¹ Legislative System (New York, 1962), 221.

² Similar background characteristics refer to traits that members have in common, and which may be a base for sharing norms and viewpoints on issues that arise in Council. These shared norms may in turn lead to a stronger relationship between members which would be sufficient to influence votes on its own. For our purposes these background characteristics include sex, age, length of service, occupation, religion, and national party affiliation, as made available in biographical data on members.

³ The Legislative System, 223.

⁴ “Patterns of Personal Relations in a Legislative Group,” Public Opinion Quarterly, XXIII (1959), 101–18.

⁵ Matthews, D. R., U.S. Senators & Their World (New York, 1960), 117.Google Scholar

⁶ Seniority or the characteristic of length of service, which is defined as years served on Council, operates similarly to the above mentioned background characteristics in terms of friendship. However, it is thought to be decisive in party alignment as it provides a criterion of status on Council, as well as one of shared norms.

⁷ Small Groups and Political Behavior (Princeton, 1961), 128.

⁸ Ibid., 186.

⁹ Formal leadership on Council is defined as being the mayor and the Board of Control, and it is hypothesized to be in the mainstream of the group's thinking.

¹⁰ Status in the external community is defined as popularity at the polls. People falling into this category are those members who headed the polls in each ward, as well as the two members of the Board of Control gaining the most votes and the mayor – in other words members of Metro Council.

¹¹ Eulau, H., Zisk, B. H., and Prewitt, K., “Latent Partisanship In Nonpartisan Elections” in Jennings, M. K. And Zeigler, L. H., eds., The Electoral Process (Englewood Cliffs, NJ, 1966), 208–37.Google Scholar

¹² Urban Political Systems: A Functional Analysis of Metro Toronto (New York, 1967), 188–90.

¹³ Williams, O. P. and Adrian, C. R., “The Insulation of Local Politics under the Nonpartisan Ballot,” American Political Science Review, LIII (Dec. 1959), 1056.Google Scholar

¹⁴ “Local Party Systems: Theoretical Considerations and a Case Analysis,” American Journal of Sociology, LXIV (Nov. 1958), 289.

¹⁵ Eulau, H.et al., “The Role of the Representative: Some Empirical Observations on the Theory of Edmund Burke,” American Political Science Review, LIII (Sept. 1959), 744.Google Scholar

¹⁶ Miller, W. E. and Stokes, D. E., “Constituency Influence in Congress,” ibid., LVII (March 1963), 45–56.Google Scholar

¹⁷ For examples of literature addressed to further psychological role orientations including occupational and religious orientations, the reader is directd to: Wahlke, J. C., ‘Behavioral Analysis of Representative Bodies,” in Ranney, A., ed., Essays on the Behavioral Study of Politics (Urbana, Ill., 1962), 173–90Google Scholar; Matthews, D. R., The Social Background of Political Decision-makers (Garden City, NY, 1954Google Scholar); and Schubert, G., “The 1960 Term of the Supreme Court: A Psychological Analysis,” American Political Science Review, LVI (March 1962), 91.Google Scholar

¹⁸ The possible definition of constituency interest is more flexible than that of the other variables mentiond. The best possible definition would take into account the perceptions of the residents of each particular ward as to what is in their interest. However, for our purposes the measure of constituency interest is determined by whether Council members supported challenges to spot rezoning by organized resident groups in their constituency as not being in the interest of their district.

¹⁹ “The Relation between Roll Call Votes and Constituencies in the Massachusetts House of Representatives,” in Eulau, H., Eldersveld, S. J., and Janowitz, M., eds., Political Behavior (Glencoe, Ill., 1956), 317–24.Google Scholar

²⁰ For example, Wilson, J. Q. and Banfield, E. C., “Public-regardingness as a Value Premise in Voting Behavior,” American Political Science Review, LVIII (Dec. 1964), 876–87.CrossRefGoogle Scholar

²¹ Similarity of constituency characteristics takes into account such traits as the average family income and the region within the city of the particular member's constituency.

²² Within the limits of this study no qualitative or issue distinction was made among the Council votes.

²³ A few gaps that were left in the biographical information presented there were filled by information gathered in the News Library of the Toronto Telegram.

²⁴ Although the vote was by secret ballot, all but three Council members disclosed their vote, as was reported by the Globe and Mail of July 6, 1967.

²⁵ Due to certain inconsistencies between census tract and ward boundaries, some boundary approximations were necessary in these cases.

²⁶ To exemplify this in practice, Alderman Horace Brown had positive scored deviations with each of the five members of the Progressive cluster, and each of the other nine members of the Old Guard cluster. However, his total scored deviation with the Old Guard members as shown in Table II was 107.9, whereas his total scored deviation with the Progressives was only 51.8, and he was thereby included in the Old Guard cluster.

²⁷ The scored deviations were used to create an index of interagreement which is used in the testing of hypotheses in the following section. However, this same principle can be applied to show that the total deviation of the pairings of each member of one cluster, the five Progressives, with each member of another cluster, the six Conservatives, produces a total scored deviation of −357.8 which when divided by the thirty pairings results in an average deviation of −11.9 as shown in Table III, which is greater than the −2.0 average deviation between Old Guard members and Progressives. In effect, these figures mean that a Council member known to be a Progressive will agree with a Conservative on any specific vote 11.9 per cent of the time less than is the expected norm on Council. Whereas by comparison, if two members are known to be Progressives they are likely to agree approximately 16.1 per cent of the time more often than is the norm on Council.

²⁸ The factor analysis program was constructed so that the twenty-three members’ measures of agreement with each other could be taken into account. This produced a 23 × 23 matrix similar to that in Table I, with two exceptions based upon the limitations of the factor analysis program. Firstly, because the program did not read negative numbers the mean agreement score of 51.8 was added to each of the scored deviations in the table. Secondly, in allowing for the program's inability to read missing data cells which occur where each Council member's voting performance was recorded against himself, two steps were taken to provide data for those missing cells. This served to create two subtables of Table Iv, one of which was to substitute the mean scored agreement based upon the mean deviations which is an average of the twenty-two deviation scores for each member taken with the other twenty-two members of Council. The second subtable which when taken together with the first tended to reduce the error inherent in using either method singly, had had 99.9 substituted for the missing data cell on the theory that this figure represents perfect agreement for a member with himself. However, on the basis of the cumulative proportion of the total variance that was accounted for, this latter method seems to have introduced a greater amount of error than did the method utilizing mean scores. In actually verifying the clusters, it should be noted that the factors that become apparent were commensurate with the cluster groupings that had been delineated previously. For each factor the highest stated number of scores equal to the number in the commensurate cluster was taken, that is, ten for the Old Guard, six for the Conservatives, five for the Progressives, and two for the Independents, and in both subtables, although the ordering was somewhat different, the same Council members corresponded to the factor grouping that had been in each cluster. This was accomplished in a factor analysis program in which maximum variance of the factors was the criterion for factor rotation. For further reference, see Rummel, R. J., “Understanding Factor Analysis,” Journal of Conflict Resolution, XI (Dec. 1967), 444–80.CrossRefGoogle Scholar

²⁹ A more detailed discussion of the tau A measure may be found in Siegel, S., Nonparametric Statistics for the Behavioral Sciences (New York, 1956), 213–23.Google Scholar

³⁰ Average family income is calculated from data in Metropolitan Profile and the standard for division is based on whether the average ward income was above or below the city average of $5,055 per family.

³¹ The regional boundary divisions employed for separating sections of Toronto are somewhat arbitrary. The Don River divides east from central; Spadina Avenue divides central from west-central; and Dufferin Street divides west-central from west.