Hostname: page-component-7c8c6479df-94d59 Total loading time: 0 Render date: 2024-03-28T03:16:10.424Z Has data issue: false hasContentIssue false

Bases of Support for Mexico's Dominant Party*

Published online by Cambridge University Press:  01 August 2014

Barry Ames*
Affiliation:
Stanford University

Extract

The purpose of this paper is to elucidate the bases of support for Mexico's Partido Revolucionário Institucional. A model is developed which identifies the major and minor variables affecting changes in PRI support in the six elections between 1952 and 1967. Throughout the paper the unit of analysis is the state; the dependent variables are voter turnout and the percentage of the total vote in each state received by the PRI.

Type
Research Article
Copyright
Copyright © American Political Science Association 1970

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

Footnotes

*

I am indebted to Bashirrudin Ahmed, John Ferejohn, Richard Fagen, Robert Packenham, Barry Rundquist, and Richard Winters for advice and comments.

References

1 Huntington, Samuel P., “Political Development and Political Decay,” World Politics, XVII (April, 1965), 385430Google Scholar; Emerson, Rupert, Political Modernization: The Single Party System, Monograph No. 1, (Denver: University of Denver, 19631964)Google Scholar; Zolberg, Aristide R., Creating Political Order: The Party States of West Africa, (Chicago: Rand McNally, 1966)Google Scholar.

2 Brandenburg, Frank, The Making of Modern Mexico, (Englewood Cliffs: Prentice-Hall, 1964), pp. 223243Google Scholar.

3 Scott, Robert, Mexican Government in Transition, (Urbana: University of Illinois Press, 1964), pp. 223243Google Scholar.

4 In the 1958 election the PRI lost nine elections for seats in the Chamber of Deputies, and more recently, the mayoralties of the cities of Hermosillo and Mérida. See Scott, ibid., p. 243. Also see Anderson, Bo and Cockcroft, James, “Control and Cooptation in Mexican Politics,” International Journal of Comparative Sociology, VII, (March, 1966)Google Scholar; and Tuohy, William and Ronfeldt, David, “Political Control and the Recruitment of Middle-level Elites in Mexico: an example from Agrarian Politics,” Western Political Quarterly, 22 (June, 1969)CrossRefGoogle Scholar.

5 I also assume that the reported voting data are meaningful. A variety of error types is possible. Random error is not too important since it tends to lower correlations. A more significant problem stems from the frequent charges of fraud by opposition leaders. If there are inflationary biases in the vote totals, they could take three forms. Perhaps all PRI totals are padded by approximately the same amount. This would have the effect of adding a constant and would not seriously affect correlational measures. If the amount added is directly proportional to the percentage of the vote received by the opposition, gross inter-state differences would be reduced, but the rank order would remain the same. If, however, the amount added is related to some unknown variable not of these two types, my results might be seriously affected. At this stage of the research I have no alternative to using reported data.

6 Scott, op. cit., pp. 225–226.

7 Election data sources are as follows: 1952, 1955, 1958, 1961 are from Pablo Gonzales-Casano-va, La Democracia en México (Ediciones Era, S.A., 1965), pp. 240, 242–243, 241247Google Scholar, in that order. The election of 1964 is from Politica, (Sept. 15, 1964), p. xixGoogle Scholar. The election of 1967 is from El Universal, July 11, 1967, p. 9Google Scholar. The last election data are unofficial, but changes were very minor. Objection may be made to the combining of Presidential and Congressional election. While the fit of the model does vary among the elections, the basic model is not seriously affected.

8 Zolberg, op. cit., p. 20.

9 Gonzalez-Casanova, op. cit., pp. 106–107.

10 Turnout has been operationalized as the six-election average of the percentage of eligible voters who actually voted. Mexico does have a registration process, and there are differences between the percentage of the population registered and the percentage actually voting. But registration figures for all elections were unavailable, so registration had to be ignored. Turnout is defined as the percentage of the population legally eligible—irrespective of registration—who voted.

Turnout rates were calculated as follows: the actual number of people who voted in each state was computed from the election results (see footnote 7). The percentage of the population eligible to vote was arithmetically extrapolated from 1950 and 1960 population per state on the basis of each state's growth rate from 1950 to 1960 (these differ widely). For the 1952 election males over 20 were used as the base. In succeeding elections both sexes were used. There are two sources of error: arithmetic rather than geometric progressions, and the assumption of state equality in age distributions. The latter error may be significant but is unavoidable. The population figures come from the Anuário Estadística de los Estados Unidos Mexicanos (Secretaria de Industria y Comercio, Direction General de Estadística, Mexico, 1965)Google Scholar. (for 1962–63), p. 28.

11 Blanksten, George, “The Politics of Latin America,” in Almond, and Coleman, (eds.), The Politics of the Developing Areas, (Princeton: Princeton University Press, 1960), p. 477Google Scholar.

12 Ridker, Ronald, “Discontent and Economic Growth,” Economic Growth and Economic Change XI (October, 1962), p. 11Google Scholar. Some of Ridker's argument is certainly debatable, especially in regard to new immigrants, but the essential point seems valid. See Cornelius, Wayne A. Jr., “Urbanization as an Agent in Latin American Political Instability: The Case of Mexico,” this Review, 63 (Sept. 1969)Google Scholar.

13 Shils, Edward, Political Development in the New States, (London: Mouton, 1965)Google Scholar; and Eisenstadt, S. N., “Breakdowns of Modernization,” Economic Development and Cultural Change, 12, pp. 345367CrossRefGoogle Scholar.

14 Lenski, Gerhard, Power and Privilege, (New York: McGraw Hill, 1966), p. 65Google Scholar.

15 D'Antonio, William and Form, William, Influentials in Two Border Cities, (South Bend: University of Notre Dame Press, 1965), p. 3738Google Scholar.

16 Olson, Mancur, “Rapid Economic Growth as a Destabilizing Force,” Journal of Economic History, (Dec., 1963), p. 533Google Scholar.

17 Padgett, L. Vincent, The Mexican Political System, (Boston: Houghton Mifflin, 1966), p. 83 fnGoogle Scholar. Padgett's emphasis on the special characteristics of less-developed areas reinforces the notion that absolute levels of development may be more important than the relative positions of two or more states. For example, the difference between the levels of urbanization of Idaho and California, and Tlaxcala and the Federal District (Mexico City) may be equal; what is more relevant is the absolute position of all these places on a traditional-modern continuum.

18 There is also some inconclusive evidence suggesting a relationship between urbanization and party competition in the United States. For a review of this literature, see Coulter, Philip and Gordon, Glen, “Urbanization and Party Competition,” Western Political Quarterly, XXI (June, 1968), 274288Google Scholar.

19 Lerner, Daniel, The Passing of Traditional Society, (Glencoe: Free Press, 1958), p. 46Google Scholar.

20 Gonzalez-Casanova, op. cit., p. 106–107.

21 Kurt Steiner, Stanford University, unpublished paper, 1966. Similar findings about Malaya are presented by Richard Winters, Dartmouth College, in an unpublished paper.

22 Dupeux, Georges, “France,” International Social Science Journal, XII (No. 1, 1960), pp. 4647Google Scholar.

23 Objection may be raised to the use of this indicator of level of development. First many scholars cited in the above theoretical section explicitly use urbanization as their variable. Second, as Schnore found, urbanization is highly related to many alternative indicators. He concluded that “Urbanization is an intrinsic part of modernization in general.” (See Schnore, , “The Statistical Measurement of Urbanization and Economic Development,” Land Economics, 38, 1961Google Scholar. I found, using Mexican data, that a variety of independent variables are all intercorrelated averaging .80.) Third, data for some alternative indicators, like GNP, is unavailable. As to the use of absolute levels rather than rates of change, we rely partly on Raymond Tanter's argument: “Although most of the hypotheses above are stated temporally, implying the need for a change measure of urbanization, empirical research suggests that the level and rate of urbanization operate similarly vis-à-vis other processes such as measures of social mobilization:” Tanter, Raymond, “Toward a Theory of Conflict Behavior in Latin America,” (University of Michigan, unpublished paper, 1967), p. 8Google Scholar. Although I believe that the “true” causal variable is a composite of many indicators of level of development, the variable will hereafter be specified as “urbanization.”

24 Wilkie, James, The Mexican Revolution (Berkeley: University of California Press, 1967), pp. 218219Google Scholar. A square root transformation was performed on the variable.

25 Scott, op. cit., p. 41.

26 Ibid., p. 40.

27 Ibid., p. 39.

28 Since historical nonintegration has no easily measurable attributes, a rough test of the hypotheses could be made by testing the adequacy of the model without the variable. Such a procedure, however, fails to reveal interactions between this variable and other variables in the model. My procedure instead will be to construct a dichotomous (dummy) variable which will be allowed to enter the equations freely. Hence I coded Oaxaca, Yucatán, Baja California (Norte) and Baja California Territory as “1” and all the other states as “0” for this variable.

29 D'Antonio and Form, op. cit., p. 218.

30 Ibid., p. 219.

31 Six states border on the United States: Baja California, Sonora, Chihuahua, Coahila, Nuevo Leon, and Tamaulipas. Using the same technique as with the previous variable, these states are coded “1”, all other states are coded “0”.

32 This variable has been operationalized by counting all non-PRI parties that received any votes in 1955, 1961, and 1967, and summing the three totals. The scores range from three to ten. A square root transformation was performed on this data.

33 Among other useful measures: economic conditions at the time of the election, especially cost-of-living indices and unemployment rates; climatic conditions on election day; and what some American scholars euphemistically call “Deaths from civil violence;” i.e., dead students. Quantification of these variables requires more data.

34 Tuohy, William, “Institutionalized Revolution in a Mexican City: Political Decision-Making in Xalapa,” unpublished Ph.D. dissertation, Stanford University, (August, 1967), p. 27Google Scholar. The sample survey was conducted by Professor Richard Fagen of Stanford University.

35 Gonzalez-Casanova, op. cit., p. 238. A logarithmic transformation was performed. The number of officially reported PRI members was available only for 1962. Census data was arithmetically extrapolated to give a population base.

36 Some speculation on “rational” allocation policies is worthwhile. In a two-party system with high party discipline the members of the executive's party should receive the lion's share of patronage, because there is nothing to gain by helping opposition party members. When party discipline is weaker, as in the United States, influential opposition leaders may be wooed, but opposition members in marginal districts would only be strengthened by receiving government benefits. In a dominant-party system we would expect the strong party organizations to struggle for larger shares of outputs. But the government might find it more rational to help the weaker party organizations in order to strengthen them. The ability of the government to pursue such a strategy depends on the alternatives open to the stronger organizations. Is the national party (and hence the government) dependent on the state organizations; do the state organizations have any plausible sanctions, etc.?

37 James Wilkie, op. cit., p. 217, 248. I recognize that among the imperfections of this and the previous measure are the non-coincidence of the time spans. This will be part of the analysis of time-dependent problems.

38 Almond, Gabriel and Verba, Sidney, The Civic Culture, (Princeton: Princeton University Press, 1963)CrossRefGoogle Scholar.

39 Pryor, Frederick, Public Expenditure in Communist and Capitalist Nations, (Homewood: Richard Irwin, Inc., 1968), p. 54Google Scholar.

40 The actual independent variables are the residuals of a regression between urbanization and the percentage of the population benefitting from drinkable-water programs.

41 Suppose, for example, that of our three hypotheses, one was found to be associated with an increase in PRI vote, one had no effect on the vote, and the third was associated with a decrease in vote. The only conclusion we could draw from such a finding is that if the government followed the first policy, its vote might have increased, but if it followed some other policy, its vote either was unaffected or dropped. These hypotheses are mutually exclusive; we cannot conclude that government policy is “a little bit of every-thing.”

42 The results are the same if capital spending between 1958 and 1963 is used as the measure of benefits. The actual hypothesis testing was done by comparing the standardized slopes, or Beta coefficients.

43 In this diagram all possible arrows are included. The signs refer to the direction of the predicted relationships. Double signs refer to directions that are uncertain. Circled intersections between the direct and non-direct factors indicate that the effect of each direct factor is measured controlling for all non-direct factors.

44 The correlation matrix refers to the following variables:

1. The percentage of the population living in communities over 2500.

2. A history of nonintegration and separatism in a state.

3. Location of state on border with United States.

4. The number of opposition parties in elections for Congress.

5. The percentage of the population affiliated with the PRI.

6. The percentage of the population benefitted by drinkable-water programs.

7. The favoritism of a state over its level of demands for benefits.

8. The percentage of probable electors voting in six elections.

9. The percentage of the vote received by the PRI in six elections, 1952–67.

** The lack of correlation between urbanization and the measure of favoritism does not mean that more urban states are not really getting more than rural states, but only that the relationship is linear; i.e., the tendency in favor of urban areas does not increase with urbanization.

45 Levels of significance are not being used to test hypotheses in this paper but only to indicate the strength of relationships, because we have a universe of all cases rather than a sample. No relationships are assumed to occur by chance; both weak and strong correlations are theoretically significant. The computational analysis was done on the IBM 360/67 with the Stanford Statistical Package for the Social Sciences.

46 One of the interesting problems to be explored in the future concerns the interrelationships between turnout, opposition voting, and government outcomes from election to election. This problem has been treated in the United States by, among others, Burnham, Walter Dean, “The Changing Shape of the American Political Universe,” this Review, LIX (March, 1965)Google Scholar, and Campbell, Angus, “Surge and Decline: A Study of Electoral Change,” Public Opinion Quarterly, 24 (1860)Google Scholar.

47 For example, the percentage gain from 1952 to 1967, the gain from 1955 to 1967, or the algebraic sum of changes in rank order between 1952 and 1967.

48 The median benefit ratio was used because the Federal District (Mexico City) had a score almost five times as high as the next highest unit. Capital spending figures are being used instead of water programs because of correspondence between their time period and the elections.

49 Cornelius, op. cit., found that in 1958, 9 percent of urban-born people in urban areas voted for opposition parties, while only 3 percent of migrants into urban areas did so.

50 Almond and Verba, op. cit., p. 17–26. A parochial political culture implies a low frequency of orientations toward the political system in such areas as inputs, outputs, participation, the “system” as a general object, etc.: “In a subject political culture, there is a high frequency of orientations toward a differentiated political system and toward the output aspects of the system, but orientations toward specifically input objects, and toward the self as an active participant, approach zero … In a participant culture, members are oriented to both the input and output sides of the system.”