¹ See, for instance, Eulau, Heinz, Wahlke, John C., Buchanan, William, and Ferguson, LeRoy C., “The Role of the Representative: Some Empirical Observations on the Theory of Edmund Burke” this Review, 53 (09, 1959), 742–56Google Scholar; or Miller, Warren E., and Stokes, Donald E., “Constituency Influence in Congress,” Review, 57 (03, 1963), 45–56. Google Scholar

² For a fuller discussion of this point of view, see Eulau, Heinz, “Changing Views of Representation,” in Pool, Ithiel de Sola (ed.), Contemporary Political Science: Toward Empirical Theory (New York: McGraw-Hill, 1967), pp. 53–85 Google Scholar.

³ We may note, in making this statement, that of 474 state legislators interviewed in the late fifties, 38 per cent failed to articulate any kind of representational role in response to an open question about how they would describe the job of being, a legislator. See Wahlke, John C., Eulau, Heinz et al. The Legislative System (New York: Wiley, 1962), p. 281 Google Scholar. Ten years later, of 435 city councilmen interviewed in connection with this study, as many as 59 per cent failed to make any spontaneous mention of their putative representa-tional role in response to the same question. As far as we know, no student of representational behavior has as yet examined the implications of the evidently low salience of thinking about representation among political practitioners. The matter will be treated in a forthcoming monograph by Katherine Hinckley.

⁴ Pitkin, Hanna Fenichel, The Concept of Representation (Berkeley: University of California Press, 1967), p. 224 CrossRefGoogle Scholar.

⁵ Ibid., pp. 221–222.

⁶ Ibid., p. 224.

⁷ Ibid., p. 209.

⁸ Pitkin's treatment of responsiveness appears to stress the condition in which the representative assembly stands ready to be responsive when the constituents do have something to say. An assembly may, therefore, be responsive whether or not there are specific instances of response. Our analysis, as will be clearer shortly, stresses the actual act of response rather than simply the potential for it. The difficulties of empirically working with a concept stressing the possibility of an act rather than the act itself dictated our decision to modify Pitkin's theoretically suggestive definition.

⁹ The coding procedure used was as follows: Both investigators, reading jointly, read through all parts of the interview schedules pertinent to how councilmen defined their relations with the public for all members of any given council. If the councilmen seldom mentioned any groups or groupings in the public or if they failed to describe an actual case where they had been responsive to public pressures or if they simply asserted (a not unusual occurrence) that they knew what was best for the community and acted upon it, the council was placed in the “self-defined image” code. If the councilmen made references to neighborhood groups or to transitory groups wanting, say, a stop-light at a given corner, or to election groups and if the councilmen indicated that they responded to pressures from such groups and attempted to placate them, then the council was coded in the “responsive to issue-groups” category. If the councilmen defined for us a fairly well organized public, attentive to what the council was doing, and if the councilmen indicated (usually by citing an illustrative case) that they were responsive to these attentive publics, the council was placed in the “responsive to attentive publics” code. The procedure, then, used the councilmen as individual informants about the response style of the council. It is quite possible, though not a frequent occurrence, for a given individual councilman to not feel responsive to, say, attentive publics but to describe the council as acting in that way.

¹⁰ That size is an adequate indicator of “social pluralism” may not be self-evident. We refer the reader to Hadden, Jeffrey K. and Borgatta, Edgar F., American Cities: Their Social Characteristics (Chicago: Rand McNally, 1965)Google Scholar for evidence of the correlative power of size as an indicator of a city's demographic and ecological diversity and pluralism.

¹¹ In Table 1 and all following tables our interpretation of the data is largely based on comparison of the distributions in “high” and “low” categories of the “independent” (column) variable. However, we are attaching to each table two statistics: the raw chi square score adjusted for degree of freedom which can tell us something about the relative order of the data; that is, by dividing chi square by the table's degree of freedom, it is possible to compare tables of different numbers of cells as long as the “N” remains the same or nearly so. Because we are not essentially dealing with a sample but with a universe (82 out of 90 cases in the defined universe), we are not concerned with the sampling problem of whether the distribution in any table is due to chance or not at some set level of confidence. Gamma is introduced as a measure of relationship because it seems especially suitable to data ordered by ordinal or weak ordinal scales.

¹² Dahl, Robert A., A Preface to Democratic Theory (Chicago: University of Chicago Press, 1956), p. 131 Google Scholar, italics added. At another point Dahl argues, “The effective political elites, then, operate within limits often vague and broad, although occasionally narrow and well defined, set by their expectations as to the reactions of the group of politically active citizens who go to the polls.” Ibid., p. 72.

¹³ Schlesinger, Joseph, Ambition and Politics (Chicago: Rand, McNally & Co., 1966), p. 2 Google Scholar. Schlesinger's study is a very careful and ingenious examination of how the political opportunity structure in the U.S. might facilitate or impede political ambitions and thus affect the working of democracy. He does not, however, consider the consequences for democratic politics if men in public office are not ambitious.

¹⁴ Pitkin, op. cit., p. 57.

¹⁵ Hyneman, Charles S., “Tenure and Turnover of Legislative Personnel,” Annals of the American Academy of Political and Social Science, 195 (1938), p. 30 CrossRefGoogle Scholar. Hyneman also remarks that his finding “completely knocks out the supposition that the transiency of legislative personnel is due to the fickleness of the voter at the polls…. Only 16.1 per cent of the 1,965 House members and 14.7 per cent of the 511 senators who quit service during this period were eliminated by defeat in the general election.” Pp. 25–27.

¹⁶ Possibly one of the reasons Hyneman's findings have had such little impact on theories about elections is that he was concerned with the implications of turnover for questions of legislative experience. Students who followed Hyneman's lead also addressed themselves to this question. As far as we have discovered, no political scientist has yet considered how the high rates of voluntary retirement might affect the attention of law-makers to voter preferences. This question is addressed more systematically in Prewitt, Kenneth, “Democratic Elitism From Another Angle,” paper presented at the Midwest Political Science Association Meetings, Ann Arbor, Michigan, 04 24–26, 1969 Google Scholar.

¹⁷ The mean percent of voluntary retirements is .53; the standard deviation is .18. The rate of voluntary retirement is not related to any major demographic characteristic of the city, not to size, population density, per cent of the working force in manufacturing occupations, nor to median income. The stability of this rate across all types of cities suggests that it is a very permanent, even institutionalized, feature of nonpartisan city politics in the Bay Area. By the way, only 3 of the 82 cities studied have limitations on tenure. The survey data which help us understand the reasons for the high rates of voluntary retirement are presented and analyzed in Prewitt, Kenneth, The Recruitment of Political Leaders: A Study of Citizen Politicians (Indianapolis: Bobbs-Merrill, in press)Google Scholar.

¹⁸ Appointed incumbents were excluded because of the high rate of appointment to the councils—24 per cent for all cities averaged over the ten year period. Appointment can be a strategy designed, in this context, to assure election. Omitting these appointed incumbents therefore strengthens the index of forced turnover. The aggregate election data which were used in constructing these analyses were initially collected by Gordon Black, now at the University of Rochester, in collaboration with William Hawley, Institute of Governmental Studies, Berkeley. We are indebted to both Black and Hawley for their help.

¹⁹ Four of the five items used in the support index were coded by using the informant procedure described in footnote 9. Councils were classified according to whether they reported 1) the public to have a respectful view toward councilmen, 2) the public to be in agreement with the council's definition of its duties, 3) the public to include disruptive and unfriendly elements, and 4) the public to be generally supportive in its behavior toward the council. The fifth item, whether there are critical groups in the community, was initially an aggregate measure of individual responses to a question about the number of critical groups. Councils were ranked in terms of this aggregate measure and those above the median were said to have supportive publics, those below, to be operating in a non-supportive environment. Each council was given a score of 1 for each plus on the five items. The support scores were then dichotomized to provide the “relatively high” and “relatively low” classifications used in the analysis.

²⁰ Bryce, James, Modern Democracies (New York: The Macmillan Co., 1924), p. 542 Google Scholar.

²¹ Wahlke, Eulau, et al, op cit., p. 269.

²² The literature, of course, is quite large. Representative studies are reviewed in Prewitt, , “Political Socialization and Leadership Selection,” Annals of the-American Academy of Political and Social Science, 361 (09, 1965), pp. 96–111 CrossRefGoogle Scholar.

²³ See, for instance, the collection of articles in Marvack, Dwaine (ed.), Political Decision-Makers (Glencoe: The Free Press, 1961)Google Scholar; Matthews, Donald R., The Social Background of Political Decision-Makers (Garden City: Doubleday & Co., 1954)CrossRefGoogle Scholar; and the chapter by Dye, Thomas R. in Jacob, Herbert and Vines, Kenneth N. (eds.), Politics in the American States (Boston: Little, Brown & Co., 1965)Google Scholar. This issue is explored with the city council data in chapter 2 of the book by Prewitt cited in footnote 17.

²⁴ The measure of sponsorship is particularly problematic since we are summing not just individual experiences to get a group score but individual experiences which took place over a considerable span of time in some cases. It may be that the aggregation of individual career experiences into a council recruitment measure disguises more variance in the original data than the index should be burdened with. Councilmen enter the council at very different points in time and recruitment, as a system property, may have under-gone major changes since the entry experiences of the older members. For the present, however, we are trapped by our own data; when we began the study we still were thinking of recruitment as an individual attribute and thus mainly collected data about individual careers. Despite the relatively weaker nature of our sponsorship measure, we are reluctant to give up our theoretical posture. We simply note, then, that the weaker relationships in tables using the sponsorship measure may be traced to these methodological difficulties. A council was given a sponsorship measure by com-puting the mean of six alternate paths to office. The “sponsorship continuum” ranged from the case in which an outside challenger initiates his own career and attains a council seat with minimum contact between himself and those already in established positions, to the case in which a councilman was deliberately selected—either asked to run or appointed to the council—by those already in office. The means were then ranked and, for present purposes, the lowest quartile in the rank constitutes the low sponsorship councils; the highest quartile constitutes the high sponsorship councils; the remainder we assigned to the middle group.

²⁵ Pennock, J. Roland, “Political Representation: An Overview,” in Pennock, and Chapman, John W. (eds.), Representation: Nomos X (New York: Atherton Press, 1968), p. 8 Google Scholar.