\ \ [1") ABSTRACT WHOSE PREFERENCES COUNT? A STUDY OF THE EFFECTS OF COMMUNITY SIZE AND CHARACTERISTICS ON THE DISTRIBUTION OF THE BENEFITS OF SCHOOLING By‘ George R. McDowell This thesis examines the arguments and issues surround- ing the basis for drawing or redrawing lines of legal auth- ority around groups of people on maps and causing them there- fore and thereafter to act together as a body politic--the so-called "boundary question." The major dimensions of com- munities which can be affected by boundary changes such as size, community homogeneity, and the position and power of groups of diverse preferences are investigated to examine their effect on the articulation of group preferences as seen via the distribution of the benefits of a publicly pro- vided service. The school district was selected as a unit of government amenable to this investigation because it is a single purpose political entity and because measures of the output of schooling are likely as good as measures of any other public output. A review of the consolidation reform tradition of Amer- ican political science thought makes clear that the major I“..- i..- .u‘1 :AF‘AI wvv‘.-'. q a \ “‘ no "V U- . Q :1" fp .‘vu NU . . V ”A v a hy- V.A\ 3.1" C‘n ‘A.‘~ a: (I) (7 I ) ’) 1‘ I“ o On' a u A V" L)?" a A “§c"1fii MK.” V “ “uh“! George R. McDowell arguments on behalf of the consolidation of local government are efficiency in production of publicly produced goods and services and reduced transactions costs between governmental units. Without questioning whether these in fact exist, the economics of public choice approach was introduced and led to questions of the effect of consolidation on the artic- ulation of diverse preferences.' In order to be able to examine and interpret the effects of size and other community characteristics on the distri- bution of benefits of school districts, a framework of analy- sis of the schooling process was deve10ped. This analysis led to three assumptions about the current public provision of schooling necessary to an interpretation of this research. First it was assumed that, whether true or not, there exists a widespread belief that an individual's performance in school makes a difference in future achievements and ap— proach to "success." Secondly, it was assumed that gener- ally under the existing educational norm, children with ex- periences substantially different from those of the majority middle class are on average disadvantaged in the schools. The third assumption is that parents or groups of parents in a community will seek to make the schools responsive to their preferences and/or particular needs. Based on the evidence that socio-economic class is a consistant influence on the performance of school children. the conceptual basis for an indicator of school district .- .ag-u‘-‘ . . d ' “‘.v-- .vvv...‘ c ‘ ”‘an .~'~-‘\» ...v.‘_‘_\ .IAQ O. a . v...“ "" o-.. . ‘ A, ““‘«~ Ad“ a. .‘Vqu a u ‘rnws“‘& U. .VH ‘5- .n!‘ on“ 5 LL l" I all n (f O ‘5- ~“o ‘m. (\Lr . vy‘ "Fa‘ u‘~’1 ‘5 V“ . C 5o“- 5.: - "up. 0. k " "A George R. McDowell responsiveness to those disadvantaged under the prevailing educational norm was developed. Drawing on the public choice approach and the analysis of the schooling process, three hypotheses were proposed for testing. The first related to the effects of school district size on the responsiveness to the disadvantaged group. The second argued that as the power of the disadvantaged group increases, school districts would be more responsive to their needs. The third argues that as community concensus with regard to schools is more diverse from the preferences or needs of the disadvantaged, the schools will be less re- sponsive to that group. A public choice model of school district behavior was developed and proxy variables were specified to approximate the general model in the empirical portion of the research. Using a socio-economic index for individual pupils, a group felt to be clearly disadvantaged was selected. Two direct measures of the responsiveness of school districts to that group were created. One measured the mean level of achieve- ment obtained by the disadvantaged group. The other meas- ured the relative mean achievement of the low SES group to that of the balance of their grademates. In addition to these two direct measures, the Coefficient of Variation of achievement was selected as an indicator of the responsive- ness of school districts to socio-economically disadvantaged pupils based on the earlier conceptual analysis. pa;- ltd-4- \ 1 J l '1 (u (I) (D 4 George R. McDowell Multiple regression equations approximating the model were fitted to data for town and rural districts, urban fringe districts, cityanuimetrOpolitan districts, and to the pooled data by the ordinary least squares method. The major data sources were (1) the Michigan Assessment Program results for 1969-70 and 1970-71 obtained from the State of Michigan Department of Education, and (2) U.S. Census Fourth Count (Population) data made to coincide with school districts by the National Center for Educational Statistics. Results of this investigation indicate that as school districts are smaller, ceteris paribus, they are more re- sponsive to the disadvantaged group of pupils. This is seen as evidence that increases in community size such as are accomplished by the boundary changes associated with con- solidation efforts result in a decline in the ability of an individual or a group whose preferences are different from the majority to have their preferences felt in the enlarged body politic. Support for the second hypothesis could not be clearly demonstrated although there were indications that the hypo- thisized relationship between group power and distribution of schooling benefits may still be valid. The results with respect to community concensus indicate that as communities are more heterogeneous the schools are less responsive to disadvantaged pupils. Further, as the George R. McDowell majority position of a community is more diverse from the socio-economically disadvantaged, the schools are less responsive to that group. These results are seen as implying that while consoli- dation of communities may reduce the cost of providing a specific service, it will also effect the distribution of the benefits of the service and will influence the character of the demand for that service. Stated again, consolidation may result in lower costs of delivery of public services but may not result in increased community satisfaction with those services. Those who stand to gain from consolidation are thosevflm) are more likely to be closer to the majority of the enlarged community, or those whose relative power in the community will be enhanced. Those who stand to lose are those whose preferences are further from the majority undem'consolidation, whose relative power is reduced, or whose power is unchanged but who will experience higher costs in articulating their preferences. WHOSE PREFERENCES COUNT? A STUDY OF THE EFFECTS OF COMMUNITY SIZE AND CHARACTERISTICS ON THE DISTRIBUTION OF THE BENEFITS OF SCHOOLING BY o‘OQS x George R. McDowell A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Agricultural Economics 1975 @ COpyright by George Robert McDowell 1975 This work is dedicated to Quincy Alonzo McDowell and to William John McDowell. From Quincy McDowell, father and friend, I learned by example of the power of love, of sacri- fice, and of personal consistancy, and that I am my brother's keeper. From John McDowell, uncle and friend, who died while this work was in progress, I learned by example a larger de- finition of who is my brother. ii r- 5. Odofi ' on ‘ ” :n h ~o...u u‘ ‘ n. fin I . 4 0.- “L.‘ -.‘.'. ~- I -—- n -: ~¥ -- n ,_ 5 y... '35: ‘A.\ "1-:- w ”h! "" ‘ CVE 9.. '5 “0 LE"- y.‘\ ACKNOWLEDGEMENTS The history and saga of a PhD program and research is filled with many participants and contributors. This page will acknowledge but a few of those who shared themselves with me in this effort. To Al Schmid, teacher, friend, and Major Professor I am most grateful for the intellectual stimulation and invest- ment which provided much of the basis of this research and of an exciting and rewarding study program. To Ron Edmonds who shared with me his struggle and insights gained from grappling with the Detroit desegregation plan, I owe a major intellectual debt for helping me maintain an overview of the issues being dealt with. The members of my guidance committee, James Shaffer, Byron Brown and Lester Manderschied, all spent many hours questioning, probing, guiding and always encouraging me in this venture. To them I am deeply grateful. There are always "those without whose help..." and this is no exception. Diane Hutchinson, Sandy Suggett, and Kathy Ely all spend many hours, sometimes round the clock sustained only by coffee and pizza, in making this a reality. To them my deepest thanks. Finally, to Renie, Laura and Kim I owe the most. They iii n...‘ u.... (D loved me, supported me physically and emotionally, and mostly cared about me. Thank you. All these and many more have helped me and sustained me. Such errors of fact, logic, method or omission as remain are mine and mine alone. iv .o-.. 3": u «-3; wt A "“F' H v- ‘ O'- .- dov. ~ ':“*cv ’wk‘uva O Y #1" in v}. an... , ,.! (‘1‘... Y >2. ‘1' TF1: D: a... 0‘ Fr 1"I‘:r-p £4le TABLE OF CONTENTS LIST OF TABLES O O O O O O O O O I O O O O O O O O O O 0 LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . Chapter I. COMMUNITY BOUNDARIES AND SIZE A PUBLIC CHOICE QUESTION. 0 o o o o o o o o o o o o o o o o o o 0 Introduction. . . . . . . . . . . . . . . . . The Consolidation Reform Tradition. . . . . . The Public Choice Approach to Governmental organization. . . . . . . . . . . . . . . . Methodological Individualism. . . . . . The Nature of Goods and Services and the Organization of Their Provision . . . . The Effect of Decision Structures on Col- lective Action and Public Provision of Goods and Services. . . . . . . . . . . Public Supply of Goods and Services . . . The Boundary Issue--A Public Choice Critique of Consolidation Reform . . . . . . . . . . Multilayered and Overlapping Governmental Jurisdictions . . . . . . . . . . . . . Economies of Scale. . . . . . . . . . . . Articulation of Preferences . . . . . . . II. THE BENEFITS OF EDUCATION - A PUBLIC CHOICE QUESTION 0 C O O C O O O O O O O C O O O O O O O 0 Introduction. . . . . . . . . . . . . . . . . Research Objectives . . . . . . . . . . . . . The Educational Problem Setting . . . . . . . Do Schools Matter?. . . . . . . . . . . . Genes or Social Class? An Educational Norm. O O O O O O O O O O O O O O O O O 15 17 19 23 25 26 27 28 32 32 33 34 35 36 ’.‘ a...“ I I -" .1” n... h‘ V. .‘ o~.. I,' .1 v..'Afl Q]. y‘h‘ \ -.d~ I“ a _.v ‘_‘1 TABLE OF CONTENTS, continued Chapter What Does Schooling Produce?. . . . . . . The Schooling Process - A Perspective . . Socioeconomic Status and Educational Cogni- tive Styles - A Plausible Explanation of the Role of Social Class in Classroom Resource Allocation . . . . . . . . . . . . . . . . . Research Assumptions . . . . . . . . . . . . . Assumption 1. . . .'. . . . . . . . . . . Assumption 2. . . . . . . . . . . . . . . Assumption 3. . . . . . . . . . . . . . . The Assumptions in Perspective. . . . . . As Others Have Seen It - A Review of the Literature . . . . . . . . . . . . . . . . . The Influence of Size . . . . . . . . . . The Influence of Community POpulation Characteristics . . . . . . . . . . . . THE DISTRIBUTION OF SCHOOL DISTRICT BENEFITS -- A MEASURE OF SCHOOL DISTRICT RESPONSIVENESS. . . . . Introduction . . . . . . . . . . . . . . . . . The Level of Benefit Versus the Distribution of Benefits -- The Brown-Saks Model. . . . . The Distribution of Achievement -- A Measure of School District Responsiveness. . . . . . WHOSE PREFERENCES COUNT? - A PUBLIC CHOICE MODEL . The Development of Hypotheses. . . . . . . . . Hypothesis 1. . . . . . . . . . . . . . . Hypothesis 2. . . . . . . . . . . . . . . Hypothesis 3. . . . . . . . . . . . . . . Beyond Hypothesis Testing. . . . . . . . . . . The Model. . . . . . . . . . . . . . . . . . . Measuring the Output of Schools -- The Bene- fits of Education. . . . . . . . . . . . . . Cognitive and Affective Skills. . . . . . Standardized Achievement Tests. . . . . . To Use or Not To Use -- No Question . . . The Michigan Assessment Program . . . . . vi 44 53 53 54 54 55 57 60 63 65 65 66 74 84 84 88 88 89 89 9O 94 94 96 98 99 M n ""11 \u, y .N. ‘ '0 ~ r. b - TABLE OF CONTENTS, continued Chapter Page Development and Discussion of Proxy Variables. . . . . . . . . . . . . . . . . . 101 Responsiveness to Disadvantaged Under Edu- cational Norm . . . . . . . . . . . . . . 101 Alternative Measures of School District Responsiveness to Disadvantaged Pupils. . 104 Degree of Implimentation of Differentiation Norm in the Production of Educational Experiences . . . . . . . . . . . . . . . 107 The Degree of Community Concensus . . . . . 112 Power of Disadvantaged in School District Community . . . . . . . . . . . . . . . . 116 Community Size. . . . . . . . . . . . . . . 117 Sponsors Vs. Consumers of Education . . . . 118 Summary of Variables. . . . . . . . . . . . 119 V. STATISTICAL METHODOLOGY. . . . . . . . . . . . . . 121 Introduction . . . . . . . . . . . . . . . . . 121 The Empirical Methodology. . . . . . . . . . . 122 The Choice of Multiple Regression Analysis. 122 The Statistical Model . . . . . . . . . . . 122 Estimation Technique. . . . . . . . . . . . 125 VI. RESULTS AND CONCLUSIONS. . . . . . . . . . . . . . 127 Introduction . . . . . . . . . . . . . . . . . 127 Coefficient of Variation of Achievement - (Standard Deviation/Mean) . . . . . . . . 127 Low SES Group Achievement Mean. . . . . . 128 Relative Mean Achievement of Low SES Group to Balance of Grademates. . . . . . . . . 129 The Indicator of School District Responsive- ness to Disadvantaged Pupils . . . . . . . . 147 The Hypothesis Tested. . . . . . . . . . . . . 150 Hypothesis 1. . . . . . . . . . . . . . . . 150 Hypothesis 2. . . . . . . . . . . . . . . . 155 Hypothesis 3. . . . . . . . . . . . . . . . 161 The Sponsors vs. the Consumers of Public Education. . . . . . . . . . . . . . . . . . 170 The Differentiation Educational Norm Revisited --The Effects of Educational Strategies on Disadvantaged Pupils . . . . . . . . . . . . 172 vii TABLE OF CONTENTS, continued Chapter Page Community Types - What Differences?. . . . . 181 \EI. SUMMARY AND IMPLICATIONS. . . . . . . . . . . . 194 A Summary of the Problem and Research Approach . . . . . . . . . . . . . . . . . 194 The Boundary Question and Associated Issues. 197 Results and Conclusions With Respect to Community Size . .‘. . . . . . . . . . 197 Results and Conclusions With Respect to Group Power. . . . . . . . . . . . . . 197 Results and Conclusions With Respect to Community Consensus. . . . . . . . . . 198 Implications . . . . . . . . . . . . . . 199 Issues Related to the Organization and De- livery of Public Schooling . . . . . . . . 203 Results and Conclusions with Respect to Instructional Materials. . . . . . . . 203 Results and Conclusions with Regard to Characteristics of Staff of Schools. . 204 Results and Conclusions with Respect to Class Size . . . . . . . . . . . . . . 206 Implications . . . . . . . . . . . . . . 206 Suggestions For Further Research . . . . . . 210 viii -q . :5 ‘M-e 30 \‘L, 4‘ Q I “v ‘1. n- . ‘ .3” J~;e S o 7'.‘ . .Z'J‘e 6. M .‘ \ ‘4F‘ uv‘e 7 Jar/2n; Table Table Table Table Table Table Table Table Table LIST OF TABLES Page Comparisons of Collective Cognitive Styles of Elementary Teachers with Collective Cog- nitive Styles of Elementary Pupils (K-3) by Pupils' SES as Determined by Residential Areas. . . . . . . . . . . . . . . . . . . . 49 Comparisons of Collective Cognitive Styles of Elementary Teachers with Collective Cog- nitive Styles of First Graders by Pupils' SES as Determined by Residential Area. . . . 50 Ordinary Least Squares Regression of Measures of School District Responsiveness to Dis- advantaged in Michigan School Districts (Fourth Grade Mathematics Achievement Score Measures). . . . . . . . . . . ... . . . . . 131 Variable Means and Standard Deviations for Equations Fitted To Data for All Available Michigan School Districts. . . . . . . . . . 133 School District Responsiveness to Disadvant- aged in Michigan Towns and Rural School Dis- tricts (Fourth Grade Mathematics Achievement Score Measures). . . . . . . . . . . . . . . 135 Variable Means and Standard Deviations for Equations Fitted to Data for Michigan Town and Rural School Districts . . . . . . . . . 137 School District Responsiveness to Disadvant- aged in Michigan Urban Fringe School Dis- tricts (Fourth Grade Mathematics Achievement Scores Measures) . . . . . . . . . . . . . . 139 Variable Means and Standard Deviations for Equations Fitted to Data for Michigan Urban Fringe Districts . . . . . . . . . . . . . . 141 School District Responsiveness to Disadvant- aged in Michigan City and Metropolitan School Districts (Fourth Grade Mathematics Achievement Score Measures). . . . . . . . . 143 ix Table Table Table Table Table Table Table Table 10. ll. 12. 13. 14. 15. 16. 17. Page Variable Means and Standard Deviations for Equations Fitted to Data for Michi- gan City and Metropolitan School Districts. 145 Elasticities of Response with Respect to Size of Measures of School District Re- sponsiveness to Disadvantaged Fourth Grade Pupils. . . . . . . . . . . . . . . . . . . 152 Elasticity of Response with Respect to Disadvantaged Group Power of Measures of School District Responsiveness to Dis- advantaged Fourth Graders . . . . . . . . . 157 Distribution and Level of Power Measures by Data Groups. . . . . . . . . . . . . . . 159 Elasticities of Response with Respect to Various Community Characteristics of Mea- sures of School District Responsiveness to Disadvantaged Fourth Grade Pupils . . . . . 163 Characteristics of the Distribution of Per- centage Black by Data Groups and Direction and Significance of Influence of Percent Black on Measures of School District Re- sponsiveness. . . . . . . . . . . . . . . . 168 Elasticities of Response with Respect to Percentage of School Consumers of Measures of School District Responsiveness to Dis- advantaged Fourth Grade Pupils. . . . . . . 171 Elasticities of Response with Respect to Various Measures of School District Instruct- ional Strategy of School District Responsive- ness to Disadvantaged Fourth Grade Pupils . 174 an ' .‘u- 'I, -I . b ‘v-‘v "‘3‘. ‘ » ~w,' v .‘J c.1 t (D W\, :ure Figure 1. Figure 2. Figure 3. Figure 4. Figure 5. Figure 6. Figure 7. Figure 8. Figure 9. Figure 10. Figure 11. LIST OF FIGURES Decision rules and the costs of collect- ive action. . . . . . . . . . . . . . . . . Median demand for services in homogeneous and heterogeneous groups. . . . . . . . . . Recursive model of intergenerational ef- fects on income or occupation. (Bowles 1973) o o o o o o a o o o o o o o o o o o o The Bowles model of intergenerational ef- fects on income or occupation modified to illustrate research assumptions . . . . . . Types of welfare indifference curves for achievement mean and standard deviation . . Production possibilities set for two pupils. . . . . . . . . . . . . . . . . . . Production choices under elitist and lev- eler preferences with two alternative edu- cational technologies . . . . . . . . . . . The effects of alternative educational strategies on pupil outcomes. . . . . . . . The effects of alternative educational strategies on pupil outcomes. . . . . . . . Strategies available to parents of disad- vantaged pupils where educational technology is fixed and is marginally more productive on advantaged pupils. . . . . . . . . . . . Strategies available to parents of disad- vantaged pupils where community preferences for pupil achievement mean and standard deviations are fixed. . . . . . . . . . . . xi Page 21 23 55 56 68 70 71 74 76 79 81 an“ &v'. -‘DPI a-‘bofi -‘Q (I! ~---~ .. 'vlv“ 4 Wm Q ...:q' ~‘*..-V. , _y I‘V'--u‘ ‘Dw. c . <..._ a I‘ "v\. ‘Vb‘v .0 ‘ . T‘.:_.".. u..‘\" . ~:.. 5“ ...v‘, a 1.. V “\R‘I ‘ ' N §‘¥‘ ‘ . 47-,“- ‘Q‘ “r . \,_ :VP-‘,.a "\ u" u “u o ‘5 A \.“.: ~y CHAPTER I COMMUNITY BOUNDARIES AND SIZE - A PUBLIC CHOICE QUESTION Introduction This thesis is aimed primarily at issues related to the so-called "boundary question" of the organization of local governmental units. Yet a major part, perhaps even a major- ity, of the thesis will deal with an analysis of the distri- bution of the standardized achievement test scores of ele~ mentary school children. It is my intent in this introduc- tory chapter to show the relationship between these seemingly diverse questions and how an understanding of the factors affecting the distribution of school children's math scores may shed light on issues of concern in the organization of local units of government. The "boundary question" as here used refers to the arguments and issues surrounding the bases for drawing lines around-groups of people and causing them therefore and there- after to act together as a body politic. Boundaries are thus seen as the institutional rule or set of rules which deter- mine group membership. Boundaries may have physical and spatial dimensions as in the definition of the physical juris- diction of a regional government, or may have social, occu- pational, financial or ideological dimensions as in the case of churches, unions, lobby groups, clubs, and other such ass“ 1 --.... J u ..A ~ F- Y ...» .. 4 wv-A “-5 Ivy...’ '. ugh. . . ~---- F . 5...“... "‘ v. -. — _' I. v- . u“ D 3 " ~ vb . "& w T" .. "’N‘ v o -F\-.- __ x y.- - u-‘V‘a V -~\ “9 V ‘A‘ ~'~.«. 'u. [U f « organizations. The clearest effect of the boundary institu- tion is to determine who is ixltjuegroup and who is outside timagroup. Thus boundaries are a major influence on the size <1fgroups and on the characteristics of the people who hold group membership. Changes in boundaries may thus result in changes in group size and in the characteristics of thelgroup. The boundary issue as it is investigated in this thesis pertains to community boundaries as they affect size and population characteristics. To some extent boundaries are only an issue when some effort is made to change them. The most concerted, and perhaps most frustrated, effort at chang- ing community boundaries has been the movement on behalf of metropolitan reform through consolidation of governmental units primarily in urban areas. The effort at consolidation of rural school districts is in the same tradition and has been somewhat more successful. The arguments for consolidation are those of efficiency in the production of publicly provided services, i.e., an economies of scale argument, and simplicity in the admini- strative boundaries of governmental units, also an efficiency argument. The latter is directed against the profusion of special purpose districts and other locally empowered units resulting in a multilayered structure with many overlapping jurisdictions. A major element overlooked in the consolidation liter- ature is the affect of boundary changes on the articulation 5 w” u. .. 1 vi v . .. . .3 r; O n» a a» a. ..l v. O 0‘ U «1| e to Q a: a . Pu ...n p. . K . a: 3L 1J4 . u .5 NJ 1: F. \ ”I ‘ F‘. P» -. . ~\U AH.» Au.‘ ... . .H u ‘ D 3V‘un- “*EK Qy «Q L1. 3» h" v a v a «In and assertion of diverse preferences for the services of government. In terms of the arguments for consolidation the overlooked question is "on whose behalf is the new efficiency?" It is the intent of this thesis to examine the effects of various characteristics of school districts which are associated with district boundaries, on the distribution of the math scores of school children. The objective is to seek to understand the effect of boundaries on whose pre- ferences count in terms of a quantified, albeit limited, measure of service output. The Consolidation Reform Tradition The "one community--one government" or consolidation thesis for organizing metropolitan areas is a dominant argument in current discussions of local government reorgan- ization. The 1972 Report of the Governor's Special Commis- sion on Local Government in Michigan argues for the consoli- dation of the smaller units of government (villages) into larger units in the interest of "more effective service areas" [38] with regard to publicly produced services. In the Fall of 1973 discussants of the commission re- port in a Public Policy Forum on the Alternatives for Mich- igan Local Government [50] generally endorsed the consolir dation theme of the report. Of the fifteen speakers, among whom were journalists, academicians, and representatives of ...v‘ . -. A‘ yd ‘ AI -V. ’4‘ Pu. -‘.¢ “A. .- Lyln b I l h“. _ any . . A: one AU LL . A .C - fit nlu «J .J 15. {.1 Tia Lk Pu.» I? t *5 .«d uxu .~d 11‘ L» r. h b lobby organizations, only one explicitly questioned the con- solidation aspects of the report.1 The consolidation arguments, which are alive and well in 1973 as evidenced in the above discussion, can be right- fully called a tradition of American political science thought. Bish and Ostrom [7] in describing the consolida- tion approach to government reform attribute its origins to a political science made popular by Woodrow Wilson and his contemporaries nearly a century ago. Warren in tracing the history of the literature specific to consolidation reform points out that much of the current dialogue is based on assumptions formulated in the 19205 as part of an initial response by political scientists to metro- politanization and the fractionation of municipal government within population centers. According to Warren the recur- rent and identifiable assumptions in the literature of this tradition are as follows: a) "metropolitan areas represent a single community linked by social and economic ties, but are arti- ficially divided governmentally;" [51] b) "The public needs of such a community cannot be 1 It is not the intent of this discussion to indicate that the Report of the Governor's Special Commission on Local Government [22] dealt only with problems where the pro- posed amelioration was consolidation. It is true that the report was in the consolidation reform tradition. satisfactorily met by the collective action of num- erous units of government, rather chaos and break- down will result;" [51] c) "...the welfare of the metropolitan community can only be realized through an integrated governmental structure in which municipal decision making auth- ority is formally centralized in a single juris- diction [51]." Bussard in her review of the rural school consolidation movement quotes a 1897 committee of the National Education Association making a very similar argument as follows: ...the necessity of adopting a larger unit than the district, as the township or the county, is very strenuously insisted upon by two or more subcommittees....It would conduce to effective- ness and simplicity of organization; to economy in the use and distribution of funds; to the equalization of the burdens of taxation, and to a system of supervision which would produce better results from the instruction given in rural schools [20]. Thus it is that analysts in this reform tradition when considering the problems of local government view small units as unprofessional and inefficient. The commitment of small jurisdictions to local interests is seen as parochial and standing in the way of achieving the overall public interest of the larger community. Fragmented authority and multi- layered overlapping jurisdictions among numerous units of local government are diagnosed as the fundamental sources of institutional failure in the governance of many areas, particularly urban areas. I. u: ‘ 1 H! O—‘ ...,“ b--.‘ I‘ "1- “~'-. 0“;‘ L.‘ - < ‘ VA a 1‘:a ‘ h r, ‘0‘ H ‘\ From this perspective, overlapping jurisdictions imply duplication of services produced. This duplication implies waste and inefficiency in government. According to these analysts, efficiency is enhanced by eliminating the many small jurisdictions and by consolidating all authority in one jurisdiction with general authority to govern each major urban area as a whole. Such consolidations vest ostensibly enlightened leaders and professional administrators with authority to coordinate all aspects of regional affiars through a single integrated structure of government. Policy analysts in this tradition assume that in addi- tion to the economies of scale gains to be made, consolida- tion of all smaller jurisdictions into a single, overall unit of government for each urban region or metropolitan area will clearly fix political responsibility, making it possible for citizens to hold officials accountable for their actions. By contrast the requirement to deal with numerous, overlapping jurisdictions is seen as overloading citizens, confusing responsibility, and frustrating citizens in their efforts to control public policy. Further, efforts to solve regional problems are frustrated by chaotic bicker- ing between these small selfinterested units. The persuasiveness of this approach to local government policy analysis is indicated by its use in a report by the Committee for Economic Development, Modernizing Local Govern- ment, published in 1966. The CED reported the following ,. O‘flfl‘ .-.-Eli ‘5‘ ‘»{:‘ 1 “bl.“n ~ V. . ‘ O ’0 0 (Pt. ‘Uah I L. E“; . n . CI findings as a basis of its diagnosis: 1. Very few of the local units [of government] are large enough--in population, area or taxable resources--to apply modern methods in solving current and future problems....Even the largest cities find major problems insoluble because of the limits on geoqraphic areas, their taxable re- sources, or their legal powers. 2. Overlapping layers of local government--muni- cipalities and townships within counties, and independent school districts within them--are a source of weakness....This [overlapping] impairs overall local freedom to deal with vital public affairs; the whole becomes less than the sum of its parts. 3. POpular control over government is ineffective and sporadic, and public interest in local poli- tics is not high....Confusion from the many-lay- ered system, profusion of elective offices without policy significance, and increasing mobility of the population all contribute to disinterest. 4. Policy-making mechanisms in many units are notably weak. The national government [by con- trast] has strong executive leadership, supported by competent staff in formulating plans that are then subject to review and modification by a rep- resentative legislative body.... 5. Antiquated administrative organizations hamper most local governments. Lack of a single execu- tive either elective or appointive is a common fault. Functional fragmentation obscures lines of authority....The quality of administration suffers accordingly [23]. Bish and Ostrom in describing this reform tradition point out that the arguments suggest the following causal relationships among variables: "(1) Increasing the size of urban governmental units through consolidation will be assoc- iated with improved output of public services, increased efficiency, increased responsibility of local officials and increased confidence among citizens about their capacity to affect public policies. (2) Reducing the multiplicity of jurisdictions serving an urban area through consolidation will also be associated with improved output of public ser- vices, increased efficiency, increased responsibility of local officials and increased confidence among citizens about their capacity to affect public policies [7]." Before critiquing the arguments of the consolidation reform tradition, it is perhaps useful to examine the per- formance record of this view in terms of its acceptance and implementation as a basis of governmental reform. Bussard [20] in her discussion of the consolidation of school dis- tricts reports the Digest of Education Statistics, 1970 as showing a decline from approximately 127,000 districts in 1930 to an estimated 17,000 districts in 1971. Much of this consolidation activitv involved rural districts and that movement appears to have been strongly influenced by the real and argued economies of size related to production of sec- ondary school education. The record of consolidation reform of metropolitan government is much less clear than that of school district reform. Warren writes, "...the ideal of centralization ap- pears to be deferred rather than compromised. Few large cities and adjacent suburban areas have not been the subject of at least one such proposal. However, the results of these efforts have been exceedingly limited [51]." as p u a o ”4:... ...» «a 71 n“ fl» 2 «a .l 0 l .Fu 1hr” r. a. a n 2 e C a. a: 3 2‘ r. .n. ..u I w: b . (w 1n n o . r2. r. .- . .. ”I a.» .J flu Illl.‘ .4 VA: ~ A: u «d 3 PB .6 i tits [’11. !,\~ ~«ices w :51 “'7 Writing in 1966, Warren reports that of twenty-four Standard Metropolitan Statistical Areas with a million or more inhabitants (1960 Census) only two had any actions taken which could be interpreted as governmental reform. In the other 190 metrOpolitan regions with populations ranging from 50,000 to 1,000,000 (1960 Census) only three plans of re- organization have been affected according to Warren. As a result of repeated defeats at the polls and public resistance generally to consolidation reform, the literature in this tradition in recent years has included discussions of strategy aimed at the "successful" implementation of re- form adoption. A good example of this response of political scientists to successive public rejections of their position is a mono- graph by Booth, "MetrOpolitics: The Nashville Consolidation" [11]. In the foreward, Charles Adrian, formerly of Michigan State University Institute for Community Development and Services writes as follows: "Among the many other questions raised in this study are some related to strategies for metro- politan change. What are optimum strategies for metropolitan change? What are optimum strategies for victory in a metropolitan inte- gration proposa1?...Reformers, being educated persons who view their approach as highly rational, have strongly tended to accept the eighteenth century more as a model for the metropolitan voter. The "pro"--who knows better than to do so--thus reached more peOple and reached them in a telling fashion....No one has yet written a monograph dealing with such ques- tions as: What are the emotional and symbolic assets available to the advocate of metropoli- tan reorganization? How can they be brought foll' LEE flit-u .104: .‘10 10 effectively into use in a campaign?.... What aspirations of citizens can be appealed to effectively? That is, how can people be led to believe that there is a causal relationship between an integrated governmental structure and their personal hopes for the future? Some reformers will be reluctant to ad— mit that the ordinary tool kit of the practicing politician should be used to repair metropolitan governmental machin- ery. The story of Nashville hints that perhaps they would be wise to cease be- ing squeamish [11]." The committee for Economic Development in their national policy statement Modernizing Local Government, in 1966 made the following exhortation on behalf of consolidation reform: "Each citizen, therefore, has a civic responsibility, as well as an enlightened self-interest, in securing the moderniza- tions now long overdue [23]."2 While the consolidation reform tradition represents a substantial school of thought in modern political science, it is appropriate to point out that there are within the main stream of political science those who take a more studied view of the role of community size. The recent (1973) book Size and Democracy [25] by Dahl and Tufte sets the issue of size and democratic objectives in rather clear view. Within the first chapter of their work these authors set forth the following main claims and counterclaims of the respective sides in the small versus large size democratic units of government debate. 2 Emphasis added. l 11 ON CITIZEN PARTICIPATION 1. Smaller democracies provide more opportunity for citizens to participate effectively in decisions. 2. But: Larger democracies provide opportunities for citizens to participate, at least by voting in elections, in the decisions of a political system large enough to control all or most of the major aspects of their situation that can be controlled. ON SECURITY AND ORDER 3. Smaller democracies make it easier for citi- zens to internalize norms and values, hence to in- crease voluntary compliance and reduce coercion. 4. But: Larger democracies provide an opportun- ity to extend the rule of law (as opposed to violence among states) over a larger area. 5. And: Larger democracies are better equipped to prevent damage to the internal life of the soc- iety from outside forces-- such as invasion and economic threats. ON UNITY AND DIVERSITY 6. Smaller democracies are likely to be more nearly homogeneous with respect to beliefs, values, and goals. 7. Conversely: Larger democracies are likely to exhibit more diversity in beliefs, values, goals, social and economic characteristics, occupations, etc. ON THE COMMON INTEREST 8. Smaller democracies make it easier for a citizen to perceive a relation between his own self- interest or understanding of the good and a public or general interest, the interests of others, or general conceptions of the good. 9. Conversely: Larger democracies provide more Opportunity for divergence of views on individual, group, and general interests and goals. 10. And: Larger democracies reduce the likeli- hood that a single interest of one segment of the members will dominate the whole system (as happens, for example, in company towns). 12 ON LOYALTIES 11. Smaller democracies are more likely to generate loyalty to a single integrated community. 12. Conversely: Larger democracies are more likely to generate multiple loyalties to various "communities." ON EMOTIONAL LIFE 13. Citizens of smaller democracies are more likely to invest civic relationships with high levels of affect. 14. Conversely: Citizens of larger democracies are more likely to divest civic relations of affect, to make civic relations more impersonal and emotion- ally neutral. 15. The citizens of smaller democracies are likely to consider each other friends or enemies, according to whether they agree or disagree on politics. 16. Conversely: The citizens of larger democrac- ies are less likely to consider their fellow citizens either friends or enemies for political reasons. 17. Smaller democracies produce stronger pres- ures for conformity to collective norms. 18. Conversely: Larger democracies weaken pres- ures for conformity to collective norms. 19. But: Alienation and anomie--1oss of communi- ty--are much more likely in larger democracies. ON RATIONALITY 20. Smaller democracies make possible greater speed and accuracy of communication among all members of the system. 21. And: Smaller democracies provide more oppor- tunities for all citizens to gain knowledge needed for decisions by direct observation and experience. 22. But: Larger democracies provide more oppor- tunities for all citizens, acting collectively, to exercise control over a broader range of important matters---and hence over their situation. 13 23. And: Larger democracies provide greater opportunities for individuals to develop special- ized skills, hence to deve10p skills needed for rational solutions to problems. 24. In general, then, citizens in a smaller democracy are likely to understand their poli- tical problems better than citizens in a larger democracy (consider propositions l, 6, 8, 20, and 21). 25. On the contrary, citizens in a larger democracy have greater opportunities for ex— ploring a bigger set of alternatives than citi- zens in a smaller democracy; hence they are the more likely to understand their political pro- blems better and to control their situations more completely (consider propositions 2, 4, 5, 9, 10, 12, l6, 17, 22, and 23). ON CONTROL OF LEADERS 26. Leaders are likely to be more responsive to citizen views in smaller democracies (consider propositions l, 3, 6, 8, ll). 27. On the contrary, leaders are likely to be more responsive to citizen views in larger demo- cracies (consider propositions 10, 16, 18) [25]. Dahl and Tufte distill these claims and counter claims into two major substantive issues or goals which appear in conflict or must be traded off in size considerations. These twin goals are "citizen effectiveness" and "system capacity" in the language of the authors. Thus there appears to be the acknowledgement of the affect of size on the articulation of citizens preferences which is missing in the consolidation reform literature. Dahl and Tufte proceed to examine some of the influence of size on aspects of these goals using data from a number of small European countries. In general they have used whole quv- .. iv 5". . u “I“ I...‘ III-”~0- aim. “~~. -.-: .. .‘ .- ‘Ib 14 countries as their units of observation and have attempted to interpret such measures as degree of representation, governmental form, party membership and party competition, and citizens sense of knowing about current issues. While such between countries comparisons are obviously difficult and perhaps suspect in a statistical sense, these writers have nevertheless made some thoughtful observations. In their conclusions they make clear that their inves- tigation persuades them that no single type or size of gov- ernmental unit is optimal for achieving both citizen effec- tiveness and system capacity. Thus they argue that both large and small units are needed - an argument that would seem to imply the continuation of multilayered government so vigorously opposed by the consolidation reform tradition. Unfortunately these authors offer little in the way of sug- gesting a basis for determining or debating what activities might be carried out at what level of government organization. The Public Choice Approach to Governmental Organization According to Bish and Ostrom [7] the intellectual roots of the public choice "way of thinking" are contained in the essays by Alexander Hamilton and James Madison in The Federalist and the writing of Alexis de Tocqueville in his Democracy_in America. The more recent intellectual precur- sors of the public choice approach are the welfare and polit- ical economists concerned with public expenditure and public Nka v \u;1 , -v‘fi ..fi. bu'vx ‘AV’ ‘0“. I" y a. \fiu ad 5 15 investment decisions 3 la benefit-cost analysis and Program Performance Budgeting (PPB) systems. Methodological Individualism.--The B/C-PPB analyst and the traditional public administration approach, which argues for clear lines of authority, share a common methodological perspective different from that of the public choice approach according to the Drs. Ostrom [41]. The B/C-PPB analysts and the public administrator take the perspective of "omniscient observer" and assumes he/she can know the will of the state or "the public interest,” i.e., the social welfare function. The public choice analyst begins with a methodological individualism. Buchanan makes this point in contrasting the public choice approach with what he calls the "open" system anal— yzed in traditional economic theory [17]. According to Buchanan traditional theory is a highly deve10ped system of market interaction. Beyond the limits of market behavior the analysis is left open. "The 'public choices' that define the constraints within which market behavior is allow- ed to take place are assumed to be made externally or exo- genously, presumably by others than those who participate in market transactions and whose behavior is subjected to the theory's examination" [17]. The public choice approach as set forth by Buchanan as- sumes that the same individual who acts in the market place . ~ ,. o ..H 5615‘ 7! cu In (I) (h = r , «“1 q 4¢Q~~ ‘w. ‘5. “ .q“ .. \- «2-, ‘ h““: ‘\ 9:.) ‘~‘ l J «h‘ y u] Q‘ ‘x. ‘l .a-h . 16 also is an actor in the political process. "Individuals be- have in market interactions, in political-governmental inter- actions, in cooperative-nongovernmental interactions and in other arrangements" [17]. Unlike the traditional theory where the behavioral system is artificially closed by some action of an omniscient government or rule maker, the public choice approach with its individualistic methodology is be- haviorally closed by the actions of the same people who act in the market place. The public choice approach is explicitly a democratic model, that is a model where the rulers are also the ruled. In acting or behaving as a "public choice" participant, the individual is presumed to be aware that he/she is in part selecting results which affect others than him or herself. While the public choice approach goes well beyond the traditional theory of markets it remains nevertheless a con- struct based on an economic model of behavior. As with the traditional economics as applied to individuals Operating in the market, the individual operating as a citizen is assumed to be motivated by the desire to maximize his/her own utility. In this pursuit different people are assumed to have different preferences. Based on the information available each will weigh and choose between alternative possibilities in relation to their preferences. Since the acts of choosing may in varying degrees affect others, the individual's self- uvl on L) O '3 “y. “do. ~ Rn;- woo. ‘1‘ (I) v' a “I ' i (T) ‘1) 'I 1 L." (b I i): A‘! ‘v (9‘. H I} ’1 1 (V 17 interest may include a concern for the well-being of others. The Nature of Goods and Services and the Organization of Their Provision.--As elaborated above, individuals are presumed to assert their preferences both in market activi- ties and as citizens. Among the array of possibilities are choices of different particular goods and services as well as choices as to whether a particular good or service will be produced in the private market or in the public sector. The public choice approach argues that the characteristic of the good itself instructs the choice between government or private provision. The distinction is thus made between "public goods" and "private goods" as two ends of a continuum. The private good is such that when exchanged the benefits of the good are the exclusive domain of the purchaser. Goods such as apples, bread or automobiles which are packageable can be withheld from all but those who pay the price and thus fit the exclusion definition of a private good. Such goods can be dealt with effectively by the private market with public action required only to assure that contracts are enforced and to resolve disputes between participants in market transactions. Other goods and services such as national defense and control of disease, among others, are such that when avail- able to one individual are available to all within the com- munity. These goods with the characteristic making exclusive 18 benefits to a single individual virtually impossible are de- fined as "public goods." As stated earlier public goods and private goods are extremes on a continuum. The measure of how "public" or "private" a good is will depend on how costly exclusion will be. For example, highways do have in some degree the characteristics of public goods. Nevertheless, in some situations like limited access toll roads, the exclusion of non-paying individuals is undertaken, albeit not without some additional expense. The public choice literature argues that the costs as- sociated with exclusion of a particular good or service is a substantial influence on the choice between private or public provision of the good. If the costs of exclusion are great, private entrepren- eurs may simply not undertake to produce the good or service. If payments for public goods are sought on a voluntary basis then each individual will find it in his/her interest to withhold payment while continuing to enjoy the benefit. One response to this "free rider" problem is for citizens through government to use the coercive power of taxation to insure that each individual contributes his/her share. While the "public good" characteristics of a commodity do seem to instruct the choice between public and private provision, many relatively private goods, i.e., where exclu- sion is not costly, are also provided publicly. Such is the “V ' Ab 19 case of education where although widely available as a privately produced commodity, it is also generally publicly provided. The public-private characteristics of goods also are seen by public choice analysts as instructive to the level of government which organizes provision. In addition to being producing organizations of goods and services, govern- mental organizations are seen as a mechanism through which citizens can communicate their preferences. Thus different goods or services, dependent on the spatial domain of their public good characteristics, will affect different groups. National defense whether produced by an individual, a city, a state or a national governmental body, will affect all within the national boundaries. Aside from production is- sues, the national level of a democratic government is re- presentative of the "community" affected by national defense and where individual preferences for that service are articulated. The Effect of Decision Structures on Collective Action and Public Provision of Goods and Services.--As indicated earlier, the costs of exclusion associated with public goods may preclude their private provision. The "free rider" problem substantially works against their continued volun- tary collective provision. Yet, many goods and services, both private and public, are provided by collective action through governmental bodies which do not rely strictly 11: (I! u— 5* .5 'v I I I“ m. n.. .A 9v ,1 ‘1 rt. - . L. b 20 upon the voluntary consent of all who are affected. The calculus an individual seeking public provision of a good or service would need to go through is developed by Buchanan and Tullock in their Calculus of Consent [18]. Ac- cording to Buchanan and Tullock such an individual would need to consider two types of costs: (1) political external- ity costs--the costs which an individual expects to bear as a result of decisions which deviate from his own particular preferences, and (2) decision making costs--the expenditures of time, effort, and opportunities foregone in decision making. Both kinds of costs are affected by the decision rules which specify the proportion of citizens required to agree to a future collective action-- the constitutional decision rule. Where the rule is one of unanimity, the political externality costs will be zero since any one individual can defeat the decision if his/her preferences are not met. Similarly, the decision making costs will be extremely high in finding that view (or level of service) over which there can be unanimous consent. Where one individual is permitted to decide on behalf of the group, the Opposite would obtain. Decision costs would be low but political externality costs would be high. The self-interested individual would seek to minimize both of these costs. When, as in Figure 1a, the two cost curves which the individual perceives are symmetrical, some o: 1 a t‘U A\v Isy s . 21 form of simple majority rule would be indicated. When as in lb the opportunity costs involved in decision making are very large in comparison to political externality costs, the individual is expected to prefer a less than majority deci- sion rule. In an extreme case where a rapid decision re- sponse is desired decision making even may be vested in a single individual. Similarly, as in lc when the political externality costs are perceived to be high relative to decision costs, the preferred rule would require a substan- tial majority. Figure 1. Decision rules and the costs of collective action. Expected Costs Political Externalities Decision Making gdp-- a D 0% Proportion necessary to decide la 100: Expected Costs Political Externalities Decision Making 0% 100% Expected Costs Political xternalities ‘D-- Decision "akin 0% 501 In commenting on these notions, the Drs. Ostrom write "An optimal set of decision rules will vary with different situations and we would not expect to find one good rule that would apply to the provision of all types of public goods and services" [41]. 1001 no. "V n- ‘0- b‘. RA ‘v II o v. (I- b. '1 ~l . ‘- 22 A loqical question following from these notions is "does the decision structure affect the articulation of individual preferences and instruct public response to those preferen— ces?" The literature examining voting under majority rule seems to give some insight to this question. Duncan Black in The Theory of Committees and Elections [8] has demonstrated that if the preference orderings of a community are single-peaked, then a choice reflecting the median preference position will dominate all others in a majority vote. The Drs. Ostrom [41] credit Edwin Haefele among others with pointing out that this solution has impli- cations for the strategy of those who must win approval Of an electorate. If representatives are aware of their constituents' pre- ferences, the task of developing a winning coalition depends on the formulation of a program which occupies the median position of voter preferences, providing that voters are mak- ing a choice between two alternatives. Similarly, political leaders or administrators of public programs would have an incentive to develop programs oriented to the median prefer- ence position of their constituents. Voters, if given a choice, would then choose the alternative which most closely approximates the median position. Bish [6] employs this median position notion to illus- trate the effect of a community's homogeneity or heterogen- eity of preference for a particular good on the political :— ?. ‘u. ‘oai 23 externality and decision making costs incurred. Figure 2a represents the individual demands of a homogeneous group and Figure 2b illustrates the demands of a heterogeneous group. Figure 2. Median demand for services in homogeneous and heterogeneous groups. P '4) Individuals' individuais‘ w Demands kDemands .——- --- I I I l l I l I , , I I I | . . Q I l I n 2}. 3‘.‘ ' .c. I I “1 j Qaugo Range Homogeneous Group Hetrogeneous Grouo 2a 2b While this analysis based on voting behavior and consti- tutional rules is instructive, it assumes that the major means of expressing preferences is via citizens' voting. Implicit, therefore, is the assumption that the power to as- sert preferences is uniformly distributed within the relevant community. More on this issue will be discussed subsequently. Public Supply of Goods and Services.--Previous discus- sion Of the public choice approach has dealt with issues which instruct the choice between public versus private O u .n ‘v “I b" 24 provision of goods and services, the level of government which defines the community of affected citizens, and the articulation of citizen preferences. In addition to these, public choice analysts include supply and management con- siderations when assessing the organization of governmental units and public service delivery. Again a characteristic of some goods is instructive to both the public versus private provision choice and to the scale of production. Some goods, such as electrical power, telephone service, and sewer service, have the characteris- tic associated with their production and delivery such that the marginal cost to an additional user is very small or negligible. Many such goods are produced under technologies which are highly capital intensive and require a large market to justify their provision as either a publicly or privately produced good. Since in the case of many such goods, exclu- sion is feasible, their production under private circumstan- ces is possible. However, because of the declining marginal cost to additional users, production is profitable at prices below private market equilibrium conditions. This is the natural monopoly argument and is the analytical basis for many "regulated industries." It also serves to introduce the economies of scale argument. Public choice analysts and economists generally would argue that economies of scale are very much a function of the technology employed in the production process, i.e., are ..~ In. ‘- ... \ .11“ \ 25 a function of the physical and technical input-output rela- tionship. As such, in a society with a rapidly changing technology, the particular optimal scale Of production of a particular good is of a rather transcient nature. Similarly, in a public sector which provides a large number of goods and services which are characterized by a high degree of interpersonal interaction, the possibility exists for dis- economies of scale to occur. Finally, each product or service produced will have its own technical production relationship; each will have its own optimal scale and that scale will change as technology changes. The Boundary Issue--A Public Choice Critique of Consolidation Reform Thus far an effort has been made to present synopses of the Consolidation Reform Tradition and of the Public Choice Approach to governmental reform. The former has been, as is representative of that literature, an argument for primarily one solution--consolidation. The public choice approach has been a setting forth of a number of analytic principles and propositions. It is the intent of this section to make ex- plicit some of the conflicts between the two approaches and to make clear the context of the research undertaken and presented in this thesis. n‘h Ayn ~\U PM I. A ~~~‘ ‘ s \ 26 Multilayered and Overlapping Governmental Jurisdictions.-- As set forth earlier, the consolidation reform tradition sees the fragmentation and overlapping of governmental juris- dictions as a major source of inefficiency and institutional failure in the governance of many areas. Improvement of government performance is only possible in this context if, for the region concerned, an integrated governmental struc- ture is developed in which decision making is formally cen- tralized in a single jurisdiction. The public choice approach challenges both the proposed solution and the analysis of the problem. As is pointed out by the Drs. Ostrom [41], from the logic of constitutional decision making, public agencies are not viewed simply as bureaucratic units which perform those services which someone at the tOp instructs them to perform. Rather, they are viewed as a means for allocating decision making capabilities in order to provide goods and services responsive to the preferences of individuals in different social contexts. That is, they are not simply supplying or producing organizations but are for the purpose of facili- tating citizens' articulation of preferences. While, according to the Ostroms centralized decision making may have the potential of reducing substantially de- cision making costs, the possibility of reducing the ex- pected political externality costs under such circumstances can be realized only if the following conditions are met: 27 l) "appropriate decision-making arrangements are avail- able to assure the integrity Of substantial unanim- ity at the level Of constitutional choice." 2) "methods of collective choice are continuously available to reflect the social preferences of mem- bers of the community for different public goods and services [41]." While the possibility of reducing decision making costs by consolidation may exist, there are those who would argue that there is no a priori basis for expecting intergovern- mental transactions to be more costly (pre consolidation) than intragovernmental transactions (post consolication) [47]. Using these arguments directed at the consolidation "solution" to multilayered and overlapping jurisdictions, and adding to them the notions of 1) public goods affecting different communities, and 2) different optimal scales for different goods in production, the public choice approach could argue that further decentralization would lead to more effective government. Economies of Scale.-—Much of the consolidation argument is based on a notion of economies of scale in production of goods and services. Implicit in this notion is an assumption that the consuming or demand unit of government must also be the producer of the service. Indeed, this is the general practice and experience of most U.S. communities. 2‘ .-.c 28 The experience of the Lakewood Plan communities in the Los Angeles area of California as described by Warren [51] and Bish [6] is seen by public choice analysis as a prime example that this need not be the case. In the Lakewood Plan area, consuming communities of varying sizes purchase many or even all of their public services by contract from nearby producing communities. This suggests that in order to capture such gains as are available under economies of scale in production, a community need not give up its local identity through annexation or consolidation. Although not based on economies of scale, there is an additional potential gain that public choice analysts see to the contracting approach. As generally practiced, the pro- duction and marketing of publicly provided goods and services is under local monopoly conditions. The potential for man- agement inefficiencies under this price giving situation is at least conceptually greater than under competitive con- ditions. The contract purchasing of services by nearby communities introduces some potential competition among pro- ducers and therefore can result in improved performance of producing agencies. Articulation of Preferences.--A major dimension of the public choice approach to the organization of local govern- ment which has been repeatedly included in the previous discussions is that government is a mechanism for the articu- lation of citizens' preferences; a product of citizens' 29 actions; a dependent as well as an independent variable. The omission of this proposition or any related notion is a major criticism of the consolidation reform tradition coming out of the public choice frame of analysis. It seems in fact a strange irony that the political science tradition of metropolitan reform should be based on "efficiency" in production arguments and apparently not un— til economists focused on the problem was there much of a notion of articulation of citizens' preferences explicitly introduced. In his "General Paradigm of Choice and Power," Samuels [45] sets forth several notions of relevance to understanding the conceptual issues involved in the "articulation of citi— zens preferences.“ Samuels first of all makes clear that society--presumably also communities—-Operates under con- ditions of scarcity and that actors in society are interde- pendent. Thus the conduct or choices of one group of individ- uals has an impact on other groups. In the process of choosing, individuals do not choose from among only those realistic or available alternatives-~the opportunity set. Samuels points out that traditional economic theory and welfare economics have concentrated on the choice of individ- uals from among the alternatives of their respective oppor- tunity sets where each alternative has its opportunity cost. The individual's welfare is attained by maximizing within the constraints of the opportunity set. In a process Of N a :4 30 public choice this traditional analysis is inadequate because it says nothing to the structure of the opportunity sets which in fact comprise the decision making process. In developing his notion of interactional choice, Samuels argues that the way in which one individuals choice affects the choices of others is by changing the structure or array of those other persons' Opportunity set. This impact of the behavior and choices of others on the struc- ture of one's opportunity set is coercion--a neutral term in Samuels usage. Thus society is a system of mutual coerc- ion in which the choices of each individual have an impact on the opportunity sets of others and therefore on the range of possible choices available to them. Finally, Samuels defines pager as "the means or capacity with which to exercise choice, with which therefore to coerce.“ The reciprocal of power is the exposure to others' coercive capacity or power. By "means or capacity" Samuels means "the de jure or de facto bases by, with, or on which one acts as a chooser." This could include a person's property, position in an organization, accumulated human capital, or skill at negotiation among others. "The articulation of preferences" is seen as a process of interactional choice where when one person or groups's preferences count, those of another will not count. It is the intent of the research reported on in this thesis to explicitly examine the effect of community size, 31 community homo or heterogeneity, and other associated char- acteristics on the articulation of citizens' preferences. In this regard, public school districts are seen as political systems in addition to being organizations for the production of educational experiences for the children of the district. Unlike multiservice governmental units, school districts purport to have a single functione-the organization and de- livery of educational experiences. This characteristic sug- gests that school districts may be particularly useful in examining the responsiveness of governmental units to pre- ferences and diversity of preferences for its services in the community. Thus it is that much of the remainder of this thesis will deal with the distribution of the performance of elementary school children on standardized achievement tests. ~ 3‘ .«i CHAPTER II THE BENEFITS OF EDUCATION - A PUBLIC CHOICE QUESTION Introduction As described in the preceding chapter, the historical record of movement toward consolidation of governmental units as well as current moves in that direction have been under- taken based substantially on arguments of economies in pro- duction of services. This view of the size of self-govern— ing communities and implicitly of the redrawing of community boundaries, completely ignores the issues related to the articulation of preferences and the effects that such bound- ary changes may have on whose preferences will count. In the area of education, the past movement by the lower courts toward redistricting of school districts to achieve racial balance was substantially a movement towards consoli- dation. Decisions to that end either ignore or accept the effects that such a move will have on the articulation of diverse needs or preferences for schooling experiences with- in the consolidated district. Since, in this case, the con— solidation is being undertaken on the basis of a racial composition formula, the clear implication is that parents of children of color will send their children to public schools in districts where, with respect to color, they can never be anything but a minority. 32 33 It is interesting to note that the redrawing of Congres- sional districts was undertaken explicitly because of a re- cognition that existing boundaries gave inordinate weight to the preferences of rural people in the body politic at the federal and state levels. Research Objectives The objectives of the research here reported is a study of the institutional boundary issue through an examination of the ways in which characteristics of school districts associated with present boundaries affect the distribution of benefits of schools as measured by standardized achievement test scores. It is believed that this examination will shed light on the institutional performance implications of changing boundaries. It may be suggestive of alternatives to current institutional settings and arrangements in which activities organized by governmental units can better meet the diverse preferences and expectations of citizens who consume the goods and services thus produced. While the "Pareto better" type of outcome implied in the above state- ment may not be attainable, it is believed that the results of this research may at least be suggestive of the directions which individuals and groups should go in seeking redress of grievances with regard to increasing their share of the benefits of publicly produced goods and services. 34 A further objective of the research is to add to the body of knowledge on social indicators, specifically with regard to their distributive dimensions. To this end the conceptual basis for employing measures of the dispersion of achievement test scores as a new social indicator will be developed. Further, a model will be advanced and empirical— ly tested which will attempt to explain in part the public choice implications of variations in the magnitude of the new indicator. Such an exercise, in addition to shedding light on the boundary question and the distribution of educational bene- fits, may suggest ways in which greater insight may be ob- tained from existing and future indicator data. Further, the notion that the investigation of a clearly institutional question such as that of boundaries may be amenable to reason- able quantitative analysis, may suggest ways of approaching other institutional questions which have so long eluded analysis. The Educational Problem Setting While the objectives of this research as stated deal with the boundary question, many of the measurements, models and research assumptions are explicitly about the schooling process. It is appropriate, therefore, to make explicit the context and perspective of the schooling process which has influenced this research. l. (I) 35 Do Schools Matter?--Ever since the early 19605 there has been a sustained political assault against economic inequality in the United States. Blacks, women, poor people, students and workers have brought the issue into the streets, forced it onto the front pages, and thrown it into the legis- lative bodies and the courts. A major response to this assault has been to focus on the schools as a major societal institutiontflmnmfln:capable of ameliorating the widespread inequality of Opportunity. Implicit in this focus was the recognition that the schools seemed to serve some groups in the society better than others. The basis of this attention to the schools rests on the belief that performance in schools is in some way re- lated to subsequent personal success. Indeed, investigations and research in the economics of human capital formation have been able to show that as the amount of schooling as measured in years increases, so does the subsequent stream of income [4]. Based on this belief and on the empirical evidence sup- porting it, massive governmental programs were undertaken to improve the ability of schools to correct the unequal distribution of economic opportunity. Where the resources of the schools themselves appear to be distributed on the Imasis of race, school integration through busing and re- rk on educational production functions has employed socio- e(monomic and other background variables with a View to con- tlilt‘olling for such influences while seeking to quantify the (zeitisality relationships between educational outcomes and in- EDIJtzs which are amenable to manipulation by administrative E><>Zlicy decisions. While some work has been done to examine t:}1€3 different responses to school inputs by children of 37“~‘— Eseee H. Averch, et a1. [2], for a comprehensive reivew of 1:}118 literature. 65 66 diverse social classes [49], work which has undertaken to employ such variables in studying the political economy of educational outcomes has not been found. As stated earlier, it is the purpose of this research to explicitly examine the relationship between characteris— tics of communities affected by boundaries and the distri- bution of benefits of education as measured by achievement tests. The Level of Benefit Versus The Distribution of Benefits-- The Brown-Saks Model In the rapidly growing literature which attempts to apply economic concepts to the analysis of education, the school or school district is thought to be a firm whose out- put is measured by the mean score of its pupils on some standardized achievement test. Most of these studies have employed regression techniques to estimate the productivity of various school inputs. The conclusion of reviews of this literature is that there is no consistently identified school input which appears to matter. As stated earlier, a growing number of educators and researchers have agreed in fact that school inputs do not matter. Brown and Saks [16] have re-examined the microeconomics of the problem by viewing the school or school district as a Inultiproduct firm where each pupil's performance is of in- ‘berest to the school. Within this perspective, the central questions relate to the allocation of resources between 67 individual pupils. Further, Brown and Saks argue and demon- strate that in such a model, the mean of pupil performance can be unaffected by variation in inputs even when those inputs are productive. As set forth in the Brown-Saks explication, the school's decision regarding the distribution of outputs will depend on the following variables: a) the marginal products of the available inputs when applied to particular pupils; b) the marginal utility of particular pupil's perfor- mance as evaluated by the school authorities or the community; and c) the amount of any input exhausted when that input is applied to a particular pupil under the existing production conditions. in dealing with the community preferences for trade offs between the level of achievement (mean) and the dispersion of achievement (variance), Brown and Saks have borrowed from notions set forth in portfolio theory. This borrowing is possible if individual pupil scores enter the social welfare function in quadratic form which can be re-expressed in terms of the mean and standard deviations. Figure 5 sets forth the possible general welfare indif- ference in curves in mean and standard deviation space. The upward sloping curves, W and W', represent the L ;preferences of a school district which in selecting between Exalicies A and B which result in the same mean performance knxt different dispersions, prefers the policy with the smal- len: dispersion, polity A. (Wt being a higher indifference curve th as "leve averte rs Fig? achieveme 68 curve that WL.) Brown and Saks dub these types of districts as "levelers" and point out their similarity with rist- averters in the parlance of portfolio theory. Figure 5. Types of welfare indifference curves for achievement mean and standard deviation. Achievement Mean Achievement Standard Deviation Similarly, the downward sloping curves We and Wé re- present the preferences of districts which when choosing between policy A and B prefer policy B and its associated greater dispersion. (Wé being a higher indifference curve that We.) Brown and Saks designate this type of district as "elitists." These authors point out that indifference curves could jbe straight horizontal lines implying that a school district {ways no attention to the dispersion of its students' per- formance . Producti' IOWS‘ con: pilsl “’3“ Ring of 1 be a diff to the t!" tie Pupil puts avai The vexent 0f seats the venent 19' puts were all of W 69 Having thus set forth the possible utility functions, the authors develop the arguments necessary to depict the production choices in mean-standard deviation space as fol- lows: Consider a school or school district with just two pu- 1 who differ in their achievement levels at the begin- pils ning of the school year. Consider further that there will be a different marginal productivity of inputs when applied to the two pupils and that the production relationships for the pupils are linear. The district is limited in total in- puts available. Figure 6 illustrates the case. The ordinate and abissa of the figure measure the achie- vement of the respective pupils. The triangle ABC repre- sents the production possibilities given the initial achie- vement levels of the two pupils. If all of the limited in— puts were applied to pupil A, point B would be obtained. If all of the resources were applied to pupil B, then point C would obtain. The line BC represents the production possi- bilities if none of the school's resources are wasted. It is from this set that the school must choose. As drawn the absolute value of the slope of the line BC is greater than one implying that the marginal physical product of the inputs are greater when applied to A than when applied to B. ‘1 The analysis is unchanged if, rather than two pupils, two homogeneous groups of pupils are used. 70 Figure 6. Production possibilities set for two pupils. Achievement Score Pupil A c I I I I max “A Score =Score A B .0 O m o. 0'. 0.4 ..0 initial S L. . —I A A = .0. :C .0. I Q. : .0 E Q... : ... L .0 = : .0 .0 : : ... . 11 Pupil B . . . I . lnltld1 max 58 m Achievement Score ’8 The next step is to map the production possibilities set into mean-standard deviation space. In order to accom- plish this, two additional lines are drawn on the figure. The 45° line from the origin represents all points where the scores of the two pupils are equal. This also represents the locus of points where the standard deviation of their scores are zero . The line mm' drawn perpendicular to the 450 line and ithersecting point D on the production possibilities curve, represents a locus of points with a constant mean. Lines 71 parallel to mm' further from the origin would have higher means and those closer would have lower means. Thus line mm' represents a whole family of isomean curves. As we move away from point D toward C along the pro- duction possibilities curve, the standard deviation increas- es and the mean decreases. Similarly, movement from D to- ward B results in a rise in the mean and an increase in the standard deviation of achievement. It is thus now possible to transfer the production possibilities set into mean- standard deviation space. Figure 7a illustrates the choice situation in mean- standard deviation space for the conditions set forth in Figure 6. Figure 7. Production choices under elitist and leveler preferences with two alternative educational technologies. ml 2 305 SD 7a 7b 72 When the indifference curves are added to the graph only the D-B portion of the production set is relevant. If the school board of a district were elitist then their highest level of welfare would be obtained by applying all their in- puts to pupil B. If the district was made up of levelers, then resources would be employed to produce both a lower mean and lower dispersion of outcomes as represented by point f. In the case where the marginal productivity of the in- put is greater for the pupil with the lower initial start- ing point (Pupil B), the production set BC will be flipped about point D as in Figure 7b. In this case only the D-C segment of the production set is relevant except for the case where the district is sufficiently elitist to still choose to devote all resources to the pupil starting off with the highest score (Pupil A). Brown and Saks point out two extensions of the model worth noting. First, if the number of inputs are increased and they possess differing comparative advantages, the pos- sibility that a school will specialize in better students is reduced regardless of how elitist the district is. The second extension of the model is to increase the number of pupils considered. According to the authors this creates a situation where there is no longer a one-to-one Inapping between student achievement space and mean-standard deviation space. Under this circumstance the production set becox ity f of th had a Partic 12g 3 the an Out of tiOn E} SE 3m Ea: Mtg a that C“ Of Licl rC~ ‘ ‘iarflle 73 becomes an area in mean-standard deviation space. The util- ity function does guarantee that only points on the boundary of the area need be considered. A major contribution of the Brown-Saks explication be- yond enlightening the debate over the productivity of inputs, is the clear demonstration that measures of the dispersion of pupil achievement are of considerable consequence. The authors make clear that even if a school district had a "leveling" set of goals, it could, if faced with a particular technology and student body, still end up produc- ing a wider dispersion of outputs if that were the effect of the available technology. Thus they argue, the separating out of the dispersion effect in production from the consump- tion effects is extremely difficult if at all possible. Several comments on this latter point are appropriate. The earlier discussion of the differentiating educational norms as set forth by Brookover, §E_El-r [47] is evidence that current educational technology embodied in the training of teachers, classroom materials and acceptable teaching practice may well be biased towards pupil differentiation regardless of the desires of communities. While this model views the consumption and production biases as distinct and separate phenomenon, it is, I believe, logical to argue that the choice of a particular type or com- bination of inputs is itself a function of the preferences of the community. The introduction of black studies and 74 other special interest programs are evidence in support of this argument. The Distribution of Achievement--A Measure of School District Responsiveness An extension of the Brown—Saks model by including the evidence associating social-economic status and achievement, leads to the use of measures of the dispersion of achieve- ment as indicators of the responsiveness of school districts to groups other than the middle class majority. Consider two school districts, each with two pupils. One pupil is advantaged and the other is disadvantaged. The definition of "advantage" is that used earlier in the second general assumption of this research. Figure 8 illustrates the model. Figure 8. The effects of alternative educational strategies on pupil outcomes. Achievement Achievement ' I ' I _ A A ‘fij. - A 5" ' Ail- A} D | PD I l I to inputs/time t1 t0 inputs/time t1 District A District 8 75 The difference between the two districts is that Dis- trict A employs an educational strategy which serves only the advantaged pupil, while District B employs a strategy which serves only the disadvantaged pupil. As illustrated, if the growth in achievement of the two affected pupils in the respective districts are equal, then, given the starting conditions, the means of the districts at time tl will be identical. The distribution of achievement will, however, be quite different. It will be recognized that this model when presented in the context of the production possibilities portion of the Brown-Saks Model represents a case where the marginal pro— ductivities of the inputs are equal in the two pupils but where, because of the achievement differences at the initial conditions, the entire production possibility set occurs in the achievement space above the zero variance line (450 line). Figure 9 presents the same analysis as Figure 8 in the man- ner of Brown and Saks. This is obviously a special case both in terms of the starting conditions and in terms of the marginal productivi- ty of the inputs. It does, however, begin to develop some notion of the ends which parents of disadvantaged children would seek and the directions that school districts would go in responding to the wishes of these parents. For both, as argued thus far, this would represent a reduction in the dispersion of achievement. 76 Figure 9. The effects of alternative educational strategies on pupil outcomes. D m. ‘—————————district A @ t ’D] / / II / . 77 -—¥L-———district B @ t1 5 N\ ‘f a)» \ / \ m \’ ./ §I E 2’ “Ix < / A\ \ 'o\. L, \ *7) t“) Achievement The goals of the parents of disadvantaged children or of the school system responding to those parents can be fur- ther defined by making two additional assumptions and by re- working the Brown-Saks model in the light of these assump- tions. The first additional assumption is the elimination of the "elitist" type preference function. While communities may vary considerably in the degree to which they are adverse to inequality of educational outcomes, i.e., leveling, the extreme solutions implied by the "elitist" indifference cur- ves do not appear reasonable within a democratic society. Even under the most elitist conditions of racially segrega- ted schools in recent history, there did not appear to exist the extremes of resource allocations implied by the "elitist" solutions of the Brown-Saks model. 77 The second additional assumption is that the marginal productivity of inputs applied to advantaged children will be greater than or equal to their productivity with dis- advantaged children. Thus, the model is restricted to cases where in student achievement space the negative slope of the production possibilities set is greater than or equal to that of the isomean lines. This assumption is intended to reflect the view that there is not presently available educational inputs which serve the children of disadvantaged parents better than those of the advantaged. Some may view this as a heroic assumption particularly when viewed from the common exper- ience of clearly observed differences among children, with regard to proclivities for different skills and activities. However, when it is considered that this assumption is with regard to the inputs employed in the instruction of a parti- cular set of skills, i.e., cognitive academic, and is here considered at elementary school levels, it is not particu- larly unreasonable given the technology available at that level. Brookover's contention of a differentiating educa- tional norm is supportive of this assumption. The evidence that the impact of Sesame Street, an effort at developing an instructional technology specifically designed to meet the special needs of disadvantaged children, has had an even greater impact on advantaged children, further supports the assumption [3] [10]. 78 One of the implications of these additional assumptions to the model is that they restrict the area of choice to the portion of the production area which falls above the zero variance line (the 45 degree line out of the origin of the achievement axes) in student achievement space. Several other implications of the additional assumptions will be developed as strategies available to parents of children who are disadvantaged within the schools are examined and ex- plained within the model. Within the context of this model parents of disadvant- aged children have several alternatives available in seeking to Obtain improved levels of achievement for their children. The alternatives are discussed here under circumstances where either community preferences are fixed or where the educa- tional technology is fixed. The purpose of using these two circumstances is illustrative and is not intended to repre- sent this writer's view of reality. The latter is that real- ity likely lies somewhere between the two extremes. If faced with a fixed educational technology which gen- erally serves their children less well than other children, parents of children who are disadvantaged within the schools have two types of strategies open to them. They can seek to change the preferences of the community such that more of the elitist inputs are allocated to their children; they can seek an increase in achievement outcomes of all pupils; or they can support both strategies. Figure 10 illustrates these alternatives. 79 Figure 10. Strategies available to parents of disad- vantaged pupils where educational technology is fixed and is marginally more productive on advantaged pupils. s 4‘ l A I ml 4-) C 0 E > G) IE U < SD 5 s A hievement D Mean-Standard Deviation c Space Pupil Achievement Space 10a 10b As Figure 10b shows, a movement from starting position A to either point B, point C or point D represents an in- crease in achievement for the disadvantaged pupil or group of pupils. Figure 10a demonstrates that movement from point A to point B would be accomplished by a change in community preferences (WA to WB). Such a case might be where disad- vantaged parents organize to get a person sympathetic to their needs or preferences elected to the school board. Fig- ure 10 also illustrates that an overall increase in output (TA to TA), i.e., an expansion of the production possibili- ties, will result in improved achievement of the disadvant- aged pupil as represented by movement from point A to point C3. Movement from point A to point D as the figure makes 80 clear, is accomplished by both a change in preferences and an expansion of the production possibilities. If faced with a situation where community preferences are fixed, and the educational technology employed serves their children less well than the advantaged children in the schools, parents of disadvantaged pupils again have several strategies available to them within this model. They can seek to change the instructional technology employed to one that is more productive with their children. Such a change might be represented by the persuading of school Officials to employ a workbook for reading which is more representative of the experiences of disadvantaged children than that pre- viously used, or by advancing the class through reading ma- terials at a pace more conducive to disadvantaged pupils. A second strategy available to the parents of disadvant- aged pupils under these circumstances is to support efforts which result in an increase in achievement of all pupils, i.e., an expansion of the production possibilities. A third alternative is a combination of both of the first and second strategies. Figure 11 illustrates these cases. The movement in Figure 11a from point E to point F re- presents the strategy of obtaining a change in the instruc- tional technology to one which is relatively more productive With the disadvantaged pupil, (TA to TB) given the fixed .Preferences as represented by W -'W'. As Figure llb makes Cilear the shift from E to F represents a gain in achievement fr>r the disadvantaged pupil. 81 Figure 11. Strategies available to parents of disadvan- taged pupils where community preferences for pupil achieve- ment mean and standard deviation are fixed. D Mean-Standard Deviation Pupil Achievement Space Space 11a 11b Similarly, the movement from point E to point G in Figure 11a and 11b represents the strategy of accomplishing an increase in the achievement of all pupils by an outward shift of production possibilities from TA to TA (W' indicates a higher preference indifference curve). The movement from point E to point H in Figure 11 represents the results of both modifying the instructional technology employed and of expanding the production possibilities. What is apparent in both of these illustrations is that a reduction in the dispersion Of pupil achievement scores represents a schooling situation more responsive to the dis- advantaged pupil. It is likely that parents of disadvantaged 82 children who are dissatisfied with the treatment of their children in school will seek both greater community acknow- ledgement of their child's right to more equal achievement and the employment of inputs in the schools which are more appropriate to their children's special needs. The reality of the situation is likely that a parent or group of parents seeking a more responsive school system will raise their con- cerns in a variety of ways in a variety of situations. Such activities may raise the consciousness of the community to the problem, may simply cause school officials or teachers to change their instructional approaches, or both may occur. The aggrieved parents need not know the specific inputs which are more appropriate, they may need only make clear that they want something done. Within this argument it is not necessary to distinguish between the influences of a shift in community preferences and those of changes in technology since the effect of both influences are measured by changes in the dispersion of achievement. Therefore, it is the argument of this section that ceterus parabus, the lower the dispersion of achievement, the more responsive a school district to the educationally disadvantaged children and their parents in the community. The balance Of this thesis will seek to examine the communi- ty circumstances under which this will occur. It is apprOpriate to point out that of the six strateg— ies discussed in the preceeding paragraphs as available to 83 parents of disadvantaged pupils in seeking to obtain improved achievement for their children, only four can be considered as "Praeto better" states within this model. Thus, it seems clear that there is the potential for substantial conflict between those whole interests are improved achievement for disadvantaged pupils and those whose interests are on behalf of those already advantaged within the schools. T' tual v of res econonj that 1 Set fo, lieved CHAPTER IV WHOSE PREFERENCES COUNT? - A PUBLIC CHOICE MODEL The previous chapters have described the major intellec- tual views of the boundary question and have argued in a set of research objectives that an examination of the political economy of schooling at the district level may shed light on that issue. A perspective to the schooling process has been set forth as a basis for a set of assumptions which it is be- lieved will make the original objective empirically testable. The conceptual basis for an indicator of school district re- sponsiveness to the socio-economically disadvantaged hastxxni developed and will be employed in the empirical work. Much of the empirical investigation of this research is exploratory and does not fit a strict hypothesis testing framework. There are, however, several hypotheses which sug- gest themselves and in order to test them some kind of model is required. It is the objective of this chapter to make explicit the hypotheses to be tested, the models to be em- ployed and the additional areas of investigation. The DevelOpment of Hypotheses The public choice approach set forth in Chapter I argues that the individual in seeking public provision of a good or service would need to consider both political externality 84 85 costs and decision making costs. SO too, a group seeking to obtain a larger share of the benefits of publicly provided services would consider both of these types of costs. Within this framework of analysis the notion of "dis- advantaged" as used in this thesis and research is to have a greater than average political externality cost imposed by the structure and processes of society. In the case of the group disadvantaged within a particular school district, this would be a group per capita share of benefits less than the mean for the school district. While, as Chapter II argues, this is generally true for those disadvantaged under the educational norm, it is likely that there will be some com- munity circumstances under which they will fare relatively better and other circumstances under which they will fare relatively worse. The objectives of this research are to examine those circumstances insofar as they relate to or are affected by changes in boundaries. The hypotheses here developed are predictions about the effect of those circumstances. Under the public choice approach two major factors would be identified as influencing the ability of a group seeking to increase its share Of the benefits of publicly produced goods and services. These would be the decision making costs they face and the power the group can muster in the body politic within which they act. The notion of de— cision making costs is here seen as including the costs of 86 "voice" set forth by Hirschman [30]. The idea of group pow- er may include the power deriving from coalitions, but the costs of the formation of coalitions would be included in decision making costs. The notion set forth by Buchanan and Tullock that de- cision making costs will rise increasingly as the percentage of a group required to decide increases is suggestive of several ideas about the community circumstances which are more or less favorable to the group seeking a change in the delivery of public services. It suggests that, as the pro- portion a group is of the total community remains constant and the size of the community increases, the numerical size of the group will increase and will thus face higher decis- ion making costs. This could mean that other things constant, smaller groups may be proportionately more effective in ob- taining a response from the body politic than larger groups. The Buchanan and Tullock notion is suggestive but some- what less than complete on the issue Of size. Bish [6], in discussing the size of school districts identifies an in- crease in bureaucratization of school administration and an increasingly dominant role of professional educators as size increases. Such circumstances could make "voice" more cost- ly. Similar arguments could be made with regard to the or- ganization of other publicly produced goods or services be- sides schools. Spatial dimensions of size, both with regard to distances 87 between group members and distances between individuals and the locus to which they must exercise their voice, would also suggest increased costs as size increases. Samuels [45] in describing his general paradigm of choice and power defines power as "the means or capacity with which to exericse choice." By means or capacity is meant the de jure or de facto basis by, with, or on which one acts as a chooser. Thus one's property, position in an organization, accumulation of human capital, occupation, skill at negatation, etc. would have an influence on the individuals resource of power. As pointed out in Chapter I, the voting models and constitutional rule paradigms assume implicitly that power in a community is distributed on the basis of "one person, one vote"--a highly unlikely circum- stance. The notions set forth as to the ways in which in- fluence may be brought to bear on a school system includes more than voting or electoral activities. While the esti- mation of an individual's or a group's power is extremely elusive, the notion of power as set forth by Samuels is nevertheless useful. With regard to the response of the body politic to the preferences of individuals or groups, the power idea argues that, other things constant, as power increases the expected response would be greater. Since the notion of power is a relative one, efforts to express power would be in prOpor- tionate terms to the relative whole, e.g., a group's voting 88 power in a community would be approximated by the prOportion the group is of the total electorate. The combining of decision making cost notions with the power concept is suggestive of several hypotheses about the size of communities and thus the boundary question which are worth testing. Hypothesis l.--As the size of a school district increa- ses, ceteris paribus, the share of the benefits a particular societal minority within the community Obtains will decrease. Restated in terms of the indicator of school district responsiveness to the disadvantaged, this hypothesis would be as follows: As the size of the school district increases while other things are held constant, the school district responsiveness to disadvantaged pupils would decline as in- dicated by measures of the dispersion of achievement. Hypothesis 2.--As the power of a societal minority with- in the community increases, other things constant, their per group member share of the benefits of education will increase. Restating this hypothesis in terms of the pupils dis- advantaged under the educational norm and the indicator of school district responsiveness to that group would give the following: As the power of the disadvantaged group in the community increases, the schools will be more responsive to that group as indicated by the measure of the dispersion of achievement. 89 Hypothesis 3.--The more diverse the preferences and/or needs for educational experiences, i.e., the greater the degree of school district heterogeneity, ceterus paribus, the less responsive the schools will be to societal minority groups within the community. Beyond Hypothesis Testing In addition to explicitly testing the hypotheses set forth, the following additional question will be included in the investigation: The Sponsors vs. the Consumers of Public Education. In Michigan and in most other states in the U.S. local school boards are empowered to organize public education in the community by general election and referendum approval. The sponsors of public education are thus the electorate of the school district community. The consumers of public financed education can be viewed as parents of public school children--a subset of the electorate. It would be interesting to know to what extent the con- sumers of public school education differ from the sponsors of education and if school systems are more responsive to one or to the other. Further, it would be of some interest to examine the general direction of that influence in con- texi.of the questions being raised in this thesis--i.e., does the influence of the nonconsuming sponsors cause the schools to be more or less responsive to the disadvantaged. 90 While no particularly definitive answers to these ques- tions are expected from this research, these notions are relavant in examining the responses of communities to citi- zens' preferences and will be included in the empirical in- vestigations. The Model A general model is prOposed to examine the political economy of the response of school districts to the dis- advantaged and to test the hypotheses set forth. In the sense that the model here described sets forth an argument for the inclusion of certain variables in the investigative work it is a "model" in the usual sense of the word. A further comment is appropriate before presenting and discussing the model. The individual and aggregate achie- vement outcomes of the schooling process which are used in this research as a measure of the benefits of schools are the result of technical production relationships, i.e., pro- duction functions. The Brown-Saks paper discussed in the previous chapter is a substantial contribution to a better understanding of how to approach investigations of that pro- cess. The model described here and the empirical research employing the model are not viewed as describing the pro- duction of achievement--they are Egg seen as descriptive of the educational production function. In large part the actual technical production function (functions--one for each 91 pupil) is viewed in this research as a black box, the work- ings of which are extremely uncertain. The black box is under the management of a heirarchy of teachers, principals, district administrators and school board members, and is influenced by both the sponsors and the consumers of public education. The social-economic-political behavior of the community is viewed as a constraining influence on the be- havior of those who manage the black box. Neither the be- havior of school district staffs or that of the sponsors or clients of the schools have been observed directly in this research. Some general assumptions about behavior of the respective groups have been made and some predicitions (hypo- theses) about the effects of their behavior under certain circumstances have been set forth. An analogy may be helpful. There is within this writer's experience at least one dairy farmer in the U.S.--there may be others--who does not make silage out of "grass" sub- stantially because his wife cannot stand the smell of grass silage in her house during much of the winter months.l/ A preliminary investigation of the influence of attitudes to- wards the smell of grass silage held by farmers' wives on the management choice between grass and corn silage would not require full specification of the respective production 1/ -— Grass silage, usually made from a mixture of a leguminous crop such as alfalfa and grass has a particularly pungent odor which is picked up by any material such as shoes or clothing which comes in contact with it. 92 functions (corn and grass) in order to be useful, however more desirable the complete model may be. The model here employed is as follows: R = f(N, C, P, S) where: R = responsiveness of school district to the dis- advantaged under the educational norm. N = the degree of implementation of the norm in the management of schooling experiences. C = the degree of community consensus, i.e., the heterogeneity of the school district community. P = the power of the disadvantaged group in the school district community. 8 = size of the school district community. Each element in the model has been discussed in some depth elsewhere in the thesis. The model is primarily a be- havioral one and, as was pointed out above, no direct obser- vation of behavior has been made in this research. Thus the empirical application of the model must be based on proxy variables which infer behavior rather than measure it direct- ly. The choice of proxy variables is constrained by the availability of data and its form. In specifying this model it is assumed that the inde- pendent variables employed are determined substantially out- side Of the system which is being described. That is, the independent variables are predetermined. The degree of implementation of the educational norm is substantially a function of the educational strategies chosen 93 by school system administrators and of the behavior of staff, particularly teachers. In fact, some of the specific strat- egies such as the employment of nonteaching instructional staff are often chosen ostensibly to serve the disadvantaged but, as is argued later, may actually Operate in the opposite direction. The characteristics of the school district community are determined by a large number of influences probably even the performance of the school system itself. However, the major determinants of family locational decisions are more likely to be such as employment opportunities or family cir- cumstances. This would be particularly true for the dis- advantaged under the educational norm. The size of school district appears rather clearly to be determined outside of the behavioral system being here described. There is a real dilemma in specifying a static model of a highly dynamic process where, in a general equilibrium sense, everything is in part a function of everything else. In that sense the model here specified can be viewed as a reduced form of a larger system of relationships, the struc- tures of which are generally unknown. Data Sources Data on pupil achievement is from the 1969-70 and 1970- 71 records of the Michigan Educational Assessment Program obtai catio for t? peri0< C 0mm 'u‘l Which Co \ .0 indi 94 obtained through the State of Michigan, Department of Edu- cation. Records of individual pupil achievement scores were for the 1969-70 school year and district summaries from both periods were used. Data on population characteristics for school district communities is from the 1970 Census Fourth Count(POpulationL which was made to coincide with school district boundaries by the National Center for Educational Statistics. The data covers 500 of the more than 600 school dis- tricts in the state of Michigan for which there was both assessment information, census information, and which had a full kindergarten through twelfth grade (K-12) program in the l969~1970 school year. Because several isolated data items were missing for some variables, fewer than the full 500 districts were used for most aspects of the empirical work. Because the Detroit Metropolitan School District was in the order of 9-10 times larger than the next largest dis- trict in the data base, it was eliminated in order to avoid the tendency of that observation to dominate all investi- gations related to size. Measuring the Output of Schools -- The Benefits of Education Cognitive and Affective Skills.--Educational outcomes to individuals are, in the literature of education, generally divided into two categories--"cognitive" and "noncognitive" 95 (affective) skills.l/ Noncognitive factors are said to in- clude maturation, attitudes, learning styles, social skills, self awareness, and other such vague concepts as happiness and quality of life. Cognitive skills are generally defined as those skills measured by standardized achievement tests and include mathematics, reading, writing, and other langu- age skills. That there is not clear agreement on what is cognitive and what is noncognitive is illustrated by the kinds of items included in the "cognitive style mapping" described earlier. Further, there is considerable disagreement by educators as to whether cognitive or affective skills are the more impor- tant effect of schooling. One view-holds that noncognitive skills are important because they are a major determinant of cognitive achievement. Another view holds that docility and sobriety may be more important to subsequent success than reading and writing and such "skills" may be the more important product of schools. As mentioned in an earlier chapter, the literature on human capital which indicates a relationship between the amount of education and the amount of income earned has not shed any light on this debate over the content of education. 1/ — Averch, et a1. [2], devote a section of their review to to a discussion of this distinction between educational outcomes. 96 What remains is a general acknowledgement that non- cognitive or affective skills are important but little re- search is devoted to their measurement or to discovering their determinants. Most educational effectiveness research is, according to Averch,e£_al., directed almost entirely towards explaining cognitive skills, which are in fact de- fined by the tests that measure them. Standardized Achievement Tests.--The first constraint on the use of standardized achievement as a measure of the benefits of education is the point made in the preceding section--those effects of schooling measured by such tests are by no means the only outcomes or benefits of the school- ing process. Michaelson [49] in a study of the association of teacher characteristics with the achievement test per- formance of children of different characteristics, goes to some pains to make this precise point. He points out that in a production analysis of multiple products, variations in inputs do not correlate with either output when the other output is not accounted for. On the other hand, the argument regarding the access to school inputs set forth in Chapter II is basically that af- fective skills are a substantial determinant of which pupils will be served in the acquisition of the cognitive skills measured by achievement tests. Such measures would thus re- flect both the cognitive skills attained and the influence of affective or noncognitive skills on their acquisition. 97 Several other issues with regard to the use of achieve- ment test scores are important. Of particular concern is the question of cultural bias embodied in such tests. A nationally normed achievement test is one that is based on a sample representing the characteristics of the national population. To be accurate the sample population on which the test is developed must be stratified by race, income, and other such characteristics in the same proportion as those characteristics appear in the general population. As pointed out by Averch, ep_al., "This means that any nation- ally normed test primarily reflects the characteristics of white, middle-class America, simply becuase there are so many of them" [2]. When such tests are used on groups which do not reflect the normative population on which the test is based, the best that one can say is that the test measures how well the particular group has achieved U.S. average goals which to some means white, middle-class goals. Another issue in the use of standardized achievement tests is the degree to which the test is consonant with the objectives of the instructional program employed. That is, does the test actually measure what has been taught. One criticism of standardized tests in this regard is that they tend to emphasize some outcomes at the expense of others, such as abstract reasoning or creativity, which are not measured by the tests. Further, standardized tests may co- ver all of the objectives equally, or may include objectives 98 not included in the instructional program at all. Thus the summary test score may misrepresent the actual performance of the instructional program in meeting its objectives. In summarizing their view of the use of standardized achievement tests, Averch, et al., wrote as follows: Using standardized tests to evaluate student achievement has become a major enterprise in the schools; but in spite of the wide use and reliance on these tests, they are generally inadequate. This is alarming in light of the growing activity in evaluation of educational outcomes based on standardized test scores. Standardized tests, even when properly used and interpreted, evaluate only a limited number of educational objectives. At best, generally used tests measure only limi- ted aspects of cognitive performance, while higher cognitive abilities and achievements go untested. Noncognitive achievement is sometimes talked about, but the evaluation of these factors is still in a very crude state. Inasmuch as schools and inno- vative education programs terms of such limitations, are being evaluated in there is a crucial need for immediate improvements in test design, concept, scoring interpretation, and administration [2]. To Use or Not To Use -- No Question.--While acknowledg- ing the severe limitations of standardized achievement tests to measure in any complete way the benefits of schooling, the licly provided services appear tations. Further, in terms of search, the examination of the some multipurpose governmental the output of schools and thus output measures of other pub- to have even more serious limi- the objectives of this re- boundary question employing unit would require the use of a large number of such output measures, most, if not all, of which would be considerably less reliable than achievement test scores. 99 Thus, the decision to use the results of the State of Michigan, Department of Education, state-wide assessment program as a limited measure of the benefits of education was in fact no decision at all. The phrase "benefits of ed- ucation as measured by standardized achievement tests" is widely used in this writing to acknowledge the limitations discussed above. The Michigan Assessment Program.--In the 1969-70 school year the State of Michigan Department of Education initiated a state-wide program of testing of all fourth and seventh graders. Tests were developed by the State Department of Education and the Educational Testing Service, Princeton, New Jersey, in the area of 1) verbal analogies, 2) reading, 3) mechanics of written English, and 4) mathematics. The tests were normed on the Michigan population and were based on educational objectives of the State Board of Education. The tests were constructed to be made as consistent as pos- sible with the various curricula employed in the schools of the state. In reporting on the results of assessment tests the Department of Education reports scores for each of the afore- mentioned tests plus a composite score composed of the scores 1/ in reading, mechanics of English and mathematics.— y This information is by way of personal conversation with Mr. Robert Huyser, Director of Assessment Programs, State of Michigan Department of Education, Lansing, Michigan. 100 In determining which of the available measures to use as a measure of the benefits of schools, a natural candidate is the composite achievement score. This measure has also been used by other researchers as a measure of the output of schools [15] [16]. However, a New York Times article reporting on a con- ference held at Harvard University suggests that other mea- sures may be more appropriate [42]. The conference held in November 1973 was a discussion of the preliminary results of an international study of education conducted by the Inter- national Association for the Evaluation of Educational Achie- vement (IEA). According to the newspaper account and from personal discussion with a conference participant, early results indicate that achievement in science and foreign languages are less influenced by the home environment and more by the school than is achievement in reading, literature and civics. While no measure of science achievement is avail- able, the logic of the argument would seem to extend to math- ematics. Therefore, that score was seen as perhaps a better measure of learning which actually was accomplished as a re- sult of schooling. Throughout the empirical portion of the research, both mathematics and composite score results were used for fourth graders. The choice between grade 4 and grade 7 was rather arbi- trary. There was some notion that if, as has been argued, schools tend to widen class differences, the identification 101 of such a trend in the earlier of the two grades is strong evidence of that influence throughout the schooling process. DevelOpment and Discussion of Proxy Variables Responsiveness to Disadvantaged Under Educational Norm.-- Chapter 3 has set forth the theOretical argument fOr the use of an indicator which combines the mean level of achievement and the dispersion of achievement scores as an indicator of school district responsiveness to the disadvantaged under the prevailing differentiating educational norm. Several conventionally used measures of inequality combine the mean with measures of dispersion. Among the more commonly used measures are the following: 1. The coefficient of variation 2. The relative mean deviation 3. The Gini coefficient 4. The standard deviation of logarithms Much of the work on measuring inequality, usually with respect to income, where these measures are commonly used, has concentrated on the relative merits of one index as over against another. Frequently the debate relates to issues of ease of computation or the particular ordering of disbri- butions obtained by one measure or another. In a recent article, Atkinson [1] points out that the conventional approach in most empirical work is to adopt some summary statistic or index of inequality with no 102 particular reason being given for choosing one measure over another. In his article he makes clear that underlying any summary measure of inequality there is implicit an assump- tion about the nature of a social welfare function. In developing this notion Atkinson draws on the work being done in the area of decision theory and demonstrates from that literature that the notion of risk aversion in its various forms (relative and absolute) has analogous appli- cation to measures of inequality in terms of "inequality- aversion." In his examination of several of the popular measures of inequality in terms of the implicit assumption about in- equality aversion, Atkinson demonstrates that the use of the variance implies increasing inequality aversion as income rise. Other measures which employ the mean of the distri- bution in their estimation, e.g., Gini coefficient; standard deviation of log; coefficient of variation; and relative mean deviation; imply a welfare function with a constant in- equality aversion. Atkinson then compares several Of the measures having constant inequality aversion with indeces of his own where he is able to specify the degree of constant inequality aver- sion employed in the index. The long-known characteristics of the Gini coefficient, the standard deviation of logs and the coefficient of variation are thus seen as implicit as- sumptions about the degree of inequality aversion of the 103 associated social welfare function. The coefficient of variation which attaches equal weight to transfers at different income levels and the Gini coefficient which attaches more weight to transfers affecting middle income classes are seen to give rankings of income distributions similar to those given by Atkinson's indices which had a relatively low degree of inequality aversion. The standard deviation of logs ranked income distributions consistent with a much higher degree of inequality aversion. Atkinson pointed out that if the degree of inequality aver- sion were equal to zero that distributions would be ranked on the basis of total income alone. Atkinson's paper makes clear that in the absence of empirical knowledge of the nature of a society's social welfare function with respect to inequality aversion, the choice of an index or measure of inequality is a highly norma- tive issue. To the extent that the researcher has either em— pirical evidence or indications from the body politic as re- gard inequality aversion, he can take those normative ques- tions as given. It may be, for example, that the legislation and court decisions regarding equal educational opportunity can be interpreted as an indication of a relatively high de- gree of inequality aversion with regard to education and that the measure most consistent with that position would be the standard deviation of the log of education outcomes. 104 The use of inequality measures employing the mean and standard deviation in this research have been given a some- what different meaning than that discussed by Atkinson, i.e., the degree of school district responsiveness. While the characterization of the degree of inequality of educational outcomes consistant with societal values may be best accom- plished by a measure such as the standard deviation of the logs of achievement scores, there is, at the present state of knowledge of the determinants of educational outcomes, no basis for placing a higher value on movement in one area of the range of the data than in another. Thus the coefficient of variation which as stated earlier weights shifts at all levels of a distribution equally is the measure employed as the empirical index of school district responsiveness. A further rational for the use of the coefficient of variation as an indicator of school district responsiveness to disadvantaged pupils is precisely that it is an "indicator". Thus the questions of ease of computation compared with other measures are indeed relevant. The calculation of the stan- dard deviation of the logs of achievement may be normatively more appropriate in American society at this time but the computation of this measure requires access to each pupil's achievement score. Alternative Measures of School District Responsiveness to Disadvantaged Pupils.--In order to provide comparative examination of the coefficient of variation as an indicator 105 of school district responsiveness to disadvantaged pupils, several additional measures of the way in which disadvantag- ed pupils are being served by the schools were employed. Of a variety of measures considered the two selected have the advantage of being rather direct measures of the absolute and relative achievement performance of disadvantaged pup- ils. The following discussion defines and describes the de- velopment of these alternative measures. Individual pupil files made available by the State of Michigan Department of Education contain, in addition to achievement scores, a number of classifying characteristics of the individual pupil. Of particular interest to this re- search was the measure of the socio-economic status of the pupil. The SES (socio—economic status) score was structured such that its statewide distribution approximated a normal distribution with a mean of 50 and a standard deviation of 10. In defining a representative group of disadvantaged pupils within the gpgpg, the ten percent of the pupils low- est on the state SES scale were selected. Given the charac- teristics of the distribution of the SES score, this repre- sented pupils with scores of 37 or less. In addition, of the approximately 500 districts with complete data, this definition of disadvantaged pupils provided data from 461 districts. The direct measures of the performance of dis-’ advantaged pupils were based on this definition and the 106 analysis employed those districts which had pupils fitting this criteria. The first direct measure of the way in which disadvant- aged pupils are served by the school districts was the dis- trict mean achievement level of the disadvantaged pupils, i.e., the mean achievement score of those pupils whose SES score was less than or equal to 37. This measure permits an examination of the circumstances under which the group fares better or worse state wide. The second direct measure employed was the ratio of the low SES group achievement mean to the achievement mean of the rest of the children in the district in that grade. The notion involved in specifying this measure was that while the state wide level of achievement of the disadvantaged group is of interest, the relative achievement of that group to the rest of the pupils in each district is of equally im- portant concern. Knowing that my daughter is in the 90th percentile in reading nationwide is of little consolation when whe is at the bottom of her class. Neither Of these measures, however, convey the notion of "share of the benefits" since that is in fact related to the cumulative proportion of total achievement obtained by one group or another. Such a measure must consider both the relative size or proportion the subgroup is of the total number of pupils. While several measures of share of achie- vement benefits Obtained by disadvantaged pupils were 107 developed, their use in the empirical investigation intro- duced violations of the assumptions of the statistical model employed such that clear interpretation of the results was not possible. It is important to note again that while the two alter- native measures of school district responsiveness to dis- advantaged pupils can be instructive, neither fully captures the sense of the issue involved in the distribution of the benefits of the schools. Degree of Implimentation of Differentiation Norm in the Production of Educational Experiences.--In describing the school policies and practices of the differentiation oriented school system, Brookover, et al., [14] identify the following characteristics: (1) the systematic identification of differences among pupils, (2) the formal classification or labeling of students, (3) an emphasis on individualized instruc- tion, and (4) differentiated instructional programs. It should be clearly stated that these policies and practices in and of themselves need not lead to a wider dis- persion of pupil outcomes or to an acceleration of the rate at which the dispersion increases. Further, when such pro- grams and practices do lead to a wider dispersion of pupil outcomes, it need not be the result of purposeful discrimin- atory decision making on the part of school district leaders 108 or educational practioners; they may only be practicing quality education as they understand it without fully under- standing the consequences of that practice. The characteristics of the differentiation oriented school system described above and the practices associated with those characteristics are manifestations of the norm insofar as they are based on the assumptions that vast dif— ferences in innate ability exist, pad that those differences are identifiable and substantially unchangeable. It is Brookover's contention that the degree to which such policies are employed do reflect the acceptance of the differentiation norm and its embodied assumptions. In The Sixtyffirst Year- book of the National Society for the Study of Education, entitled Individualized Instruction, Cook and Clymer write, "The net result of a good instructional program is growth for all but greater growth for the capable. Hence, there is greater spread at the end of an effective instructional per- iod than there was at the beginning [24]." In seeking proxy variables to characterize the degree to which the norm is accepted by a school district, none of the characterizations used by Brookover and group were avail- able. What is available in the data set from the State of Michigan Assessment Program are a number of variables which make it possible in part to characterize the resources em- ployed by school districts in their instructional program. 109 The variables used and the rationale for their use in charac- terizing the degree to which the educational norm is imple- mented are set forth below: 1. Kindergarten - 12th Grade Instructional Expenses, Net of Teachers' Salaries, Per Pupil. (Inst. Materials Expendi- tures/Pupil) ' This variable was constructed by subtracting teachers' salaries from total K-12 instructional expenses per pupil. While it is unclear whether the salaries of instructional specialists other than classroom teachers are included, it was felt that this was an indication Of the instructional resources employed in the classrooms beyond the provision of classroom teachers. A major shortcoming of using an expenditure item as an indicator of inputs is the issue of varying prices by region. However, the use Of this variable in this case is more modest than what is attempted in the estimation of production func- tions. This variable is used as one of a group of variables indicating the degree of implementation of the differentia- tion norm. A major characteristic of individualized instructional programs is that they are more costly in terms of materials and support systems. Thus this measure is here used as an indicator of the degree to which such a program is implemented. 2. Average Teacher Experience Since the instruction of teachers and educators in this 110 norm takes place primarily in colleges of education, this variable can be a reasonable approximation of the average period since teachers were in full-time attendance at a col* lege of education. Empirical evidence from Brown and Saks [16] is that as average teacher experience increases in the district, the standard deviation of achievement decreases. Thus in this context a decline in average teacher experience is seen as the degree to which the norm is actively accepted by the teachers and at least functionally by the school district. 3. Percentage of Master's Degree Teachers The logic of using this variable is similar to that of the previous one. The completion of a Master's degree re- flects a longer period of association with a college of ed- ucation--for teachers frequently a substantial period of a part-time association--and thus a more complete imbuement with the differentiation norm. The empirical evidence of Brown and Saks here is that as the percentage of Master's degree teachers increases, the standard deviation of school district achievement scores also increase. The decision to hire Master's degree teachers may reflect the norm or, as with the previous variable, the measure at least functionally appears to indicate the degree to which the norm is accepted by teachers of the district. 4. Pupils per Classroom Teacher Another dimension of the higher costs associated with 111 individualized instructional strategies is the effort to re- duce class size. This may occur for reasons other than in- structional strategy such as a rapid decline in pupil popu- lation With an existing commitment to teachers or the bargain- ing position of teachers unions. It is, however, included as an additional variable indicating the general commitment to individualized instruction. 5. Pupils per Nonteaching Professional Instruction Staff This variable is constructed from the previous variable, pupils per classroom teacher and from a variable giving the pupils per professional staff. The latter includes in addi- tion ot classroom teachers, principals, assistant principals, other administrative staff excluding district-wide admini- strative staff, consultants, librarians, guidance and coun- selling staff, psychological staff, and other specialists. This measure is included as an indicator in response to several of the dimensions of the differentiating norm de- scribed by Brookover, et a1. First it includes the variety of specialists involved in the support system of the individ- ualized instruction approach. Second, it also includes the variety of personnel involved in the identification and class- ification of differences among pupils. The interpretation of this variable is that as the num- ber of pupils per nonteaching professional declines, i.e., their proportionate number increases, the school district will be more oriented toward the differentiation norm. 112 The Degree of Community Concensus.--The data from the 1970 Census Fourth Count made possible the development of a number of variables which characterize the general socio- economic level of school district communities and the degree of community socioeconomic hetrogeneity. Among the character- istics which it was felt might distinguish political prefer- ences and educational preferences within communities and be- tween communities were those related to incomes, educational attainment, mobility or stability of the population, occu- pations, and race. 1. Income variables Several attempts were made a various summarizations of the level and distribution of income. Part of the diffi- culty in dealing with the income data derive from the way in which it is presented in the census income table. The in- come classes are shown below: Under $1,000 $11000 - $11999 2,000 - 2,999 9,000 - 9,999 10,000 - 11,999 15,000 - 24,999 25,000 - 49,999 50,000 and over The intervals (classes) are uniform in $1,000 units from zero to $10,000. Above that value the intervals vary con- siderably. In addition to frequencies associated with these 113 c:laisses, total family income was available as a separate statistic. Initially the median income and an approximation of the standard deviation of income were selected to characterize the: .level and dispersion of incomes respectively. The latter estimate was made by considering the midpoint of each class ix) h>€3 the actual income associated with the frequency for pmrp>§g normality does not hold in these communities and that in féiczt the skewedness of income is not positive as would be the Case with a distribution which is normal in the logs, but 543 skewed in entirely the Opposite direction. At this point the interquartile range of income, the difference between the income at the bottom of the second quartile and the income at the tOp of the third quartile, was selected as the measure of dispersion. This measure was estimated assuming a uniform distribution within the classes 114 in which the quartile breakpoints occurred. The information that there appeared to be differences in skewedness of the distributions of income suggested that this might be an interesting characteristic of school dis- trict communities to examine. An estimate of the degree and direction of skewedness was obtained from the following com- putation:i/ Skewedness = 3(mean - median)/standard deviation In the final results the skewedness variable is not used be- cause a programming error resulted in a meaningless esti- mate of the variable. In summary, the statistics used as variables in char- acterizing the level and distribution of income in school district communities are as follows: Median income Interquartile range of income 2. Educational Attainment Variables The census provides data on the years of school com- pleted by individuals 25 years and over. While this does not completely describe the voting age population in 1970 (age 21 or above), it was considered to provide a reason- able approximation. The mean number of years of education attained and the standard deviation associated with it were estimated for use as variables describing the level and 1/ This method of estimation was obtained from Kane [33], p.85. 115 distribution of education in the adult population of the school district community. 3. Community Stability The 1970 Census Fourth Count, Table 28 provides infor- mation on the 1965 residence of persons 5 years and Older. From this table a measure of the "stability" of a school district community was calculated. The variable represents the number of individuals who had lived in the same house or the same county in 1965 as a proportion of the total num- ber of individuals in the category (5 years and above). As the proportion of individuals who meet this criteria in- creases in a community, the community is considered to be more "stable." There is some potential bias in the statistic because the unit of observation in this research, school districts, often do not conform to other legal boundaries, including counties. Thus an individual may have moved within a county and have changed school districts, or have moved across coun- ty lines and remained in the same school district. The mea- sure was considered a reasonable representation of the degree to which the particular population had stayed within the community. 4. Occupation Variables The 1970 Fourth Count Census Table 58 provides data on 42 occupational classes. Unfortunately, the development of 116 a continuous variable or variables which would characterize the distribution of occupations from this data does not ap- pear feasible. The various efforts at developing occupation- al status scores such as that of the Bureau of the Census, Reiss [43] or the effort of Blau and Duncan [9], use much more detailed definitions of occupations than are available in the data set. No summary scores appear available for the 42 categories which are employed in this data set. As a result a single variable was developed which would characterize the proportion of total employed persons 16 years and over who were "professional, technical, and kind- red workers" or "managers and administrators (except farm)." The variable was felt to be the only easily identified and intuitively appealing subdivision of the 42 occupational classes which was obvious. 5. Racial Composition Variables A small amount of data from the Census First Count was copied from microfilm records. Included was the data neces- sary to compute the proportion of population which is racial- ly "black" for each district. This variable, percent Black is the only characterization of the racial composition of communities employed in this research. Power of Disadvantaged in School District Communipy.-- The distinction between who is and who is not disadvantaged under the educational norm is not clear or precise. In using the census data to approximate the power of the "disadvantaged" 117 group, the census category of 'below poverty level' suggests a group which may be considered disadvantaged. The Fourth Count Census tape documentation [39] states in a footnote that the definition of "poverty level" is that defined by the Social Security Administration Poverty Index. Two variables were constructed using this definition of disadvantage. First the proportion Of total families who were below the poverty level was caluculated. This variable thus estimates the numerical proportion of adults who are below the poverty level and may be considered the voting strength of the poor. Since the unit of measure is "families" and there may be some expectation that the incidence of single parent fam- ilies is higher among the poor, the variable may overesti- mate the proportion of adults who are poor. The second variable used to indicate the power of those below the poverty level is an estimate of the proportion of total family income held by poor families. Since this vari- able is based on the report of total income received by fam- ilies below the poverty level and total family income, it is considered an accurate estimate. Communipy Size.--A variety of candidates for indicating the size of school districts were possible from the available data. Pupil enrollment is an often used measure. Since this research was focusing on the articulation of preferences and the political process affecting the schools it was determined 118 that the measure of size would be the number of voting age (21 years or above in 1970) adults in the community. Sponsors Vs. Consumers of Education.--In an effort to examine whether the distinction between sponsors and con- sumers of education discussed earlier was meaningful or not, a variable was developed which attempted to express the per- centage of sponsors who were also consumers of public edu- cation. First an estimate of the number of parents of school age children was developed. To accomplish this the number of female headed families with related children under 18 years old was added to 33133 the number of similar male headed households. While by no means are all single parent families female headed, it was determined that in the absence of additional information the male headed households would be treated as two parent units. In order to account for sponsors who sent their children to private or parochial schools, the above estimate of the number of parents of School age children was reduced in the same proportion as the parochial and private schools atten- dance is of total school attendance. This is the same as assuming that all families are Of the same size--obviously not the case-- but more refined estimates were beyond the capacity of the census categories. This estimate of the consumers of public education was . ~— ___ is: (T) n; then the c Meas: SQre 119 tinen expressed as a percentage of the voting age adults in -t11e community or the sponsors of public education. Summary of Variables.-- pdeaaasures of Responsiveness to Disadvantaged Pupils Coefficient of Variation of Achievement Low SES Group Achievement Mean Relative Mean Achievement of Low SES Group to Balance of Grademates Me51£511res of Degree of Implimentation of Differentiation Norm Instructional Materials Expenditures Per Pupil Average Teacher Experience Percentage of Masters Degree Teachers Pupils Per Classroom Teacher Pupils Per Nonteaching Professional Instruction Staff Measures of Community Consensus Mean Years of Education Standard Deviation of Years of Education Median Income Income Interquartile Range Community Stability Percentage Professional Workers Percentage Blacks Measures of Disadvantaged Group Power Poverty Families As Percentage of Total Families Poverty Families' Income as Percentage of Total Income 1 120 Measure of Community Size Voting Age Population Measure of Consumers Vs. Sponsors of Education Consumers as a Percentage of Sponsors The use of these proxy variables in the estimation of the generalized model is presented in the following chapter. CHAPTER V STATISTICAL METHODOLOGY Introduction The generalized model described in Chapter IV was ap— proximated through the use of multiple regression equations estimated by the ordinary least squares technique. The pro- xy variables described earlier were employed in estimating components of the model. The model was estimated using methematic achievement scores as measures of the benefits of education. Of the various measures of school district responsive- ness to disadvantaged pupils which were developed the fol- lowing three were retained and employed in the regression equations as dependent variables: 1. Coefficient of Variation - an indicator of school district responsiveness. 2. Mean achievement of low SES group. 3. Relative mean of low SES group achievement to mean achievement of the balance Of grademates. It had been hOped that the census data would make it possible to rather completely account for the differences between communities. It was soon recognized that the char- acterizations of communities which was in fact possible was less than complete. For this reason the equations were esti- mated for several community type subsets of the data in 121 122 addition to the estimations made using all of the available complete Observations. The community types considered were (1) Towns-Rural, (2) Urban-Fringe (Suburbs), and (3) Metro- politan-Cities. The Empirical Methodology The Choice of Multiple Regression Analysis.--The gen- eral model was approximated and estimated using multiple regression was the need to disentangle the various influen- ces of different components of the model on the measures of school district responsiveness, i.e., the need to meet the ceterispparabus conditions of the hypotheses. As Kane [33] points out statistical analysis provides two possible ways of accomplishing this. Experimental design approaches meet the "other things constant" con- ditions by the structure of controlled experiments. In cases where data is not from controlled experiments, multiple regression is employed to estimate the influence of variables other than those of primary interest and, in an ex post fashion, meet the ceteris parabus conditions. The Statistical Model.--The general model of Chapter IV was approximated by the following multiple regression model: 123 R = a constant + 31 (Inst. Materials Exp./Pupil) + 82 (Avg. Teacher Experience) measures of imp- + 33 (Pupils Per Teacher) lementation Of + 34 (Masters Degrees %) differentiation + 85 (Pupils/Non Teaching Staff norm ‘ + 36 (Mean Years of Education + 37 (Std. Dev. Years of Education measures of + 88 (Median Income) degree of com- + 89 (Income Interquartile Range) munity'consensus + 810 (Community Stability) measures of de- + 811 (Professional Workers %) gree of commun- + 812 (Black %) ity concensus + 313 (Poverty Families %) + 814 (Poor Income %) measures of dis- + 815 (Pov. Fam. %) (Poor Income %) advantaged power + B (VOting Age Population) 16 2 size + 817 (Voting Age Population) + 818 (Schooling Consumers %) consumers vs. sponsors + error Where: R = responsiveness of school district to disadvant- aged and is measured by the previously described variables. = constant to be estimated. = coefficients associated with specific variables and which are to be estimated. 124 (Variable) = as previously defined. Several points must be made about the statistical model. The several components of the general model were esti- mated by several variables which were independently and ad- ditively included in the statistical model. A variety of analytical techniques such as analysis of principle com- ponents or factor analysis might have been employed to com- bine the various sets of multiple variables into a few or a single estimator of the respective components in the general model. The difficulty with such an approach is that com- posit variables or factors thus created are extremely dif- ficult to interpret in terms of the original variables. Since the original variables are part of the existing infor- mation system, the interpretation of the results in terms of generally understood measures adds substantially to an under- standing Of the results. Further, in a policy context, the reason that many measures are in fact included in public information systems is because they are amenable to change by policy decisions. Such would be the case for the per- centage Master's degree teachers. In the case of this re- search on the boundary question, size of district could be similarly viewed. For these reasons, the several elements of the various components were generally entered into the statistical equa- tions in an additive fashion. In the case of the "power" 125 component and the "size" component of the general model which are of primary interest in this research, further refinements of the component variables were included as is indicated. Estimation Technique.--The various equations were esti- mated using ordinary least squares estimation techniques after examining the several equations for possible violations of the usual OLS assumptions 1. Heteroskedasticity Since much of the data were summary statistics from group data there was some concern that the OLS estimation procedure might violate the homoskedasticity assumption of the procedure.l/ This is particularly true when the depen- dent variable is a group mean, the error of which is a func- tion of the group size. This would apply in this research where the mean achievement level of the low SES group is used as a dependent variable. In order to examine the equations estimated for prob- lems of heteroskedasticity deriving from the use of the group data, an examination of the residuals was accomplished. The residuals of the equations employing each of the re- spective dependent variables were both plotted and regressed against the primary suspect variable, -the size of the group from which the particular dependent variable was derived. l/ See Kemente [35], pages 322-336. 126 In the attempt to fit a line to the residuals by regression a squared term for group size was included. Neither approach indicated any systematic relationship with group size. In the case of the regressions, the F values of the fitted lines were nonsignificant, as were the individual coefficients, i.e., the constant and variables for group size. It was thus concluded that the homoskedastic assumption of the OLS was likely met. 2. Multicolinearity An examination of the simple correlations between inde- pendent variables indicated little concern of multicolinear- ity in the data. Indeed, several of the variables had sim- ple correlations which were in the order of plus or minus 0.8 - 0.9. In all such cases, however, the two variables are being interpreted as a single influence. Such is the case with the linear and the squared terms for size (Voting Age Population) as well as the several measures related to the power of the disadvantaged group within the community. In summary, the equations were fitted using ordinary least squares estimation proceedures after examining the data to determine whether the assumptions underlying that technique were reasonably met. Three equations - one for each of the dependent variables - were estimated using four data sets based on community types. The resulting twelve equations are presented and discussed in the subsequent chapter. CHAPTER VI RESULTS AND CONCLUSIONS Introduction The results of the empirical analysis and their discus- sion are primarily directed toward the questions associated with the boundary issue. There are in addition findings which are specifically related to the performance of school systems which are also presented and discussed. As stated earlier, the analysis was first undertaken across all dis- tricts and subsequently done on subsets for (1) rural- town districts, (2) urban fringe districts, and (3) city-metro- politan districts. The presentation of the results will fol- low this general organization. While the meanings attached to the several variables have been set forth in some detail already, it iszappropriate to restate here the specific interpretation attached to the three dependent variables used as measures of schoolidistrict responsiveness to disadvantaged pupils. Coefficient of Variation of Achievement - (Standard Deviation/Mean).--The definition of this variable as argued in Chapter 3 is that it is a direct indicator of school dis- trict responsiveness to those disadvantaged under the edu- cational norm. A decline in the magnitude of the variable 127 128 which is accomplished by a rise in the mean, a decline iJIthe standard deviation, or both indicates either a more egali- tarian preference for schooling outcomes; a choice of inputs which generate more egalitarian outcomes; or both of these situations. The range of this variable in the data set which in- cludes all of the districts is from 0.098 to 0.241 with a mean of 0.179. The lower bound of the range is a district where the Coefficient of Variation results from a mathematics achievement mean score of 55.9 and a standard deviation of 5.5 (5.5/55.9 = 0.098). Similarly, the actual upper bound of the range is calculated from a mean of 40.7 and a standard deviation of 9.8. Several numerical examples of the calculation of this indicator and its magnitude may further assist in a more in- tuitive understanding of the results reported in this chapter. Using a value of this variable of 0.2, the following com- binations of achievement mean and standard deviation will yield the same value: 10/50 = 0.2 12/60 = 0.2 8/40 = 0.2 Appendix I lists the school districts in this data set and ranks them according to the coefficient of variation of fourth grade math scores. Low SES Group Achievement Mean.--The definition and in- terpretation of this variable is relatively straightforward. The value of this variable for each district is the mean math achievement score for those fourth grade pupils who meet the 129 criteria of being in the lowest 10 percent on the basis of socio-economic status state wide. Thus, an increase in the magnitude of this variable indicates a situation where the average achievement level of this group is better. The con- verse would also be true. The range of this variable in the data set is from 36.0 to 64.8, with a mean of 49.1 and a standard deviation of 3.74. Relative Mean Achievement of Low SES Group to Balance of Grademates.--This variable is calculated as a ratio of the Low Group Mean divided by the mean achievement of the balance of the grademates. Thus as the magnitude of this variable is greater than 1.0 the mean achievement of the low group is greater than the mean of the balance of their grade- mates. Similarly, as the value of this variable is less than 1.0 the low SES pupils mean achievement is lower than the achievement mean of the balance of their grademates. This variable is seen as an indicator of the relative responsiveness of school districts to the low SES pupils as compared to the balance of pupils in the grade. The range of this variable in the data set is from 0.656 to 1.21 with a mean of 0.96 and a standard deviation of 0.065. In interpreting results employing these three dependent variables it is important to remember that increases in the magnitude of the Low Group Mean and the Relative Mean indi- cate a schooling situation more responsive Us the disadvant-— aged group of pupils. On the other hand, a decline in the 130 magnitude of the Coefficient of Variation indicates a simi— lar situation. Thus the expected signs in equations employ- ing the Coefficient of Variation will be Opposite those of the other two dependent variables. Tables 3, 5, 7, and 9 present the results of the equa- tions fitted to data from (1) all districts; (2) town and rural districts; (3) urban fringe districts; and (4) city and metropolitan districts respectively. Variable means and standard deviations for each grouping of data follow the regression results in Tables 4, 6, 8, and 10. As was pointed out earlier the Detroit Metropolitan School District which is on the order of ten times the size of the next largest district was eliminated from the data because of the tendency of that Observation to dominate all considerations of school district size. 131 Table 3. Ordinary Least Squares Regression of Measures of School District Responsiveness to Disadvantaged in Michigan School Districts matics Achievement Score Measures) (Fourth Grade Mathe- Coefficient Relative of Variation Mean Low Group Mean Constant Inst. Materials Expenditures/Pupil Average Teacher Experience Pupils Per Teacher Masters Degree (%) Pupils Per Non- teaching Staff Mean Years of Education Std. Dev. - Years of Education Median Income ($1000) Interquartile Range ($1000) Community Stability (%) Professional Workers (%) Black (%) Poverty Families (%) numbers in parentheses are values of‘t .165045 1.15280 (4.13)*** (7.06)*** .000077 -.000034 (2.93)*** (0.32) .001538 -.000551 (3.81)*** (0.33) .001035 .000602 (3.08)*** (0.44) .000128 .000160 (1.33) (0.42) .000014 -.000040 (1.87)* (1.34) .001913 -.021775 (0.73) (2.02)** .001743 -.012708 (0.43) (0.79) .001730 .012530 (1.28) (2.16)** .000750 -.007390 (0.76) (1.71)* .000073 -.000314 (0.67) (0.73) .000669 .000081 (2.66)*** (0.07) .000516 .000012 (3.69)*** (0.02) .000712 .000106 (1.09) (0.04) 52.5346 (5.70)*** -.008393 (1.40) .039228 (0.41) .021190 (0.27) .050188 (2.32)** .002252 (1.32) .720527 (1.19) .261071 (0.29) .479850 (1.46) .376500 (1.54) .009513 (0.39) .001481 (2.40)** .123993 (3.50)*** -.005467 (0.03) 132 Table 3 (cont'd.) Coefficient Relative Low Group of Variation Mean Mean numbers in parentheses are values of t Poor Income (%) -.001535 .017879 .642906 (0.56) (1.38) (0.88) %POV x %POORY -.004305 -.034672 1.02972 (0.68) (0.65) (0.34) Voting Age Popula- tion (thousands) .000507 .000292 -.050343 (2.63)*** (0.38) (1.16) (Voting Age Pop.)2 -.000003 .000002 .000363 (1.45) (0.26) (0.80) School Consumers (%) .000179 .000283 -.004625 (1.29) (0.49) (0.14) R2 .262 .069 .117 Number of Observations 499 463 463 * significant at .10 ** significant at .05 two-tailed tests *** significant at .01 133 Table 4 Variable Means and Standard Deviations for Equa- tions Fitted To Data for All Available Michigan School Districts. Dependent Variables Coefficient Relative Low Group of Variation Mean Mean mean .17929 .96346 49.0566 standard deviation .01912 .06453 3.7368 Independent Variables Inst. Materials 116.432 117.083* Expenditures/Pupil 45.860 45.175 Average Teacher 9.275 9.263 Experience 2.384 2.353 Pupils Per Teacher 24.049 24.163 2.417 2.275 Masters Degrees (%) 21.735 21.458 10.951 10.864 Pupils Per Non-teach" 231.447 230.710 ing Staff 109.300 107.657 Mean Years of 10.684 10.682 Education 0.717 0.687 Std. Dev. - Years 2.949 2.949 of Education 0.266 0.267 Median Income ($1000) 10.060 10.071 2.307 2.157 Interquartile Range 7.388 10.913 ($1000) 1.944 2.457 Community Stability (%) 78.300 78.244 9.007 9.090 Professional Workers 18.656 18.489 (%) 6.883 6.424 Black (%) 2.129 2.087 6.669 6.074 Table 4 (cont'd.) 134 Dependent Variables Coefficient of Variation Relative ,Low Group Mean Independent Variables Poverty Families (%) Poor Income (%) %POV X %POORY Voting Age Population (thousands) (Voting Age Pop.)2 (thousands) School Consumers % 8.517 4.900 1.736 1.659 0.223 0.477 8.226 12.605 226.221 1013.470 47.132 7.473 8.425 4.567 1.677 1.345 0.200 0.273 8.422 12.870 236.210 1049.645 4.286 6.280 * Data set for equations employing Relative Mean and Low Group Mean are identical thus the mean and standard deviation for the independent variables are identical. 135 Table 5 School District Responsiveness to Disadvantaged in Michigan Towns and Rural School Districts (Fourth Grade Mathematics Achievement Score Measures) Coefficient Relative Low Group of Variation Mean Mean numbers in parentheses are values©MmHo may Ou coHumouwm mafiaoonom m>flmcommmu whoa m mmumofiocfi magmaum> many CH mafiaomp <« .mucmflofimmmoo uCMUAMHcmfimcoc mo mucosamsw mo coHuomuHo may mumoflocfi mcmflm .ucmoHMHc Imam maamoflumflumum mumz noesB mucmfiowmmmoo mmocu waco How pmofl>oum mum mmwuwoflummHm I I moo.I moo.I I + I + I + moo. moo. Awo xomam I aha. + Hoo. I + I + I + boo.I obo.I mumxuog Hmcowmmwmoum Amy + + + + I + I I I I + I suflaflnmum suflcsesoo 3 m. 8638 + + + + + + + HmH.I + I I I wEoocH Guano: cofiumosom + moo.I I I I oao.I I va.I + mmn.I + + no whom» c002 Aoooamo mEoocH no I I I I I I I qmo.I + I hma. + mmcmm mafluumovumucH cowumooom mo mumme + mom.I I I + omm.I I I I Hmv.I I I coflumw>0o pump:0pm ouumz mono Hmuom muoflnu ZIU now mIB Haé ZIO bow mIB Had Immfluflo Imam IQ3OB Imam Had com: msouo 30A cam: w>flumflmm IGOHumflum> mo ucmflOHmmmoo .mflflmsm momma nuusom commucm>©0mHQ ou mmwcm>wmcommmm unflnumflo Hoonom mo mmuommmz mo mowumflumuomumnu muwcoEEou msowum> ou pommmmm nua3 oncommmm mo mmfiuflofiummam «a wanna 164 group (the Coefficient of Variation and the Relative Mean) and in terms of the mean or absolute level attained by the low SES group. In general the signs of the nonsignificant coefficients further support this position. The results in Table 14 indicate an apparent difference in the influence of heterogeneity with respect to education from that with respect to incomes. Both variables indicate that as the community is more hetrogeneous the Low Group Mean will decline. Similarly, increases in both measures of heterogeneity result in a decline in the Relative Mean. How- ever, communities which are more heterogeneous with respect to education appear to be relatively more responsive to dis- advantaged pupils whereas communities which are more hetero- geneous with respect to income are relatively less respon- sive to the disadvantaged. The overall indication of these results are that insofar as individuals' preferences or needs for schooling are meas- ured by income and/or education, the more heterogeneous the community preferences, the less responsive the schools will be to minority groups. 2. Measures of Community Diversity From the Low SES Group The statistically significant results from Table 14 generally support the corollary to Hypothesis 3. That is, as the majority preferences or needs move away form those of the low SES group, the schools are less responsive to them. 165 The opposite indication from the results of the Coefficient of Variation in Urban Fringe districts is unexpected and unexplained. Whether this represents a difference in the behavior of the suburban communities, a dimension of the Coefficient of Variation which has not been accounted for or an inadequacy of the model is not known. While these results are hardly overwhelming in their support of this ammended hypothesis, they are seen as tenta- tively supporting the argument that as a minority group's preferences are more diverse from the majority, the body politic will be less responsive to their preferences. 3. Community Stability The complete lack of statistically significant results with regard to the influence of the stability of a community on the distribution of achievement precludes any inference with regard to this dimension of the characteristics of com- munities. 4. Occupational Characteristics As was pointed out earlier, the only occupational characterization of communities which was easily available was the prOportion of professional, managerial and technical workers in the community. Again referring to Table 14, the statistically significant results indicate that as the Per- cent Professional Workers increase the school district is 166 more responsive to the disadvantaged pupils as measured by the Coefficient of Variation and by the Low Group Mean. Whether the lack of statistically significant results in the City and Metropolitan districts indicates different behavior of this group of people in those community types is not known. 5. Racial Composition Only the percentage black was available as a measure to characterize the racial composition of communities. In four of the twelve equations fitted the coefficient was statis- tically significant and in all such cases the direction of influence was the same--as the prOportion of blacks increase, the schools are less responsive to the disadvantaged pupils. The low SES group is not a racial characterization of pupils. In fact the simple correlation between percent black and prOportion low SES pupils is very low--about 0.15. How- ever, it is not unlikely that where the prOportion of blacks in the community is high, a large number of the low SES pupihs are black. It is interesting to note that none of the coefficients for percent black in equations fitted to cities-metro data were significant. In the case of the Coefficient of Vari- ation the signs for percent black in the cities-metro data was reverse those for the other data groups. These results are more definitive than are the results involving the power of poor families reported earlier, even though the mean 167 percent black is lower than the percentage of poor families throughout the data. A plausible explanation of this result is that a kind of visibility phenomenon is at work. The argument is that for groups who are the likely object of discrimination, where their numbers in a community are very small such that they constitute no threat, a sort of nobless oblige relationship occurs. At least they are not particularly singled out for discrimination. As the group numbers increase at these low levels they become more of a threat and the discrimination 1/ against them becomes more systematic.— When the group gains a majority or becomes a significant political, social or economic force the power relationship would hold. In general the number of communities in these data where blacks are in the majority or even a politically significant minority and where the visibility effect would conceivably be overcome are few. A count of such communities by some critical mass criteria (say 1 30 percent) and the proportion those communities are of the total number of observations in each data group might give some insight to an interpretation of the variability of results by community type. The ex- pectation would be that as the proportion of communities with greater than 30 percent blacks increased, the visibility 1/ This notion was suggested in personal conversation with Dr. Joseph Spielberg of the Department of Anthropology, Mich- igan State University, who has observed the phenomenon in oc- cupational ceilings among Latino Americans. 168 effect would be moderated and even overcome in the fitted data. Since this procedure has not been accomplished some in- sight that this result would be forthcoming can be had by an examination of the characteristics of the distribution of percentage black in the respective data groups. Table 15 provides this information. Table 15 Characteristics of the Distribution of Percentage Black by Data Groups and Direction and Significance of Influence of Percent Black on Measures of School District Responsiveness Range Mean SD SDs Coefficient' Relative Low Data % % % to of Group 30% Variation Mean ' Mean All 0-70.9 2.1 6.7 4.1 + * +NS -* Town-Rural 0-50.5 1.2 4.0 7.2 + * -NS -* Suburban 0-70.9 2.6 8.9 3.1 +NS +NS -NS Cities- 0-55.3 9.3 12.2 1.7 -NS -NS -NS Metro *significant coefficient NS - nonsignificant coefficient The probability that a district with 30 percent black or greater are in a particular data group (1 - [probability such a district is not in the data set]) is negatively related to the number of standard deviations above the mean necessary to get to 30 percent. As Table 15 shows, the City-Metro 169 data has the highest likelihood (lowest number of SD to 30 percent) of having observations with greater than 30 percent black and also is the data set in which signs of coefficients have changed in the data. The data set with the next highest likelihood of haVing observations of greater than 30 percent black is that for the suburban districts. Again this is associated with a change from significant coefficients to nonsignificant co- efficients. This discussion of the interpretation of the data for the percent black and its influence on school district re- sponsiveness is not seen as proof of a theory of "visibilityf' i.e., that at low levels of a minority presence increases in their numbers result in more systematic discriminatjrwIagainst them. It is, however, viewed as a plausible explanation of the data which may in fact be a future testable hypothesis. One last comment is appropriate on the results with re- gard to community racial composition. The possibility that these data can be interpreted under the genetic endowment hypothesis requires a strictly racial interpretation of gen- etic endowment, i.e., the assumption that the differences in endowment are primarily on the basis of race. Without that assumption it is impossible to resolve the data on race with even the limited data on community power of the poor. It is further interesting to compare the response to race with the response to poor peOple. While the measures of 170 school response are not explicitly to race, it is not too great a stretch of the data to argue that there is some evidence that race is the basis of greater discrimination than is socio-economic class. In summary, these data show that where the percentage of blacks is generally small, increases in their proportion result in a decline in responsiveness to low SES pupils both relatively and absolutely. There is some inferential evi- dence which suggests that where blacks are a substantial prOportion of the community, the above reported influence may be moderated and even reversed. The Sponsors vs. the Consumers of Public Education In an effort to examine the question as to whether school systems are more responsive to consumers or sponsors of education, equations were fitted using the parents of school age children as a measure of school district size. Generally, the consumers variable explained less of the vari- ation than did the voting age pOpulation variable. Although no statistical test was accomplished on the comparison, the conclusion can be tentatively reached that in the character- ization of size of school district, the parents of school children are no better a measure than are the sponsors of the schools. To examine the kinds of influence on school district responsiveness to the disadvantaged affected by non-consumer 171 sponsors a variable representing the proportion of total sponsors that are also consumers was used (percent consumers). Table 16 presents the limited results of this portion of the investigation. Table 16 Elasticities of Response with Respect to Percent- age of School Consumers of Measures of School District Responsiveness to Disadvantaged Fourth Grade Pupils Coefficient Relative Low Group of Mean Mean Variation All Districts + + — Town-Rural .087 + - Suburbs - + + Cities-Metro + - + Direction of influence (sign) is indicated for all nonsignificant coefficients. In the single case where this variable was statistically significant the results indicate that as the proportion of consumers in the community increased, the schools were less responsive to the disadvantaged. Stated the other way'around, the possible influence of the non-consumer sponsors of public education is to cause the schools to be more responsive to the disadvantaged. The conclusion from this result may be that citizens are for equality when it does not affect them (or their children) personally--that is, magnanimous behavior is more likely where it is costless or, at least, less costly. 172 The Differentiation Educational Norm Revisitedo-The Effects of Educational Strategies on Disadvantaged Pupils It is appropriate to again state that the differentia— tion educational norm described by Brookover and others [14] and which is substantially the basis of the interpretation of the variables used here, is not seen as some conspiracy of educators in schools and colleges of education with mid- dle Americans to systematically turn the screw on groups who have different experiences, needs or preferences than their own. The argument is rather that the philos0phy of public education in current prominence implicitly holds children responsible for their past experiences and preparation for schooling, rather than assigning responsibility to educators and teachers to find where children are and start with them there. An instructional program which would best accomplish this would perhaps be even more "individualized" than the current vogue in education. Such efforts as the Teacher Improvement Program which generated the cognitive style map- ping data reported on earlier would appear to be a step in that direction. It is claimed, however, that regardless of the in- tentions of the principle participants in the education pro- cess, the implications of the behavior and strategies which flow from the differentiation norm are such as to frustrate efforts to make the schools more neutral with regard to soc- ial class. Further, since such changes as may be necessary to cause schools to be an institution capable of ameliorating 173 inequality in American society may not be costless to the children of the majority of the current consumers of schools, there is little incentive to make these changes. Five variables characterizing the educational strategy employed by school districts were used in the equations. They were selected on the basis of their expected contri- bution to the implementation of a differentiation strategy of education. An examination of the results will indicate whether the measured influence of the respective variables does indeed coincide with the predicted influence. Further, the measured influences of these variables are suggestive of some of the changes in educational strategies which will lead to more equal distribution of the benefits of education as measured by achievement test scores. These results, already presented in Tables 3 through 10, are again displayed in Table 17 in terms of mean value elasticities for those variables which had significant co- efficients. This discussion will deal successively with each of the variables included under the arguments related to the differentiation educational norm. 1. Nonteaching Instructional Expenses per Pupil This measure was entered into the equation as a char- acterizetion of the degree to which individualized instruc-' tion is implemented and thus the degree to which the differ- entiation norm is implemented. As is seen in Table 17 the 174 .Hm>ma wo.aa 02» um uCMOHmwcmHm mm3 unmaowmmmoo mH£B* .Ommmucm>ommflc may Op coaumoufim mafiaoonom O>Hmcommou mHOE m wouMOHGOH manmflum> many cw OOHHOOU ¢« .mpcwwoflmmmoo ucmowwwcmflmcoc MOM pouMOfiocfl mum Acmwmo mucosamcfl mo :Owuomuwo maamofiumflumum mum3 nofl£3 mustOflmmmoo mmosu waso .ucmOHMflcmHm MOM UOOH>OHQ mum mmwuflOHummHm + I I I + mHo.I I I I + mmo.I mao.I mmmum mcfisommuIcoz mom maflmom + + omo. mmo. I I + + I moo. omo.I I lay mmmuo Ion mumummz Monomme I I I I I I + + I + HRH. mma. \mafimsm mocmflummxm Honomma I + I + I I I I I mmH.I mko.I omo.I 06696>¢ Hamsm\mmuou Iwocmmxm mama I wmmof + I I mmo.I hmo. + + omo. + omo. Inmumz.umaH ouuwz mow: Hmnsm mpOfiuu zIO now mIB Ham SIU now mIB Had Imwfluflo Inom IOBOB Imflo Ham moan0fium> com: moouw 30A mumnm muwmmu mom moouw 304 IcoHumHum> mo ucmHOflmmoov maflmom mpmuw numsom commucm>ommwo ou mmocm>wmcoommm uowuumwo Hoonom mo mmwumuum HmaowuosuumcH uoauumflo Hoonow mo mousmmmz mooflum> Op uommmom nuflz oncommmm mo mmfluwowummam 5H magma 175 general direction of influence of this variable is as pre- dicted, i.e., an increase in non-teacher instructional ex- penses results in a schooling situation which is less re- sponsive to those pupils disadvantaged under the educational norm as measured by all three dependent variables. There is some indication that in Town and Rural districts this may not hold at least as indicated by the Relative Mean equation and the sign associated with the Low Group Mean. With these two exceptions the balance of the results support the pre- dicted outcome. As was stated earlier, this good not be the outcome of an individualized instruction program characterized by larger expenditures for classroom and instructional materials, how- ever, under the prevailing educational norm this appeara to be the case. A possible interpretation of this result is that the instructional materials employed in the materials support system to individualized instructional programs are of great- er assistance to advantaged pupils than to disadvantaged pupils, and that their employment may in fact reduce the positive learning experiences of disadvantaged pupils. It will be recalled that in the development of the indicator of school district responsiveness to disadvantaged pupils in Chapter III, this issue was raised and included as one of the assumptions. 176 This should not be interpreted as meaning that expendi- tures on instructional materials are of no consequence in educational performance, since clearly the Coefficient of Variation was influenced by this variable. The change in the Coefficient of Variation is such as to indicate that there was a greater increase in the overall standard devi- ation of achievement than there was in the mean level of achievement. Neither should this be interpreted as an argu- ment against expenditures for classroom materials. The real question is with the 513g Of materials used. In summary, these data give the clear impression that increased expenditures on instruction net Of teachers' sal- aries, presumably including mostly materials support to in- struction, result in a wider dispersion of pupils' outcomes and a less responsive schooling situation for the low SES group of pupils. This would seem to argue that materials in current use in individualized instructional programs are, whether by design or otherwise, more productive with advant- aged children than with disadvantaged pupils. There is some evidence that they do not increment achievement of disadvant- aged pupils at all. 2. Average Teacher Experience The meaning attached to the average years of experience Of school district teachers and its basis for inclusion in the equations was that it represents the average period of 177 time since teachers were full-time students and exposed to the philOSOphy of education embodied in the differentiating norm. As seen in Table 17 this variable was statistically significant only in those equations employing the Coefficient of Variation. The nature of its influence in those cases was consistant with that predicted, i.e., as the average teacher experience increased the school district was more responsive to those disadvantaged under the prevailing educational norm. An examination of the means of Average Teacher Exper- ience increased the school district was more responsive to those disadvantaged under the prevailing educational norm. An examination of the means of Average Teacher Exper- ience in the respective community types indicates that there are significant differences. The means and t values of those differences are provided below: Mean - Average Teacher Experience t value Town - Rural 9 . 50 Districts y 5.48 ‘Urban Fringe 8.12 3.05 Districts ’1; 6.99 City-Metro 10.74 Districts This information would suggest that children who are sent from city or metropolitan districts to urban fringe dis- ‘tricts will not gain, on average, from the experience of tea- chers in the urban fringe districts. 178 3. Pupils Per Teacher Because individualized instructional programs generally are characterized by smaller class size, this variable was included as another indication of the degree to which the differentiation norm was implimented. There was no partic- ular direction of influence expected although a decline in responsiveness to disadvantaged pupils with smaller classes would not have been a surprise under the arguments of this research. As the statistically significant results with regard to this variable in Table 17 indicate, a decline in class size results in a schooling situation relatively more responsive to disadvantaged pupils. While the only statistically sig— nificant results were from equations employing the Coeffic- ient of Variation, the prepondernace of the signs associated with results which were not significant do indeed support this interpretation. A plausible explanation Of this influence is that in smaller classes, teachers simply have more time with all pupils including the disadvantaged pupils. In a schooling situation emphasizing individualistic behavior it may very well be that disadvantaged pupils benefit more from this change than do pupils who are highly individualistic in their learning styles. While it is still seen as a characteristic of individual- ized instructional programs, it also occurs for quite 179 different reasons as was pointed out in earlier chapters. It is clear that this aspect Of individualized instruction is Of benefit to the disadvantaged pupils. 4. The Percentage of Masters Degree Teachers The inclusion of this variable as an indication Of the degree of implementation of the differentiation norm was rather similar to that argued for the average experience Of teachers. The argument is that since teachers who have at- tained Masters degree training have been exposed more com- pletely to the educational philosophy embodied in the dif- ferentiation norm, this variable represents, at least func- tionally, the degree of acceptance of that norm by schOol districts. The evidence from Table 17 on this variable seems to counter the argument made with regard to its expected in- fluence. Although the overall results appear to support the argument that as the percentage of Masters degree teachers increase the schools are more responsive to the disadvantaged pupils, there is some evidence, including some that is statistically significant, which supports the converse argu- ment. The prevailing view within the educational community as reflected in incentive structures is that Masters level training of teachers is desirable for quality schooling. In so far as "quality" includes responsiveness to disadvantaged 180 pupils, these results suggest that further investigation of the effects Of Masters level training of teachers is appro- priate. 5. Pupils Per Nonteaching Professional Instructional Staff As a characterization of the differentiation norm, non- teaching professionals were seen as participating in the sorting of pupils by perceived differences and in the support system of the individualized instruction approach. The ex- pected direction of influence Of this variable was that as the relative number of this type of personnel increased, the schools would be less responsive to the disadvantaged. The results of Table 17 are internally contradictory with regard to this variable. In the case of the equations employing the Coefficient of Variation, the influence is as expected. On the other hand the indications from the equa- tions employing the Low Group Mean and the Relative Mean do not support the expected outcome. Several explanations appear plausable in resolving these results. It may be that the apparent differences in outcomes is the result Of the differences between what is measured by the respective dependent variables. As has been discussed previously, the Coefficient of Variation is seen as more completely measuring the relative responsiveness to disadvantaged pupils than do the other dependent variables which do not deal with the share of total achievement obtained by the low SES group. 181 A second plausable explanation of the results is that the differences in the results relate to different behavior by community types. A somewhat different aspect of the same argument is the possibility that the composition of "non- teaching professionals", a rather aggregate variable, is sys- tematically different by community types. These results leave the question of the influence of this variable somewhat unclear. They do suggest, however, that further investigation Of the effect of this type Of edu- cational professional would be a worthy endeavor. Community Types - What Differences? The segmentation of the data into subsets for town and rural districts, urban fringe districts and city and metro- politan districts was based on early work with the data which indicated that the demographic and other characterizations Of communities employed were insufficient to account for much of the differences between communities. The particular groupings were based on the availability of a classification of school districts by the State Department of Education which identified Metropolitan, City, Town, Urban Fringe and Rural districts. A judgement was made that for reasons of number of ob- servations, MetrOpolitan and City districts should be com— bined. Another judgement was made that Town and Rural dis- tricts were likely quite similar and those were combined into a single grouping. 182 Confirmation that the three subsets or groupings used are indeed statistically different was indicated using a test suggested by Kementa [35]. The test is used to deter- mine whether regression equations run on different samples are significantly different.1 The F statistic calculated on the equations reported in this research was significant at the 1% level. While there were no a priori hypotheses or expectations as to the differences between community types or the differ- ent behavior of the model by community grouping, an examin- ation of such differences as are apparent is of interest. A brief review of the results already presented with partic- ular reference to community type differences is given below. Community Size The most Obvious difference between the respective com- munity type groupings related to size are the mean and dis- tribution of size within the groups. That information from earlier tables is repeated here. Size - Number of Voting Age Persons Mean Standard Deviation Town and Rural Districts 3,911 3,146 Urban Fringe Districts 14,182 13,272 City and Metro Districts 30,457 27,371 Whether these differences in the data groupings related i/ See Kementa [35], p. 373-374. 183 to size explain the different results obtained - only the Town and Rural equations gave statistically significant re— sults with respect to size - or whether some other explan- ation is apprOpriate is not known. Poor Power Data differences with respect to the power of poor fam- ilies between the respective community type groupings were presented in Table 13. That information indicates that on average the poor in Town and Rural districts and in City and Metro districts when compared to their counterparts in the Urban Fringe districts have more than twice the numerical proportion of total families and from two to three times the proportion of total income. As was pointed out earlier, the level of poor power in all Of the data sets are so low that these differences may be of little or no consequence. The limited results with respect to these variables which were statistically significant make any inferences about different behavior between community types and the in- fluence of poor families on the schools impossible. Community Heterogeneity Differences between community types exist using both the standard deviation of educational attainment and the inter- quartile range Of income as the following information from earlier tables indicates: 184 Std. Dev. Years of Interquartile Range Education of Income Mean Std. Dev. Mean Std. Dev. Town and Rural 2.94 .26 6,877 1,050 Districts Urban Fringe 2.89 .22 8,864 3,123 Districts City and Metro 3.24 .27 7,639 1,231 Districts Statistical tests of the differences between means in- dicate that City - Metro districts are more heterogeneous with respect to education than either of the other two com- munity types which are not significantly different from each other. The Urban Fringe districts are most heterogeneous with respect to incomes with Town and Rural districts least heterogeneous in that regard. Those differences are also statistically significant. Again the limited information from the equations makes identification of anything but the above descriptive differ- ences impossible. Divergence of Majority From Low SES Group Median income and mean years Of education shown below indicate that the low SES group are further from the majority in the Urban Fringe districts both with reSpect to income and education. (The respective means are statistically dif— ferent from each other.) 185 Mean Years Of Median Income Education Mean Std. Dev. Mean Std. Dev. Town and Rural Districts 10.52 .54 9,226 1,711 Urban Fringe Districts 11.15 .91 12,600 2,273 City and Metro Districts 10.81 .87 10,073 1,418 The mean years of education was statistically signifi- cant in equations employing all three dependent variables when fitted to data from Urban Fringe districts and was non significant in virtually all other situations. The generally greater variation in that variable within the Urban Fringe data set may account for this difference although alternative explanations would be difficult to rule out. Community Stability There are no Obvious differences between the community types with respect to community stability. This is truelxfifli with regard to the discriptive differences of the communi- ties and in the apparent influence Of this variable in the fitted equations. It will be recalled that this variable had non significant coefficients in all cases. Percentage Professional Workers The following information indicates the differences be- tween community types with respect to the percentage of pro- fessional workers: 186 Percentage Professional Workers Mean Std. Dev. Town and Rural Districts 17.1 4.7 Urban Fringe Districts 22.4 10.0 City and Metro Districts 21.5 7.2 There is no significant difference between the means of the Urban Fringe and the City-Metro districts. There is, however, a significant difference between the means of both of those data subsets and the mean of the Town and Rural dis- tricts for this variable. There is within the results presented in Table 14 an in- dication that the influence Of professional workers on school district responsiveness to the disadvantaged is different in the Urban Fringe districts from the other two community types. The sign associated with the Coefficient of Variation in the Urban Fringe districts indicates that increases in the per- centage Of professional workers results in a less responsive schooling situation for disadvantaged pupils. The Opposite influence is exhibited in the other community types. The statistically significant indication that the Low Group Mean is also incremented by this variable in Urban Fringe districts would lead to the conclusion that in these districts the in- fluence of professional workers is to improve the level of educational achievement generally out to improve the achieve- ment Of advantaged pupils more. In Town and Rural districts the apparent influence is to improve the general level of 187 achievement and to make the schools relatively more respon- sive to the disadvantaged pupils. In City and Metro dis- tricts the influence appears to be to make the schools more responsive to the disadvantaged pupils but also lowers the mean of the Low SES Group and implicitly the over all level of achievement. It should be made clear that these interpretations are based only on the signs of the coefficients for percentage professional workers in the respective equations and not on statistically significant results. These possible differen- ces are nevertheless worth noting. Percent Black The major differences apparent in the data with regard to community type differences with reSpect to race have been elaborated in the initial reporting of the results. Table 15 and the discussion involving that table make clear that the range of the magnitude of Percent Black is greatest in the Metro-City districts and smallest in the Rural and Town districts. There appears to be some relationship between this information and the possible influence of the variable in the respective community types. While the overall in- fluence Of an increase in the Percentage Black is a schooling situation less responsive to disadvantaged pupils, that in- fluence is less clear as the range of date in the respective data groups includes observations which have larger magni- tudes of this variable. 188 Comsumers of Schooling Differences between the community types exist in terms of the proportion of voting age adults who are consumers of the services of the schools. The following information on the mean and standard deviation of this variable indicates those differences: Percentage Consumers mean std. dev. Town and Rural Districts 47.2 7.1 Urban Fringe Districts 49.3 7.4 City-Metro Districts 39.6 6.5 Beyond these discriptive differences, the general lack of statistically significant results makes the inference of further differences questionable. Instructional Materials Expenditures Per Pupil The mean level of school district expenditures for in- structional materials by community type and the respective standard deviations are provided below. In all cases the differences between the means are statistically significant. Per Pupil Materials Expenditures -$ mean std. dev. Town and Rural Districts 98.99 29.47 Urban Fringe Districts 105.58 46.37 City - Metro Districts 175.22 61.56 189 The general impact of increases in this variable as in- dicated in Table 17 is to make for a schooling situation which is less responsive to disadvantaged pupils. There is some indication that in Town and Rural districts the overall level of achievement may be improved but that advantaged pupils achievement is improved more. In the other community types the indications are that the influence of this vari- able is to make the schools both relatively and absolutely less responsive to the disadvantaged pupils. Whether the difference in the level of expenditures for instructional materials of the Town and Rural districts or some other influence explains this difference is not known. It may very well be that Town and Rural districts tend to buy different kinds of materials than do districts in the cities and suburbs. Average Teacher Experience Differences between community types do exist on average with respect to the average experience of teachers as the following information from earlier tables shows: Average Teacher Experience - Years mean std. dev. Town and Rural Districts 9.5 2.4 Urban Fringe Districts 8.1 2.1 City - Metro Districts 10.7 1.7 While the differences between these means are statistic— ally significant there does not appear to be much difference 190 between community types with respect to the influence of this variable. Generally, as was reported earlier, as the average experience of teachers increases, the schools are more re- sponsive to disadvantaged pupils. Whether the mean differences between communities with respect to this measure reflects employment strategy or practice, or the behavior of teachers in seeking employment is not known. Pupils Per Teacher - Class Size The following information by community type on class size and statistical tests on the respective means indicates that there are no significant differences between community types with respect to this variable: Pupils Per Teacher mean std. dev. Town and Rural Districts 24.1 2.5 Urban Fringe Districts 24.0 2.4 City - Metro Districts 23.9 1.7 The general influence of increases in the size of clas- ses, as reported earlier, is a schooling situation which is less responsive to disadvantaged pupils. There is little evidence of differences between community types in this re- gard. Whether the uniformity of class size between community types reflects school district initiatives or the initiatives of teachers bargaining organizations is not known. The latter 191 would seem a logical candidate to explain this lack of dif- ferences between community types. Percentage Masters Degree Teachers The differences between community types with respect to the percentage of Masters degree teachers are statistically significant. The means and standard deviations are shown below. Percent Masters Degrees mean std. dev. Town and Rural Districts 18.3 9.1 Urban Fringe Districts 28.6 11.3 City - Metro Districts 32.9 7.6 The results of Table 17 indicate that the influence of increases in this variable is to increment the achievement of the low SES pupils in all of the community types. How- ever, there are indications that in the Urban Fringe dis- tricts the achievement of advantaged pupils is incremented even more and that the schools are relatively less responsive to disadvantaged pupils as the percentage of Masters degree teachers increases. In the Town and Rural and in the City and Metropolitan districts the schools are made relatively more responsive to disadvantaged pupils by increases in the number of Masters degree teachers. Again, explanations of this apparent difference between community types are difficult to infer from this data and may very well be outside of these data sets entirely. 192 Pupils Per Nonteaching Instructional Professional The information on this variable by community type which is provided below indicates that Town and Rural dis- tricts employ relatively fewer Of this type of staff than do either the other two groupings of districts. The differ- ence between the means of the Urban Fringe and City - Metro districts are not significant whereas the difference between the means of both of these district types and the Town and Rural districts are statistically significant. Pupils Per Nonteaching Professional mean std. dev. Town and Rural Districts 246.6 109.2 Urban Fringe Districts 201.3 109.8 City - Metro Districts 182.2 69.2 As the initial reporting of results with respect to this variable indicated there is some apparent difference in the influence of this variable in the respective commun— ity types. On the basis of the signs of the respective coefficients reported in Table 17 an increase in the proportionate numbers of this type of staff (a decline in this variable) in City and Metro districts results in a schooling situation which is less responsive to disadvantaged pupils as measured by all three dependent variables. In Town and Rural districts, the achievement level of low SES pupils is incremented by an increase in the use of nonteaching instructional professional 193 but the schooling situation is relatively less responsive to that group of pupils. In Urban Fringe districts the school- ing situation is both absolutely and relatively more respon- sive to disadvantaged pupils as more of these professional staff are employed in the schools. Since this variable does not distinguish between the kinds of professionals who are included, it is difficult to account for these apparent differences between community types. It is suggestive however, that further investigation of the influence of the variety of specialized staff used to augment instructional programs would be a fruitful under- taking. CHAPTER VII SUMMARY AND IMPLICATIONS A Summary of the Problem and Research Apppoach This thesis has examined the arguments and issues sur- rounding the basis for drawing or redrawing lines of legal authority around groups of peOple and causing the, therefore, and thereafter, to act together as a body politic--the so— called "boundary question." The major dimensions of bound- aries, size and other community characteristics, are inves- tigated to examine their effect on the articulation of group preferences as seen via the distribution of the benefit of a publicly provided service. The school district was selec- ted as a unit of government amenable to this investigation because it is a single-purpose political entity and because measures of the benefits of schooling are likely as good as measures of any other public output. A review of the consolidation reform tradition makes clear that the major arguments on behalf of the consolidation of local government are efficiency in production and in ad- ministration. The former is an economies of scale argument. The latter is a notion of reduced transactions costs when, under consolidation, several local governmental units are combined into one larger government for the region. Without 194 ques app: of c of s buti sis was the tati exis in s Proa unde ces dle disa that to m Part SiSt thus disa deVe 195 questioning whether these in fact exist, the public choice approach was introduced and led to questions of the effect Of consolidation on the articulation of group preferences. In order to be able to examine and interpret the effects Of size and other community characteristics on the distri- bution of benefits of school districts, a framework of analy- sis of the schooling process and the effects of education was deve10ped. This analysis led to three assumptions about the public provision of education necessary to an interpre- tation of this research. First it was assumed that, whether true or not, there exists a widespread belief that an individual's performance in school makes a difference in future achievements and ap- proach to "success." Secondly, it was assumed that generally under the existing educational norm, children with experien- ces substantially different from those of the majority mid- dle class, particularly low SES children, are on average disadvantaged within the schools. The third assumption is that parents or groups of parents in a community will seek to make the schools responsive to their preferences and/or particular needs. Based on the evidence that socioeconomic class is a con- sistent influence on the performance of school children and thus on the distribution of benefits, the conceptual basis for an indicator of school district responsiveness to those disadvantaged under the prevailing educational norm was developed. 196 Drawing on the public choice approach and the analysis of the schooling process, three hypotheses were proposed for testing. The first related to the effects of school district size on the reSponsiveness to the disadvantaged group. The second argued that as power of the disadvantaged increased, school districts would be more responsive to their needs. The third argued that as the preferences and/or need for educational experience within the community are more diverse, the less responsive will schools be to minority groups. A corollary to the third hypothesis argued that as preferences and/or needs of a group are more diverse fgom the majority, the schools will be less responsive to that group. A public choice model of school district behavior was developed and proxy variables were spedified to approximate the general model in the empirical portion of the research. Using a socioeconomic index for individual pupils, a group felt to be relatively disadvantaged was selected. Two direct measures of the responsiveness of school districts to that group were created and a specific "indicator of respon- siveness to see disadvantaged pupils" was selected based on the earlier conceptual analysis. The multiple regression equations approximating the model were fitted to data for town and rural districts, urban fringe districts, city and metropolitan districts and to the pooled data by the ordinary least squares method. 197 A summary of the results, conclusions and implications are presented below dealing first with those issues related to the boundary question and then with those issues of con- cern in the organization and delivery of public schooling. A final section identifies and enumerates questions for future research raised by this investigation The Boundary Question and Associated Issues Results and Conclusions With Respect to Community Size.—- The results of this investigation as reported indicate that over much of the range of the data, as school districts are smaller, they are relatively more responsive to the disad- vantaged group of pupils. This is seen as evidence that increases in community size such as are accomplished by the boundary changes associated with consolidation efforts may result in a decline in the ability of individuals or groups who are not a part of the majority on a particular issue to have their preferences count in the enlarged body politic. Results and Conclusions With Regpect to Gropp Power.-- There was evidence, though it was not clearly conclusive, that as the power of the socio-economically disadvantaged in the community increase, the schools are relatively more re- sponsive to that group. The evidence on the power of blacks in the community tended to reinforce this notion. There was also evidence that when either the poor or blacks were a small proportion of the community (had a small 198 percentage Of total power), increases in their strength re- sulted in a reduction in the responsiveness of the schools to socio—economically disadvantaged pupils. This appeared to be a kind of negative influence of the increased visa- bility of a group which is the likely Object of discrimin- ation. The phenomenon was clearly more distinct in response to blacks than to the poor and can be viewed as evidence of greater discrimination on the basis of race than on the basis of class. These results are seen as evidence that changes in boundaries which modify the relative power of groups within the community affect the ability of those groups to articu- late their preferences for publicly provided services. This would apply to those already disadvantaged within the soc- iety and perhaps also to those who are advantaged. Results and Conclusions with Respoct to Community Con sensus.-- There was evidence that as communities are more heterogeneous in their preferences and/or needs for school- ing they are less responsive to minority groups within the communities. This evidence was both in terms of the relative responsiveness to the disadvantaged group and in terms of the group's mean level of achievement. Furhter, there was evidence that as the consensus of the community is farther away from the preferences or needs of the disadvantaged group, the less responsive are the schools to that group. 199 In terms of the boundary issue, these results support the following two conclusions: (1) As the character of a community is changed by changed in the boundary, groups which are further from the majority of preferences under the new boundary than they were formerly can expect to have greater difficulty in having their preferences count. 1(2) As the character of a community is made more heterogeneous by changes in the boundary, minority groups will have great- er difficulty in having their preferences count. The influence of the proportion of persons in profes- sional occupations indicates that generally the schools are more responsive to disadvantaged pupils as the proportion of these workers in the community increases. This result is both in relative and in absolute terms. There was evi- dence that in urban fringe districts increases in the pro- portion of professionals resulted in an increase in the mean achievement of the low SES pupils but that the achievement of the advantaged pupils increased more. Implications.--The rather clear implications of these results to the boundary question is that while consolidation gay result in lower cost of delivery of public serivces, it may not result in increased satisfaction for all citizens. Stated again, while consolidation of communities may reduce the cost of providing a specific service, it will also ef- fect the distribution of the benefits of the service and will influence the character of the demand for that service. 200 Those who stand to gain by consolidation are those who are more likely to be closer to the majority of the enlarged community, or whose relative power as a group will be im- proved. Those who stand to lose are those whose preferences are further from the majority under consolidation, whose re- lative power is reduced (this would apply to each persOn as an individual), or whose relative power is unchanged but who will have greater difficulty in having their preferences felt in the larger community. When alternative mechanisms for Obtaining economies of scale in production such as contractual arrangements between governmental units are considered, the remaining argument for the consolidation position relates to the potential of consolidation to equalize the per capita resource or tax base for public expenditures of a region. While consoli- dation may indeed be an effective means to reduce location- al inequalities in the public expenditure resource base there are other ways to accomplish this same end. Among the alter- natives to consolidation are methods which would not expose groups, whose interests consolidation is ostensibly to serve under this argument, to the other implications of con- solidation here set forth. Such may be the case of a more effective and equitable revenue sharing program, or in the case of schools, federal or state leadership in the financ- ing of education. The logical extension of these conclusions appears to lead to a society organized into relatively small homogeneous 201 units of local government with overlapping jurisdictions of special interests which do not coincide with the general purpose units. The normative question as to whether that is the kind of society this writer wants to be a part of raises some real dilemmas which will be discussed shortly. Rather along the line of this implication, a recent thesis by Ann Markusen [37] questions the hypothesis ex- pounded by Tiebaut that settlement patterns are primarily the result of citizens "voting with their feet" for the kinds of qualities of services they want. The extension of the Tiebaut view also argues for small, homogeneous govern- mental units. Markusen rather persuasively challenges the Tiebaut view that locational decisions primarily reflect a revealed preference for a package of local services. She argues that in addition to the effects of public service impacts on the well being of a household, there is a substantial redistri- butional impact of income, or benefits from income, toward or away from the household via the public sector process. The latter may be equally as important in locational de- cisions of households as is the package of local public services. Markusen proceeds to show that local governments tend to be responsive to citizens and to Operate in the interests of residents as if they were maximizing the representative resident's utility function including the aversion to a 202 distribution of income or its benefits away from residents. Thus local government is seen as cOnstructing its own market for its services by manipulating entry into the community and public sector tax and expenditure structure. This she argues is the basis of the movement toward community homo- geneity and thus local government tends to "exacerbate the tendency toward stratification" [37] within the society. Although not clearly explicit, Markusen seems to point to the consolidation of communities as a means of amelior- ating social and economic stratification . The findings of this thesis are that dimensions of consolidation appear to increase the inequality in the distribution of the benefits of schooling. This would suggest that consolidation also appears to "exacerbate the tendency towards stratification." Without in any way challenging the substance of Markusen's argument with regard to the locational decisions of families or the behavior of local government in restricting entry, neither of which are the subject of this research, this writer must question whether consolidation is necessarily the most logical or even efficient way of modifying the stratification impact for those who prefer a more equal dis- tribution of benefits of the society. A clear implication of consolidation in a representative government system is a greater concentration of power in the hands of elected or appointed Officials. The effects of a sense of powerlessness (i.e., "you can't fight city hall" 203 but you may be able to fight town hall) on the individual's productivity, creativity, and sense of well being are dif- ficult to evaluate, but also difficult to ignore. To this writer the prospect Of a socity of small homo- geneous groups franchised into local governments is somewhat P less than appealing. However, that situation may be prefer- ed to large heterogeneous groupings where a larger number . of individuals are disenfranchised on the basis of race or 5 class. Further, the organization of society into small homo- i geneous groupings which interact as equals may be a more realistic intermediate step to a truly culturally democratic or pluralistic society which allows groups who are different to find some sense of personal and group integrity and re- late to others from positions of self and mutual respect. Issues Related to the Organization and Delivery of Public Schooling Results and Conclusions With Respect to Instructional Materials.--The instructional expenditures per pupil net of teachers' salaries was the best available measure of expendi- tures on instructional materials. The results indicate that in general as these expenditures increase, school districts are less responsive to the disadvantaged group of pupils. The interpretation of these results is that they are evidence that existing instructional materials are more productive of learning as measured by achievement on advantaged pupils than they are on disadvantaged pupils. 204 Within the normative view on behalf of equality if edu- cational outcomes or even a view that schools should be more class neutral, these results would argue for substantial efforts in the development of materials (technology) more precisely suited to currently disadvantaged pupils. This would require an educational norm which may be more "differ- entiating" than that described by Brookover but which would see these pupil differences as a responsibility of educators rather than as an explanation of unequal outcomes. An example may make this point clearer. The apparent view of educators (as opposed to many linguists) is that black dialect represents a language deficiency rather than a language difference. As a result, the preparation of in- structional materials in black dialect has yet to be much exploited or examined on behalf of that group of children. Results and Conclusions With Regard to Characteristics Of Staff of Schools.--The results with regard to staff characteristics indicate that as school district staffs are more recently imbued with the differentiation educational norm (average teacher experience), that the schools are rela- tively less responsive to disadvantaged pupils. Similarly, the evidence from Urban Fringe districts is that as staffs are more completely imbued with that norm (percentage Masters degrees) the schools are less responsive to disadvantaged pupils. The evidence from the Town and Rural districts and 205 the City-Metro districts with respect to Masters degree teachers is that socio-economically disadvantaged children in those districts are better served as the relative number of Masters degree teachers increase. The implications of this evidence strongly suggest that substantial changes in the philOSOphy of education underly- E ing the educational norm are needed within teacher training institutions, if the schools are to ameliorate inequality in American society or even just cease to exacerbate it. It also suggests as Michaelson [49] points out, that a re-examination of the training and preparation of teachers is of some merit. It may very well be that for some school- ing situations those experiences which lead individuals to become teachers' aides or other such paraprofessionals may indeed be better preparation than is the formalized instruc- tion of colleges of education. In fact, it may very well be that personality traits or other nonacademic characteristics may be more important and more productive resources in ameliorating the inequality of educational outcomes than any formalized training. There is already some evidence of this with regard to the verbal ability or proclivity of teachers and the acquisition of language skills. One further implication that this evidence suggests is that for the short run those wishing to make the Urban Fringe schools more responsive to the disadvantaged might seek to have incentives to teachers to obtain Masters degrees neduced 206 or replaced with incentives to gain experiences which will clearly interrupt the perspectives embodied in the prevailhx; educational norm. The evidence with regard to instructional professionals other than classroom teachers is not clear. There is tenta- tive evidence they serve to increase inequality of education- a1 outcomes and have a negative effect if any on the level of achievement of the low SES group. One possible conclusion of this result is that if such professionals are employed ostensibly to better serve the disadvantaged-~the prOportionate numbers are on the average greater in the City and Metro districts--the strategy is worth questioning. Results and Conclusions with Respect to Class Size.-- These data indicate that as classes are smaller the schools are more responsive to the disadvantaged group of pupils. Implications.--If seeking to reallocate the resources of a school district to achieve a more equal distribution of benefits, the reduction of class size and incentives to teachers to remain in the system are more likely productive strategies than are the employment of nonteaching profession- als, increased expenditures in instructional materials now available, or incentives to teachers for further formalized training. There are several implications to the organization of 207 school districts to be drawn from the results dealing with the boundary issue. The first general implication for school districts is rather similar to that made with regard to the boundaries of local government. If the goal is a more equal distribution of benefits within a school district, the organ- ization of schools into smaller, more homogeneous districts is clearly implied from this research. Such a change does not, however, speak to differences between these small homo- geneous districts. It does spggest that the distribution of benefits between groups when consolidated may_be no more egual than before consolidation. Further, the level of bene- fits of a particular group depending on their power and posi- tion in the community may be lower. The finding of Brown and Saks [16] that school inputs do make a difference in achievement outcomes was also in- ferred from this research. This result would imply that a more equal distribution of resources between schools within districts and between school districts may be more effective in ameliorating inequality in the distribution of educational outcomes in the larger society, than would be the consoli- dation Of school districts. Neither approach is likely to affect in any way the ap- parent differences in the impact of instructional materials. However, the behavior of teachers and school district staffs is more likely to be affected on behalf of the disadvantaged under the smaller, relatively more homogeneous organization 208 Of districts than in consolidated districts. It would seem that a similar argument can be made with regard to schools within a large district. An explicit extension of this implication reflects on the legal efforts of such groups as the NAACP—Legal Defense Fund to achieve a more equitable treatment of pupils of color or race differences. Where these efforts imply redistrict- ing to achieve racial balance, the implications of this re- search are that there is no a priori basis for expecting the desired results to be forthcoming. Depending on the changes in the character of the consolidated district, there may be some reason to believe that the opposite effect will be ob- tained. The argument is similar with regard to legal efforts which result in intradistrict busing. In both cases where busing is the mechanism, the evi- dence is that proportionately larger numbers of black pupils are bused than white pupils. Where this is the case the rise in transport costs to the black community in articulat- ing its preferences can be expected to be higher than to the white community. As a proportion of total resources (time and income) this increased cost may be even more regressive to the black community and could influence the relative ef- fort of the groups and thus the responsiveness of the schools to them. It should be recalled that what is being described here are the effects on the distribution of the acquisition of cognitive skills of fourth graders as measured by achie— 209 -vement scores. The impact on high schoolers' continuance into college or on dropout rates, or other such valid con- cerns have not been dealt with in this research. This implication would seem to argue that those seeking legal redress for unequal benefits from school districts may find that those ends can better be served by several alter- native strategies. The continued pursuit of a legal basis for more equitable educational financing would seem appro- priate. The legal insistence on a more equal distribution of resources between schools within a district, e.g., class size and teacher experience might be implied by this research. Clearly more difficult, but perhaps not beyond the realm of possibility, might be the legal challenge to materials which are obviously not suitable to a particular racial or class group. Even a close judicial defeat on this issue might be highly productive. One further statement on these comments with respect to school busing is apprOpriate. It is with considerable hesi- tation that they are made at all. It appears clear to this writer that there are many who Oppose busing for racist rea- sons. It also appears that there are many who support busing because Of the racist position of those who oppose it. It is appropriate in the world this writer would like to live in to challenge racism. However, the ideology of the major spokespersons for a position does not necessarily change the consequences or implications of actions or policy implied by 210 that position. The personal dilemma seems to be to choose what appears to be a racist position or to throw the baby out with the bath water. Suggestions For Further Research This research has been an investigation of the effect of the structure of schools and school districts on the per- formance of schools with respect to a particular group of pupils. While the results reported do provide some indica- tion of the direction of influence of some of the variables, they cannot be seen as clearly conclusive. Virtually all aspects of this research are suggestive Of areas for future research. Of particular note would be the following: I l. The results with respect to group or community size are clearly tentative. They are, however, interesting in the light of the other work that has been done with respect to economies of scale in education. The results here reported suggest that future work on economies of scale could be made richer by the inclusion of distributive issues. The examination of interactions between group size and boundary rules other than geographical ones as in this re- search would also seem a fruitful area for future efforts. Along that line, there was no consideration in this research effort of a possible interaction between size and homogeneity. Since the apparent affect of many local government ordinances and tax structures are, according to Markusen [37], to make 211 for more homogeneous communities, some future examination of that interaction would seem useful. 2. The notion of a "visibility" effect identified as a plausible explanation of the influence of increases in the power of the poor and blacks when their numbers are small is certainly a worthy research question. ' 3. Following the work of Brown and Saks [16] referred to earlier and the results of this research, there seems a clear implication that research with regard to understanding the education process must focus on individuals as the unit of observation. The ideas embodied in the cognitive style map- ping used in this research are suggestive of possible methods of distinguishing between individual children in a schooling situation and thus identifying more clearly the effects of different inputs to the education process. 4. The results of this research clearly identified that differences between the community type groupings used do in- deed exist. Beyond the discriptive differences there was little explanation of apparent response differences between types of communities in this data. Any effort to examine the production, delivery, and demand for publicly provided services must come to grips with the possibility of system- atic differences in the behavior of communities. This re-’ search has done little to shed light on that question and does suggest this as a fruitful area of endeavor. 212 5. There has been little in this research which is sugges- tive of Pareto better solutions. In fact in some regards this has been a research investigation of conflicts between diverse interests. To the extent that this effort has been productive it is suggestive that perhaps more research which explicitly identifies the interests which will be served may be more useful in elevating public debate than where those issues are disguised within some notion of "efficiency." 6. Intergovernmental service contracting has been suggested as an alternative to consolidation as a means of capturing such benefits as are possible from lower cost methods of production associated with scale or size. Research related to the structure, conduct and performance of contract ar- rangements as well as other arrangements between governmental units would be a most fruitful area of investigation. 7. A part of this research effort has been the development Of an indicator of school district responsiveness to dis- advantaged pupils. The appendix includes a listing of dis- tricts ranked by that indicator. Should this indicator be used in the evaluation of schools or districts, particu- larly if it is used as a basis of allocating resources be- tween districts, an investigation of the changes in the in- dicator over time would be a most important area of ongoing research. APPENDIX APPENDIX I Ranking of Michigan School Districts By Coefficient of Var- iation of 1970 Fourth Grade Mathematics Achievement Test Scores. (Ranking is from lowest value of CV - most respon- sive to disadvantaged pupils — to highest value of CV - least responsive to disadvantaged pupils.) Rank Coefficient School Size - Number Community of Variation District of Voters Type * l .0984 MADISON NO 2 2002 4 2 .1145 ADLANTA COMM 1159 5 3 .1150 BURT TWP 240 5 4 .1171 FOWLER 1091 5 5 .1176 ISHPEMING 5205 3 6 .1182 WHITE PINE 810 5 7 .1199 PEWAMO-WESTPHALIA 2032 5 COMM 8 .1232 CHASSELL TWP 860 5 9 .1241 SUTTONS BAY 1547 5 10 .1244 IRON MOUNTAIN 5727 3 11 .1298 CARNEY-NADEAU 956 5 12 .1356 NORTH MUSKEGON CITY 2665 4 13 .1359 NATIONAL MINE 672 5 14 .1383 IRONWOOD CITY 7441 2 15 .1386 BURR OAK COMM 1098 5 16 .1389 VANDERBILT AREA 779 5 17 .1394 BRECKENRIDGE COMM 3099 5 18 .1431 GROSSE ILE TWP 46832 4 19 .1439 FRANKFORT AREA 2053 5 20 .1447 LAKE LINDEN-HUBBELL 2324 3 21 .1453 BLOOMFIELD HILLS 19042 4 22 .1466 PICKFORD 1090 5 23 .1478 ROCK RIVER TWP 706 5 24 .1480 ROGERS UNION 3727 3 25 .1484 REPUBLIC-MICHIGAMME 1314 5 26 .1491 ALMONT COMM 2398 5 27 .1498 GENESEE NO 6 1093 4 28 .1502 NORWAY-VULCAN AREA 2910 3 29 .1504 INLAND LAKES 1778 5 213 214 APPENDIX I (cont'd.) Rank Coefficient of Variation School District Size - Number of Voters Community Type * 30 .1504 31 .1507 32 .1509 33 .1510 34 .1512 35 .1513 36 .1518 37 .1518 38 .1521 39 .1531 40 .1533 41 .1541 42 .1546 43 .1547 44 .1550 45 .1552 46 .1562 47 .1566 48 .1567 49 .1569 50 .1569 51 .1571 52 .1573 53 .1573 54 .1577 55 .1577 56 .1584 57 .1590 58 .1590 59 .1593 60 .1594 61 .1596 62 .1596 63 .1599 64 .1599 65 .1600 66 .1600 67 .1601 68 .1602 69 .1603 70 .1604 71 .1607 72 .1610 73 .1611 74 .1611 75 .1614 HANCOCK CITY CHARLEVOIX NEW BUFFALO AREA NEGAUNEE CAMDEN-FRONTIER BESSEMER CITY WESTWOOD HEIGHTS EAST GRAND RAPIDS ONEKAMA CONS GOODRICH AREA NEWAYGO NORTH DICKINSON HOMER COMM GERRISH HIGGINS HART GLADWIN RURAL MESICK CONS BEAL CITY MUNISING MACKINAW CITY CARO COMM VESTABURG COMM MEMPHIS COMM ST CHARLES COMM 4138 3707 3521 5240 2022 1771 4472 7654 1876 2596 2525 1144 2564 2282 2199 4318 1357 1148 3220 294 6341 1513 1986 2961 MASON COUNTY CENTRAL 3326 CLAWSON CITY BENDLE FORSYTH HILLMAN COMM SOUTHGATE COMM BIG RAPIDS GALESBURG-AUGUSTA WEST IRON CO BELLAIRE FREMONT MIDLAND CITY ALLEN PARK HOLLAND CITY WHITEFORD WAKEFIELD TWP SUMMERFIELD CONSTANTINE AKRON-FAIRGROVE ST JOSEPH CITY FLUSHING COMM BAD AXE 9520 5174 4246 1510 19660 7755 3815 5844 1505 5904 23181 17970 16349 2194 2079 2075 2988 2245 11825 11149 2565 upwmmmwmwbwwmwwwpmmnpmmmmwmummwmmmmmmmnhwwwwww 215 APPENDIX I (cont'd) Rank Coefficient School Size - Number Community of Variation District of Voters Type * 76 .1615 MONTROSE TWP 3273 5 77 .1617 IDA 3215 5 78 .1619 SCHOOLCRAFT COMM 2036 5 79 .1620 CENTER LINE 14771 4 80 .1621 REESE 3053 5 81 .1626 ROCKFORD 6643 3 82 .1629 MIO-AU SABLE 1870 5 83 .1630 SHELBY 2197 5 84 .1637 OAK PARK CITY 18004 4 85 .1639 ADDISON COMM 2913 5 86 .1640 OSCEOLA TWP 1083 5 87 .1641 EAST CHINA TWP 12044 3 88 .1642 KINGSLEY AREA 1240 5 89 .1644 BIG BAY DE NOC 1199 5 90 .1647 CENTRAL LAKE 1379 5 91 .1650 MERRILL COMM 1950 5 92 .1651 STANDISH-STERLING 4114 5 COMM 93 .1652 FOREST PARK 3331 3 94 .1654 ESSEXVILLE-HAMPTON 4807 4 95 .1660 HARBOR SPRINGS 1988 5 96 .1661 SAGINAW TWP COMM 15703 4 97 .1663 BEAR LAKE 972 5 98 .1664 FITZGERALD 12658 4 99 .1667 DETOUR TWP 859 5 100 .1667 NORTH HURON 1336 5 101 .1667 GRASS LAKE COMM 2545 5 102 .1667 ROCHESTER COMM 16497 3 103 .1667 MASON COUNTY 1553 5 EASTERN 104 .1667 EWEN-TROUT CREEK 1760 5 105 .1670 TRAVERSE CITY 22571 2 106 .1670 PORTAGE TWP 12324 4 107 .1670 CASS CITY 4004 5 108 .1670 YALE 3954 5 109 .1670 CROSWELL-LEXINGTON 5318 3 COMM 110 .1670 WALKERVILLE RURAL 613 5 111 .1670 TRI-COUNTY AREA 3078 5 112 .1673 BRIGHTON AREA 6678 3 113 .1676 LAKEVIEW 13379 4 114 .1676 COLON COMM 2610 5 115 .1679 STEPHENSON 3105 5 116 .1682 PLYMOUTH COMM 23096 2 117 .1682 BYRON COMM 3618 5 118 .1683 FOWLERVILLE COMM 3355 5 APPENDIX I (cont'd.) 216 Rank Coefficient School Size - Number Community of Variation District of Voters Type * 119 .1684 PENTWATER 837 5 120 .1684 EAST LANSING 25957 4 121 .1686 MONABELLA COMM 3514 5 122 .1686 CARSONVILLE COMM 1689 5 123 .1686 BENTLEY COMM 5772 4 124 .1687 SARANAC COMM 2204 5 125 .1687 BRONSON COMM 4150 5 126 .1687 NORTH BRANCH AREA 3207 5 127 .1688 DEARBORN CITY 68317 4 128 .1691 CALUMET 5023 5 129 .1691 WHITE CLOUD 2338 5 130 .1692 CHARLOTTE 8329 3 131 .1692 SPARTA AREA 5012 3 132 .1693 BIRMINGHAM CITY 38188 4 133 .1695 MORENCI AREA 2671 5 134 .1696 MARCELLUS 2785 5 135 .1698 LESLIE 3052 5 136 .1698 LUDINGTON AREA 8427 3 137 .1700 QUINCY COMM 3588 5 138 .1700 MASON CONS 3665 5 139 .1701 RUDYARD TWP 4154 5 140 .1703 CHIPPEWA VALLEY 5603 4 141 .1703 CHELSEA 5414 3 142 .1703 ADAMS TWP 1637 5 143 .1705 ASHLEY COMM 1174 5 144 .1705 GLEN LAKE COMM 1707 5 145 .1707 WILLIAMSTON COMM 3624 3 146 .1709 VASSAR 4026 3 147 .1711 GRAND HAVEN 12912 2 148 .1711 TAYLOR 39842 4 149 .1711 LAKE CITY AREA 1756 5 150 .1712 WEST OTTAWA 8025 5 151 .1712 MANISTIQUE CITY 4559 3 152 .1713 MONA SHORES 13599 4 153 .1714 REDFORD UNION 20932 4 154 .1715 MANISTEE CITY 7108 3 155 .1715 HOUGHTON LAKE 4195 5 156 .1716 TROY 14646 4 157 .1718 CHEBOYGAN 6339 3 158 .1721 KENTWOOD 12239 4 159 .1721 ZEELAND 7550 3 160 .1721 LITCHFIELD COMM 1668 5 161 .1722 NORTHVILLE 8960 3 162 .1723 NAPOLEON 3302 5 163 .1724 BENZIE COUNTY 3559 5 CENTRAL 217 APPENDIX I (cont'd.) Rank Coefficient School Size - Number Community of Variation District of Voters Type * 164 .1725 MANCELONA 1990 5 165 .1725 OVID-ELSIE 4968 5 166 .1726 MARYSVILLE 4386 4 167 .1730 LINCOLN PARK CITY 31406 4 168 .1730 COMSTOCK PARK 4288 4- 169 .1731 NORTH DEARBORN 6742 4 HEIGHTS ‘ 170 .1732 ST JOHNS 8128 3 171 .1734 OTSEGO 6039 3 172 .1736 HASTINGS 8799 3 173 .1737 HUDSONVILLE 5355 3 174 .1740 FENTON AREA 6275 5 175 .1740 STOCKBRIDGE COMM 3852 5 176 .1740 LAPEER 13941 3 177 .1740 ROMEO COMM 6999 3 178 .1743 IMLAY CITY COMM 3909 5 179 .1743 PORTAGE TWP 4681 3 180 .1745 WHITTEMORE-PRESCOTT2241 5 AREA 181 .1747 BERKLEY CITY 20892 4 182 .1749 BEVERTON RURAL 3152 5 183 .1749 MASON 6813 4 184 .1749 ANCHOR BAY 5948 3 185 .1751 SWARTZ CREEK COMM 7062 3 186 .1752 MONTAGUE 3543 3 186 .1753 ROSEVILLE 30303 4 187 .1753 HARBOR BEACH COMM 2817 3 188 .1754 FREELAND COMM 2732 5 189 .1755 PETOSKEY 6217 3 190 .1755 ADRIAN 15502 2 191 .1756 RAPID RIVER 1268 5 192 .1757 MARQUETTE CITY 12308 2 193 .1758 KINGSTON COMM 1441 5 194 .1758 KEARSLEY COMM 9970 4 195 .1759 BULLOCK CREEK 3549 5 196 .1760 PORTLAND 3065 3 197 .1761 CRAWFORD AU SABLE 3357 5 198 .1763 GALIEN TWP 1924 5 199 .1764 CARROLLTON 4550 4 200 .1764 TECUMSEH 6911 3 201 .1764 HOWELL 9915 3 202 .1766 BARAGA TWP 1400 5 203 .1766 MARION 1810 5 204 .1766 ONAWAY AREA 2469 5 205 .1767 KENOWA HILLS 6130 4 206 .1768 VANDYKE COMM 31394 4 APPENDIX I (cont'd.) 218 Rank Coefficient School Size - Number Community Of Variation District of Voters Type * 207 .1768 OWENDALE—GAGE- 1444 5 TOWN AREA 208 .1769 SOUTH LAKE 17472 4 209 .1769 REETHS PUFFER 6836 4 210 .1773 LAINGSBURG COMM 1619 5- 211 .1773 ENGADINE CONS. 1030 5 212 .1774 BLISSFIELD COMM 4378 3 213 .1774 ALLEGAN 6720 3 214 .1775 BELDING AREA 5511 3 215 .1776 LITTLEFIELD 863 5 216 .1777 BATH COMM 2760 4 217 .1777 BROWN CITY COMM 2218 5 218 .1777 MATTAWAN CONS 2993 5 219 .1778 HESPERIA COMM 2069 5 220 .1779 VANDERCOOK LAKE 3273 4 221 .1779 WATERVLIET 3516 5 222 .1780 CHERRY HILL 9238 4 223 .1781 MARSHALL 7407 3 224 .1782 SOUTH REDFORD 19377 4 225 .1782 WARREN CONS 54372 4 226 .1784 STURGIS CITY 8111 3 227 .1785 WHITE PIGEON COMM 2262 5 228 .1786 AIRPORT COMM 7074 5 229 .1788 COMSTOCK 10694 4 230 .1788 JEFFERSON CONS 5808 5 231 .1789 TWIN VALLEY 18145 4 232 .1789 HAZEL PARK 18145 4 233 .1792 OAKRIDGE 3486 5 234 .1792 CONCORD COMM 2099 5 235 .1792 FOREST HILLS 6110 4 236 .1793 CEDAR SPRINGS 4110 5 237 .1793 WEST BRANCH-ROSE 6123 5 CITY AREA 238 .1794 EDWARDSBURG CONS 4440 5 239 .1794 LAKEVIEW 17308 4 240 .1794 WYANDOTTE CITY 24816 4 241 .1795 MORLEY-STANWOOD 2648 5 COMM 242 .1795 CLARKSTON COMM 10378 5 243 .1797 OWOSSO 14390 2 244 .1798 TRENTON 12404 4 245 .1798 DURAND AREA 5299 3 246 .1800 AU GRES-SIMS 1190 5 247 .1801 BUCHANAN 5261 3 248 .1801 SOUTHFIELD 38449 4 249 .1803 CLARE 3767 3 219 APPENDIX I (cont'd.) Rank Coefficient School Size - Number Community of Variation District of Voters Type * 250 .1804 AVONDALE 8675 4 251 .1804 BREITUNG TWP 5172 3 252 .1805 NORTHVIEW 5912 4 253 .1805 LES CHENEAUX COMM 1174 5 254 .1808 ALMA CITY 6870 3~ 255 .1809 LAWRENCE 1600 5 256 .1809 MORRICE AREA 1649 5 257 .1809 MAPLE VALLEY 3619 5 258 .1809 OLIVET COMM 2902 5 259 .1809 ST IGNACE 1615 3 260 .1811 DECKERVILLE COMM 2225 5 261 .1811 FENNVILLE 2823 5 262 .1811 EVART 2648 5 263 .1813 SANDUSKY COMM 3744 5 264 .1815 GRAND LEDGE 10653 4 265 .1815 VICKSBURG COMM 8921 3 266 .1816 SPRING LAKE 4737 4 267 .1818 ELK RAPIDS 1760 5 268 .1818 ORCHARD VIEW 6238 4 269 .1820 LANSE TWP 3034 3 270 .1820 GRANT 3086 5 271 .1820 ONTONAGON AREA 3157 3 272 .1822 LAKESHORE 7466 4 273 .1822 WAVERLY 10145 4 274 .1823 CENTRAL MONTCALM 4153 5 COUNTY 275 .1824 DAVISON COMM 8532 4 276 .1824 NEW HAVEN COMM 3341 5 277 .1825 FRANKENMUTH 3798 5 279 .1827 SAND CREEK COMM 2547 5 280 .1827 BERRIEN SPRINGS 6457 5 281 .1827 ITHACA 4037 3 282 .1827 RIVERVIEW COMM 6562 4 283 .1828 ALCONA COMM 3088 5 284 .1829 CARMAN 17076 4 285 .1829 PINCKNEY COMM 4744 5 286 .1829 CALEDONIA COMM 5693 5 287 .1829 ELKTON-PIGEON-BAY 4931 5 PORT 288 .1831 KENT CITY COMM 2417 5 289 .1831 HOLLY AREA 7133 3 290 .1832 BEECHER 8857 4 291 .1832 FLAT ROCK COMM 3233 4 292 .1833 DRYDEN COMM 1144 5 293 .1833 ST LOUIS 4204 3 294 .1833 KALKASKA 2437 5 220 APPENDIX I (cont'd.) Rank Coefficient School Size - Number Community Of Variation District of Voters Type * 295 .1834 HILLSDALE COMM 6965 3 296 .1835 ALGONAC COMM 6818 3 297 .1835 HUDSON AREA 3486 5 298 .1835 FARMINGTON 35082 4 299 .1836 NORTHWEST JACKSON 9532 5_ TRAIL 300 .1838 BANGOR TWP 8941 4 301 .1838 HOLT 7982 4 302 .1839 HARPER CREEK 6526 4 303 .1840 LAKE ORION COMM 10827 3 304 .1841 DEERFIELD 1109 5 305 .1842 BAY CITY 46670 1 306 .1842 MELVINDALE—N 13375 4 ALLEN PARK 307 .1842 COLDWATER CITY 11419 3 308 .1842 MAYVILLE COMM 2408 5 309 .1844 MERIDIAN 3188 5 310 .1846 FULTON 2449 5 311 .1847 BARK RIVER-HARRIS 1226 5 312 .1847 CHESANING UNION 6106 3 313 .1848 CLIMAX-SCOTTS COMM 2003 5 314 .1850 UTICA COMM 36713 4 315 .1850 WEBBERVILLE COMM 1551 5 316 .1851 OSCODA AREA 8754 5 317 .1851 VAN BUREN 12906 3 318 .1851 LIVONIA 64718 4 319 .1852 ARMADA AREA 2724 5 320 .1853 GRANDVILLE 10500 4 321 .1854 THORNAPPLE-KELLOGG 4353 5 322 .1854 LAKE FENTON 3564 5 323 .1854 HALE AREA 1839 5 324 .1856 GULL LAKE COMM 6431 5 325 .1857 PENNFIELD 5468 4 326 .1857 HARPER WOODS CITY 7539 4 327 .1860 MILAN AREA 7991 3 328 .1861 BRANDYWINE 3634 5 329 .1862 PINCONNING AREA 5695 5 330 .1863 MADISON HEIGHTS 8437 4 331 .1864 MONROE CITY 24569 2 332 .1864 COOPERSVILLE 4847 5 333 .1865 RICHMOND COMM 3842 3 334 .1865 CRESTWOOD 12899 4 335 .1866 WHITMORE LAKE 2292 5 336 .1867 GRAND BLANC COMM 14261 4 337 .1869 HURON VALLEY 16803 3 338 .1869 TAWAS AREA 4725 3 APPENDIX I (cont'd.) 221 Rank Coefficient School Size - Number Community of Variation District of Voters Type * 339 .1871 SAUGATUCK 1751 5 340 .1871 EAST DETROIT 32166 4 341 .1871 KALEVA-NORMAN- 1813 5 DICKSON 342 .1871 HOPKINS 2693 5 343 .1873 BRIDGEPORT COMM 8913 4 344 .1873 . CORUNNA 5839 3 345 .1873 PARCHMENT 4912 4 346 .1876 CLINTON COMM 2264 5 347 .1876 MENOMINEE 8273 2 348 .1876 CLINTONDALE 10115 4 349 .1876 GODFREY LEE 3871 4 350 .1877 GIBRALTER 5713 4 351 .1882 LINDEN COMM 3883 5 352 .1882 LiANSE CREUSE 18702 4 353 .1882 BEDFORD 10963 3 354 .1884 MICHIGAN CENTER 3830 4 355 .1884 GREENVILLE 8147 3 356 .1884 HARTFORD 3130 3 357 .1885 HARTLAND CONS 3894 5 358 .1886 FRUITPORT COMM 6426 4 359 .1887 ESCANABA CITY 12557 2 360 .1888 HEMLOCK 4285 5 361 .1892 WAYLAND UNION 4065 5 362 .1892 POSEN CONS 1343 5 363 .1896 FRASER 12147 4 364 .1896 NORTH ADAMS 1683 5 365 .1897 HARRISON COMM 3789 5 366 .1898 MCBAIN RURAL 1491 5 367 .1899 CADILLAC CITY 8419 2 368 .1900 MARTIN RURAL 1788 5 369 .1902 IONIA 8734 3 370 .1903 HAMTRAMCK CITY 19133 1 371 .1904 MT MORRIS CONS 7062 4 372 .1904 BYRON AREA 1896 5 373 .1905 BIRCH RUN AREA 3662 5 374 .1906 PINE RIVER AREA 2104 5 375 .1907 JONESVILLE COMM 2779 5 376 .1908 RAVENNA 2538 5 377 .1912 DELTON-KELLOGG 3846 5 378 .1913 CLIO AREA 8898 3 379 .1914 NEW LOTHROP AREA 2037 5 380 .1915 DUNDEE COMM 3979 3 381 .1917 FARWELL AREA 3100 5 382 .1920 WESTERN JACKSON CO 4505 5 383 .1921 OXFORD AREA 5496 3 1 APPENDIX I (cont'd.) 222 Rank Coefficient School Size - Number Community of Variation District Of Voters Type * 384 .1921 MILLINGTON COMM 3948 5 385 .1922 COLOMA COMM 5723 5 386 .1922 BRIDGMAN 2346 5 387 .1922 SALINE AREA 6040 3 388 .1925 HOLTON 1752 5 389 .1925 LAKEVIEW COMM 3996 5 390 .1925 MANCHESTER 2585 5 391 .1927 MENDON COMM 1814 5 392 .1928 BRITTON-MACON AREA 1482 5 393 . 1928 WEST BLOOMFIELD TWP 6857 4 394 .1928 PAW PAW 5061 3 395 .1930 EAST JACKSON CO 4911 4 396 .1932 SOUTH LYON COMM 6517 5 397 .1935 COVERT 1714 5 398 .1936 CASSOPOLIS 5763 3 399 .1937 SAULT STE MARIE 10533 2 400 .1937 COLEMAN COMM 2506 5 401 .1937 WATERFORD TWP 33173 5 402 .1940 LAKEVILLE COMM 4840 5 403 .1940 NORTHPORT 767 5 404 .1942 HASLETT 4160 4 405 .1944 PERRY 3396 3 406 .1944 HURON 4485 3 407 .1947 DECATUR 2851 5 408 .1949 INKSTER CITY 8614 4 409 .1949 GARDEN CITY 21348 4 410 .1950 RIVER VALLEY 5493 5 411 .1952 NILES 16151 2 412 .1952 ROCK 683 5 413 .1952 HANOVER-HORTON 2250 5 414 .1954 MT PLEASANT CITY 14554 2 415 .1956 WALLED LAKE CONS 17093 3 416 .1959 UNION CITY COMM 3343 5 417 .1960 ARENAC EASTERN 1295 5 418 .1963 OKEMOS 8180 4 419 .1964 CARSON CITY- 3479 5 CRYSTAL AREA 420 .1967 BRANDON TWP 3123 5 421 .1970 WALDRON AREA 1149 5 422 .1973 PORT AUSTIN 1192 5 423 .1973 PITTSFORD RURAL 2075 5 424 .1976 BLOOMINGDALE 2673 5 425 .1977 ROYAL OAK CITY 55259 4 426 .1978 BUENA VISTA 6955 4 427 .1980 PORT HURON CITY 34274 2 428 .1983 ECORSE 9315 4 223 APPENDIX I (cont'd.) Rank Coefficient School Size - Number Community of Variation District of Voters Type * 429 .1984 SHEPHERD 3084 5 430 .1988 CAPAC COMM 2551 5 431 .1992 CLARENCEVILLE 11213 4 432 .1992 THREE RIVERS 7976 3 433 .1992 DE WITT 2454 4 434 .1992 WYOMING 16602 4 435 .1996 GODWIN HEIGHTS 7169 4 436 .1996 EATON RAPIDS 6726 3 437 .1996 ALPENA CITY 16752 2 438 .1996 ALBION CITY 8681 2 439 .1996 HAMILTON COMM 4523 5 440 .2000 MUSKEGON HEIGHTS 9483 1 CITY 441 .2000 ANN ARBOR CITY 69700 1 442 .2000 GOBLES 2199 5 443 .2004 SOUTH HAVEN 7653 3 444 .2008 ONSTED COMM 3260 5 445 .2012 CHIPPEWA HILLS 5092 5 446 .2013 WESTWOOD COMMUNITY11096 4. 447 .2013 WOODHAVEN 3450 4 448 .2024 READING COMM 2207 5 449 .2026 TEKONSHA COMM 1294 5 450 .2035 LINCOLN CONS ‘5235 5 451 .2036 COLUMBIA 4411 5 452 .2036 DOWAGIAC UNION 8887 3 453 .2037 ATHERTON COMM 4219 4 454 .2041 WAYNE COMM 45317 4 455 .2041 VAN BUREN 12906 3 456 .2049 SPRINGPORT 2575 5 457 .2052 MARLETTE COMM 3545 5 458 .2054 CHAMPION-HUMBOLT- 544 5 SPURR 459 .2058 LOWELL AREA 6101 3 460 .2058 WHITEHALL 4334 3 461 .2061 LAKEWOOD 6073 5 462 .2065 LAMPHERE 11818 4 463 .2070 DEXTER COMM 4194 5 464 .2074 DETROIT CITY 939485 1 465 .2074 CENTREVILLE 2189 5 466 .2075 PELLSTON 1572 5 467 .2076 PLAINWELL COMM 5682 3 468 .2077 BANGOR PUBLIC 3643 5 469 .2081 YPSILANTI CITY 23908 2 470 .2086 _ EAU CLAIRE 2305 5 471 .2086 PECK COMM 1185 5 472 .2087 JOHANNESBURG- 1532 5 CENTRAL 224 APPENDIX I (cont'd.) Rank Coefficient of Variation School District Size - Number of Voters Community Type * 473 .2087 BATTLE CREEK 30075 1 474 .2088 ATHENS AREA 1913 5 475 .2093 SPRINGFIELD CITY 2169 4 476 .2094 PONTIAC CITY 58680 1 477 .2096 WILLOW RUN 8169 4 478 .2105 RIVER ROUGE 10294 4 479 .2107 FERNDALE CITY 23990 4 480 .2107 KALAMAZOO CITY 62042 1 481 .2113 LAWTON COMM 2032 5 482 .2113 ROMULUS COMM 12578 3 483 .2115 SAGINAW CITY 54036 1 484 .2117 UBLY COMM 2538 5 485 .2119 MUSKEGON CITY 27926 1 486 .2122 MT CLEMENS COMM 14969 2 487 . 2124 HIGHLAND PARK CITY 21913 1 488 .2128 KELLOGGSVILLE 5601 4 489 .2131 GAYLORD COMM 4445 3 490 .2136 LANSING 82319 1 491 .2149 FLINT 110605 1. 492 .2175 BELLEVUE COMM 2331 5 493 .2188 JACKSON UNION 41055 1 494 .2192 POTTERVILLE 1132 4 495 .2206 GRAND RAPIDS 116398 1 496 .2210 BENTON HARBOR 23645 2 497 .2216 DANSVILLE 1833 5 498 .2319 MANTON CONS 1582 5 499 .2330 BRIMLEY 1055 5 500 .2408 BALDWIN. 2643 5 * 1 Metropolitan Districts 2 City Districts 3 Town Districts 4 Urban Fringe Districts 5 Rural Districts BIBLIOGRAPHY 'I I .. .J‘ iall [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] BIBLIOGRAPHY Atkinson, Anthony B. "On the Measurement of Inequality." Journal of Economic Theory 2 (Sept. 1970) :244-263. Averch, Harvey; Carroll, Stephen J.; Donaldson, Theodore S.; Kiesling, Herbert J.; and Pincus, John. How Effective Is Schooling? A Critical Review and Syn- thesis of Research Findings. Santa Monica, Cali- fornia: The Rand Corporation, 1972. Ball, Samuel and Bogatz, Gerry Ann. The First Year of Sesame Street: An Evaluation. Princeton, N.J.: Educational Testing Service, 1970. Becker, Gary S. Human Capital. New York: Columbia Uni- versity Press, 1964. Benson, Charles. State and Local Fiscal Relationships in Public Education in California. Report of the Senate Fact Finding Committee on Revenue and Taxa- tion, Senate of the State of California, Sacramento, March 1965. Bish, Robert L. The Public Economy of Metropolitan Areas. Chicago: Markham Publishing Co., 1971. Bish, Robert L. and Ostrom, Vincent. Understanding Urban Government. Washington, D.C.: American Enterprise Institute for Public Policy Research, 1973. Black, Duncan. The Theory of Committees and Elections. Cambridge, England: Cambridge University Press, 1958. Blau, Peter M. and Duncan, Otis Dudley. The American Occupational Structure. New York: John Wiley & Sons, 1967. Bogatz, Gerry Ann and Ball, Samuel. The Second Year of Sesame Street: A Continuing Evaluation, Vol. 1 & 2. Princeton, N.J.: Educational Testing Service, 1971. 225 [ll] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] 226 Booth, David A. Metropolitics: The Nashville Consoli- dation. East Lansing: Institute for Community Development and Services, Michigan State Universi- ty, 1963. Bowles, Samuel. "Understanding Unequal Economic Oppor- tunity." The American Economic Review 63 (May 1973); 346-356. Bowles, Samuel and Gintis, Herbert. "I.Q. in the U.S. ta Class Structure." Social Policy 3 (Nov.-Dec. ' 1972): 63-75. ‘ Brookover, Wilbur B.; Gigliotti, Richard J.; Henderson, Ronald D.; Niles, Bradley E.; and Schneider, Jeffrey M. "Quality of Educational Attainment, ,_ Standardized Testing, Assessment, and Account- El ability." Uses of the Sociology of Education, The Seventy-third Yearbook of the National Society for the Study of Education. Chicago: The Uni- versity of Chicago Press, 1974. Brown, Byron. "Achievement, Costs and the Demand for Public Education." Western Economic Journal 10 (June l972):193-219. Brown, Byron and Saks, Daniel. "The Production and Dis- tribution of Cognitive Skills Within Schools." Journal of Political Economy, (forthcoming). Buchanan, James M. Theory of Public Choice. Ann Arbor: University of Michigan Press, 1972. Buchanan, James M. and Tullock, Gordon. The Calculus of Consent. Ann Arbor: University of Michigan Press, 1962. Burkhead, Jessee; Fox, Thomas G.: and Holland, John W. Input and Output in Large City High Schools. Syracuse: Syracuse University Press, 1967. Bussard, Mary Rachel. "Economy, Efficiency and Equal- ity: The Myths of Rural School Consolidation." Qualifying Paper, Harvard University, 1972. Cohn, Elchanan. "Economies of Scale in Iowa High School Operations." The Journal of Human Resources 3 (Fall 1968):422-434. [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] 227 Coleman, James 8.; Campbell, Ernest A.; Hobson, Carol J.; McPartland, James: Mood, Alexander M.; . Weinfeld, Frederic D.; and York, Robert L. Egual- ity of Educational Opportunity. Washington, D.C.: National Center for Educational Statistics, 1966. Committee for Economic Development. Modernizing Local Government. New York: Committee for Economic DevelOpment, 1966. Cook, Walter W. and Clymer, Theodore. "Acceleration and Retardation." Individualized Instruction: The Sixtyfirst Yearbook of the National Society for the Study of Education. Chicago: University of Chicago Press, 1962. Dahl, Robert A. and Tufte, Edward R. Size and Demo- cracy. Stanford: Stanford University Press, 1973. DeLoach, Joseph F. Cognitive Styles and Cognitiye Dissonance. Oakland, Michigan: Oakland Community College Press, 1970. Gintis, Herbert. "Education, Technology, and the Char- acteristics of Worker Productivity." The American Economic Review 61 (May 1971):266-279. Hill, Joseph E. Michigan: The Educational Sciences. Oakland, Oakland Community College Press, 1971. Hirsch, Werner Z. "Expenditure Implications of Metro- politan Growth and Consolidation." The Review of Economics and Statistics 41 (Aug. 1959):232—240. Hirschman, Albert 0. Exit, Voice and Loyalty. ridge, Massachusetts: 1970. Cam- Harvard University Press, Jencks, ChristOpher; Smith, Marshall; Acland, Henry; Bane, Mary Jo; Cohen, David; Gintis, Herbert; Heyns, Barbara; and Michelson, Stephan. Inequality. New York: Harper & Row, 1972. Jensen, Arthur R. "How Much Can We Boost IQ and Scho- lastic Achievement?" Harvard Educational Review 39 (Winter 1969):1-123. Edward J. New York: Kane, Economic Statistics and Econometrics. Harper & Row, 1968. [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] 228 Katzman, Martin T. "Distribution and Production In a Big City Elementary School System." Yale Econ-v omic Essays 8 (Spring 1968):201-256. Kementa, Jan. Elements of Econometrics. New York: The Macmillan Co., 1971. Kiesling, Herbert J. "Measuring A Local Government Service: A Study of School Districts in New York State." The Review of Economics and Statistics 19 (Aug. 1967):356-367.. Markusen, Ann Roell. "The Economics of Social Class and Metropolitan Local Government." Ph.D. disser- tation, Michigan State University, 1974. Michigan State Department of Housing and Urban Develop- ment. Report of the Governor's Special Commission on Local Government (1972). National Center for Educational Statistics. User's Manual for 1970 Census Fourth Count (Population) School District Tapes. Washington, D.C.: Nation- al Center for Educational Statistics, 1973. Osburn, Donald D. "Economies of Size Associated With Public High Schools." The Review of Economics and Statistics 52 (Feb. 1970):113-115. Ostrom, Vincent and Ostrom, Elinor. "Public Choice: A Different Approach to The Study of Public Admini- stration." Public Administration Review 31 (March 1971):203-216. Reinhold, Robert. "Study Questions Belief That Home Is More Vital To Pupil Achievement Than The School." The New York Times, 18 November 1973, p. 59. Reiss, Albert J. York: Occupations and Social Status. New The Free Press of Glencoe, Inc., 1961. John. ations." 48 (Aug. Riew, "Economies of Scale In High School Oper- The Review of Economics and Statistics l966):280-287. Samuels, Warren J. "Welfare Economics, Power, and Pro- perty." Edited by Wunderlich and Gibson, Perspeg- tives of Property. University Park, Penn.: The Pennsylvania State University Press, 1972. [46] [47] [48] [49] [50] [51] [52] [53] 229 Schmandt, Henry J. and Stevens, Ross G. "Measuring Municipal Output." The National Tax Journal 13 (Dec. 1960):369-375. Schmid, A. Allan and Faas, Ronald. A Research Approach to Institutionalization and Administration of Agrarian Develgpment. Michigan State University, Department of Agricultural Economics, 1974. (Mimeographed). Thomas, James Alan. "Efficiency in Education: A 1 Study of the Relationship Between Selected Inputs ° and Mean Test Scores in a Sample of Senior High Schools." Ph.D. Dissertation, Stanford University, 1962. (microfilm) U.S. Department of Health, Education, and Welfare. 3“ "The Association of Teacher Resourceness With E Children's Characteristics." Do Teachers Make A Difference by Stephan Michelson. Washington, D.C.: U.S. Government Printing Office, 1970. Verburg, Kenneth and Church, Gerald M., eds. Proceed- ings of the Public Policy Forum on the Alternatives for Michigan Local Government. East Lansing: Institute for Community DevelOpment, Michigan State University, 1973. Warren, Robert 0. Government in Metropolitan Regions: A Reappraisal of Fractionated Political Organiza- tion. Davis: Institute of Governmental Affairs, University of California, 1966. Wasser, Laurence. Are Marks Larks? Oakland, Michigan: Oakland Community College Press, 1971. White, Fred and Tweeten, Luther. "Optimal School District Size Emphasizing Rural Areas." American Journal of Agricultural Economics 55 (Feb. 1973): 45-53.