¹ Deutsch, Karl W., “Social Mobilization and Political Development,” this Review, 55 (09, 1961), 493–515 Google Scholar. Also, particularly, Lerner, Daniel, The Passing of Traditional Society (New York: Free Press of Glencoe, 1958)Google Scholar. For important analysis which in part contradicts the social mobilization hypothesis, see Burnham, Walter Dean, “The Changing Shape of the American Political Universe,” this Review, 59 (03, 1965), 7–28 Google Scholar. Burnham indexes political participation with voter turn-out; we deliberately exclude voting from our scale of participation (see footnote 7). It is not clear, therefore, whether our general findings are in opposition to Burnham's. There is some reason for presuming that voting and other types of political participation are much more independent than a previous generation of scholarship has assumed.

² Almond, Gabriel A. and Verba, Sidney, The Civic Culture (Princeton: Princeton University Press, 1963)CrossRefGoogle Scholar; Dahl, Robert A., Who Governs? (New Haven: Yale University Press, 1961), pp. 282–301 Google Scholar; Key, V. O., Public Opinion and American Democracy (New York: Alfred A. Knopf, 1961)Google Scholar; Milbrath, Lester W., Political Participation (Chicago: Rand McNally Co., 1965)Google Scholar.

³ For example, Huntington, Samuel P., “Political Development and Political Decay,” World Politics, 17 (04, 1965), 386–430 CrossRefGoogle Scholar.

⁴ Some of these data were analyzed by Almond and Verba, op. cit. For description of the methods employed and the sampling problems, see Chapter II and Appendices A and B of that work.

⁵ See, for example, the introductory chapter in Lipset, Seymour M. and Rokkan, Stein, Party Systems and Voter Alignments (New York: Free Press, 1967)Google Scholar.

⁶ For the relationship between economic development indicators and various indices of urbanization and class structure, see the data in Russett, Bruce M., et al., World Handbook of Political and Social Indicators (New Haven: Yale University Press, 1964)Google Scholar. Also see McCrone, Donald J. and Cnudde, Charles F. “Toward a Communication Theory of Democratic Political Development,” this Review, 61, (03, 1967), 72–79 Google Scholar. The literature on general processes of change in the social structure during modernization is, of course, very large. See, among other analyses, Eisenstadt, S. N., “Social change, Differentiation, and Evolution,” American Sociological Review, 29 (06, 1964), 375–387 CrossRefGoogle Scholar; and Parsons, Talcott, “Evolutionary Universals in Society,” American Sociological Review, 29 (06, 1964), 339–357 CrossRefGoogle Scholar.

⁷ See Appendix B for specific scale items and methodology of scale construction. The urban residence variable is size of place of residence for individuals reporting two or more years of residence in that location. (Adding the length of residence control increases the correlation slightly.) It should be noted that the political participation scale does not include voting turnout, partially because the Almond-Verba questions on voting do not lend themselves well to participation analysis, partially because electoral participation seems to include a slightly different dimension than these other factors. There are, of course, variances among these different participation items themselves. Although factor analysis does justify a single scale, future analysis must explore differences among the items. Some separation of the political participation variable is undertaken at various points below.

⁸ For readers unfamiliar with correlation coeffcients in assessing the strength of relationships, Table 4 below presents some of the same patterns in terms of differences between percentages. The reader may wish to review that Table before continuing.

⁹ It is interesting that social status is a stronger predictor of political participation in the United States than in any of the other nations. However, it still lags behind organizational involvement. For all the nations, of course, the status measure and the involvement measure are moderately correlated, with the U.S. showing the strongest relationship.

¹⁰ When the individual items composing these summary indices were correlated with political participation, the coefficients obtained were almost always considerably lower than those shown in Table 1. Further, no single item accounts for a disproportionate amount of the correlation between the summary indices and the participation scale. This indicates (1) that the unidimensionality indicated by the factor loadings is justified and (2) that no single variable, such as education, for instance, is the “real” explanation for the strong correlations which appear in Table 1. Social status, in other words, seems to be the common dimension being tapped by the varied findings linking participation to education, income, and the like.

¹¹ For a discussion of the general problem of ecological fallacies see Robinson, W. S., “Ecological Correlations and the Behavior of Individuals,” American Sociological Review, 15 (1950), 351–357 CrossRefGoogle Scholar. Also see Alker, Hayward R., “A Typology of Ecogical Fallacies: Problems of Spurious Associations in Cross-level Inferences,” International Social Science Council: Symposium on Quantitative Ecological Analysis in the Social Sciences, Evian, France, 09, 1966 Google Scholar.

¹² Although the effort was to tap the same basic variables of social status, organizational involvement, and participation as in the 1960 study, different specific questions were believed appropriate to the Indian context. The Indian sample is also particularly useful for testing urbanization hypotheses because interviews were conducted in towns as small as 200 persons.

¹³ The slight falling off of the relationship between participation and organization membership might be due to the differences in the measures used in India, on the one hand, and the original five nations, on the other. In India, only information on number of memberships was available and this had to be used in place of a composite involvement scale. The correlations between participation and organizational membership only in the five nations are comparable to the correlation produced by the India data.

¹⁴ See citations under footnote 1 above.

¹⁵ Dahl, Robert A., “The City in the Future of Democracy,” this Review, 61, (12, 1967), p. 960 Google Scholar. For a very different perspective on size of city and citizen representation in local matters, see Prewitt, Kenneth and Eulau, Heinz, “Political Matrix and Political Representation: Prolegomenon to a New Departure from an Old Problem,” this Review, 63 (06, 1969) pp. 427–441 Google Scholar.

¹⁶ Methodologically it is quite possible and, for that matter, common for items to demonstrate a strong relationship to a principle factor (e.g., participation level) and at the same time contain uncorrelated portions which are strongly related to different orthogonal factors. This may be what is taking place with the local-national distinction.

¹⁷ The very strong correlation between social class and urban residence and the quite substantial correlation between urban residence and organizational involvement in India may help explain why the aggregate correlation between urbanization and political participation is so powerful when the entire spectrum of the world's nations are included in the analysis. The fact that the correlations between all three of the independent variables and particularly between urban residence and organizational involvement are much stronger in India than in the five more developed nations under study, suggests the magnitude of the changes in social structure which take place in the development process.

¹⁸ The ranking was achieved by calculating the national means on the participation variable. This is the same variable used in Table 1. Medians were also calculated and in no case did this alter the ranking.

¹⁹ The Mexican sample contains no respondent from communities with populations less than 10,000. The .232 correlation between size of place of residence and social status in Mexico may account for the reversal in rankings between Italy and Mexico. Indeed, the means are so close at present that it is highly probable that they would reverse with a representative Mexican sample.

²⁰ As an additional check stemming from our concern for the accuracy of the rankings among the more developed nations a second dichotomous variable was created giving each respondent in the three developed nations a score of “2” and those in the two less developed nations a “1.” The same analysis was then performed on this variable.

The results of this analysis are similar to those reported in Table 3. The simple correlation between the development index and level of participation is .142. The partial for this relationship controlling only for SES is .073. That controlling for only organizational involvement is .030. The second order partial controlling for both SES and ORG. IN. is an insignificant .002.

²¹ The same absolute cutting points were used in each nation. Cutting points were chosen to equalize, as far as possible, the size of the groups. Initial tables showed a trichotomization of both variables, but the size of the groups obtained was too small, especially among upper class respondents in the less developed nations, and the results of that procedure were somewhat misleading. On the other hand, a three group division along organizational involvement lines does help show visually, more effectively than a simple dichotomy, the strength and consistency of that variable.

²² Politically active citizens are arbitrarily defined as citizens participating at or above the grand mean of the participation variable when all five national samples are combined.

²³ Table 4 also reveals a systematic pattern for citizens of the three developed nations, at any given status and involvement level, to be slightly more active in politics than their coharts of the two lesser developed nations. This finding is in direct contradiction to Table 3 which reports, and was so interpreted, that once social status and organizational involvement have been controlled there are no significant differences in participation rates directly related to the development process. Table 4 appears to suggest just the opposite.

A brief methodological excursion is necessary to explain the seeming contradiction. The explanation turns on how sub-categories are constructed. What is happening in Table 4 is that those classified as upper status in the more developed nations contain a much higher proportion of very high status citizens than does the cohart group in the less developed nations. The converse is true as well. There is a higher concentration of very low status persons in the low status categories in Italy and Mexico than in the three more developed countries. There is no easy way to compensate for this artifact of data manipulation. Irrespective of where these continuous variables are cut, there are greater proportions of very highs or very lows in comparable categories when shifting from the more to less developed nations. This reflects the social structures of the different nations—a reality which derives directly from overall level of economic development. The unequal concentrations of very highs and very lows explains most, if not all, of the differences between most and least developed nations detected in Table 4. It also alerts us to the methodological problems of making inferences from category groupings.

²⁴ See, for example, Dahl, Who Governs?, op. cit.

²⁵ A more detailed discussion of these five attitudes as well as the logic connecting them to socio-economic traits and to political participation can be found in Part II, appearing in the September 1969 issue of this Review.

²⁶ See footnote 19 above.

²⁷ Further research will have to distinguish, of course, between the different dimensions of the variables here considered. For example, the “information” questions here only cover very basic items like the names of national party leaders. It might be that detailed information would tend to follow, rather than proceed, participation. The various dimensions of political efficacy are also a matter for investigation. At a very abstract level, efficacy seems to become more related to general national affect items. At the more concrete and personalistic level, it follows the pattern here outlined: linked to personal status and involvement, varying little by national pattern. Thus, the only item on which an underdeveloped nation scored higher than a developed one was on an estimate as to whether possible national participation might succeed in influencing outcomes. Here, Mexicans scored higher than Germans, although very few felt they understood local or national events, or could name any potential strategy of influence. Again, the factor analysis justifies the dimension of the scale, but there are obviously additional dimensions to explore.

²⁸ Almond and Verba, op. cit.

* Data analysis for this paper was performed at the Stanford Computation Center under an NSF grant to Stanford University for unsponsored research. Most of the analysis was accomplished with programs from the Statistical Package For the Social Sciences. See Nie, Norman H., Bent, Dale, and Hull, C. Hadlai SPSS: Statistical Package For the Social Sciences (New York McGraw Hill Book Co., forthcoming)Google Scholar. Several special purpose programs required by our analysis were written by C. Hadlai Hull, systems programmer for Stanford's institute of Political Studies.

** We are grateful to Professor William Paisley of the Communications Department at Stanford for consulting with us on this matter and for suggesting this checking procedure.

* For a discussion of the basic concepts of principle component factor indexes see Hagood, M. J., and Price, P. O., Statistics for Sociologists (New York, Henry Holt and Company, Inc., 1952), Chapter 26Google Scholar.