Reformism and Public Policies in American Cities Author(S): Robert L

Reformism and Public Policies in American Cities Author(s): Robert L. Lineberry and Edmund P. Fowler Source: The American Political Science Review, Vol. 61, No. 3 (Sep., 1967), pp. 701-716 Published by: American Political Science Association Stable URL: http://www.jstor.org/stable/1976089 Accessed: 29/09/2010 12:45

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=apsa.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

American Political Science Association is collaborating with JSTOR to digitize, preserve and extend access to The American Political Science Review.

http://www.jstor.org REFORMISM AND PUBLIC POLICIES IN AMERICAN CITIES* ROBERT L. LINEBERRY University of Texas AND EDMUND P. FOWLER York University

A decade ago, political scientists were istics of their governments. Our central re- deploring the "lost world of municipal govern- search concern is to examine the impact of ment" and calling for systematic studies of political structures, reformed and unreformed, municipal life which emphasized the political, on policy-making in American cities. rather than the administrative, side of urban political life.' In recent years, this demand has I. POLITICAL CULTURE, REFORMISM AND been generously answered and urban politics POLITICAL INSTITUTIONS is becoming one of the most richly plowed fields The leaders of the Progressive movement in of political research. In terms originally intro- the United States left an enduring mark on the duced by David Easton,2 political scientists American political system, particularly at the have long been concerned with inputs, but state and municipal level. In the states, the more recently they have focused their attention primary election, the referendum, initiative on other system variables, particularly the and recall survive today. The residues of this political culture3 and policy outputs of munici- Age of Reform,6 as Richard Hofstadter called it, pal governments.4 persist in municipal politics principally in the The present paper will treat two policy out- form of manager government and at-large and puts, taxation and expenditure levels of cities, nonpartisan elections. The reformers were, to as dependent variables. We will relate these borrow Banfield and Wilson's phrase, the origi- policy choices to socio-economic character- nal embodiment of the "middle class ethos" in istics of cities and to structural character- American politics. They were, by and large, White Anglo-Saxon Protestants reacting to the * The authors are indebted to Professors Rob- politics of the party machine, which operated ert T. Daland, James W. Prothro, William R. by exchanging favors for votes.' Keech and James Q. Wilson for comments on an It is important that we understand the earlier draft of this paper. For assistance in statis- ideology of these reformers if we hope to be tical and methodological questions, the advice of able to analyze the institutions which they Professor Hubert Blalock and Mr. Peter B. Har- created and their impact on political decisions. kins has been invaluable. The authors, of course, The reformers' goal was to "rationalize" and assume responsibility for all interpretation and "democratize" city government by the sub- misinterpretation. stitution of "community oriented" leadership. 1 Lawrence J. R. Herson, "The Lost World of To the reformers, the most pernicious char- Municipal Government," this REvIEw, 51 (June, acteristic of the machine was that it capitalized 1957), 330-345; Robert T. Daland, "Political on socio-economic cleavages in the population, Science and the Study of Urbanism," ibid., 491- playing on class antagonisms and on racial and 509. religious differences. Ernest S. Bradford, an 2 David Easton, "An Approach to the Analysis early advocate of commission government of Political Systems," World Politics, 9 (April, 1957), 383-400. 3 Edward C. Banfield and James Q. Wilson, REvIEw, 60 (June, 1966), 306-326; Edgar L. City Politics (Cambridge: Harvard University Sherbenou, "Class, Participation, and the Council- Press and the MIT Press, 1963); see also James Q. Manager Plan," Public Administration Review, Wilson and Edward C. Banfield, "Public-Regard- 21 (Summer, 1961), 131-135; Lewis A. Froman, ingness as a Value Premise in Voting Behavior," Jr., "An Analysis of Public Policies in Cities," this REvIEW, 58 (December, 1964), 876-887. Journal of Politics, 29 (February, 1967), 94-108. 4 See, for example, Thomas R. Dye, "City- 5 (New York: Alfred A. Knopf, 1955.) Suburban Social Distance and Public Policy," 6 John Porter East, Council Manager Govern- Social Forces, 4 (1965), 100-106; Raymond Wol- ment: The Political Thought of Its Founder, Rich- finger and John Osgood Field, "Political Ethos ard S. Childs (Chapel Hill: University of North and the Structure of City Government," this Carolina Press, 1965), p. 18. 701 702 THE AMERICAN POLITICAL SCIENCE REVIEW

with at-large elections, defended his plans for penetrating analysis of American suburbia. at-large representation on grounds that The "no-party politics of suburbia" is char- ... under the ward system of governmental repre- acterized by "an outright reaction against sentation, the ward receives the attention, not in partisan activity, a refusal to recognize that proportion to its needs but to the ability of its there may be persistent cleavages in the representatives to 'trade' and arrange 'deals' with electorate and an ethical disapproval of perma- nent group fellow merribers. . . . Nearly every city under the collaboration as an appropriate aldermanic system offers flagrant examples of this means of settling disputes."'0 This ideological vicious method of 'part representation.' The opposition to partisanship is a product of a commission form changes this to representation of tightly-knit and homogeneous community, for the city as a whole.7 "nonpartisanship reflects a highly integrated community life with a powerful capacity to The principal tools which the reformers induce conformity."" picked to maximize this "representation of Considerable debate has ensued over both the city as a whole" were the commission, and the existence and the consequences of these later the manager, form of government, the two political ethics in urban communities. nonpartisan election and the election at-large. Some evidence has supported the view that City manager government, it was argued, reformed governments'2 are indeed found in produced a no-nonsense, efficient and busi- cities with higher incomes, higher levels of ness-like regime, where decisions could be education, greater proportions of Protestants implemented by professional administrators and more white-collar job-holders. Schnore rather than by victors in the battle over and Alford, for example, found that "the spoils. Nonpartisan elections meant to the popular image of the manager city was veri- reformer that state and national parties, whose fied; it does tend to be the natural habitat of issues were irrelevant to local politics anyway, the upper middle class." In addition, manager would keep their divisive influences out of cities were "inhabited by a younger, more municipal decision-making. Nonpartisan elec- mobile population that is growing rapidly."13 tions, especially when combined with elec- More recently, Wolfinger and Field cor- tions at-large, would also serve to reduce the related socio-economic variables-particularly impact of socio-economic cleavages and minor- ethnicity and region-to political structures. ity voting blocs in local politics. Once estab- They concluded that "the ethos theory is ir- lished, these institutions would serve as bas- relevant to the South . . . inapplicable to the tions against particularistic interests. West . . . fares badly in the Northeast . . . " Banfield and Wilson have argued that the and that support for the theory in the Midwest "middle class ethos" of the reformers has be- was "small and uneven."'4 Region proved to be come a prevalent attitude in much of political a more important predictor of both government life. The middle class stands for "public forms and of policy outputs like urban renewal regarding" virtues rather than for "private expenditures than did the socio-economic com- regarding" values of the ethnic politics of position of the population. machines and bosses. The middle class searches In our view, it is premature to carve a head- for the good of the "community as a whole" rather than for the benefit of particularistic 10 Robert C. Wood, Suburbia: Its People and interests.8 Agger, Goldrich and Swanson, in Their Politics (Boston: Houghton Mifflin Co., their study of two western and two southern 1959), P. 155. communities have documented the rise of a 1 Ibid., p. 154. group they call the "community conserva- 12 We refer to cities characterized by commis- tionists," who "see the values of community sion or manager government, nonpartisan elec- life maximized when political leadership is tions, and at-large constituencies as "reformed." exercised by men representing the public at Our use of the term is historical and no value posi- large, rather than 'special interests.'"' Robert tion on reformism's merits is intended. To refer to Wood has taken up a similar theme in his reformed cities as "public regarding" or "middle class" is, it seems, to assume what needs to be 7Ernest S. Bradford, Commission Government proved. in American Cities (New York: Macmillan, 1911), 13 Leo Schnore and Robert Alford, "Forms of p. 165. Government and Socio-Economic Characteristics 8 Banfield and Wilson, op. cit., p. 41. of Suburbs," Administrative Science Quarterly, 8 9 Robert Agger, Daniel Goldrich, and Bert E. (June, 1963), 1-17. See also the literature cited in Swanson, The Rulers and the Ruled (New York: Froman, op. cit. John Wiley and Sons, 1964), p. 21. 14 Wolfinger and Field, op. cit., pp. 325-326. REFORMISM AND PUBLIC POLICIES IN AMERICAN CITIES 703

TABLE 1. INDEPENDENT VARIABLES variables were gathered for each of the two hundred cities in the sample.16 1. Population, 1960 Our principal theoretical concern is with 2. Per cent population increase or decrease, the consequences of variations in the structural 1950-60 characteristics of form of government, type of 3. Per cent non-white constituency and partisanship of elections. The 4. Per cent of native population with foreign variable of government form is unambiguous. born or mixed parentage Except for a few small New England towns, all 5. Median income American cities have council-manager, mayor- 6. Per cent of population with incomes below council or commission government. There is, $3000 however, somewhat more ambiguity in the 7. Per cent of population with incomes above classification of election type. By definition, a $10,000 "nonpartisan election is one in which no candi- 8. Median school years completed by adult date is identified on the ballot by party affilia- population tion."'7 The legal definition of nonpartisan- 9. Per cent high school graduates among adult ship conceals the wide variation between population Chicago's and Boston's nominal nonpartisan- 10. Per cent of population in white collar ship and the more genuine variety in Minne- occupations apolis, Winnetka and Los Angeles.'8 We will 11. Per cent of elementary school children in quickly see, though, that formal nonpartisan- private schools ship is not merely an empty legal nicety, but 12. Per cent of population in owner-occupied that there are very real differences in the politi- dwelling units cal behavior of partisan and nonpartisan cities, even though we are defining them in legal terms only.'9 stone for the ethos theory. It is our thesis that Our classification of constituency types into governments which are products of the reform only two groups also conceals some variation movement behave differently from those in the general pattern. While most cities use which have unreformed institutions, even if either the at-large or the ward pattern of the socio-economic composition of their popu- constituencies exclusively, a handful use a lation may be similar. Our central purpose is to combination of the two electoral methods. determine the impact of both socio-economic variables and political institutions (structural 16 We used a random sample of 200 of the 309 variables) on outputs of city governments. American cities with populations of 50,000 or By doing this, we hope to shed some additional more in 1960. All information on the forms of illumination on the ethos theory. government and forms of election are drawn from The Municipal Yearbook, 1965, op. cit. II. RESEARCH DESIGN 17 Banfield and Wilson, op cit., p. 151. 18 Variables. The independent variables used For Minneapolis, see Robert Morlan, "City in this analysis, listed in Table 1, consist of Politics: Free Style," National Municipal Review, relatively "hard" data, mostly drawn from 48 (November, 1949), pp. 485-490; Winnetka, the U. S. census.'5 These variables were selected Banfield and Wilson, op. cit., p. 140; Los Angeles, because they represent a variety of possible Charles G. Mayo, "The 1961 Mayoralty Election social cleavages which divide urban popula- in Los Angeles: The Political Party in a Non- tions-rich vs. poor, Negro vs. White, ethnic partisan Election," Western Political Quarterly, 17 vs. native, newcomers vs. old-timers, etc. We (1964), 325-339. assume that such social and economic char- 19 At least one other variable may produce a acteristics are important determinants of given institutional form in a city-the legal re- individual and group variations in political quirements of a state government, which vary preferences. Data oi1 each of these independent from state to state and may even vary for different kinds of cities within the same state. We have not taken account of this variable because sys- 15 The source for the first nine variables is The tematic information on comparative state re- City and County Data Book (Washington: United quirements in this area was unavailable to us. States Bureau of the Census, 1962). For the last However, Wolfinger and Field consulted several three variables, the source is Orin F. Nolting and experts and eliminated cities which are not given David S. Arnold (eds.), The Municipal Yearbook free choice over their institutions. Nevertheless, a 1965 (Chicago: International City Managers' comparison of our figures with theirs revealed no Association, 1965), pp. 98 ff. important differences. 704 THE AMERICAN POLITICAL SCIENCE REVIEW

For our purposes, we classified these with This hypothesis focuses on the intention of district cities. the reformers to minimize the role of particu- The dependent variables in this study are laristic interests in policy making. two measures of public policy outputs. A growing body of research on local politics has III. REFORMED AND UNREFORMED CITIES: utilized policy measures as dependent vari- A COMPARISON ables.20 The present research is intended to The economic and social contrasts between further this study of political outputs by relat- reformed and unreformed cities have been the ing socio-economic variables to expenditure subject of much research,22 and for our pur- and taxation patterns in cities with varying poses we may be brief in our treatment. We political structures. divided the independent variables into three The dependent variables are computed by a groups, one measuring population size and simple formula. The measure for taxation was growth, a second containing social class indi- computed by dividing the total personal cators and a third including three measures of income of the city into the total tax of the social homogeneity. The means and standard city, giving us a tax/income ratio. Similarly, deviations for each variable by institutional dividing expenditures by the city's aggregate category are found in Table 2. personal income gave us an expenditure/in- It should initially be noted that population come ratio as the measure for our second size and growth rate fairly clearly separate the dependent variable. These measures, while reformed from the unreformed cities. As Alford admittedly imperfect,2' permit us to ask how and Scoble have amply documented,23 the much of a city's income it is willing to commit larger the city, the greater the likelihood of its for public taxation and expenditures. being unreformed; the faster its growth rate, Hypothesis. Much of the research on city the more likely a city is to possess manager politics has treated reformed institutions as government, nonpartisan and at-large elec- dependent variables. Although we shall briefly tions. These differences are largely accounted examine the social and economic differences for by the fact that very large cities are most between reformed and unreformed cities, our likely to (1) have unreformed institutions and principal concern will be to explore the conse- (2) be stable or declining in population. Since quences for public policy of political institu- neither of these variables emerged as particu- tions. From our earlier discussion of the politi- larly important predictors of our output cal culture of cities we hypothesized that: variables, we relegated them to secondary 1. The relationship between socio-economic importance in the rest of the analysis. cleavages and policy outputs is stronger in The data in Table 2 indicate that reformed unreformed than in reformed cities. cities (at least those over 50,000) do not appear to be "the natural habitat of the upper middle 20 See footnote 4, sutpra. class." While reformed cities have slightly more 21 We recognize that these are only rough indi- educated populations and slightly high pro- cators of city finance policies. Definitions of taxa- portions of white collar workers and home tion vary from city to city and what may be ownership, unreformed cities have generally financed from taxes in one city may be financed high incomes. In any case, whatever their from fees in another. Expenditures present a more direction, the differences are not large. What complex problem because the types and amounts is striking is not the differences between the of state transfer payments vary from state to cities but the similarities of their class composi- state according to state laws, the division of gov- tion. ernmental labor in a state, the incomes and sizes of cities, not to mention political factors at the 22 See, for example, Robert Alford and Harry state level. We think it important, however, that Scoble, "Political and Socio-Economic Character- our independent variables explain a large propor- istics of American Cities," The Municipal Year- tion of the variation in municipal outputs as we book 1965, op. cit., pp. 82-97; Sherbenou, op. cit.; measured them. No doubt one could explain an John H. Kessel, "Governmental Structure and even larger proportion of the variation in mea- Political Environment," this REVIEw, 56 (Sep- sures which specify different functional responsi- tember, 1962), 615-620. bilities of cities. At least these measures constitute 23 Alford and Scoble, op. cit. The particularly a starting point, and we hope others will improve large differences found between the populations of on them. reformed and unreformed cities reflect the fact The source of our output measures was the that New York City and several other urban County and City Data Book, op. cit. giants are included in the sample. REFORMISM AND PUBLIC POLICIES IN AMERICAN CITIES 705

TABLE 2. COMPARISON OF THE MEANS (AND STANDARD DEVIATIONS) OF SOCIO-ECONOMIC CHARACTERISTICS OF REFORMED AND UNREFORMED CITIES

Government Type Independent Variable Mayor-Council Manager Commission

Population: Population (103) 282.5 (858.6) 115.7 (108.0) 128.6 (115.2) % Change, 1950-60 36.4% (118.8) 64.1% (130.4) 18.5% (36.7) Class: Median Income $6199. (1005.0) $6131. (999.6) $5425. (804.4) % under $3000 15.3% (7.0) 17.3% (6.9) 21.5% (7.9) % over $10,000 16.9% (7.2) 17.5% (6.7) 12.5% (3.7) % high school graduates 40.7% (10.8) 48.1% (8.9) 41.6% (10.4) Median Education (yrs.) 10.7 (1.1) 11.4 (.89) 11.0 (2.1) % Owner-Occupied Dwelling Units 54.9% (15.1) 57.3% (13.6) 54.6% (13.7) % White collar 44.1% (9.0) 48.1% (7.1) 44.2% (7.6)

Homogeneity: % Nonwhite 10.6% (11.5) 11.6% (10.8) 16.5% (14.9) % Native with Foreign Born or Mixed Parentage 19.7% (9.9) 12.4% (8.3) 11.7% (10.7) % Private School Attendance 23.5% (11.9) 15.3% (11.8) 16.6% (11.8) N=85 N=90 N=25

Election Type Independent Variable Partisan Nonpartisan

Population: Population (103) 270.8 (1022.1) 155.8 (198.7) % Population Increase 1950-1960 17.1 (40.1) 58.3% (136.1)

Class: Median Income $5996 (904.5) $6074 (1045.5) % under $3000 16.8% (7.1) 17.2% (7.2) % over $10,000 16.1% (6.1) 16.7% (7.0) % High School Graduates 40.5% (9.2) 45.3% (10.6) Median Education (yrs.) 10.6 (1.1) 11.2 (1.2) % Owner-Occupied Dwelling Units 51.5% (14.4) 57.7% (13.8) % White Collar 43.5% (7.5) 46.7% (8.3)

Homogeneity: % Nonwhite 13.0% (11.9) 11.5% (11.8) % Native with Foreign Born or Mixed Parentage 17.5% (10.7) 14.7% (9.6) % Private School Attendance 24.1% (13.6) 16.9% (11.3) N=57 N=143

Constituency Type

Independent Variable District At-Large

Population: Population (103) 246.9 (909.8) 153.6 (191.2) % Population Increase 1950-1960 23.1% (36.4) 59.1% (143.7) 706 THE AMERICAN POLITICAL SCIENCE REVIEW

TABLE 2-(Continued)

Constituency Type

Independent Variables District At-Large

Class: Median Income $6297 (965.2) $5942 (1031.9) % under $3000 14.7% (6.5) 18.2% (7.6) O over $10,000 17.7% (7.1) 16.0% (6.6) % High School Graduates 43.6% (10.9) 44.4% (10.4) Median Education (yrs.) 10.9 (1.1) 11.2 (1.2) % Owner-Occupied Dwelling Units 55.1% (14.4) 56.9% (14.5) % White Collar 45.2% (9.4) 46.3% (7.5)

Homogeneity: % Non white 9.8% (10.6) 13.0% (12.3) % Native with Foreign Born or Mixed Parentage 18.9% (9.4) 13.4% (9.7) % Private School Attendance 23.2% (12.5) 16.6% (11.7) N=73 N=127

Homogeneity is easily one of the most am- First, Wolfinger and Field argued that what biguous terms in the ambiguous language of differences there are between unreformed and the social sciences. We have followed Alford reformed cities disappear when controls for and Scoble who used three measures of region are introduced: "The salient conclusion homogeneity: for ethnicity, the per cent of to be drawn from these data is that one can do population native born of foreign born or a much better job of predicting a city's political mixed parentage; for race, the per cent non- form by knowing what part of the country it is white; and for religious homogeneity, the per in than by knowing anything about the com- cent of elementary school children in private position of its population."24 Since regions schools. The last measure, while indirect, was have had different historical experiences, con- the only one available, since data on religious trols for region are essentially controls for affiliation are not collected by the Census history, and more specifically, historical varia- Bureau. tion in settlement patterns. The problem with With the exception of race, reformed cities this reasoning, however, is that to "control" appear somewhat more homogeneous than for "region" is to control not only for history, unreformed cities. While the differences in but for demography as well: to know what homogeneity are more clear-cut than class region a city is in is to know something about differences, this hardly indicates that reformed the composition of its population. Geographical cities are the havens of a socially homogeneous subdivisions are relevant subjects of political population. Although the average nonpartisan inquiry only because they are differentiated on city has 16.9 per cent of its children in private the basis of attitudinal or socio-economic schools, this mean conceals a wide range- variables. The South is not a distinctive from 2 to 47 per cent. political region because two surveyors named Our findings about the insignificance of Mason and Dixon drew a famous line, but class differences between reformed and unre- because the "composition of its population" formed cities are at some variance with Alford differs from the rest of the country. and Scoble's conclusions. There is, however, It is therefore difficult to unravel the mean- some support for the argument that reformed ing of "controlling" for "region" since regions cities are more homogeneous. While we used are differentiated on precisely the kinds of cities with populations of over 50,000, their demographic variables which we (and Wolfinger sample included all cities over 25,000; and and Field) related to reformism. Cities in the varying samples may produce varying con- Midwest, for example, have a much higher clusions. The only other study to analyze proportion of home ownership (64%) than cities over 50,000 was Wolfinger and Field's cities in the Northeast (44%), while north- and our conclusions are generally consistent eastern cities have more foreign stock in their with theirs. We differ with them, however, on two important questions. 24 Op. cit., p. 320. REFORMISM AND PUBLIC POLICIES IN AMERICAN CITIES 707

population (27%) than the Midwest (16%). larger the size of a group in the city's popula- Hence, to relate ethnicity to political re- tion, the easier it can enforce its choice of formism and then to "control" for "region" is political forms. At least one prominent urban in part to relate ethnicity to reformism and sociologist, however, has found empirical then to control for ethnicity. Consequently, support for precisely the opposite proposition. we have grave reservations that the substitu- Hawley concluded that the smaller the pro- tion of the gross and unrefined variable of portion of middle class persons in a city, the "region" for more refined demographic data greater their power over urban renewal poli- adds much to our knowledge of American cies.26 Similarly, it may also be dubious to cities. "Controlling" for "region" is much more assume that the size of an ethnic population is than controlling for historical experiences, an accurate indicator of influence of ethnic because region as a variable is an undifferen- groups. Although we recognize the importance tiated potpourri of socio-economic, attitudinal, of describing the socio-economic correlates of historical and cultural variations.25 political forms, the logical problems involved We also differ with Wolfinger and Field in suggest the need for a good deal of caution in their assertion that their analysis constitutes interpreting these differences as explanations." a test of the ethos theory. As we understand it, In any case, the question of why the city Banfield and Wilson's theory posits that par- adopts particular structures is of less interest ticular attitudes are held by persons with vary- to us than their consequences for public ing sociological characteristics (ethnic groups policy. It is to this analysis that we now turn. and middle class persons, in particular) and that these attitudes include preferences for one IV. POLICY OUTPUTS AND THE or another kind of political institution. But RESPONSIVENESS OF CITIES relating the proportion of middle class persons We are now in a position to take three in a city's population to its form of government additional steps. First, we can compare the says nothing one way or another about middle differences in policy outputs between reformed class preferences. An important part of under- and unreformed cities. Second, we can assess standing, of course, is describing and it is the cumulative impact of socio-economic certainly useful to know how reformed cities variables on these policy choices. Finally, we differ from unreformed cities. can specify what variables are related in what In our view, however, such tests as Wolfinger ways to these output variables. In essence, we and Field used cannot logically be called can now treat political institutions, not as explanations, in any causal sense. The most dependent variables, but as factors which obvious reason is that they violate some impor- influence the level of expenditures and taxation tant assumptions about time-order: inde- and the relationship between cleavage variables pendent variables are measured with con- and these outputs. temporary census data, while the dependent Differences between reformed and unreformed variables are results of decisions made ten to cities' outputs. Contrary to Sherbenou's con- fifty years ago. Moreover, this problem is clusions about Chicago suburbs,28 our data multiplied by the difficulty of inferring con- indicate that reformed cities both spend and figurations of political power from demographic tax less than unreformed cities, with the excep- data. Presumably, their assumption is that tion of expenditures in partisan and non- there is a simple linear relationship between partisan cities. It appears that partisan, mayor- sheer numbers (or proportions) of, say, middle council and ward cities are less willing to com- class persons and their political power: the mit their resources to public purposes than their reformed counterparts. What is of more a5 In statistical parlance, the problem with "re- importance than the difference in outputs, gion" as an independent variable might be de- however, is the relative responsiveness of the scribed as treating a complicated background two kinds of cities to social cleavages in their variable as the first variable in a specific develop- population. mental sequence. But, as Blalock argues, ". . . one should avoid complex indicators that are related in 26 Amos Hawley, "Community Power and unknown ways to a given underlying variable. Geo- Urban Renewal Success," American Journal of graphical region and certain background variables Sociology, 68 (January, 1963), 422-431. appear to have such undesirable properties": 27 See also the exchange between Banfield and Hubert M. Blalock, Causal Inferences in Nonex- Wilson and Wolfinger and Field in "Communica- perimental Research (Chapel Hill: University of tions," this REVIEw, 60 (December, 1966), 998- North Carolina Press, 1964), p. 164 (italics in orig- 1000. inal). 28 Sherbenou, op. cit., pp. 133-134. 708 THE AMERICAN POLITICAL SCIENCE REVIEW

TABLE 3. MEAN VALUES OF TAX/INCOME AND Matthews recognized implicitly what political EXPENDITURE/INCOME RATIOS, BY scientists would now call the "interest aggrega- STRUCTURAL CHARACTERISTICS tion" function of political parties.3l Parties in a democracy manage conflict, structure it, and encapsulate social cleavages under the rubric Structural Variables ,JcmTaxes ExpendituresJcm /Income /Income of two or more broad social cleavages, the parties themselves. "Parties tend to crystallize Election type: opinion, they give skeletal articulation to a Partisan .032 .050 shapeless and jelly-like mass ... they cause Nonpartisan .030 .053 similar opinions to coagulate .. "32 The parties "reduce effectively the number of Government type: political opinions to manageable numbers, Mayor-Council .037 .058 bring order and focus to the political struggle, Manager .024 .045 simplify issues and frame alternatives, and Commission .031 .057 compromise conflicting interests."33 Since parties are the agencies of interest aggregation, so Constituency type: the argument goes, their elimination makes for Ward .036 .057 greater, not lesser, impact of social cleavages At-large .027 .049 on political decisions. Political scientists have recently confirmed Matthews' fears, at least with regard to elec- The responsiveness of cities. We have argued toral behavior in partisan and nonpartisan that one principal goal of the reform movement elections. Evidence points to the increased was to reduce the impact of partisan, socio- impact of socio-economic cleavages on voting economic cleavages on governmental decision when a nonpartisan ballot is used than when making, to immunize city governments from the election is formally partisan. Gerald "artificial" social cleavages-race, religion, Pomper studied nonpartisan municipal elec- ethnicity, and so on. As Banfield and Wilson tions and compared them with partisan elec- put their argument, the reformers "assumed tions for the New Jersey State Assembly in that there existed an interest ('the public Newark. He concluded that the "goal of non- interest') that pertained to the city 'as a whole' partisanship is fulfilled, as party identification and that should always prevail over competing, does not determine the outcome. In place of partial (and usually private) interests."29 The party, ethnic affiliation is emphasized and the structural reforms of manager government, result is 'to enhance the effect of basic social at-large, and nonpartisan elections would so cleavages.' "34 If (1) this is typical of other insulate the business of governing from social American cities and if (2) electoral cleavages cleavages that "private regarding" interests can be translated effectively into demands on would count for little in making up the mind the government in the absence of aggregative of the body politic. But amid the calls of the parties, then we might assume that the re- reformers for structural reforms to muffle the formed institutions would reflect cleavages impact of socio-economic cleavages, a few hardy souls predicted precisely the opposite 31 For a discussion of the concept of interest consequence of reform: instead of eliminating aggregation, see Gabriel Almond, "Introduction: cleavages from political decision-making, the A Functional Approach to Comparative Politics," reforms, particularly the elimination of parties, in Gabriel Almond and James S. Coleman (eds.), would enhance the conflict. Nathan Matthews, The Politics of Developing Areas (Princeton: Jr., a turn-of-the-century mayor of Boston, Princeton University Press, 1960), pp. 38-45. issued just such a warning: 32 Maurice Duverger, Political Parties (New As a city is a political institution, the people in the York: Science Editions, 1963), p. 378. end will divide into parties, and it would seem 33 Frank J. Sorauf, Political Parties in the extremely doubtful whether the present system, American System (Boston: Little, Brown and Co., however illogical its foundation be, does not in 1964), pp. 165-166. fact produce better results, at least in large cities, 34 Gerald Pomper, "Ethnic and Group Voting than if the voters divided into groups, separated in Nonpartisan Municipal Elections," Public 30. (Srijng. 1966). -. 90: see }wmoz~s ~ucisa ers r~ai,@ml ,22XH Q 0 OPinionrQuarterlu. also, J. Leiper Freeman, "Local Party Systems: 29 Op. cit., p. 139. Theoretical Considerations and a Case Analysis," 30 Quoted in Banfield and Wilson, op. cit., p. American Journal of Sociology, 64 (1958), 282- 154. 289. REFORMISM AND PUBLIC POLICIES IN AMERICAN CITIES 709

DIAGRAM 1. Proportion of variation explained (R2) in taxation policy with twelve socio-economic variables, by institutional characteristics.&

Independent Variables Structural Variables Dependent Variable

Reformed Institution: Government: Commission 62% Government: Council-Manager 42% Election: Nonpartisan 49% Constituency: At-Large 49%

Twelve Socio-Economic Variables Tax/Income Ratio

Unreformed Institution: Government: Mayor-Council 52 % Election: Partisan 71% Constituency: Ward/Mixed 59%

a In the total sample, the twelve independent variables explained 52% of the variation in taxes.

DIAGRAM 2. Proportion of variation explained (R2) in expenditure policy with twelve socioeconomic variables, by institutional characteristics.b

Independent Variables Structural Variables Dependent Variable

Reformed Institution: Government: Commission 59% Government: Council-Manager 30% Constituency: At-Large 36% Elections: Nonpartisan 41%

Twelve Socio-Economic Variables Expenditure/Income Ratio

Unreformed Institution: Government: Mayor-Council 42% Constituency: Ward/Mixed 49% Elections: Partisan 59%

b In the total sample, the twelve independent variables explained 36% of the variation in expenditures. more, rather than less, closely than unreformed economic variables in partisan, mayor and ones. ward cities than in nonpartisan, manager and Essentially, then, there are two contrasting at-large cities. Operationally, we will test this views about the consequences of municipal hypothesis by using multiple correlation co- reform. One, the reformers' ideal, holds that efficients. Squaring these coefficients, called institutional reforms will mitigate the impact "multiple R's," will give us a summary measure of social cleavages on public policy. The other of the total amount of variation in our depen- argues that the elimination of political parties dent variables explained by our twelve inde- and the introduction of other reforms will make pendent variables.35 The results of the correla- social cleavages more, rather than less, important in political decision-making. The measurement of responsiveness. We have 35 It is possible that the difference between any hypothesized that socio-economic cleavages two correlations may be a function of very differ- will have less impact on the policy choices of ent standard deviations of the independent vari- reformed than unreformed governments. Thus, ables. A quick look at Table 2, however, suggests one could do a better job of predicting a city's that this is not likely to affect the relationships we taxation and expenditure policy using socio- find. 710 THE AMERICAN POLITICAL SCIENCE REVIEW tion analysis are summarized in Diagrams 1 flected than in the taxation and expenditure and 2. patterns of cities. A generation ago, Charles On the whole, the results of the correlation Beard wrote, "In the purposes for which analysis strikingly support the hypothesis, appropriations are made the policies of the city with the exception of commission cities. Thus, government are given concrete form-the we can say, for example, that our twelve socio- culture of the city is reflected. Indeed, the his- economic variables explain 71 per cent of the tory of urban civilization could be written in variations in taxation policy in partisan cities, terms of appropriations, for they show what and 49 per cent of the variation in nonpartisan the citizens think is worth doing and worth cities. In commission cities, however, socio- paying for."38 Pressures to expand and contract economic variables predict substantially more government regulations and services are al- variation in both taxes and expenditures than most always reflected one way or another in the in the unreformed mayor-council cities.36 The municipal budget. Labor, ethnic groups, the anomaly of commission governments is in- poor and the liberal community may press for teresting, for they present, as we will see, additional services and these must be paid for; marked exceptions to virtually every pattern the business community may demand munici- of relationships we found. The substantial ex- pal efforts to obtain new industry by paring planatory power of these socio-economic vari- city costs to create a "favorable business cli- ables is not altered, but confirmed, by examin- mate"; or businessmen may themselves de- ing the variables independently. The rest of mand municipal services for new or old busi- the correlations show a consistent pattern: ness. In any case, few political conflicts arise reformed cities are less responsive to cleavages which do not involve some conflict over the in their population than unreformed cities. budget structure. If one of the premises of the "political ethos" Class variables and public policies. Part of the argument is that reformed institutions give political rhetoric associated with the demand less weight to the "private regarding" and for a decrease in the scope of the national "artificial" cleavages in the population, that government is the argument that the initiative premise receives striking support from our for policy-making should rest more with the analysis. Our data suggest that when a city state and local governments. Opposition to adopts reformed structures, it comes to be high federal spending levels, as V. 0. Key has governed less on the basis of conflict and more demonstrated, is found more often among on the basis of the rationalistic theory of ad- persons with middle class occupations than ministration. The making of public policy among blue-collar workers.39 It is not incon- takes less count of the enduring differences ceivable that the middle class argument about between White and Negro, business and labor, state and local responsibility might be more Pole and WASP. The logic of the bureaucratic than political rhetoric, and that at the local ethic demands an impersonal, apolitical settle- level, middle class voters are willing to under- ment of issues, rather than the settlement of take major programs of municipal services, conflict in the arena of political battle. requiring large outlays of public capital. Wilson and Banfield have argued that the "public V. TO SPEND OR NOT TO SPEND regarding" upper-middle class voters in metro- If efforts to expand or contract the scope of politan areas are often found voting for public government stand at the core of municipal policies at variance with their "self-interest political life,37 they are nowhere better re- narrowly conceived," and that "the higher the income of a ward or town, the more taste it has 36 Wolfinger and Field, op. cit., p. 312, " ... for public expenditures of various kinds."40 omit the commission cities from consideration Similarly a longitudinal study of voting pat- since this form does not figure in the ethos the- terns in metropolitan Cleveland found that an ory." Historically, however, commission govern- index of social rank was positively correlated ment was the earliest of the structures advocated by the Progressives and is quite clearly a product 38 Charles A. Beard, American Governmentand of the reform era. While history tells us that com- Politics (New York: Macmillan, 1924, 4th edi- mission cities can not legitimately be excluded tion), p. 727. from the fold of reformism, they appear to be its 39 V. 0. Key, Public Opinion and American black sheep, characterized by low incomes, low Democracy (New York: Alfred A. Knopf, 1961), p. population growth and large proportions of non- 124. whites. In fact, they present a marked contrast to 40 Wilson and Banfield, op. cit., p. 876. Footnote both mayor-council and manager cities. 5 in the same article conveniently summarized an Agger et al., op. cit., pp. 4-14. research supporting this proposition. REFORMISM AND PUBLIC POLICIES IN AMERICAN CITIES 711

TABLE 4. CORRELATIONS BETWEEN MIDDLE CLASS CHARACTERISTICS AND OUTPUTS IN REFORMED AND UNREFORMED CITIES

Government Type Election Type Constituency Type Correlations of Mayor- Manager Com- Partisan Non- Council mission Partisan Ward At-large

Taxes with: Median income -.13 -.24 -.19 .03 -.19 -.17 -.22 White collar -.23 -.12 -.62 -.21 -.33 -.30 -.32 Median education -.36 -.22 -.08 -.45 -.24 -.48 -.18

Expenditures with: Median income -.19 -.32 -.43 -.04 -.32 -.23 -.34 White collar -.24 -.23 -.58 -.18 -.39 -.32 -.35 Median education -.32 -.36 -.26 -.36 -.38 -.44 -.32 with favorable votes on welfare referenda.4' If groups on the other."42 If the relationships these data reflect middle class willingness to between income and expenditure is curvilinear, spend on a local level, they might indicate that then we should expect to find that proportions the "states' rights" argument was more than of both low and high income groups were ideological camouflage: middle class voters positively correlated with outputs. Our data, stand foursquare behind public expenditures at however, lend no support to this notion of a the local level even when they oppose those "pro-expenditure" alliance. Rather, the pro- expenditures from the national government. portion of the population with incomes below Therefore, we hypothesized that: $3000 is positively correlated with expenditures 2a. The more middle class the city, measured in all city types (although the relationships are by income, education and occupation, the small) and the proportion of the population in higher the municipal taxes and expendi- the above $10,000 bracket is negatively cor- tures. related with expenditures. Summing the two measures and correlating the combined mea- In line with our general concern of testing the sure with outputs produced no correlation impact of political structures on municipal greater than .15 and the relationships were as policies, we also hypothesized that: likely to be negative as positive. Tests for non- 2b. Unreformed cities reflect this relationship linearity also suggested that no such coalition more strongly than reformed cities. exists in the cities in our analysis. To be sure, aggregate data analysis using With respect to hypothesis 2a, the data in whole cities as units of analysis is no substitute Table 4 on three middle class indicators are for systematic survey data on middle class unambiguous and indicate a strong rejection attitudes, but it is apparent that cities with of the hypothesis. However we measure social larger middle class population have lower, not class, whether by income, education or occupa- higher expenditures. As we emphasized earlier, tion, class measures are negatively related to the "ethos theory" deals with attitudes and the public taxes and expenditures. behavior of individuals, while our data deal It is possible, however, that income does not with cities and their behavior. The coalition have a linear, but rather a curvilinear relation- suggested by Banfield and Wilson, however, is ship with municipal outputs. Banfield and not discernible at this level of aggregation in Wilson argue that "In the city, it is useful to these cities. think in terms of three income groups-low, Hypothesis 2b is not consistently borne out middle, and high. Surprising as it may seem by the data. In fact, the relationships between to Marxists, the conflict is generally between middle class variables and outputs are, if any- an alliance of low-income and high-income thing, stronger in the reformed cities than in groups on one side and the middle-income in their unreformed counterparts. One would not want to make too much of the data, but a 41 Eugene S. Uyeki, "Patterns of Voting in a large body of literature on city politics, which Metropolitan Area: 1938-1962," Urban Affairs Quarterly, 1 (June, 1966), 65-77. 42 Banfield and Wilson, op. cit., p. 35. 712 THE AMERICAN POLITICAL SCIENCE REVIEW

TABLE 5. CORRELATIONS BETWEEN OWNER OCCUPANCY AND GOVERNMENT OUTPUTS IN REFORMED AND UNREFORMED CITIES

Government Type Election Type Constituency Type Correlations of Owner Occupancy Mayor- Non- with: Council Manager Commission Partisan Partisan Ward At-large

Taxes -.57 -.31 -.73 -.64 -.45 -.56 -.48 Expenditures -.51 -.23 -.62 -.62 -.40 -.50 -.40

we discuss below, suggests that reformed insti- expenditures.43 If this is symptomatic of a tutions maximize the power of the middle class. generalized association of ethnic and religious We originally assumed that the proportion minorities with higher expenditures, we might of owner-occupied dwelling units constituted find support for the hypothesis that: another measure of middle class composition, 4a. The larger the proportion of religious and but it soon became apparent that it was only ethnic minorities in the population, the weakly related to income, occupation and higher the city's taxes and expenditures. education measures. Nevertheless, it emerged as the strongest single predictor of both ex- And, if our general hypothesis about the impact penditure and taxation policy in our cities. We of political institutions is correct, then: hypothesized that: 4b. Unreformed cities reflect this relationship 3a. Owner-occupancy and outputs are nega- more strongly than reformed cities. tively correlated, and The correlations between ethnicity, religious 3b. Unreformed cities reflect this relationship heterogeneity and outputs (see Table 6) are, more strongly than reformed cities. with one exception, positive, as predicted by hypothesis 4a. These associations may reflect Hypothesis 3a is consistently borne out in the substantial participation by ethnic groups the data presented in Table 5. These relation- in municipal politics long after the tide of ships were only slightly attenuated when we immigration has been reduced to a trickle.44 controlled for income, education and occupa- The relatively intense politicization of ethnic tion. No doubt self-interest (perhaps "private groups at the local level,45 the appeals to na- regardingness") on the part of the home owner, tionality groups through "ticket balancing" whose property is intimately related to the tax and other means, and the resultant higher structure of most local governments, may turnout of ethnic groups than other lower account for part of this relationship. Moreover, status an on home ownership is correlated (almost by defi- groups,46 may produce influence far out of to their nition) with lower urban population density. city government proportion number. High density, bringing together all manner of We found when we related all twelve of our men into the classic urban mosaic, may be independent variables to outputs in various itself correlated with factors which produce city types that the associations were much demands for higher expenditures-slums, in- weaker in cities we have labeled reformed. The creased needs for fire and police protection, correlations for ethnicity and religious homo- and so on. In confirmation of hypothesis 3a, the unmis- 43 Richard E. Dawson and James A. Robinson, takable pattern is for unreformed cities to "The Politics of Welfare," in Herbert Jacob and reflect these negative relationships more Kenneth Vines (eds.), Politics in the American strongly than the manager, nonpartisan and States (Boston: Little, Brown and Co., 1965), pp. at-large cities, although commission cities show 398-401. their usual remarkably high correlations. 44 Raymond Wolfinger, "The Development and Homogeneity variables and public policies. Persistence of Ethnic Voting," this REVIEW, 59 Dawson and Robinson, in their analysis of (December, 1965), 896-908. state welfare expenditures, found strong posi- 45 Robert E. Lane, Political Life (Glencoe, Ill.: tive relationships between the ethnicity of a The Free Press, 1959), pp. 236-243. state's population and the level of its welfare 46 Ibid. REFORMISM AND PUBLIC POLICIES IN AMERICAN CITIES 713

TABLE 6. CORRELATIONS BETWEEN ETHNICITY AND RELIGIOUS HETEROGENEITY AND OUTPUTS IN REFORMED AND UNREFORMED CITIES

Government Type Election Type Constituency Type Correlations ._ _ of (OayCouilM Manager Commission Partisan PNrtin | Ward At-large

Taxes with: Ethnicity .49 .26 .57 .61 .43 .56 .40 Private School Attendance .38 .15 .37 .33 .37 .41 .25

Expenditures with: Ethnicity .36 .02 .21 .48 .21 .44 .13 Private School Attendance .34 -.01 .07 .25 .24 .40 .05 geneity show a generally similar pattern, with should have an additive effect and further commission cities exhibiting their usual erratic reduce the impact of cleavages on decision- behavior. The data, then, show fairly clear sup- making. We therefore decided to treat "report for hypothesis 4b. formism" as a continuous variable for analytic The third variable of our homogeneity indi- purposes and hypothesized that: cators-per cent of population non-white-had 5. The higher the level of reformism in a city, almost no relationship to variation in outputs, the lower its responsiveness to socio-eco- regardless of city type. We found the same nomic cleavages in the population. weak correlations for the poverty income variable, which was, of course, strongly related to We utilized a simple four-point index to test the racial variable. An easy explanation sug- this hypothesis, ranging from the "least re- gests that this is a consequence of the political formed" to the "most reformed." The sample impotence of Negroes and the poor, but one cities were categorized as follows: should be cautious in inferring a lack of power from the lack of a statistical association. 1. Cities with none of the reformed institutions We have dealt in this section with factors (i.e., the government is mayor-council, elec- which are positively and negatively related to tions are partisan and constituencies are spending patterns in American cities. While wards). social class variables are associated negatively 2. Cities with any one of the reformed institu- with outputs, two measures of homogeneity, tions. private school attendance and ethnicity are 3. Cities with two of the reformed institutions. related to higher taxes and spending. Examin- 4. Cities with three reformed institutions (i.e., ing the strengths of these correlations in cities the government is either manager or com- with differing forms, we found some support mission, elections are nonpartisan and con- for our general hypothesis about the political stitiencies are at-large). consequences of institutions, especially for the We can not overemphasize the crudity of this homogeneity variables and the home ownership index as an operationalization of the complex variable. Interestingly, however, this was not and abstract concept of "reformism." Nonethe- the case with class variables. less, we think some of the relationships we found are strongly suggestive that reformism VI. REFORMISM AS A CONTINUOUS VARIABLE may in reality be a continuous variable. The central thrust of our argument has been To test this hypothesis, we took four vari- that reformed governments differ from their ables which had moderate-to-strong correla- unreformed counterparts in their responsive- tions with our dependent variables and com- ness to socio-economic cleavages in the popula- puted simple correlations in each reform cate- tion. Logically, if the presence of one feature gory. If our hypothesis is correct, the strength of the "good government" syndrome had the of the correlations in Table 7 should decrease impact of reducing responsiveness, the intro- regularly with an increase in reform scores. duction of additional reformed institutions While there are some clear exceptions to the pre- 714 THE AMERICAN POLITICAL SCIENCE REVIEW

TABLE 7. CORRELATIONS BETWEEN SELECTED INDEPENDENT VARIABLES AND OUTPUT VARIABLES BY FOUR CATEGORIES OF REFORMISM

Reform Scores Correlations of 1 24 (least reformed) 2 (most reformed)

Taxes with: Ethnicity .62 .41 .50 .34 Private School Attendance .40 .32 .28 .25 Owner-Occupancy -.70 - .39 -.54 -.44 Median Education -.55 -.27 -.32 -.13

Expenditures with: Ethnicity .51 .27 .41 .05 Private School Attendance .46 .23 .16 .08 Owner-Occupancy -.67 -.30 -.54 -.38 Median Education -.49 -.19 -.38 -.37 dicted pattern of relationships, there is some municipal reformism has been a simple one: fairly consistent support for the hypothesis. socio-economiccleavages cause the adoption of Even when the decreases in the strengths of the particularpolitical forms. A more sophisticated correlations is irregular, there is a clear differ- model would include political institutions as one ence between cities which we have labeled of the factors which produce a given output "most reformed" and "least reformed." structure in city politics. We hypothesize that a Again, we would not want to attach too much causal model would include four classes of vari- importance to the results of this rough-and- ables: socioeconomic cleavages, political vari- ready index. But, the patterns support our pre- ables (including party registration, structure of vious argument about the impact of reformism: party systems, patterns of aggregation,strength the more reformed the city, the less responsive of interest groups, voter turnout, etc.), political it is to socio-economic cleavages in its political institutions (form of government, type of elec- decision-making. tions and types of constituencies), and political outputs. Diagram 3 depicts one possible causal VII. A CAUSAL MODEL AND model. AN INTERPRETATION This study has of necessity been limited to A causal model. The implicit, or at times ex- exploring the linkages between socio-economic plicit, causal model in much of the research on cleavages, political institutions and political

DIAGRAM 3. A hypothesized causal model. Political Institutions

Socio-Econom ic Political Cleavages Outputs

Political Variables REFORMISM AND PUBLIC POLICIES IN AMERICAN CITIES 715 outputs. We found that political institutions political conflicts should not be confused with "filter" the process of converting inputs into the "responsibility" of a political system as the outputs. Some structures, particularly partisan latter term is used in the great debate over the elections, ward constituencies, mayor-council relative "responsibility" of party systems.48 In governments and commission governments, fact, the responsiveness of political forms to operate to maximize the impact of cleavage in- social cleavages may stand in sharp contrast to dicators on public policies. We conclude by dis- "responsible government" on the British model. cussing some of the reasons why different struc- Presumably, in American cities, partisan elec- tures have varying impacts on the conversion tions, ward constituencies, and mayor-council process. governments maximize minority rather than An interpretation. Three principal conclusions majority representation, assuring greater access may be derived from this analysis. to decision-makers than the reformed, bureau- 1. Cities with reformed and unreformed insti- cratized and "de-politicized" administrations. tutions are not markedly different in terms of Partisan electoral systems, when combined demographic variables. Indeed, some variables, with ward representation, increase the access of like income, ran counter to the popular hypoth- two kinds of minority groups: those which are esis that reformed cities are havens of the residentially segregated, and which may as a middle class. Our data lent some support to the consequence of the electoral system demand and notion that reformed cities are more homogene- obtain preferential consideration from their ous in their ethnic and religious populations. councilmen; and groups which constitute iden- Still, it is apparent that reformed cities are by tifiable voting blocs to which parties and politi- no means free from the impact of these cleav- cians may be beholden in the next election. The ages. introduction of at-large, nonpartisan elections 2. The more important difference between the has at least five consequences for these groups. two kinds of cities is in their behavior, rather First, they remove an important cue-giving than their demography. Using multiple corre- agency-the party-from the electoral scene, lation coefficients, we were able to predict mu- leaving the voter to make decisions less on the nicipal outputs more exactly in unreformed than policy commitments (however vague) of the in reformed cities. The translation of social con- party, and more on irrelevancies such as ethnic flicts into public policy and the responsiveness of identification and name familiarity.49 Sec- political systems to class, racial, and religious ond, by removing the party from the ballot, the cleavages differs markedly with the kind of polit- reforms eliminate the principal agency of in- ical structure. Thus, political institutions seem terest aggregation from the political system. to play an important role in the political process Hence, interests are articulated less clearly and -a role substantially independent of a city's are aggregated either by some other agency or demography. not at all. Moreover, nonpartisanship has the 3. Our analysis has also demonstrated that effect of reducing the turnout in local elections reformism may be viewed as a continuous vari- by working class groups,50 leaving officeholders able and that the political structures of the re- freer from retaliation by these groups at the form syndrome have an additive effect: the polls. Fourth, nonpartisanship may also serve greater the reformism, the lower the responsive- to decrease the salience of "private regarding" ness. demands by increasing the relative political Through these political institutions, the goal power of "public regarding" agencies like the of the reformers has been substantially fulfilled, local press.51And when nonpartisanship is com- for nonpartisan elections, at-large constituencies and manager governments are associated 48 The standard argument for party responsi- with a lessened responsiveness of cities to the bility is found in the works of E. E. Schatt- enduring conflicts of political life. Or, as Stone, schneider, esp., Party Government (New York: Price and Stone argued in their study of changes Farrar and Rinehart, 1942) and in the report of produced by the adoption of manager govern- the Committee on Political Parties of the Ameri- ments, the council after the reform "tended to can Political Science Association, Toward a More think more of the community as a whole and Responsible Two-Party System (New York: Rine- less of factional interests in making their de- hart, 1950). cisions."47 49 See Pomper, op. cit.; and Freeman, op. cit. The responsiveness of a political institution to 60 Robert Salisbury and Gordon Black, "Class and Party in Partisan and Nonpartisan Elections: 47 Harold Stone, Don K. Price and Kathryn The Case of Des Moines," this REvIEw, 57 (Sep- Stone, City Manager Government in the United tember, 1963), 584-592. States (Chicago: Public Administration Service, 61One newspaperman said of nonpartisan 1940), p. 238. politics that "You can't tell the players without a 716 THE AMERICAN POLITICAL SCIENCE REVIEW bined with election at-large, the impact of res- oration between groups and a bureaucratic identially segregated groups or groups which ob- agency, a relationship which has characterized tain their strength from voting as blocs in administrative patterns in the federal govern- municipal elections is further reduced.52 For ment. As a result of this decentralization, group these reasons, it is clear that political reforms strength in local governments may be maxi- may have a significant impact in minimizing the mized. role which social conflicts play in decision-mak- It is important in any analysis of reformism ing. By muting the demands of private-regard- to distinguish between the factors which produce ing groups, the electoral institutions of reformed the adoption of reformed institutions and the governments make public policy less responsive impact of the new political forms once they have to the demands arising out of social conflicts been established. We can offer from our data no in the population. conclusions about the origins of reformed struc- The structure of the government may serve tures, for it is obviously impossible to impute further to modify the strength of minority causation, using contemporary census data, to groups over public policy. It is significant in events which occurred decades ago. Once a city this respect to note that commission govern- has institutionalized the reformers' ideals, how- ments, where social cleavages have the greatest ever, a diffused attitude structure may be less impact on policy choices, are the most decen- helpful in explaining the city's public policy tralized of the three governmental types and than the characteristics of the institutions them- that manager governments are relatively the selves. With the introduction of these reforms, most centralized.53 From the point of view of a new political pattern may emerge in which the reformer, commission government is a failure disputes are settled outside the political system, and their number has declined markedly in re- or in which they may be settled by the crowd cent years.54 This greater decentralization of at the civic club at the periphery of the system."8 commission and of mayor-council governments If they do enter the political process, an imper- permits a multiplicity of access points for sonal, "non-political" bureaucracy may take groups wishing to influence decision-makers.56 less account of the conflicting interests and pay It may also increase the possibilities for collab- more attention to the "correct" decision from the point of view of the municipal planner. These conclusions are generally consistent scorecard, and we sell the scorecards": Banfield with the ethos theory developed by Banfield and and Wilson, op. cit., p. 157. Wilson. If one of the components of the middle 52 Oliver P. Williams and Charles Adrian, Four class reformer's ideal was "to seek the good of Cities (Philadelphia: University of Pennsylvania the community as a whole" and to minimize the Press, 1963), pp. 56-57. impact of social cleavages on political decision- 53 Alford and Scoble, op. cit., p. 84. making, then their institutional reforms have 54 In our view, the failure of the commission served, by and large, to advance that goal. government to achieve the intended reforms is more plausible as an explanation of its demise 56 Carol E. Thometz discusses the role of the than its administrative unwieldiness-the con- "Civic Committee" in decision-making in Dallas: ventional explanation. see The Decision-Makers (Dallas: Southern Meth- 66 Williams and Adrian, op. cit., pp. 30-31. odist University Press, 1963).