¹ Turner, Julius, Party and Constituency: Pressures on Congress (Baltimore, 1951)Google Scholar.

² These statements are based on a number of interviews with Members of Congress from both parties, conducted during March, 1958. They do not reflect a systematic sampling. Confidence in them rests rather upon the consistency with which these views were reported.

³ The study of party leadership in the Congress, of which this report is part, was made possible by a grant to Columbia University from the Carnegie Corporation of New York.

The author also gratefully acknowledges the help of a fellowship from the John Simon Guggenheim Memorial Foundation, which released him from his usual academic duties during 1955–56, the generosity of the Watson Scientific Computing Laboratory of Columbia University and its able staff, an indispensable resource for the study, and the creative assistance of Elinor G. Truman.

⁴ The number and size of the sets was dictated by the number of roll call votes with cohesion indexes of less than 100 in each of the sessions and by restrictions in the chosen procedure for IBM machine computation, which permitted analysis at one time of not more than and, for reasons of efficiency, not appreciably less than 74 votes. These procedures were adopted for the major purpose of a modified Rice-Beyle bloc analysis of the two party groups which would locate the formal party leadership in the voting structure and permit inferences concerning leadership roles. For the basic procedure see Rice, Stuart A., Quantitative Methods in Politics (New York, 1928)Google Scholar, ch. 16, and Beyle, Herman C., Identification and Analysis of Attribute-Cluster Blocs (Chicago, 1931)CrossRefGoogle Scholar.

⁵ The largest state party delegation was chosen from each of eight regions if it amounted to at least three men, and the second largest party delegation if it amounted to more than three. In addition, one extra Democratic delegation was selected from the South (Alabama) and one extra Republican delegation from the East North Central region (Wisconsin). The regions used were New England, Middle Atlantic, East North Central, West North Central, Border, South, Mountain, and Pacific. No Republican delegations qualified from the Border, South, and Mountain states. The states selected were, for the Republicans: Massachusetts, New York, Pennsylvania, Michigan, Illinois, Wisconsin, Iowa, Kansas, Oregon, and California; for the Democrats: Massachusetts, New York, Pennsylvania, Ohio, Illinois, Minnesota, Missouri, Oklahoma, North Carolina, Alabama, Texas, Colorado, and California.

⁶ “Agreements,” as used here, include more than actual votes. Any declared position officially recorded in the permanent edition of the Congressional Record was included along with the “yeas and nays.” Thus a Representative who voted “nay” on a roll call and one who was paired or announced against would be counted as in agreement on that vote. “Absences” thus cover all unrecorded preferences, including general pairs in which neither partner announces his position.

⁷ The formula for the number of possible pairs of items in a group of N size, N(N –1)/2, cannot be used in this case because of the wide discrepancies possible between the top scores of two men in a delegation. For example, in a delegation of ten one would be interested in the first nine scores of each man in the group. If, on 74 votes, the top score of man A is with man B of the same delegation and they vote the same way on 45 of the roll calls and if man B's ninth score is 55, A will not appear among B's intra-state agreements, although B's name will appear on A's list.

⁸ These figures are based on the sum of the actual agreement scores in the thirteen delegations divided by the sum of the theoretically possible scores.

⁹ In this connection it is important to bear in mind that the three sets of Democratic votes and the equivalent Republican ones are not identical, since the issues which divide the one party are not necessarily those which split the other. In the first session set 50 roll calls appear among the 74 votes of both parties. In the low-cohesion sets of the second session the overlap is 43 out of 74 and in the high-cohesion sets it is 29 out of 66 Republican and 62 Democratic votes. It is also true, of course, that, even if a vote appears in the equivalent set in both parties, it may occur at the upper end of the distribution in one and at the lower end in the other.

¹⁰ Speaker Rayburn was omitted from all calculations because he so rarely exercised his right to vote.

¹¹ The delegations from the two Border states, Missouri and Oklahoma, were treated in this way after inspection indicated that on the whole both were somewhat more closely tied to the non-Southern legislators.

¹² Among the many informal but continuing partisan and bi-partiaan gatherings on Capitol Hill, many of which may be of far more consequence than their place in the literature would suggest, the ostensibly social but politically functional meetings of state and regional party delegations are common. (An interesting example is discussed in Connally, Tom, My Name is Tom Connolly (New York: Crowell, 1954), pp. 89–92 Google Scholar. For a suggestive case of a somewhat different sort see Bailey, Stephen K. and Samuel, Howard D., Congress at Work (New York: Henry Holt, 1952), pp. 125–26)Google Scholar. Both Senators and Representatives testify to the value, on controversial measures likely to cause difficulty in election campaigns, especially primary elections, of solidarity in state party delegations. Unless deep personal or political differences intervene, two Senators from the same state and party may work out a common position on a set of votes about to be taken in the chamber. Much the same thing takes place, apparently, within the House delegations. On occasion such arrangements may bridge the wide gulf between the two wings of the Capitol and, of course, they may cross party lines as well.

¹³ For an illustration of this see Bailey and Samuel, op. cit., p. 132.

¹⁴ The plurality margin in the House is a more satisfactory indicator of controversy within the House, however, than the incidence of “party votes,” on which a majority of Democrats oppose a majority of Republicans. As the cluster of votes at the upper end of the distributions in the last two columns of Table VII suggests, the high-cohesion votes of both parties contain a disproportionate number of non-party votes, which testifies to the non-controversial character of many of these roll calls. When one party or both are badly split, however, the incidence of party votes need not be high, though it may be, since controversy within the House does not necessarily follow party lines. For instance, in the Democratic low-cohesion set of the second session there were 31 votes of the 74 on which the corresponding Republican cohesion indexes were relatively high. When these roll calls involved a coalition between the Republicans and the northern wing of the Democrats, who constituted a majority of the Democratic representatives, a non-party vote typically resulted. However, when the votes reflected a coalition between the Republicans and the southern wing of the Democrats, a minority of the party, the result was a party vote. Thus, though party votes slightly outnumber non-party among all the low cohesion votes of both parties, the difference—six percentage points—is not great.

One might further assume that if a considerable proportion of the high-cohesion roll calls concerns issues no longer controversial, among them there would be a concentration of votes on final passage and on the acceptance of conference reports—ratifying roll calls taken after the hotly contested phases of the legislation have passed and the final outcome is clear. Conversely, the low-cohesion votes would include a disproportionate number of votes on modifying amendments, dilatory motions, and the like—votes reflecting the unsettled, controversial character of the issues. Votes on recommittal motions might lie between, since they often provide the only opportunity in the House to register opposition to a measure but also may be merely ratifying actions.

Analysis of the roll calls classified in this fashion raises some interesting questions about the roles of the two legislative parties but contributes little to determining the degrees of controversy within the House. The Democratic distribution fits the expected pattern, but the Republican reverses it. This suggests that the Democrats draw together on the ratifying type of motion because they are the responsible majority party, whereas the Republicans are more likely to be united on the preliminary skirmishes, splitting, when the ultimate result is clear, into a group opposed to the impending majority decision and one willing to go along with it. The latter choice is for them the “tough” one, on which colleagues in the state delegation may be influential, whereas the reverse tends to be true of the majority party. Whether this pattern is indeed a feature of the roles of the majority and minority parties or whether, as the material in Duncan MacRae's forthcoming monograph, “Dimensions of Congressional Voting,” suggests, it is a reflection of persistent ideological tendencies in the two parties, only analysis of a Congress in which the majority and minority positions were reversed could determine.

¹⁵ Both these state studies and the inclusive bloc matrices underscore the lack of solidarity in the representation of the South, race issues apart, of which the Texas cleavage is illustrative.

¹⁶ Since almost all Northern Democrats come from “urban” districts, devising a test of this sort presents difficulties. “Big city” districts are here defined as ones which 1) are in or contain a Standard Metropolitan Area and 2) are in or contain a city of 100,000 or more or 3) contain a county of 100,000 or more. The per cents on which the statements in the text are based are not identical with those in Table II, since all tie scores were included in this calculation, there being no defensible way of discriminating among out-of-state scores.

¹⁷ In his forthcoming study, “Dimensions of Congressional Voting,” an application of scaling technique to roll calls in the House in the 81st Congress, Duncan MacRae, Jr., presents evidence to much the same effect, particularly in his chapter entitled “Scale Positions and Constituency Characteristics.”

¹⁸ Occasional bits of direct evidence on delegation functions are encountered, in addition to those referred to earlier. For example, the twelve-man Missouri Democratic delegation in the 81st Congress, seven of whom were newcomers to the House in 1949, reportedly made a special effort at delegation solidarity as a way of supporting their fellow Missourian in the White House. This incidentally may account for some of the peculiarities in the Missouri data in Tables II and V.

¹⁹ The peculiarly personal character of these agreements is suggested not only by the clustering within the delegations but also by the fact that the number of agreements within the delegations is unrelated to two fairly obvious formal factors, seniority and primary or general election pluralities. The data could not be presented within the limits of this paper, but analysis of the intra-state scores in these terms shows that freshman Representatives are no more likely to tie to others in the delegation than are those who have served for several terms and that occupants of “safe” seats are no more likely to behave in this way than are legislators from “close” districts.