall, this provides support for the view that organizational consists of several related subsystems (i.e. sociocultural, organizational, individual), which can be positively related to outcomes. Our proposed model suggests all three subsystems are relevant to understanding differences in patterns of belief, behavior, and outcomes among the schools investigated. Second, given the additional evidence that K-8 schools outperform middle schools at the eighth grade level, our results also help illuminate specic aspects of teacher-perceived cultural variables within K-8 and MS structures that are important and further suggest possible paths through which such perceptions about

organizational processes may impact school performance. This provides initial evidence for considering which aspects of K-8 and middle school settings may yield positive academic benets for students. Our results therefore provide initial clues about which components parts of schools organizational culture may be amenable to school-level action directed at improving outcomes. Since the examined model was found to t the teacher data from both types of school structures and because a statistically signicant difference was determined in the eighth grade performance, one can compare the specic aspects of the model across the two teacher groups and conrm which paths differ and infer that the higher performance may indeed be explained by the higher teacher perception on that particular path. Because some variables emerged as better predictors of performance according to teachers from each school type, strategic school improvement might result from becoming more aware of teachers perceptions about key organizational processes (e.g. academic orientation underlying organizational values, organizational climate, teacher attitudes) and also consider ways to change those aspects that appear to inuence school performance. In particular, in K-8 settings, we noted considerable continuity of direct and indirect relationships involving teachers perceptions of their schools academic values, their climates, and teacher attitudes in being more strongly related to outcomes. These paths were not as clearly evident in middle school settings, although we did note that the relationship between organizational values, managerial processes, and student outcomes was positive and substantial, and the indirect relationship between climate, teacher attitudes, and outcomes was also positive. It is not clear from these data, however, what might be possible reasons for systematic differences in organizational relationships across these two types of settings. We caution that simply changing leadership practices to mirror successful schools may not necessarily result in direct specic changes in outcomes; however, it is clear that discussions of best practices by leaders and staff, at least as perceived by teachers, must at least begin with an understanding of the important variables inuencing school outcomes (Heck and Marcoulides, 1996a; Leithwood et al., 2010; Leithwood and Jantzi, 1997; Leithwood and Mascall, 2008). The proposed model examined and tested in this study across teachers from two different types of middle level school structures compared quite favorably with previous studies of organizational culture and performance in for-prot organizations (Marcoulides and Heck, 1993a), at the high school level in Singapore (Heck and Marcoulides, 1996b), and across elementary and high school level data obtained from the Trends in International Mathematics and Science Study (TIMSS) data (Marcoulides et al., 2005). Although the overall model tested in this study was determined to t equally well with both types of middle school structures considered, the actual magnitude of the parameter estimates within the model reecting the effects across specic latent variables and on the outcome variable differed somewhat across the two school types. Based on results obtained in past research studies, it appears that the observed variation of the parameter estimates across the two school types found in this dissertation can most likely be attributed to different school contexts considered. This inferred difference as a function of the school context is further supported by the fact that the model t and parameter estimates obtained on the total teacher sample (i.e. when ignoring the two considered school structures) were quite similar to those

Examining culture and performance 217

IJEM 26,2

218

obtained for the originally proposed Heck and Marcoulides (1996b) model. Nevertheless, once the separate teacher groups (school conguration) were introduced into the analysis, the model did not t the data from the two school structures in the same manner. Indeed, since the actual magnitude of effects within the model also changed once school conguration was taken into account, it is clear that the school conguration is the key factor bringing about these changes. For example, similar to the originally proposed model, in the total teacher sample the sole important direct factor in explaining school performance was teacher attitudes. In contrast for the K-8 group, the most important variables explaining school performance were organizational climate and teacher attitudes. In the MS group no signicant direct effects to school performance were determined. It should be emphasized that a number of the observed variations between the middle and K-8 schools in the examined model did not follow some of the current middle level literature. For example, K-8 teachers were found to have higher perceptions of parental support (see details provided in Table III) due to the longer period of time that they work with the same parents and thus have more chances to connect with them compared to the MS teachers. Because K-8 teachers worked with the same parents for a longer period of time (nine years at the K-8 versus three years at the MS), however, our data suggest that they may become more frustrated with the lack of support by the principal to connect the parents with the school. In other words, although there are more parental contacts within the K-8 school structure, the magnitude of this path may suggest that the K-8 teachers believe that more should be done by the schools leadership to actually connect the community to the schools in order to follow what was found in the literature review. A second example refers to the fact that organizational structure (i.e. middle school versus K-8) did not appear to affect outcomes indirectly in the model. Although Becker et al. (2009) found that efcacious adult-student relationships are easier to foster within the K-8 structure, this study determined that organizational structure did not have an indirect effect on school performance for either school structure type. In fact, non-signicant parameter estimates were observed for both K-8 and MS schools. However, the lack of inuence of grade conguration on school performance found in this study is quite consistent with recent results reported by Williams et al. (2010). Because structural types do not appear to account directly for differences in outcomes, but process variables do, as Heck and Marcoulides (1993) have suggested, instead of comparing the pros and cons of different middle level structures, especially since organizational structure has been found not to affect school performance, school leaders are encouraged to focus on school climate and on school-wide student improvement goals in order to raise teacher attitudes and collegiality. Improving teacher attitudes and collegiality is important as it is those variables that have been found to impact school performance in both school structures. As they note Because time in the school day is limited, there are trade-offs associated with school management (p. 26). Hence, deciding on which variables are important in facilitating strong educational outcomes becomes a crucial aspect of school leadership aimed at facilitating school academic improvement (Leithwood et al., 2010). Our results should be considered in relation to a few limitations. First, we were not able to obtain direct information about how teachers instruct in their classrooms. It is likely that differences in teachers classroom effectiveness have a direct effect on

student learning (as well differences in their attitudes). Second, although concerns can be raised about any attempt to reduce the complexity of complex day-to-day educational relationships in schools to a model of organizational culture and estimate its relationship to outcomes, there is a benet in doing so if it helps to identify paths that might provide insight to strategic actions that can improve classroom instruction and the environment surrounding the schools instructional practices. These two areas have been identied in a recent meta-analysis of school factors that inuence outcomes (Kyriakides et al., 2010). Further research related to schools leadership, academic and social organization, and values may help illuminate strategies for improving educational practices in schools that lead to increased outcomes for early adolescent students. This is because school climate, teacher attitudes, and organizational values appear to explain school performance more than the other latent variables (e.g. structure) in the considered model.
References Alspaugh, J.W. (1998), Achievement loss associated with the transition to middle school and high school, Journal of Educational Research, Vol. 92 No. 1, pp. 20-5. Arcia, E. (2007), A comparison of elementary/K-8 and middle schools suspension rates, Urban Education, Vol. 42 No. 5, pp. 456-69. Balfanz, R., Spridakis, K. and Neild, R.C. (2002), Will Converting High-Poverty Middle Schools to K-8 Schools Facilitate Achievement Gains?, Philadelphia Education Fund, a Research Brief for the School District of Philadelphia, November. Becker, N.C., Gomez, M. and Zykowski, J. (2009), Achieving a delicate balance: a guide to the advantages, disadvantages, and unique challenges of educating middle-grade students in a contemporary K-8 school in Southern California, Graduate School of Education, University of California, Riverside, CA, May. Byrnes, V. and Ruby, A. (2007), Comparing achievement between K-8 and middle schools: a large-scale empirical study, American Journal of Education, Vol. 114, pp. 101-35. California Department of Education (2007), Fact Book 2007. Handbook of Education Information, California Department of Education, Sacramento, CA, (online), available at: www.cde.ca. gov Connolly, F., Yakimowski-Srebnick, M.E. and Russo, C.V. (2002), An examination of K-5, 6-8 versus K-8 grade congurations, ERS Spectrum, Vol. 20 No. 2, Spring, pp. 28-37. Creemers, b. (1994), The Effective Classroom, Cassell, London. Erb, T.O. (2006), Middle school models are working in many grade congurations to boost student performance, American Secondary Education, Vol. 34 No. 3, pp. 4-13. Heck, R.H. and Marcoulides, G.A. (1990), Examining the contextual differences in the development of instructional leadership and school achievement, The Urban Review, Vol. 22 No. 4, pp. 247-65. Heck, R.H. and Marcoulides, G.A. (1993), Principal leadership behaviors and school achievement, NASSP Bulletin, May, pp. 20-8. Heck, R.H. and Marcoulides, G.A. (1996a), The assessment of principal performance: a multilevel evaluation approach, Journal of Personnel Evaluation in Education, Vol. 10 No. 1, pp. 11-28. Heck, R.H. and Marcoulides, G.A. (1996b), School culture and performance: testing the invariance of an organizational model, School Effectiveness and School Improvement, Vol. 7 No. 1, pp. 76-96.

Examining culture and performance 219

IJEM 26,2

220

Heck, R.H., Larsen, T.J. and Marcoulides, G.A. (1990), Instructional leadership and school achievement: validation of a causal model, Educational Administration Quarterly, Vol. 26 No. 2, pp. 94-125. reskog, K.G. and So rbom, D. (2006), LISREL 8.80, Scientic Software, Chicago, IL. Jo Juvonen, J., Le, V., Kaganoff, T., Augustine, C. and Constant, L. (2004), Focus on the Wonder Years: Challenges Facing the American Middle School, RAND Corporation, Santa Monica, CA, available at: www.rand.org/. Koufteros, X. and Marcoulides, G.A. (2006), Product development practices and performance: a structural equation modeling-based multi-group analysis, International Journal of Production Economics, Vol. 103, pp. 286-307. Kyriakides, L., Creemers, B., Antoniou, P. and Demetriou, D. (2010), A synthesis of studies searching for school factors: implications for theory and research, British Educational Research Journal, Vol. 36 No. 5, pp. 807-30. Leithwood, K. and Jantzi, D. (1997), Explaining variation in teachers perceptions of principals leadership: a replication, Journal of Educational Administration, Vol. 35 No. 4, pp. 312-24. Leithwood, K. and Mascall, B. (2008), Collective leadership effects on student achievement, Educational Administration Quarterly, Vol. 44 No. 4, pp. 529-61. Leithwood, K., Patten, S. and Jantzi, D. (2010), Testing a conceptualization of how school leadership inuences student learning, Educational Administration Quarterly, Vol. 46 No. 5, pp. 671-706. Marcoulides, G.A. and Heck, R.H. (1993a), Organizational culture and performance: proposing and testing a model, Organization Science, Vol. 4, pp. 209-25. Marcoulides, G.A. and Heck, R.H. (1993b), Examining administrative leadership behavior: a comparison of principals and assistant principals, Journal of Personnel Evaluation in Education, Vol. 7, pp. 81-94. Marcoulides, G.A., Heck, R.A. and Papanastasiou, C. (2005), Student perceptions of culture and achievement: testing the invariance of a model, International Journal of Educational Management, Vol. 19 No. 2, pp. 140-52. Mulford, B. and Silins, H. (2003), Leadership for organizational learning and improved student outcomes: what do we know?, Cambridge Journal of Education, Vol. 33, pp. 175-95. Offenberg, R. (2001), The efcacy of Philadelphias K-to-8 schools compared to middle grades schools, Middle School Journal, March, pp. 23-9. Opdenakker, M.C. and Van Damme, J. (2007), Do school context, student composition and school leadership affect school practice and outcomes in secondary education?, British Educational Research Journal, Vol. 33, pp. 179-206, available at: www.informaworld.com Raykov, T. and Marcoulides, G.A. (2008), An Introduction to Applied Multivariate Analysis, Taylor & Francis, New York, NY. Snyder, T. (1993), 120 Years of American Education: A Statistical Portrait, National Center for Education Statistics, Washington, DC. Tyack, D. and Cuban, L. (1995), Tinkering Toward Utopia: A Century of Public School Reform, Harvard University Press, Cambridge, MA. Viadero, D. (2008), Evidence for moving to K-8 not airtight, Education Week, Vol. 27 No. 19, pp. 11-13. Weiss, C.C. (2008), Re-examining middle school effects: what the research says, Principal, Vol. 87 No. 3, pp. 60-1.

Weiss, C.C. and Kipnes, L. (2006), Reexamining middle school effects: a comparison of middle grades students in middle schools and K-8 schools, American Journal of Education, Vol. 112, pp. 239-72. Whitley, J., Lupart, J.L. and Beran, T. (2007), Differences in achievement between adolescents who remain in a K-8 school and those who transition to junior high school, Canadian Journal of Education, Vol. 30 No. 3, pp. 649-69. Williams, T., Kirst, M.W. and Haertel, E. et al. (2010), Gaining Ground in the Middle Grades: why Some Schools do Better, EdSource, Mountain View, CA. Yecke, C.P. (2006), Mayhem in the middle: why we should shift to K-8, Educational Leadership, Vol. 63 No. 7, pp. 20-5.

Examining culture and performance 221

Further reading California Teachers Association (2006), Trend toward reconguration puts middle schools in danger of extinction, available at: www.cta.org Raykov, T. and Marcoulides, G.A. (2006), A First Course in Structural Equation Modeling, 2nd ed., Lawrence Erlbaum Associates, Mahwah, NJ. Viadero, D.K-8. (2006), K-8 structure gives no academic boost, analysis nds, Education Week, Vol. 25 No. 25, pp. 5-6. Yecke, C.P. (2005), Mayhem in the Middle: How Middle Schools have Failed America and How to Make them Work, Thomas R. Fordham Institute, Washington, DC, available at: www. edexcellence.net/foundation/global/found.cfm?author &keyword mayhem in themiddle

Appendix 1
Observed variable Managerial processes Availability of resources (x8) Principal responsiveness (x9) Principal leadership (x10) Teachers attitudes Perceptions of students (x7) Organizational climate Social relationships (x2) Open communication (x3) Collegiality of teachers (x4) School performance School performance (x1) Total sample (n 628) 0.81 (0.03) 23.88 1.00 0.90 (0.04) 22.5 1.00 0.71 (0.03) 20.86 1.00 0.42 (0.03) 14.39 1.00 K-8 (n 154) 0.87 (0.07) 13.26 1.00 0.90 (0.04) 22.5 1.00 0.75 (0.06) 12.39 1.00 0.31 (0.08) 4.09 1.00 MS (n 474) 0.78 (0.04) 19.53 1.00 0.90 (0.04) 22.51 1.00 0.71 (0.04) 17.01 1.00 0.46 (0.04) 12.97 1.00

Table AI. Factor loading matrix (variables reecting endogenous latent variables)

IJEM 26,2

Observed variable Organizational structure Degree of bureaucracy (x5) Organizational values Time for collaboration (x11) Encourage innovation (x12) Participate in decisions (x13)

Total sample (n 628) 1.00 1.00 0.63 (0.02) 30.88 0.74 (0.02) 37.52

K-8 (n 154) 1.00 1.00 0.62 (0.04) 15.78 0.71 (0.04) 17.78

M.S. (n 474) 1.00 1.00 0.62 (0.02) 27.72 0.75 (0.02) 30.29

222
Table AII. Factor loadings (variables reecting exogenous latent variables)

Appendix 2
Latent variable Organizational climate Organizational structure Teacher attitudes Managerial processes Organizational values Observed variable Social relationships (x2) Communication open (x3) Collegiality of teachers (x4) Bureaucracy (x5) Teacher perception of students (x6) Teacher perception of parents (x7) Resources available (x8) Responsiveness by principal (x9) Leadership by principal (x10) Time for collaboration (x11) Innovation encouraged (x12) Participate in decision making (x13) Question on survey 40, 41, 42, 43, 44, 45 (alpha 0.75) 8, 9, 10 (alpha 0.77) 20, 21, 22 a, 23 a (alpha 0.79) 2, 3, 4, 5 (alpha 0.70), 28 a 35, 36 (alpha 0.63), 47 b 31 a, 38 a, 39 a 6, 7 (alpha 0.70), 46 b 11, 12, 13 (alpha 0.86), 14 a 24, 25, 26, 27, 29, 30 (alpha 0.79) 32, 33, 34 (alpha 0.68) 18, 19 (alpha 0.61), 37 a 15, 16, 17 (alpha 0.60)

Table AIII. Latent and observed variables included in the survey

Notes: aThese numbers are added to the survey. bThese numbers are newly constructed questions to ensure each observed variable is composed of three or more items. Dependent variable School Performance measured by: 2006-2007 CST sixth grade mean scale scores: ELA, Math, History, Science. 2008-2009 CST eighth grade mean scale scores: ELA, Algebra 1, History, Science

About the authors Martin Omar Gomez is an Assistant Principal with the San Bernardino School District. George A. Marcoulides is a Professor of Research Methods and Statistics at the University of California, Riverside. George A. Marcoulides is the corresponding author and can be contacted at: George.Marcoulides@ucr.edu Ronald H. Heck is a Professor of Educational Administration at the University of Hawaii, Manoa.

To purchase reprints of this article please e-mail: reprints@emeraldinsight.com Or visit our web site for further details: www.emeraldinsight.com/reprints