¹ See, for example, Fauvet, J., La IV^e République (Paris, A. Fayard, 1959), pp. 9–15.Google Scholar

² E.g., Aron, R., Immuable et Changeante (Paris, Calmann-Lévy, 1959), pp. 50–51.Google Scholar

³ The latter position is argued by Leites, N. in his On the Game of Politics in France (Stanford, 1959).Google Scholar

⁴ The account that follows is partly a summary of well-known historical facts and partly a generalization from the statistical analysis to be presented below. For a more detailed account through 1956, see Williams, Philip, Politics in Post-War France (London, Longmans, 2d ed. 1958).Google Scholar

⁵ See preceding article by W. O. Aydelotte, “Voting Patterns in the British House of Commons in the 1840's”.

⁶ E.g., MacRae, D., Jr., Dimensions of Congressional Voting (Berkeley and Los Angeles, California, 1958), Ch. 2.Google Scholar

⁷ (Paris, Presses Universitaires de France, annually). Referred to below as AP. The roll calls selected are those for which a numerical vote was given in the text.

⁸ Farris, C. D., “A Method of Determining Ideological Groupings in the Congress”, Journal of Politics, 20 (1958), pp. 308–338; D. MacRae, Jr., op. cit.CrossRefGoogle Scholar

⁹ It is of considerable interest to determine those characteristics of legislative bodies that make it possible to scale all their members together. Aydelotte has suggested that those bodies in which debate is conducted in direct confrontation or in a relatively small group may establish standards of comparison that extend across party lines. This hypothesis is consistent with the apparent ease of scaling the two parties together in the U.S. Senate as against the House, and with the fact that scales combining the two parties have also been constructed in the analysis of state legislative processes. See Aydelotte, op. cit.; and MacRae, D., Jr., “Roll Call Votes and Leadership”, Public Opinion Quarterly, 20 (Fail, 1956), pp. 543–558.CrossRefGoogle Scholar

¹⁰ Aydelotte, loc. cit., Tables I and II.

¹¹ See Goodman, L. A. and Kruskal, W. H., “Measures of Association for Cross Classifications”, Journal of the American Statistical Association, 49 (1954), pp. 723–764; 54 (1959), pp. 123–163.Google Scholar

¹² One convention has been to use 10 as the limit of acceptance, with additional special conditions for roll calls with nearly unanimous votes. Aydelotte uses the more stringent limit of 06. This general approach is discussed in Riley, J. W., Riley, M.W., and Toby, J., Sociological Studies in Scale Analysis (New Brunswick, N. J., Rutgers, 1954), Ch. XV.Google Scholar

¹³ Yule, G. U. and Kendall, M. G., An Introduction to the Theory of Statistics (London, Griffin, 1950)Google Scholar. For the calculation of this index, the politically significant votes on each roll call in question are grouped into two categories, “positive” and “negative”, corresponding approximately to “left” and “right” respectively. For a discussion of voting procedures in the Assembly see Lidderdale, D. W. S., The Parliament of France (London, Hansard Society, 1951).Google Scholar

¹⁴ An earlier approach to this problem was given in MacRae, D., Jr., “An Exponential Model for Assessing Fourfold Tables”, Sociometry 19 (06, 1956), pp. 84–94. The results obtained by that model are very similar to those obtained by using Q, if for k = 10 or 5 we substitute Q =.95 or.69, respectively.CrossRefGoogle Scholar

¹⁵ These routines were programmed by Mrs. Natalie Beller. The largest Q-matrices produced in this study involved about seventy roll calls.

¹⁶ Criterion of division: a dissident minority of ten per cent or more. For the Socialists and MRP, the category “did not take part in the vote” was considered dissident for this purpose; for the other parties studied, it was not normally so considered except in conjunction with another category of dissidence (“for”, “against”, or “abstained voluntarily”).

¹⁷ A similar general factor appears in interrelations among scales for the parties in the U.S. House of Representatives.

¹⁸ The distribution of Q in such a matrix is also influenced by whether it is a single party or a group of parties that are being examined.

¹⁹ This value was chosen because it reveals the major divisions in the non-Communist parties. It corresponds to a lower level of scalability than the author required in Dimensions of Congressional Voting (op. cit.). More detailed information might be conveyed by using different Q-levels for different parties, but use of a single level (Qmin = 0.7) facilitates exposition.

²⁰ This procedure is somewhat analogous to the multiple-group method of factor analysis.

²¹ This cluster will be interpreted below when the Socialist party is discussed.

²² Abbreviation for party name and approximate membership are indicated in parentheses. See Williams, op. cit., pp. 442ff.

²³ This was not true of the Peasant Party in the First Legislature, in relation to RI and PRL; it is for this reason that the Peasants are not scaled for the First Legislature. The Peasants often voted unanimously on roll calls which divided both RI and PRL.

²⁴ It would be of interest to see whether the proportion positive decreased during the life of a given government; if it did, on a particular scale or on particular subjects within a scale, this would make it possible to examine in detail the process of disaffection from that government. One could then see which legislators went into opposition first, and which joined them successively on a given issue. Fluctuations in the support for a government might be traced more clearly by separating distinct issues in this way than simply by tracing the government's majority on all issues as they succeed one another.

²⁵ See Stouffer, S. A. et al. , “A Technique for Improving Cumulative Scales”, Public Opinion Quarterly, 16 (Summer, 1952), pp. 273–291. This is a method of data smoothing which is necessitated by the number of “error” responses which our criterion of Qmin = 0.7 permits.CrossRefGoogle Scholar

²⁶ Here, as in subsequent tables, a “positive” or “leftist” vote or scale position appears at the bottom.

²⁷ This placement is consistent with the finding that elective leaders in the parties of the United States Congress tend to occupy central positions in their parties. See MacRae, Dimensions of Congressional Voting, op. cit.; Truman, D. B., The Congressional Party (New York, Wiley, 1959).Google Scholar

²⁸ Goodman and Kruskal, op. cit., pp. 749–751. For the data of this paper, y is normally somewhat higher than Kendall's tau (corrected for ties), but this difference probably does not exceed 0.2.

²⁹ The associations among these scales were Y12 = -.01, y13 = -f-.33, = +.26.

³⁰ See Lerner, D. and Aron, R. (eds.), France Defeats EDC (New York, Praeger, 1958).Google Scholar

³¹ Mutter had given some support to MRP cabinets in the First Legislature.

³² A scale was also constructed for Moderates together with the ARS for the MendèsFrance period. See Table 9.

³³ We shall see that this individualism is not characteristic of the vote of all parties on the electoral law. When the Socialists split on the electoral law in early 1951, for example, they seemed to have done so as a matter of principle. The rightists' votes on the electoral law of 1951 were closely related to their “general left-right” scale, but the connection may have been made by attitudes toward Gaullism. In 1955, with no appreciable Gaullist threat, the Moderates seem to have separated the electoral-law question from other issues.

³⁴ The first two scales have a fairly high positive association: Y12 = +.76. The third, however, is associated negatively with each of the first two: Y13 = -.51,y23 = -.14.

³⁵ Independents, Peasants, and ex-Gaullists combined.

³⁶ The associations between the scales are given by Y12 = +.39, y13 = +68, = +.62. The last two associations appear to correspond respectively to the Mollet government and the Algerian question.

³⁷ The associations in the principal cluster were y12 = +.68, y113 = +.59, y23 = +.57. Those of Scale II-4 with this cluster were Y14 = +.77, Y24 = +.40, y34 = +.53. Those of Scale II-5 with the others were +.36 or less.

³⁸ This category also included the great majority of the Faure and Morice Radicals. These deputies could be scaled on Scale III-4, but less easily on III-2.

³⁹ See Einaudi, M. and Goguel, F., Christian Democracy in Italy and France (South Bend, Ind., Notre Dame, 1952), pp. 171–172.Google Scholar

⁴⁰ This selection was done by Jack D. Liebert. Some of the scalable votes found in this way included division of the party under the premiership of an MRP member, Schuman; only one such vote (under Bidault) appears in our scales.

⁴¹ The associations between the scales are: y12 = +.54, y13 = +.44, y23 - +-32; Scale II-4 is relatively distinct: y14 =-.30, y24 = +.14, y34 =-.50.

⁴² As well as a considerable Gaullist group not included in our scales.

⁴³ Their association is Y12 = +.55.

⁴⁴ A number of Socialist militants were also expelled from the party over a vote on Communist-inspired strikes in 1947; see Williams, op. cit., p. 141.

⁴⁵ André Philip, who had opposed -Thorez, was also an active opponent of de Gaulle though outside of the Assembly at the time. The Q-value for these two roll calls is only about.4.

⁴⁶ Williams, op. cit., p. 63.

⁴⁷ Williams, op. cit., p. 75.

⁴⁸ Table 2, above, shows the Q-matrix for the Socialists in the Second Legislature.

⁴⁹ Another set of roll calls split off a few overseas deputies who favored German rearmament and supported Pleven and Faure against the party majority. These were not a genuine “right” wing, but indicate the overseas deputies' support for governments.

⁵⁰ The associations of the “de Gaulle” scale (III-3) with the other two in the Third Legislature were very low: y13=+.01, y23=+.11; the others seem to be artifacts.

⁵¹ A deviant case is Jules Moch, who served in an early de Gaulle cabinet and may have supported de Gaulle on other grounds than nationalism. For the most part, however, there were few devoted Gaullists in the Socialist party. Both Moch and Lapie may have been opposing Bourgès through support of de Gaulle (AP, 1958, 70–71).

⁵² The senior Socialists presumably had an independent base of power in their popularity with the militants; see Williams, op. cit., p. 66 n.

⁵³ It required the aid of the Socialists in the Senate to reach a favorable decision.

⁵⁴ Except in the First Legislature, when the issue of Gaullism was involved.