Published online by Cambridge University Press: 27 January 2009
The literature on Indian electoral behaviour portrays parliamentary elections since independence as a series of critical, realigning and restoring contests in which voters are repelled by and attracted to the Congress party in great waves. The model is examined from the perspective of voter mobilization. Not only is the established view misleading, it often fails even to describe accurately what actually happened. Significant differences in voting behaviour are found separating the Hindi and non-Hindi speaking areas of India. District-level voting trends are explored within the framework of a multiple regression model applied comprehensively across all of India's general elections and administrative districts. Persistence of democratic politics is to a considerable degree due to the contained volatility of the Congress and opposition party coalitions.
1 Harrison, Selig S., India: The Most Dangerous Decades (Princeton, N.J.: Princeton University Press, 1960).CrossRefGoogle Scholar
2 Joshi, Ram and Desai, Kirtidev, ‘Towards a More Competitive Party System in India’, Asian Survey, 18 (1978), 1091–116.CrossRefGoogle Scholar
3 There is a popular misconception that Britain provided India with an extensive practicum in electoral politics before independence. In fact, elections to provincial and national advisory councils were conducted with a very limited franchise in an essentially non-partisan environment for the first time in 1920, only twenty-seven years before independence. Limited franchise elections were also conducted in 1923, 1926 and 1930. Not until 1937 was there extensive party activity. The last colonial elections, in the winter of 1945–46, while enfranchising most males, were effectively non- partisan affairs in which the two premier independence agitation organizations, the Indian National Congress and the Muslim League, stood almost completely unopposed in their respective con stituencies. See Vanderbok, William G. and Sisson, Richard, ‘Parties and Electorates from Raj to Swaraj: An Historical Analysis of Electoral Behavior in Late Colonial and Early Independent India’, Social Science History, 12 (1988), 121–42.Google Scholar Prior to the 1952 general elections, then, India had experi enced only one competitive partisan election, and it took place fifteen years before with a much re stricted franchise.
4 Kothari, Rajni, Politics in India (Boston, Mass.: Little, Brown and Company, 1970), p. 170.Google Scholar
5 Palmer, Norman D., The Indian Political System, 2nd edn (Boston, Mass.: Houghton Mifflin, 1971), p. 243.Google Scholar
6 Robins, Robert S., ‘Votes, Seats, and the Critical Indian Election of 1967’, Journal of Common wealth and Comparative Politics, 17 (1979), 247–63CrossRefGoogle Scholar; Rudolph, Lloyd I. and Rudolph, Susanne H., ‘Transformation of Congress Party: Why 1980 Was Not a Restoration’, Economic and Political Weekly, 16 (1981), 811–18.Google Scholar
7 Brass, Paul R., ‘Coalition Politics in North India’, American Political Science Review, 62 (1968), 1174–91, p. 1189.CrossRefGoogle Scholar
8 Joshi, and Desai, , ‘Towards a More Competitive Party System’.Google Scholar
9 Mendhelsohn, Oliver, ‘The Collapse of the Indian National Congress’, Pacific Affairs, 51 (1978), 41–65.CrossRefGoogle Scholar This consensus is based on an analysis of campaign strategy, tactics and rhetoric. Graves employs a variance component analysis of constituency returns which disputes this, attributing 40 per cent of the variance in the Congress vote between 1957 and 1962 to national factors (Graves, Daniel R., ‘Elections and National Mobilization in India’, Comparative Political Studies, 11 (1978), 255–78CrossRefGoogle Scholar). Between 1962 and 1967 the national factor rises to 51 per cent. However, it falls to only 21 per cent for the 1967–71 transition. Elaborating Graves's findings, Bueno de Mesquita calculates that overall only 4 per cent of the variance is due to national factors. See de Mesquita, Bruce Bueno, ‘Redistricting and Political Integration in India’, Comparative Political Studies, 11 (1978), 279–88.CrossRefGoogle Scholar He attributes the difference in findings to gerrymandering. Strangely, if Graves is right, 1971 was the least national of the early elections. If Bueno de Mesquita is right, they were never very national.
10 Joshi, and Desai, , ‘Towards a More Competitive Party System’.Google Scholar
11 Hampton, Davey, ‘Polarization and Consensus in Indian Party Polities’, Asian Survey, 12 (1972), 701–16Google Scholar; Heginbotham, Stanley J., ‘The 1971 Revolution in Indian Voting Behaviour’, Asian Survey, 11 (1971), 1133–52CrossRefGoogle Scholar; Rudolph, Lloyd I., ‘Continuities and Change in Electoral Behavior: The 1971 Parliamentary Election in India’, Asian Survey, 11 (1971), 1119–32.CrossRefGoogle Scholar
12 Weiner, Myron, India at the Polls: The Parliamentary Elections of 1977 (Washington, DC: American Enterprise Institute for Public Policy Research, 1978).Google Scholar
13 Joshi, and Desai, , ‘Towards a More Competitive Party System’.Google Scholar
14 Rudolph, , ‘Continuities and Change in Electoral Behavior’.Google Scholar
15 Weiner, Myron, ‘Congress Restored: Continuities and Discontinuities in Indian Polities’, Asian Survey, 22 (1982), 339–55.CrossRefGoogle Scholar
16 Gould, Harold A., ‘The Second Coming: The 1980 Elections in India's Hindi Belt’, Asian Survey, 20 (1980), 595–616.CrossRefGoogle Scholar
17 In addition, available data are suspect for many purposes. This point is elaborated upon in the Appendix.
18 An excellent recent presentation of this approach, which has enjoyed nearly universal accept ance among analysts of Indian elections, is Rudolph, Lloyd I. and Rudolph, Susanne H., In Pursuit of Lakshmi: The Political Economy of the Indian State (Chicago: University of Chicago Press, 1987), especially chaps. 4 to 6 inclusive.Google Scholar
19 Reading this paragraph, a colleague wrote to me that ‘Common knowledge of Indian politics tells us that this was an election when Sanjay's goons were out mobilizing more than in probably any other election. Congress was mobilizing hard but lost popularity due to a host of reasons. [Measuring party performance by mobilization rather than valid vote falsely] suggests that this loss was a function of decline in Congress' mobilizational efforts.…’ This statement presents multiple misunderstandings. Firstly, there is no substantial evidence that ‘Sanjay's goons’ more than compensated for the ‘common knowledge’ that the party organization was seriously atrophying under Mrs Gandhi's efforts at centralization. Secondly, it presumes without evidence that Sanjay's supporters' efforts at intimidation generated more support than hostility. Thirdly, and most importantly, it asserts that changes in electoral mobilization are a simple function of organizational effort. Electoral mobilization is a straightforward descriptive statistic indexing how well a party has done in the electorate. Explaining changes in mobilization is a totally different matter. ‘Common knowledge’ of Indian elections has gone astray precisely because it is has only considered share of the valid vote and ignored overall voter mobilization.
20 Harrison, , India: The Most Dangerous DecadesGoogle Scholar; Rudolph, and Rudolph, , In Pursuit of Lakshmi.Google Scholar
21 Hardgrave, Robert L., ‘The Northeast, the Punjab, and the Regionalization of Indian Polities’, Asian Survey, 23 (1983), 1171–81.CrossRefGoogle Scholar
22 Baxter, Craig, The Jana Sangh: A Biography of an Indian Political Party (Philadelphia: University of Pennsylvania Press, 1969)CrossRefGoogle Scholar; Blair, Harry W., ‘Rising Kulaks and Backward Classes in Bihar: Social Change in the Late 1970s’, Economic and Political Weekly, 15 (1980), 64–74Google Scholar; Brass, Paul R., ‘Congress, Lok Dal, and the Middle-Peasant Castes: An Analysis of the 1977 and 1980 Parlia mentary Elections in Uttar Pradesh’, Pacific Affairs, 54 (1981), 5–41CrossRefGoogle Scholar; Fickett, Lewis P. Jr, ‘The Politics of Regionalism in India’, Pacific Affairs, 44 (1971), 193–210CrossRefGoogle Scholar; Hardgrave, , ‘The Northeast, the Punjab, and the Regionalization of Indian Polities’Google Scholar; Prasad, Pradhan H., ‘Rising Middle Peasantry in North India’, Economic and Political Weekly, 15 (1980), 215–19.Google Scholar
23 Rudolph, and Rudolph, , In Pursuit ofLakshmi, pp. 179–87.Google Scholar
24 Kothari, , Politics in India.Google Scholar
25 Eldersveld, Samuel J. and Ahmed, Bashiruddin, Citizens and Politics: Mass Political Behavior in India (Chicago: University of Chicago Press, 1978).Google Scholar For a discussion of the problems encountered in using existing survey data, see the Appendix.
26 Data for the period 1952–67 were originally collected for the Data Confrontation Seminar, sponsored by the Interuniversity Consortium for Political and Social Research. Data for the period 1971–84 have been collected by the author using official census and election returns.
27 Robinson, William S., ‘Ecological Correlations and the Behavior of Individuals’, American Sociological Review, 15 (1950), 351–7CrossRefGoogle Scholar; Alker, Hayward R. Jr, ‘A Typology of Ecological Fallacies’ in Dogan, Mattei and Rokkan, Stein, eds, Social Ecology (Cambridge, Mass.: MIT Press, 1974)Google Scholar; Bourke, Paul F. and DeBats, Donald A., ‘Individuals and Aggregates: A Note on Historical Data and Assumptions’, Social Science History, 4 (1980), 229–50.Google Scholar
28 Wright, Gerald C. Jr, ‘Linear Models for Evaluating Conditional Relationships’, American Journal of Political Science, 20 (1976), 349–73.CrossRefGoogle Scholar
29 An earlier effort, by Dasgupta and Morris-Jones, to combine census and electoral materials used ‘share of the valid vote cast’ rather than ‘mobilization’ and only covered the first four elections, 1952–67 (Dasgupta, Bilap and Morris-Jones, W. H., Patterns and Trends in Indian Politics: An Ecological Analysis of Aggregate Data on Society and Elections (Bombay: Allied Publishers Private Limited, 1975).Google Scholar Their central concerns were with developing a taxonomy of districts based on voteshare change from one election to the next (termed institutionalization) and describing party performance in terms of district development categories and simple correlations.
30 The scheduled castes, although they are Hindus, are officially identified (scheduled) as having suffered historical discrimination. While such listing makes these castes eligible for special govern ment benefits, it most certainly does not mean that the members of such castes have a sense of corporateness to the extent that Muslims or Sikhs presumably do. However, and in spite of their heterogeneity, the scheduled castes form a common unit in political discourse. In addition, there are no statistically consequential differences between scheduled castes and tribes in this analysis, and so they have been combined into a single variable.
31 Both tolerance and condition indices were employed to identify and eliminate collinearity problems.
32 Ever since the 1971 general elections the Congress has entered into an electoral arrangement with either the DMK or AIADMK in the southern state of Tamil Nadu. The result is the lack of a Congress candidate in a large number of constituencies. This in turn produces extraordinary vari ances in district level support for the Congress. Tamil Nadu has therefore been excluded from the analysis for the years 1971–84 inclusive. Electoral adjustments have also taken place in Kerala from time to time, so that it, too, has been dropped, beginning with 1971.
33 Wright, , ‘Linear Models for Evaluating Conditional Relationships’, pp. 356–62.Google Scholar
34 Graves's variance component analysis attributes 10 to 29 per cent of the Congress vote variation to state-level forces in the 1957–71 period, but without making any regional distinctions; see Graves, , ‘Elections and National Mobilization’.Google Scholar Bueno de Mesquita estimates the state component at 13 per cent for Congress, 33 per cent for the Jan Sangh and 16 per cent for the Communists during the same period; see de Mesquita, Bueno, ‘Redistricting and Political Integration’.Google Scholar
35 Hanushek, Eric A. and Jackson, John E., Statistical Methods for Social Scientists (New York: Academic Press, 1977), pp. 122–9.Google Scholar
36 This is not to imply that regional politics is free of communal tension. In recent years very serious Hindu-Muslim riots have occurred in Ahmedabad and Hyderabad. Assam has linguistic/immigrant tensions. Sikh separatist demands and terrorism in the Punjab regularly receive international attention.
37 Manor, James, ‘The Electoral Process Amid Awakening and Decay: Reflections on the Indian General Election of 1980’, in Lyon, Peter and Manor, James, eds, Transfer and Transformation: Political Institutions in the New Commonwealth (Leicester: Leicester University Press, 1983), pp. 87–116.Google Scholar
38 Brass, , ‘Congress, Lok Dal, and the Middle-Peasant Castes’, p. 36.Google Scholar
39 Brass pursues this theme at great length in an ecological analysis of Uttar Pradesh voting in which the dependent variable is classified in a series of fine gradations from agricultural labour and ‘dwarf’ land-holders to giant talukdars. Unfortunately, this strategy led to small sample sizes, large variances and innumerable cross-level inferences based on twenty-seven pages of product-moment correlations. Moreover, in a state that is 82 per cent rural with 84 per cent of its agricultural labour- force being owner-cultivators, it is hard to imagine any party being successful without an agricul tural base. See Brass, , Caste, , Faction and Party, p. 74.Google Scholar
40 If agricultural land-owners dominated agricultural labour there would be a high correlation between these variables, leading to one or the other being dropped on multicollinearity grounds. The correlation, 0.63, hovers at the cutoff point and so both variables were retained.
41 Converse, Philip, ‘Of Time and Partisan Stability’, Comparative Political Studies, 2 (1969), 139–71.CrossRefGoogle Scholar
42 For a discussion of partisan volatility under conditions of prolonged non-democratic interruption as it compares to the Indian case of a developing democracy, see Brown, Courtney and Lancaster, Thomas D., ‘Partisan Volatility and the Genesis of Contemporary Democracies’, presented at the Annual Meeting of the American Political Science Association, New Orleans, 1985.Google Scholar
43 Kothari, Rajni, ‘The Political Change of 1967’, Economic and Political Weekly, 6 (1971), 231–50.Google Scholar
44 Sheth, D. L., ‘Partisanship and Political Development’, Economic and Political Weekly, 6 (1971), 259–74.Google Scholar
43 Key, V. O. Jr, ‘A Theory of Critical Elections’, Journal of Politics, 17 (1955), 3–18.CrossRefGoogle Scholar
46 This leaves aside both the question of voter exchanges between the Congress and opposition parties from one election to the next and the problem of dropouts. I deal with these and related issues in Vanderbok, William G., ‘Critical Elections, Contained Volatility and the Indian Electorate’, Modern Asian Studies, forthcoming.Google Scholar
47 Weiner, Myron, ‘The Wounded Tiger: Maintaining India's Democratic Institutions’Google Scholar, in Lyon, and Manor, , eds, Transfer and Transformation, pp. 49–57.Google Scholar
48 Eldersveld, and Ahmed, , Citizens and Politics.Google Scholar