Educational Evaluation and Policy Analysis Spring 1994, Vol. 16, No. 1, pp. 21-39 Curriculum Control and Teachers' Perceptions of Autonomy and Satisfaction Douglas A. Archbald University of Delaware Andrew C. Porter University of Wisconsin-Madison High school mathematics and social studies teachers were studied to determine the influence of curriculum control policies on their sense of autonomy and job satisfaction. Control policies were found to have their largest effects on content decisions in mathematics. Nevertheless, teachers in all conditions studied reported relatively high degrees of personal control over both content and pedagogy. Further, there was little evidence that teachers felt less efficacious or satisfied about their jobs because of curriculum policy constraints. T h e state standard setting reforms of the 1980s represented an unprecedented assertion of state control over school and classroom curriculum decision making (Boyd, 1987; Pipho, 1991). The goal was to raise standards of content and performance. The means were high-stakes testing, increased graduation requirements, prescriptive curriculum policy, textbook control, and strengthened accountability and accreditation programs (Cuban, 1987; Goertz, 1988; Kirst, 1987; Reilly & Gersh, 1988). These reforms have sparked a fierce debate over autonomy and control in public education. Critics argue local control—an important tradition and functional management principle—is threatened (Brooks, 1991). Teacher empowerment proponents contend central regulation undermines the professional autonomy of teachers (Frymier, 1987; Rosenholtz, 1987) and damages morale (Boyer, 1988). Empowerment proponents also worry about negative effects of central curriculum control on pedagogical effectiveness: By prescribing curriculum and instruments of assessment, such reforms . . . separate the craft of teaching from teaching style and remove teachers' discretion from their judgments about students and what they need to know. In this de-skilled model of teaching, one teacher lamented, the teacher becomes little more than an assembly-line worker, performing mechanical tasks. (McNeil, 1988, p. 335) According to the critics of the 1980s reforms, a curricular "zone of discretion" at the school and classroom level is shrinking (Schwille et al., 1986), and this is a problem because local actors need discretion to make effective curriculum decisions and feel professionally efficacious. An assumption is that, in the absence of central curriculum control policies, local actors would make different and better content and pedagogical decisions leading to improved student achievement. The "shrinking autonomy" view and its assumptions about the reach of state policy contrast with a body of research on organizational structure and the implementation of change in education (Baldridge & Deal, 1975; Fullan, 1982; Lortie, 1975; Weick, 1976). This research depicts schools as resistant to change, especially the sort of top-down 21 Archbald and Porter change characteristic of 1980s reforms. Schools have been characterized as "loosely-coupled" "organized anarchies," with classrooms relatively impervious to external control. It is exactly this resistance to change and loose-coupling that "policy centralizers" view as a problem. Reforms using new curriculum control policies are predicated on the assumption that there has been too much discretion at the local and classroom level. The 1980s crisis-in-education reports attributed declining performance to unclear goals, lax standards, and insufficient accountability in the schools (Kirst, 1987). Both critics and supporters of central curriculum control policies assume these policies have the clout and the reach to affect core processes of content selection and pedagogy in classrooms. This study investigates two propositions related to this assumption: (a) State and district curriculum control policies reduce teachers' feelings of professional autonomy and local curriculum discretion, and (b) teachers' perceptions of diminished control over curriculum decisions resulting from control policies adversely affect their selfefficacy and job satisfaction. These propositions are examined using data from high school mathematics and social studies teachers under differing conditions of centralized control over curriculum. Conceptual Framework What is "centralized curriculum control" and what are its effects on teachers? To answer these questions, it is necessary to begin with a model of centralized curriculum control. The model used by most systems with central curriculum control policies (including the districts studied here) is based primarily on textbook adoption policies, curriculum guidelines, and testing. These curriculum control policies are intended to guide teachers in their decision making about course content and hold teachers and schools accountable for prescribed content and achievement standards. At the district level, curriculum control policies are viewed as contributing to a more coherent and efficient curriculum program. 22 The Curriculum Control Policy Model Curriculum Guides The main function of guides is to state learning goals and topics for a course. They can do this to varying degrees of specificity (Archbald, in press; Curry & Temple, 1992). Some guides state only general goals and topics. Toward the prescriptive end of the continuum are guides that contain hierarchies of goals and objectives, describe sequences of units composing a course, and state or imply a pacing schedule. Units can be described in detail, with concepts and recommended learning strategies. Textbook Adoption Textbook adoption controls course content by restricting the range of textbooks that can be used for a course. Some policies limit the approved textbooks for a course to a small number (two or three) from which the course instructor makes an individual selection; others prescribe a particular book for each course. One purpose of textbook adoption policies is to reduce the potential variability in content across different sections of a course (both within and between schools). Assuming teachers using the same book use it similarly— curriculum guides are intended to facilitate this—central adoption policies increase the likelihood that students in the same course get the same content. Adoption policies also have a quality control purpose. It is assumed a committee of selected teachers informed of district curriculum goals and representing teachers' preferences will choose better textbooks than teachers making choices at an individual or school level. Testing Policies The curriculum control policy model views tests as both prescribing content and improving performance. First, test questions, like curriculum guides, imply content goals by adding authority to selected goals and topics. Like guides, they identify certain topics and skills as essential. Second, tests are part of an inducement system encouraging teachers to teach and students to learn tested content. Most students and teachers want to perform well, or at least avoid poor performance, although this desire varies depending on how Teacher Control and Teacher Autonomy results are used (Airasian, 1988; Madaus, 1991). To the extent that curriculum control policies operate as intended, centralized districts can be expected to have greater uniformity in course content and more consistent achievement standards across schools. We should emphasize that the curriculum control policies in the districts we studied were not aimed at restructuring curriculum and instructional practice. The alignment of those policies around textbooks and standard scope and sequence guidelines and the use of relatively conventional testing formats were consistent with traditional conceptions of curriculum. In contrast, recent curriculum reports and constructivist theories of curriculum and learning are pressing for substantial changes in these traditional conceptions and practices (Farnham-Diggory, 1990; National Council of Teachers of Mathematics, 1989; Sizer, 1992). The extent to which the propositions and assumptions of the curriculum control policy model described above apply in practice remains unclear. The model has not been systematically examined. To examine these assumptions and shed light on debates over autonomy and control, this study compared teachers' beliefs and attitudes about professional autonomy and self-efficacy in districts differing in curriculum policy centralization. Study Design We compared high school social studies and mathematics teachers' ratings of control and autonomy under conditions of high, medium, and low curriculum control. The subject matter comparison is illuminating because mathematics and social studies are differentially regulated (Stodolsky, 1988). Mathematics content and achievement standards are subject to greater controls because mathematics is considered a basic skill while social studies is not, because mathematics objectives are easier than social studies objectives to specify in guides, and because students are tested more in mathematics than in social studies. Teachers' responses on questionnaire items were used to assess claims supporting and critical of the curriculum control model. If teacher responses indicate curriculum control policies influence classroom content and do not show detrimental effects on job-related attitudes, then perhaps some of the virtues of top-down curriculum control assumed by "policy centralizers" may in fact be real. If, on the other hand, teachers react negatively to centralized curriculum control, then this approach to reform—or at least the elements teachers find objectionable—might be redesigned to be more compatible with teachers' interests and professional values. Sample Design California, Florida, and New York were selected as the study states on the basis of their policy characteristics. In different ways, each of these states in the mid-1980s implemented curriculum reforms through frameworks, tests, and textbook policies. Thus, each state presented a policy context that was useful for the purposes of this study (Fuhrman, 1988; Marsh & Rowan, 1988; Shujaa & Richards, 1989; Timar & Kirp, 1988). In each state, two urban districts were studied. The main comparison investigated was degree of curriculum centralization. The study focused on urban districts because urban districts have been a priority in education reform and because urban districts have often turned to centralized curriculum policies to try to implement common content and boost achievement. Also, focusing on urban districts and excluding suburban or rural districts were useful in reducing the number of potentially confounding variables. The districts were selected based on data from telephone interviews with state and district curriculum and testing directors, from conversations with researchers and policy officials knowledgeable about policy characteristics of particular districts, and from policy studies by the Council of Great City Schools, Policy Analysis for California Education, the Center for Policy Research in Education, and the departments of education in Florida, California, and New York. Through this process, the sample of districts was progressively narrowed from the initial "universe," which consisted of all of the urban districts in California, Florida, and New York, to the sample of six, which were determined to have the ap23 Archbald and Porter propriate testing, textbook, and curriculum policy characteristics. (These are described in more detail below.) Two high schools were selected in each district for a total of 12 high schools in the sample. High schools were selected to be comparable across districts on percentage of students on free lunch and ethnic composition. In each district, one high school was selected at the 65th to 75th percentile of district high schools on the free/reduced lunch statistic and one at the 25th to 35th percentile to ensure a range on type of student background served. Based on interview information from district personnel, the high schools selected were typical of the broad middle range of urban comprehensive high schools in their curriculum programs and student composition. Selective or otherwise specialized high schools were not chosen, and none of the high schools were undergoing significant changes (e.g., restructuring or desegregation) at the time of the study. The high schools ranged in size from about 1,000 to 2,000 students. Each of the high schools had a standard departmental structure, with a department chair and a core of full-time teachers. The courses taught by these teachers spanned the full range of credit courses typical of comprehensive high schools. In mathematics, these courses included algebra 1 and 2, geometry, trigonometry, calculus, business mathematics, general mathematics, and computer mathematics. In social studies, courses included U.S. history, world history (global studies in New York), geography, civics, and economics in addition to occasional courses of a more specialized nature. (This information was derived from master schedules, which were requested of each high school.) All of the full-time teachers in the mathematics and social studies departments in the 12 high schools were surveyed in 1989-1990. Out of 221 teachers, 195 returned surveys for a response rate of 88%. Varying Degrees of Curriculum Control Among the Districts For analytical purposes we placed the six districts in three categories of control: high, 24 medium, and low, with two districts in each category. Following is a description of the policy characteristics of the districts in each category. Since the two medium control districts have some of the policy characteristics of both the high and low control districts, the medium districts are described last. Policy Characteristics of the High Control Districts Both high control districts have districtwide policies. Each district requires all high schools to offer the same set of district-prescribed courses, with detailed district guidelines on particular course topics and sequences, and with a single textbook adopted per course. Each district also uses coursebased testing on a districtwide level to monitor performance and specify districtwide standards. Except where noted, the policies are the same for mathematics and social studies. Detailed Curriculum Guides Each of the high control districts has detailed course content guides. While the districts' course content guides differ in how content is organized and the level of detail at which it is prescribed, each guide in the centralized districts prescribes sequences of units, topics, and lesson ideas. The course guides in the two districts range from 15 pages to about 100 pages of material per course. Figures 1 and 2 show excerpts of topics from curriculum guides in one of the high control districts. The mathematics excerpt shows the objectives and sub-objectives for "Operations on Rational Numbers," 1 out of 13 topics prescribed for the course. The guide indicates whether particular objectives are specified in the state's curriculum frameworks, tested in the state student assessment program, tested by the nationally norm-referenced standardized achievement tests used by the district, and tested by the district's own course-based tests. The social studies excerpt is from course content guidelines for a history course. The curriculum guides used in the other high control district are also detailed in their organization and specification of content. The district Teacher Control and Teacher Autonomy 2.0 E@#* E@#* 2.2 2.3 E@#* 2.4 * o # E @ Operations on Rational Numbers 2.1 The Number Line 2.1.1 State coordinates 2.1.2 Graph numbers 2.1.3 Compare numbers # 2.1.4 Add integers 2.1.5 Additive inverses 2.1.6 Absolute value Add Signed Numbers Subtract Signed Numbers 2.3.1 Express differences as sums 2.3.2 Compute Multiply Signed Numbers District Minimum Level Skills Test State Student Assessment Program Stanford Achievement Test State Standards of Excellence State Performance Standards FIGURE 1. Excerpt showing selected mathematics topic from a high control district guide. uses the New York state Regents syllabi for Regents courses and uses its own district-developed course content guides for districtwide non-Regents courses. The Regents course syllabi are detailed, approximately 60 to 80 pages in length. They are divided into units with goals, objectives, subunits, topics, concepts, and lesson suggestions. Single Textbook Adoption Each high control district has a single textbook adoption per course. Textbook adoption decisions are made by district textbook adoption committees composed mainly of teachers; district curriculum specialists are also on the committees. Course-Based Testing Both of the high control districts use districtwide course-based tests. These tests are developed by teachers and district specialists and reflect prescribed course content. Thus, each course (e.g., algebra 1, American history, etc.) has an end-of-course test required for all students. These tests enable teachers and administrators to assess student performance in each course on a uniform standard. In both districts, the tests have high stakes for students—they can prevent students from getting credit for courses they do not pass. The two high control districts do not have exactly the same type of district testing program. The high control district from New York uses state-developed New York Regents exams for its Regents courses and uses its own tests, patterned after the Regents exams, for courses that are not Regents courses. Both tests are graded by committees of teachers. For the Regents exams, students must pass a preset score to get Regents credit for the course (irrespective of their grades in the course). Students in non-Regents courses must pass the district test for those courses in order to receive course credit. While these passing requirements are set in policy, high school departments have discretion concerning how much weight they want to place on these tests for their own student grading policies. Results on the tests are sent to the district office for research, evaluation, and reporting purposes. Regents exam results are sent to the state, which uses the results to award (or not award) Regents diplomas and for school and district accountability purposes. In the other district, all the tests are developed and graded at the district level. Teachers can use the test scores for grading purposes at their (or their department's) discretion, but students must achieve a preset score to pass the course and receive course credit. Reports of individual student scores are sent back to teachers and principals and used by the central office for research, evaluation, and reporting purposes. 25 Archbald and Porter @*12.4 @12.5 @12.6 @12.7 13.0 @13.1 By distinguishing between underlying and immediate causes, students will demonstrate knowledge of the reasons for United States' entrance into World War II. Students will demonstrate knowledge of ways the government promoted support for the war effort by selecting a slogan or poster used during World War II. Students will investigate the problems posed in conducting a two-theater war by the following: (1) identifying the major Axis and Allied powers and their leaders; (2) contrasting the strategies and obstacles in the Pacific war with those in Europe; (3) identifying elements of Allied strategy in either the European or Pacific theaters using the legend on a map of that theater; and (4) examining the American decision to use the atomic bomb. Students will compare the "Frontier" in 1945 to that of 1919 politically, diplomatically, economically, and socially. World Leadership/Prosperity, 1946-1960 Student will study the development of American society in the postwar age by describing the following: (1) the effects of demobilization; (2) the bi-polarization of the world politically and the onset of the Cold War; (3) the commitment to world leadership and the policy of containment; (4) the effects of increasing prosperity; (5) the growth of the civil rights movement; (6) the explosion of mass culture; (7) the military and political dominance of America; and (8) the domestic and international challenges to American society. FIGURE 2. Excerpt showing selected social studies topic from a high control district guide. In addition to course-based testing, each district also administers norm-referenced standardized tests at the high school level. These tests cover mathematics, but not social studies. Policy Characteristics of the Low Control Districts The low control districts are distinguished from the high control districts by the general absence of districtwide course, textbook, and testing policies, and have much more curriculum autonomy. Note that the term "low control" is preferable to "decentralized" because the latter may suggest site-based management reforms such as have occurred in Chicago, Rochester, and other districts, but not in the low control districts in this sample. Recently, however, the two low control districts either eliminated or considered and rejected districtwide testing and more restrictive textbook policies. Nonprescriptive Curriculum Guides The two low control districts have course curriculum guides, but in sharp contrast to 26 the high control districts, they lack detail, are voluntary, and are designed independently of textbook adoption decisions and without consideration of district tests (which do not exist). One of the districts uses the state's curriculum frameworks for content guidelines for its high school courses. These state frameworks are not specific—about 10 onesentence "Intended Outcomes" per course. The state frameworks do not specify sequences of topics and leave considerable room for interpretation. The other district has no district-level curriculum guides. Use of the state's frameworks is discretionary. Course content guidelines in this district are primarily under the jurisdiction of high school departments, are generally a page or two in length, and are developed by the teachers teaching the courses. Multiple Textbook Adoptions Neither of the low control districts has a policy requiring a single textbook adoption per course. Textbook decisions are made by individual teachers or departmental committees. Teacher Control and Teacher Autonomy No Course-Based Testing Both of the districts conduct standardized testing in reading and mathematics. This testing is used for conventional student diagnostic and placement purposes, not for policy purposes. The tests are not linked to courses or aligned with curriculum guidelines. Policy Characteristics of the Medium Control Districts The two medium control districts have policies that have some of the characteristics of the two other categories. They lack the extent of course-based testing in the high control districts, but have significant, centrally prescribed course content guides. One district is in New York. Like all districts in New York, it offers Regents courses, which means students must pass Regents exams if they seek credits toward a Regents diploma. However, this district does not have districtwide course-based tests in non-Regents courses. Also, textbook adoptions and decisions about the use of curriculum guides are made at the department level. The other district has highly detailed curriculum guides—they average about 320 pages per course—and textbook adoption decisions are made centrally, but teachers can select from several approved books. The district's course guides are aligned with the selected textbooks and contain unit tests and final exams. Administration, scoring, and results of these tests are entirely under the control of teachers in the schools. Also, teachers may substitute their own tests with administrative approval at the school level. Thus, unlike the case with the two high control districts, administrators do not have direct access to students' scores. The district conducts standardized testing at every grade level. Test results are reported annually for each school and are used for identification of students for special programs and for research and program evaluation purposes. Testing covers mathematics (but not specific courses), reading, and language. Teachers' Perceptions of Policy Influences and Curriculum Control We examine teachers' perceptions of curriculum control policies in this section by comparing mean ratings from teachers in the different conditions of centralization. Also, we discuss mean ratings across the entire sample on scales that reveal differential influences of policies and individual discretion. Policy and Teacher-Discretion Influences Table 1 shows mean scores for each of the three categories of curriculum control as well as the total sample mean on four different scales.1 The external control scales consist of teachers' ratings of the influence of a number of factors "in determining the content (information, concepts, skills) of your [mathematics or social studies] course." To give survey items more specific referents, respondents TABLE 1 Means on Scales by Categories of Centralized Control Over Curriculum Centralization category Section and scale High External control S&D Guides S&D Tests* Textbook Department Self* Teacher control Content Pedagogy* Involvement Involvement* Empowerment Efficacy Satisfaction* Standards* Medium Low Total 3.15 2.78a 3.04 1.78a 3.13 2.92 2.41 a 3.11 2.47b 3.33 3.00 1.93b 3.37 2.62 b 3.38 3.02 2.34 3.18 2.35 3.29 2.92a 5.12 2.94a 4.86a 3.69b 5.36b 3.20 5.11 2.75a 3.35a 4.21 b 3.48 3.97 7.13 3.91 4.15 6.86a 3.58a 4.37 7.54b 4.12b 4.18 7.18 3.86 Note: * indicates F ratios exceeding the p = .05 level of statistical significance for differences among category means on the scale. For the ANOVAs conducted for Table 1, a Scheffe' test (a conservative multiple comparisons test) was used to determine which category means differed from each other by statistically significant margins. Thea andb indicate means differing by statistically significant margins from each other (e.g., for the S&D Tests scale, High is significantly different from Low control). Tests of significance are used as a crude sort for interpreting differences; thus no attempt has been made to control the overall error rate of the multiple F tests. External control: 1 = No Influence; 2 = Minor; 3 = Moderate; 4 = Major Influence. Teacher control: l = No Control to 6 = Complete Control. Involvement: 1 = Strongly Disagree to 6 = Strongly Agree. Empowerment: l = Low to 6 = High for Efficacy and Standards; 1 = Low to 10 = High for Satisfaction. 27 Archbald and Porter were instructed to answer questions about content influences with reference to a particular section of a particular course. Master schedules from each of the 12 schools in the sample were used to ensure that a balanced cross sample of courses were targeted by the surveys. Only standard core courses were sampled. The items have been combined into the following scales (Figure 3 presents items used to form the scales): Guides: influence of state and district curriculum guides on your course content. S&D Tests: influence of state and district tests on your course content. Textbook: influence of the main course textbook on your course content. Self: influence of your own individual decisions on your course content. Department: influence of departmental decisions and guidelines on your course content. Expectations Theoretically, ratings of policy influences (S&D Guides, S&D Tests) in the more centralized conditions should be higher and ratings of the influence of individual teacher discretion (Self) should be lower. The influence of the textbook scale on content might be expected to be higher with increasing centralization on the assumption that the textbook has been selected by the district because the book is aligned with districtrequired tests and curriculum objectives. On the other hand, whether or not there are externally imposed tests and guides, all teachers may rely equally on textbooks because textbooks make teaching easier. The model of centralization described earlier does not specify a clear role for departments, although it does not preclude departmental decisions and guidelines. Different models are conceivable: 1. Higher departmental influence occurs under centralized curriculum control. State and district curriculum control policies create needs for heightened intradepartmental coordination among teachers. This coordination serves purposes of sharing lessons and techniques, discussing common problems, and checking to see that prescribed material is being covered appropriately. In noncen28 tralized districts, there is more of a laissez faire attitude, with little need for teachers to coordinate, and so departmental influences are lower. 2. Lower department influence under centralized curriculum control because state/district curriculum control policies replace some of the roles departmental decisions and guidelines otherwise would serve. In noncentralized conditions, the department is a more dominant agent of curriculum control because there is no other organizational source of guidance. Results Concerning total sample means for the "influences on control scale," ratings on the Self scale were highest, at 3.29 on the 4.0 point scale; Textbook is rated 3.18; S&D Guides, 3.02; Department, 2.35; and S&D Tests, 2.34. That Textbook is rated nearly as high as Self attests to the heavy reliance on textbooks in high school mathematics and social studies classes. The Guides scale has a surprisingly high influence rating and the Tests scale a surprisingly low one if viewed from the perspective that is common that guides are often ignored while external tests have a powerful influence on curriculum and instruction (e.g., Haney, 1991; Neill & Medina, 1989). Comparisons of means among the categories of centralization are consistent with the expectations: The S&D Tests scale and the S&D Guides scale show higher influence with increasing centralization; and the Self scale shows lower ratings with increasing centralization. However, the pairwise differences on S&D Guides and Self fail to achieve statistical significance (p < .05). On the S&D Tests scales, statistically significant differences separate the low from both the medium and the high control districts, with a .85 point difference separating the low from the high control difference on the 4-point scale. This difference is equivalent to a standard deviation on the S&D Tests variable. There are no significant differences among the categories of control on the Textbook scale, although ratings of textbook influence increase slightly with decreasing centraliza- Teacher Control and Teacher Autonomy External Control Scales: S&D Guides (items a, b); S&D Tests (items e, f); Textbook (item d); Department (item c, g); Self (items h-k). "Rate how big an influence each factor below has in determining the content (information, concepts, skills) of your [subject] course." (Note: Respondents rated each influence (a-k) on a 1 to 4 [No Influence to Major Influence] scale next to the item; scale not shown.) a) State curriculum guides b) District curriculum guides c) Departmental decisions and guidelines d) The main course textbook e) District tests f) State tests g) Department-wide tests h) My own beliefs about what topics are important i) My own knowledge of particular topics j) What my students are capable of understanding k) What my students need for future study and work Teacher Control Over Classroom Content/Pedagogy Scales: Content (items a, b); Pedagogy (items c, d, e). "How much control do you feel you have in your classroom over each of the following areas in your planning and teaching?" (Note: Respondents rated their control over each area (a-e) on a I to 6 scale ["None" to "Complete Control"] next to the item; scale not shown.) a) b) c) d) e) Selecting textbooks/instructional materials. Selecting content, topics, and skills. Selecting teaching techniques. Determining amount of homework to be assigned. Setting standards for achievement in my classes. Staff involvement in course content decisions: "Staff are involved in making decisions about what will be taught in their courses." (Note: Respondents rated their agreement with the statement on a 1 to 6 scale ["Strongly Disagree" to "Strongly Agree"] next to the item; scale not shown.) Teacher Empowerment Scales: Self-Efficacy (a, b, c); Job Satisfaction (d, e); Standards (f, g). "Please use the scale provided to rate the extent to which you agree or disagree with the statements below." (Note: Except for item (e), respondents rated their agreement with the statements on a I to 6 scale ["Strongly Disagree" to "Strongly Agree"] next to each item; scale not shown. For items (a-c), the scale was reversed so that high efficacy was associated with big numbers. For item (e), the response categories ranged from 4 "All of the time" to 1 "Almost never"; scale not shown.) (a) My success or failure in teaching students is due primarily to factors beyond my control rather than to my own effort and ability. (b) I sometimes feel it is a waste of time to try to do my best as a teacher. (c) Teachers are not a very powerful influence on student achievement when all factors are considered. (d) I usually look forward to each working day at this school. (e) How much of the time do you feel satisfied with your job in this school? (f) Staff members maintain high standards of performance for themselves. (g) The teachers in this school push the students pretty hard in their academic subjects. FIGURE 3. Scale definitions 29 Archbald and Porter tion. Textbooks appear to be a major influence under any level of curriculum control in this study. Ratings on the Department scale are consistent with the departmental influence model 2 above: Teachers in the more centralized districts report lower departmental influences on curriculum content by statistically significant margins. The Department influence for high control is significantly less than either of the other two categories of control. The differences on the Self scale are in the expected direction (from 3.13 to 3.38), but none of the pairwise differences are statistically significant. Apparently, while teachers in the high control districts perceive stronger policy influences on content, these influences do not seem to be viewed as having comparable size effects in reducing their own decisions about content. One explanation of this finding is that teachers view the external controls as authoritative and are persuaded of their appropriateness (Porter, 1989; see also Floden, Porter, Schmidt, Freeman, & Schwille, 1981, for a study of teachers' perceptions of policy influences on mathematics at the elementary level). Teacher Control Over Classroom Content and Pedagogy Table 1 shows means on two "teacher control" scales: teachers' control over content (books, content, topics, skills) in their own classroom and teachers' control over pedagogical methods (teaching techniques, homework, standards for achievement) in their own classroom. Expectations According to the centralized curriculum control model, the greater the control over curriculum, the lower should be the reported control by teachers over both content and pedagogy in the classroom. Curriculum control policies are intended primarily as influences on content (evident in explicit prescriptions concerning topics and skills to cover), but they are not silent on pedagogy, evident in recommended lesson ideas and teaching strategies that can be found in the centralized districts' guides. Further, when pushed, the distinction between content and pedagogy 30 blurs; for example, teachers may employ active learning strategies to foster conceptual understanding. Results On the content scale, results are fairly consistent with the centralized control model. Teachers' ratings of control increase as centralization decreases. Teachers in low control conditions rate their control over curriculum content in their classrooms statistically significantly higher than both other categories; teachers in high and medium control conditions are essentially equivalent in their ratings (2.92 and 2.94, respectively). On the pedagogy scale, the results are a little more mixed. Teachers in the low control condition report the highest ratings of control (5.36), statistically significantly higher than those of the teachers in the medium control districts (4.86), but not significantly higher than the teachers in the high control districts (5.12). Nor is the difference between the medium and high control conditions statistically significant. The overall sample mean is 3.20 for teacher control over content and 5.11 for teacher control over pedagogy (on the 6-point scale). This difference of almost 2 full points clearly indicates that teachers believe they have near total control over their pedagogy but generally lower and more varying control over content. Even in the high control districts, teachers' control in the pedagogical domain was above 5 on a 6-point scale. Staff Involvement The response categories for staff involvement ("in making decisions about what will be taught in their courses") are on a 6-point scale (1 = Strongly Disagree to 6 = Strongly Agree). Since this item uses the phrase "what will be taught," it is a measure of perceptions about decisions regarding content, not teaching methods (e.g, teaching strategies and achievement standards). Also, the term "staff" (in contrast with the above scales) shifts the object of the question away from the individual respondent toward teachers in the school as a whole. Expectations According to the curriculum control model, teachers in the more centralized cate- Teacher Control and Teacher Autonomy gories should express greater disagreement with this statement, implying staff in their school have lower involvement in course content decisions. Results The ratings from each of the categories conforms with the expected pattern (Table 1). In the high control category, teachers express slight disagreement with the statement about staff involvement in what will be taught, while in the low control condition teachers express slight agreement. The differences are statistically significant between medium and low control and between high and low control. Teacher Empowerment Table 1 shows the means on three empowerment scales: Self-Efficacy: teacher's sense of confidence/effectiveness with respect to success in teaching Job Satisfaction: morale/satisfaction on the job Standards: general schoolwide expectations for student/staff performance Expectations Some analysts contend that large-scale testing, prescriptive curriculum guides, and textbook adoption regulations are antithetic to teacher professionalism. While the curriculum control model suggests that organizational actors unproblematically follow hierarchical directives, empowerment models posit a need among organizational actors to control their work and predict lower outcomes on the empowerment variables in the more centralized conditions. The items in the Self-Efficacy scale measure "personal sense of efficacy," a personality construct important in psychology (Bandura, 1977; Lewin, 1938).2 Self-efficacy is defined as an individual's expectancy of achieving valued outcomes through personal effort. Education researchers have studied teachers' sense of efficacy and believe it is an important variable influencing educational outcomes (Brookover, Beady, Flood, Schweitzer, & Wisenbaker, 1979; Newmann, Rutter, & Smith, 1989; Rutter, Maughan, Mortimore, Ouston, & Smith, 1979). There is concern that prescriptive curriculum regulations prevent teachers from feeling ownership over curriculum and diminish their sense of responsibility for learning outcomes. The question here is, what is the evidence that centralized curriculum policies affect teacher efficacy? Results Table 1 shows results on two empowerment scales assessing teachers' work-related attitudes. On the Self-Efficacy scale, the differences among the means do not achieve statistical significance. On the Job Satisfaction scale, there is a statistically significant difference between the low and medium control conditions (7.54 compared with 6.86). While this is consistent with concerns about effects of central policies on job satisfaction, the high control mean of 7.13 is not significantly different from low control mean. Overall, the means on these scales were only slightly above the scale midpoint, suggesting that efficacy and job satisfaction may be somewhat of a problem in urban high schools. These results are ambiguous concerning the contention that higher levels of curriculum control are likely to come at the expense of teachers' morale and feelings of efficacy on the job. While these items clearly cannot capture the complexity of teachers' feelings about their work and about the effects of curriculum policies, significant observed differences between the curriculum control conditions would be important information in a debate surprisingly short on empirical analysis. On the Standards scale, the same pattern occurs: There is a statistically significant difference between the low and medium control means (4.12 and 3.58, respectively, on a 6-point scale), but the difference between low and high control is not significant. If this scale is a measure of general standards of performance teachers hold for themselves and their students, then, depending upon one's perspective, these results either allay concerns or raise concerns. If one views curriculum control policies as a way to raise expectations and set higher standards, then the evidence raises concerns; if one views curriculum control policies as detrimental to high expectations and standards, then the evi31 Archbald and Porter dence allays concerns (since the differences are not large). Either way, the results are neither strong nor consistent; at least for these teachers, standards and external control are not strongly linked. All means on the Standards scale are only slightly above the midpoint of the scale, indicating that standards for teachers and standards for students are perceived by teachers to be a problem. Summary Comments The ratings given by teachers to the influence of external policies (Tests and Guides) increase under conditions of increasing curriculum centralization. However, teachers in all three categories of centralization rate their own beliefs and decisions as the dominant influence on course content. The differences between the centralization categories are most pronounced when questions elicit ratings of control over curriculum characterized as content, although teachers report at least moderate levels of control over content even in the more centralized conditions. When questions turn to methods of teaching, teachers report they are in control and differences between the centralization categories cannot be interpreted as either supporting or being inconsistent with the central control model. If there is bias in our findings related to differences among the districts in student background variables, it is in a direction that would likely lower values on all three empowerment scales for the high control condition. All of the districts in the sample are urban, and all of the high schools were selected to match on subsidized lunch and ethnic composition figures. However, based on the demographics of the cities in which the districts are located and on site visits to the schools, the low control districts seem overall not to serve as low a socioeconomic population as the schools in the other districts. If teachers in the medium and high control conditions experience a higher degree of frustration from working with larger numbers of more difficult students, this frustration could express itself in lower ratings of efficacy and job satisfaction (Heyns, 1988). If so, our data may be even less supportive of claims of the negative effects of curriculum control on teacher empowerment than they appear. 32 Mathematics Versus Social Studies Teachers Policy debates over potential consequences of increasing state and district curriculum control rarely distinguish among teachers of different school subjects. This section compares ratings between the mathematics (n = 85) and the social studies (n = 103) teachers in the six-district sample. This comparison is another way to investigate outcomes of curriculum control policies. Mathematics is subject to more curriculum policy controls: It is tested more frequently by externally mandated tests, and mathematics books and guides prescribe topics, sequences, and objectives in greater detail. In social studies there is little external testing; the discipline is marked by less consensus; and externally prescribed content may be less authoritative. Further, the enormous influence of and agreement with the National Council of Teachers of Mathematics' Curriculum and Evaluation Standards contrasts markedly with disputes among authorities and commissions in the social studies area over curriculum emphases, sequencing of topics, and instructional methods (Gehrke, Knapp, & Sirotnik, 1992; Mehlinger, 1992; Viadero, 1990). Thus, we expect to see higher curriculum control ratings in mathematics than social studies. External Control Expectations On the assumption that mathematics is more highly regulated, mathematics teachers should report stronger policy influences of S&D Guides and S&D Tests and weaker influences of Self. Because individual interests, values, and teaching styles can be expected to have greater influence on content in social studies, we expect the role of the Department to be weaker in social studies. Teachers in general in the humanities probably view their subject matter and instruction as more of an individualistic enterprise (Gehrke et al., 1992). Results The results (Table 2) are as expected: Mathematics teachers report statistically significantly higher influence ratings for the pol- Teacher Control and Teacher Autonomy TABLE 2 Means on Scales by Subject Matter Subject matter Section and scale External control S&D Guides* S&D Tests* Textbook Department* Self* Teacher control Content Pedagogy Involvement Involvement Empowerment Efficacy Satisfaction Standards* Math Social studies 3.12 2.48 3.26 2.49 3.17 2.90 2.15 3.08 2.13 3.45 3.15 5.25 3.26 4.93 3.54 3.40 4.22 7.12 4.11 4.13 7.24 3.57 Note: * indicates F ratios exceeding the p = .05 level of statistical significance for differences among category means on the scale. External control: 1 = No Influence; 2 = Minor; 3 = Moderate; 4 = Major Influence. Teacher control: l = No Control to 6 = Complete Control. Involvement: 1 = Strongly Disagree to 6 = Strongly Agree. Empowerment: l = Low to 6 = High for Efficacy and Standards; l = Low to 10 = High for Satisfaction. icy scales (S&D Guides, S&D Tests) and the Department influence scale; and social studies teachers report higher ratings on the Self scale. Differences on Textbook are not statistically significant. Teacher Control Expectations As explained above, if external tests and curriculum guidelines are more influential in mathematics, it follows that teachers of mathematics should report less control over both content (selecting textbooks and other instructional materials, content, topics, skills) and pedagogy (teaching techniques, homework, and student achievement standards) than teachers of social studies. Results The results on the content scale and on the pedagogy scale do not differ significantly between the two groups (Table 2). On the pedagogy scale, the mathematics teachers report a higher degree of control by a p value close to statistical significance, p = .063. That the mathematics teachers rate policy influences at a higher level than do the social studies teachers, while reporting equivalent levels of teacher control, may be explained by the different nature of the scales. The classroom control (content and pedagogy) questions do not make reference to or ask for ratings of comparative influences of different factors; rather, more abstractly they ask the teacher to rate his or her curriculum control in the classroom on a 6-point scale from "None" to "Complete Control." It appears that when the question is asked this way, mathematics and social studies teachers think more alike, focus less on particular policy influences, and report they "feel" (as the question asks) that they have approximately equivalent levels of control. Staff Involvement Expectations The item specifies "staff involvement in curriculum" and thus could imply the entire school staff. This interpretation would suggest mathematics and social studies teachers would not differ greatly, since the referent for both groups is the same "staff." Results The means for the two groups are close to each other and close to the middle of the scale. Whatever differences there are between the two subjects in determinants of curriculum control, mathematics and social studies teachers at these schools seem to have similar ratings of overall staff involvement in course content decisions. Teacher Empowerment Expectations for Efficacy, Satisfaction, and Standards If feelings of control over content and pedagogy in the classroom are related to the affective dimensions of empowerment measured here, we should not expect to find differences between the mathematics and the social studies teachers on the empowerment scales. Results Table 2 shows no differences between the two groups on the two scales, Efficacy and Satisfaction. Mathematics teachers com33 Archbald and Porter pared with social studies teachers in the sample appear to feel no more or less confident about or responsible for producing student learning and no more or less satisfied with their job. However, on the Standards scale, mathematics teachers report higher standards. It is difficult to assess what this means, but mathematics has a reputation of demanding standards for students and teachers. People tend not to think of social studies as either inherently difficult like mathematics or a subject that some people "are just not good at." These beliefs may produce a perception that standards in mathematics are higher. Are There Interactions Between the Centralization and Subject Variables? The possibility that differences between the subjects on the external control scales are not the same at different levels of centralization bears investigation. Hence, we assessed the significance of interactions between the categories of centralization and the subjects. No statistically significant interactions were found, so findings for control and subject main effects need not be qualified by interaction effects. It is noteworthy that the pattern of results from the analysis of interactions lends further verification to the discriminatory power of the external control scales. To illustrate, Table 3 shows the means for the categories of centralization broken down by subject on the S&D Tests scale. Although the pattern of higher influence ratings of tests in mathematics persists across levels of centralization, the gap between mathematics and social studies is substantially larger in the low control condition compared with the medium and high control conditions (.60 vs. .20, .21, respectively), and the influence rating for the lowcontrol/social studies cell is the smallest of all TABLE 3 Means on S&D Tests Scale by Categories of Centralization and Subject Subject District category Math Social studies High Medium Low 2.87 2.51 2.17 2.67 2.30 1.57 34 six cells, by a large margin. This is consistent with the complete absence of state or district testing in social studies in the low control districts, but the with presence of some testing related to social studies in the other districts, with the high control districts doing the most. However, even the low control districts do some "low-stakes" standardized testing in mathematics. It is also noteworthy that the results on the teacher empowerment scales and the classroom curriculum control scales are no different between mathematics and social studies in the low control districts. Although negative effects of standardized testing tend to be widely assumed, this finding suggests the assumption warrants additional scrutiny. Discussion State and district initiatives to raise standards and improve curriculum through test, textbook, and course content policies raise complex issues about education reform. These policies are intended to improve curriculum quality and standards but may have the unintended consequence of undercutting school-based curriculum control and the professional autonomy of teachers; they may have little effect on curriculum at all, positive or negative. Our research probed these issues. Shrinking Zones of Teacher Control? Critics of centralized curriculum control argue that teachers are experiencing a loss of control over curriculum. One concern is that central control policies prevent teachers from making content or instructional decisions that would be better for their students than the curriculum prescribed by policy. This concern assumes that curriculum control policies exercise a level of influence that can prevent teachers from following their own beliefs about content and instructional methods, and that their beliefs would differ from what is prescribed. This probably happens. Surely there are instances where, but for the curriculum policy, a teacher's content choices or instructional approach would be different. However, our results raise doubt as to whether this was the general pattern for policies and practices in our study. Teacher Control and Teacher Autonomy First, our results and other studies show teachers report almost complete control over pedagogy in the classroom (teaching techniques, achievement standards, and assignments) (Elam, 1989; Rowan, 1990).3 In the high control districts studied here, despite prescribed textbooks, specific course content guidelines, and required district tests for each course, teachers reported high levels of control over instruction, levels that did not differ significantly from those reported in the low curriculum control districts. With respect to content, teachers' ratings of control were highest in the districts without central curriculum control policies, although differences between medium and high control districts were not found. This result and the higher ratings of policy influences from teachers in the centralized districts indicates teachers believe policies of the types examined here can influence teachers' decisions about content. This finding, however, does not necessarily support contentions about loss of curriculum control. The main reason is that a policy's influence on a decision to teach a topic is not necessarily equivalent to not having control over content. A teacher may look to a policy for guidance on sequencing of topics, pacing of coverage, or inclusion or exclusion of topics, but from the teacher's perspective, these forms of influence are not likely to be viewed as deprivation of control. One reason is the large amount of discretion the teacher has in deciding when and how to follow policy.4 The policies we examined were not prescriptive enough to shape day-to-day content decisions and the policies rarely prescribed how to teach. A second reason is that the great majority of teachers are already predisposed to teach the topics prescribed in the policies we studied. Although curriculum organizations are calling for reform, mathematics and social studies courses had a fairly standard scope and sequence fixed by institutional arrangements (e.g., university-based mathematics disciplines) and traditional beliefs about what is appropriate content for these courses. Textbooks are designed with teachers' preferences in mind and teachers rely heavily on them. Thus, when policies prescribe textbooks and guidelines and tests that are aligned with those textbooks, the average teacher is unlikely to feel "controlled" by policies. Those claiming that policies deprive teachers of curriculum control should recognize that probably teachers rarely feel compelled to teach something against their will. However, this could change if curriculum control policies shift to more powerful strategies aimed more at changing than standardizing curriculum practice (Porter, Archbald, & Tyree, 1991). Impediments to Teacher Empowerment? Some contend that centralized curriculum control threatens to demoralize and de-professionalize teachers. Our survey results yield little evidence that teachers feel less efficacious or less satisfied in their work because of curriculum policy constraints. If it is true that curriculum control policies exert a fairly modest influence on teachers' curriculum decisions and practices, then it is not surprising that teachers working in the more centralized districts report ratings on the empowerment scales that do not differ significantly from those in the less centralized districts. This is not to say that teachers in the most centralized conditions do not have some questions about some curriculum control policies. Teachers' attitudes about district endof-course tests or single textbook adoption policies are likely to vary and some teachers may find fault with these policies. At the same time, it cannot be assumed that teachers unequivocally oppose these policies. The rationales for these policies reflect principles of equity, accountability, and program quality that many teachers find persuasive. Whatever individual variation in teachers' attitudes and beliefs about these particular policies, it appears that on balance, these policies are neither intrusive nor unpopular enough to engender adverse ratings of job satisfaction or personal efficacy. Curriculum Content Impacts? Our findings on the effects of central curriculum control are consistent with the loose coupling thesis (Weick, 1976). They suggest skepticism that the rather substantial differences in curriculum control policies distinguishing the centralized from the non35 Archbald and Porter centralized districts produce similarly substantial differences at the level of the classroom. While we lack observational data, clearly teachers did not feel that their professional discretion was sharply curtailed. A policy observer may be tempted to assume that a massive districtwide curriculum control program has a strong determining influence in shaping curriculum and instruction. But in teachers' classroom-centric perspectives and value systems, central policies tend to be remote, often not well-understood, and easy to ignore with impunity (especially if the teacher is beyond probationary status). A state or district test, an adoption cycle, a curriculum frameworks revision process— these policies pale in significance compared with the day-to-day curriculum planning, instructional activities, and social demands making up teachers' working lives. Does this suggest these curriculum control policies made no difference in curriculum at the classroom level? The centralized districts' policies of alignment of guides, textbooks, and tests and their uses of tests for monitoring and accountability probably produce a measurable, if not substantial, influence on curriculum. (Our data indicate this alignment is more true in mathematics than social studies.) The use by teachers of the same textbook for the same course increases the probability of more similar content coverage, especially if a guide prescribes coverage of particular topics and a test evaluates students on particular topics. That test score gains have been documented in districts with curriculum control policies also suggests influences of these policies on content coverage (Archbald & Porter, 1991).5 However, that teachers use textbooks with much discretion to pick and choose their coverage (Freeman & Porter, 1989) and that teachers exercise considerable discretion in their classrooms tend to temper the size of the effect of control policies on standardizing practice. A Caveat About Curriculum Control and Education Reform Interpretations of policy influence have a Janus-faced quality. Textbooks or curriculum guidelines adopted at the district level preclude alternative decisions with respect to 36 these matters at the school level. And while teachers have a great deal of autonomy in their classroom and much freedom to interpret guides to suit their individual interests and talents, duty requires some adherence to policy. On these grounds, one could argue a zone of discretion is shrinking. However, in relation to the overall scope of discretion teachers have, and to their ultimate "veto power" in the classroom concerning what gets taught, the policies studied here are relatively weak instruments of curriculum control. In interpreting statistical results, one is cautioned from extrapolating beyond the range of one's data. This study examined curriculum control policies within a range of centralization that well represents existing conditions. However, one should refrain from concluding that policies aimed at more fundamental changes in curriculum—whether through highly decentralized or centralized action—would not have greater impacts on the kinds of variables measured in this study. School-based management has been initiated in some places, but curriculum content impacts have not been systematically measured. Reports of impacts on staff morale and job satisfaction have been mixed (Chira, 1991; Malen, Ogawa, & Kranz, 1989; Olson, 1991). Also, while we selected the most centralized examples of districtwide curriculum control we could find, we can imagine more centralized conditions. Might more prescriptive and comprehensive policies have a large effect in driving curriculum in desired directions? Can this be accomplished while enhancing or at least not compromising teachers' professional role? These questions lie before us. Notes The research reported in this manuscript was supported by the Center for Policy Research in Education through a grant from the Office of Educational Research and Improvement (Grant No. OERI-R117G10007) and by the Wisconsin Center for Education Research, School of Education, University of Wisconsin-Madison. The opinions expressed in this article are those of the authors and do not necessarily reflect the views of the U. S. Department of Education, Office of Educational Research and Improvement, the institutional Teacher Control and Teacher Autonomy Ashton, P. T., & Webb, R. B. (1986). Making a difference: Teachers' sense of efficacy and student achievement. New York: Longman. Baldridge, V., & Deal, T. (1975). Managing chance in educational organizations. Berkeley, CA: McCutchan. Bandura, A. (1977). Social learning theory. Englewood Cliffs, NJ: Prentice-Hall. Boyd, W. (1987). Public education's last hurrah: Schizophrenia, amnesia, and ignorance of school politics. Educational Evaluation and Policy Analysis, 9, 85-100. Boyer, E. (1988). Report card on school reform: The teachers speak. Princeton, NJ: Carnegie Foundation for the Advancement of Teaching. Brooks, M. (1991). Centralized curriculum: Effects on the local level. In M. F. Klein (Ed.), Issues in centralizing the curriculum. Albany: SUNY Press. Brookover, W., Beady, C , Flood, P., Schweitzer, J., & Wisenbaker, J. (1979). School social systems and student achievement—schools can make a difference. Brooklyn, NY: Praeger. Chira, S. (1991, April 10). Rochester: An uneasy symbol of school reform. New York Times, All. Cuban, L. (1987). State-powered curricular reform, measurement driven instruction. The National Forum, 67(3), 22-25. Curry, B., & Temple, T. (1992). Using curriculum frameworks for systemic reform. Alexandria, VA: Association for Supervision and Curriculum Development. Elam, S. (1989). Second annual GallupAP/w Delta Kappan poll of teachers' attitudes toward the public schools. Phi Delta Kappan, 70, 785-795. Farnham-Diggory, S. (1990). Schooling. Cambridge, MA: Harvard University Press. Floden, R., Porter, A., Schmidt, W., Freeman, D., & Schwille, J. (1981). Responses to curriculum pressures: A policy capturing study of teachers' decisions about content. Journal of Educational Psychology, 75(12), 127-141. Freeman, D., & Porter, A. (1989). Do textbooks References dictate the content of mathematics instruction Airasian, P. W. (1988). Symbolic validation: The in elementary schools? American Educational case of state-mandated, high-stakes testing. EdResearch Journal, 26(3), 403-421. ucational Evaluation and Policy Analysis, 10, Frymier, J. (1987). Bureaucracy and the neutering 301-313. of teachers. Phi Delta Kappan, 69, 9-14. Archbald, D. (in press). Reflections on the design Fuhrman, S. (1988). State politics and education and purposes of state curriculum guides: A comreform. In J. Hannaway & R. Crowson (Eds.), parison of mathematics and social studies guides Politics of education yearbook (Special issue of from four states. Brunswick, NJ: Consortium Journal of Educational Policy, 3[5], 61-75). for Policy Research in Education. New York: Taylor & Francis. Archbald, D., & Porter, A. (1991). A retrospecFullan, M. (1982). The meaning of educational tive and an analysis of the roles of mandated change. New York: Teachers College Press. testing in education reform. Washington, DC: Fuller, B., Wood, K., Rapoport, T., & DornCongressional Office of Technology Assessment. busch, S. M. (1982). The organizational context partners of CPRE, or the Wisconsin Center for Education Research. lr The sample sizes for the three categories of control are as follows: high control, 69; medium control, 71; low control, 55. We examined differences among the categories on two variables that could potentially exert confounding influences— the teachers' years of experience teaching and their total load of students. There was not much variation among the districts on these two variables. The district means on the years of experience variable ranged from 16 years to 21 years, and the total student load ranged from a mean of 111 to a mean of 134. Neither of these variables showed any significant correlations with any of the survey variables analyzed for this study. 2 For theoretical treatments of principles of efficacy and their relationship to teachers and pedagogy, see Ashton and Webb (1986), Gibson and Dembo (1986), and Fuller, Wood, Rapoport, and Dornbusch (1982). 3 High School and Beyond survey results show 92% of a nationally representative sample of high school teachers reporting 5s and 6s on a 6-point scale where 6 denotes "total control" (over teaching techniques); 72% report 5s and 6s for control over "content and skills taught in class" (Rowan, 1990). 4 See Educational Evaluation and Policy Analysis, 72(3), 1990, for analyses and studies of teachers' interpretations and implementation of a state mathematics framework. The teachers' studies had and used substantial discretion in interpreting this curriculum policy; although the policy was not without influence, the teachers shaped it to fit their existing notions of mathematic content and pedagogy. 5 At the time of our study, one of the high control districts in the sample had posted several consecutive years of test score gains since implementing its districtwide curriculum program. 37 Archbald and Porter of individual efficacy. Review of Educational Research, 52, 7-30. Gehrke, N., Knapp, M., &Sirotnik, K. (1992). In search of the school curriculum. In G. Grant (Ed.), Review of Research in Education, 18 (pp. 51-110). Washington, DC: American Educational Research Association. Gibson, S., & Dembo, M. H. (1986). Teacher efficacy: A construct validation. Journal of Educational Psychology, 76, 569-582. Goertz, M. (1988). State educational standards in the 50 states: An update. Princeton, NJ: Educational Testing Service. Haney, W. (1991). We must take care: Fitting assessments to functions. In V. Perrone (Ed.), Expanding student assessment (pp. 142-163). Alexandria, VA: Association for Supervision and Curriculum Development. Heyns, B. (1988). Educational defectors: A first look at teacher attrition in the NLS-72. Educational Researcher, 27(3), 24-33. Kirst, M. (1987, April). Curricular leadership at the state level: What is the new focus? NASSP Bulletin, 71, 8-14. Lewin, K. (1938). The conceptual representation and the measurement of psychological forces. Durham, NC: Duke University Press. Lortie, D. (1975). Schoolteacher. Chicago: University of Chicago Press. Press. Madaus, G. F. (1991). The effects of important tests on students: Implications for a national examination system. Phi Delta Kappan, 73, 226-231. Malen, B., Ogawa, R., & Kranz, J., (1989). What do we know about school-based management: A case study of the literature—A call for research. Paper prepared for Conference on Choice and Control In American Education, University of Wisconsin-Madison, May 17-19. Marsh, D., & Rowan, B. (1988). State-initiated top-down versus bottom-up reform in secondary schools. Unpublished manuscript, National Center on Effective Secondary Schools, University of Wisconsin-Madison. McNeil, L. M. (1988). Contradictions of control, Part 1: Administrators and teachers. Phi Delta Kappan, 69, 333-339. Mehlinger, H. D. (1992, May). The National Commission on Social Studies in High Schools: An example of the politics of curriculum reform in the United States. Social Education, 56(3), 149-153. National Council of Teachers of Mathematics. (1989). Curriculum and evaluation standards for school mathematics. Reston, VA: Author. Neill, M., & Medina, N. (1989). Standardized testing: Harmful to educational health. Phi Delta Kappan, 70, 688-697. 38 Newmann, F. M., Rutter, R. A., & Smith, M. S. (1989). Organizational factors that affect school sense of efficacy, community, and expectations. Sociology of Education, 62, 221-238. Olson, L. (1991, September 25). Shortcomings in school-based management in Boston noted. Education Week, 77(4), 5. Pipho, C. (1991). Centralizing curriculum at the state level. In M. F. Klein (Ed.), Issues in centralizing the curriculum. Albany, NY: SUNY Press. Porter, A. C. (1989). External standards and good teaching: The pros and cons of telling teachers what to do. Educational Evaluation and Policy Analysis, 11, 343-356. Porter, A., Archbald, D., & Tyree, A. (1991). Reforming the curriculum: Will empowerment policies replace control? In S. Fuhrman & B. Malen (Eds.), The politics of curriculum and testing (1990 Politics of Education Yearbook). New York: Falmer. Reilly, D., & Gersh, T. (1988). The current status of educational improvement: Curricula, standards, and quality control. Report to the Southern Regional Education Board, Atlanta, GA. Rosenholtz, S. J. (1987). Education reform strategies: Will they increase teacher commitment? American Journal of Education, 95, 534-562. Rowan, B. (1990). Commitment and control: Alternative strategies for the organizational design of schools. In C. Cazden (Ed.), Review of research in education, Vol. 16. Washington, DC: American Educational Research Association. Rutter, M., Maughan, B., Mortimore, P., Ouston, J., & Smith, A. (1979). Fifteen thousand hours: Secondary schools and their effects on children. Cambridge, MA: Harvard University Press. Schwille, J., Porter, A., Alford, L., Floden, R., Freeman, D., Irwin, S., & Schmidt, W. (1986). State policy and the control of curriculum decisions: Zones of tolerance for teachers in elementary school mathematics (Research Series No. 173). East Lansing, MI: Michigan State University, Institute for Research on Teaching. Shujaa, M., & Richards, C. (1989). Designing state accountability systems to improve schoolbased organizational learning. Administrator's Notebook, 33. Sizer, T. (1992). Horace's School: Redesigning the American high school. Boston: Houghton-Mifflin. Stodolsky, S. (1988). The subject matters. Chicago: University of Chicago Press. Timar, T., & Kirp, D. (1988). Managing educational excellence. New York: Falmer. Viadero, D. (1990, August 1). History curricula stir controversy in largest states. Education Week, 90(33), 38. Teacher Control and Teacher Autonomy Weick, K. (1976). Educational organizations as loosely coupled systems. Administrative Science Quarterly, 2, 1-19. Authors DOUGLAS A. ARCHBALD, Assistant Professor, University of Delaware, 103 Willard Hall, Newark, DE 19716. Specializations: curriculum and assessment policy; magnet schools and school choice policy. ANDREW C. PORTER, Director, Wisconsin Center for Education Research and Professor of Educational Psychology, University of WisconsinMadison, 1025 West Johnson St., Madison, WI 53706. Specializations, educational policy analysis; student and teacher assessments and psychometrics. 39