1. Oksenberg discusses five major sources – the press, migrant interviews, visitor accounts, Chinese fiction and secretly obtained documents – and the strengths, weaknesses, and potential biases of each. See his “Sources and methodological problems in the study of contemporary China” in Barnett, A. Doak (ed.), Chinese Communist Politics in Action (Seattle: University of Washington Press, 1964), pp. 577–606Google Scholar. Martin Whyte covers the same ground with the sociologist's eye. He counsels a strategy centered around the search for patterns of variation in Chinese society. See his “The study of Mainland China: sociological research and the minimal data problem,” Contemporary China, Vol. 1 (03 1977), pp. 1–12Google Scholar.

2. Two areas of controversy which spring immediately to mind are the discussions about the existence of regular cycles of social change and their causes, and about the problems encountered in attempts to implement innovative organizational changes, especially in industry. For an introduction to the debate on cyclical change, see Skinner, G. W. and Winckler, E., “Compliance succession in rural Communist China: a cyclical theory,” in Etzioni, Amitai (ed.), A Sociological Reader on Complex Organizations (New York: Holt, Rinehart and Winston, 1969), pp. 410–38Google Scholar; Nathan, Andrew, “Policy oscillations in the People's Republic of China: a critique,” The China Quarterly (CQ), No. 68 (12 1976), pp. 720–33CrossRefGoogle Scholar; Edwin Winckler, “Policy oscillations in the People's Republic of China: a reply,” ibid. pp. 734–50; and Hiniker, Paul and Perlstein, Jolanta, “Alternation of charismatic and bureaucratic styles of leadership in post-revolutionary China,” Comparative Political Studies, Vol. 10 (01 1978), pp. 529–54CrossRefGoogle Scholar. For an introduction to the issues surrounding attempts at organizational change in industry, see Baum, Richard, “Technology, economic organization, and social change: Maoism and the Chinese industrial revolution” in German Association for East Asian Studies (ed.), China in the Seventies (Weisbaden: Otto Harrossowitz, 1975), pp. 131–91Google Scholar; Walder, Andrew, “Industrial organization and socialist development in China,” Modern China, Vol. 5 (04 1979), pp. 233–72CrossRefGoogle Scholar; and Whyte, Martin, “Iron law versus mass democracy: Weber, Michels and the Maoist vision” in Hsiung, James C. (ed.), The Logic of “Maoism” (New York: Praeger, 1974), pp. 37–61Google Scholar.

3. Liao and Whiting have argued these advantages in applying thematic analysis to problems in international relations; Paul Wong has presented the case for thematic analysis in domestic politics and for coding data on elites. See Liao, Kuang-sheng and Whiting, Allen, “Chinese press perceptions of threat: the U.S. and India, 1962,” CQ, No. 53 (01–03 1972), pp. 80–97Google Scholar; and two pieces by Wong, Paul: “Coding and analysis of documentary materials from Communist China,” Asian Survey, Vol. 7 (03 1967), pp. 198–211CrossRefGoogle Scholar; and Content Analysis of Documentary and Biographic Materials: Methodology (Berkeley: Survey Research Center, University of California, 1967), Monograph No. 22Google Scholar.

4. Oksenberg, , “Sources and methodological problems,” p. 605Google Scholar.

5. See Henkel, Raymond E., Tests of Significance (Beverly Hills and London: Sage Publications. Sage University Paper series on Quantitative Applications in the Social Sciences, series No. 07–004. 1976), pp. 1–43CrossRefGoogle Scholar for a summary discussion of the framework of assumptions and probability theory that underlies statistical inference.

6. Liu, Alan P. L., Communications and National Integration in Communist China (Berkeley: University of California Press, 1971), pp. 191–96Google Scholar.

7. Sullivan, Lawrence and Solomon, Richard H., “The formation of Chinese Communist ideology in the May Fourth era: a content analysis of Hsin ch'ingnien,” in Johnson, Chalmers A. (ed.), Ideology and Politics in Contemporary China (Seattle: University of Washington Press, 1973), pp. 117–60Google Scholar.

8. Liao and Whiting, “Chinese press perceptions.” See also the following exchange on the possible pitfalls of such a technique: Friedman, Edward, “Some political constraints on a political science: quantitative content analysis and the Indo-Chinese border crisis of 1962,” CQ, No. 63 (09 1975), pp. 528–38CrossRefGoogle Scholar, and Whiting, Allen S. and Liao, Kuang-sheng, “Reply,” CQ, No. 65 (03 1976), pp. 117–19CrossRefGoogle Scholar.

9. Dittmer, Lowell, Liu Shao-ch'i and the Chinese Cultural Revolution: The Politics of Mass Criticism (Berkeley: University of California Press, 1974), pp. 297–335Google Scholar.

10. Paul J. Hiniker and Jolanta J. Perlstein, “Alternation of charismatic and bureaucratic styles.” Hiniker uses the same method in his study of ideology during the Cultural Revolution. See his Revolutionary Ideology and Chinese Reality: Dissonance Under Mao (Beverly Hills and London: Sage, 1977)Google Scholar.

11. See, for example, Diao, Richard K., “The impact of the Cultural Revolution on China's economic elite,” CQ, No. 42 (04–06 1970), pp. 65–87CrossRefGoogle Scholar; Kau, Yingmao, “The urban bureaucratic elite in Communist China: a case study of Wuhan, 1949–65,” in Barnett, , Chinese Communist Politics, pp. 216–67Google Scholar; Klein, Donald W., “The next generation of Chinese Communist leaders,” CQ, No. 12 (10–12 1962), pp. 57–74CrossRefGoogle Scholar; North, Robert C. and Pool, Ithiel de Sola, “The KMT and Chinese Communist elites” in Laswell, Harold and Lerner, Daniel (ed.), World Revolutionary Elites (Cambridge: MIT Press, 1966), pp. 319–455Google Scholar; Oksenberg, Michel, “Local leaders in rural China, 1962–65: individual attributes, bureaucratic positions, and political recruitment” in Barnett, , Chinese Communist Politics, pp. 155–215Google Scholar; several papers in Scalapino, Robert A. (ed.), Elites in the People's Republic of China (Seattle: University of Washington Press, 1972)Google Scholar; and Teiwes, Frederick C., Provincial Party Personnel in Mainland China, 1956–66 (New York: Columbia University. Occasional Papers of the East Asian Institute. 1967)Google Scholar.

12. See, for example, Parish, William L. Jr, “Factions in Chinese military politics,” CQ, No. 56 (10–12 1973), pp. 667–99CrossRefGoogle Scholar; three publications by Whitson, William: “The field army in Chinese Communist military politics,” CQ, No. 37 (01–03 1969), pp. 1–30CrossRefGoogle Scholar, Chinese Military and Political Leaders and the Distribution of Power in China, 1956–1971 (Santa Monica: Rand, 1973)Google Scholar, and The Chinese High Command (New York: Praeger, 1973)Google Scholar; and Wong, Paul, China's Higher Leadership in the Socialist Transition (New York: The Free Press, 1976)Google Scholar.

13. We do not wish to minimize the difficulties faced by elite analysts. How, for example, does one interpret the quiet disappearance of a certain political figure – demotion, death or routine transfer? The point here is solely that the kind of information sought by elite analysts is, comparatively, much more reliable and far less ambiguous than that sought by those who scan newspapers for social indicators.

14. Lee, Hong Yung, “The radical students in Kwangtung during the Cultural Revolution,” CQ, No. 64 (12 1975), pp. 645–83CrossRefGoogle Scholar. See also the book based on ProfessorLee's, dissertation, The Politics of the Chinese Cultural Revolution (Berkeley: University of California Press, 1978)Google Scholar.

15. Cell, Charles P., Revolution at Work: Mobilization Campaigns in China (New York: Academic Press, 1977)Google Scholar.

16. Andors, Stephen, “Factory management and political ambiguity, 1961–63,” CQ, No. 59 (07–09 1974), pp. 435–76CrossRefGoogle Scholar; and China's Industrial Revolution: Politics, Planning, and Management, 1949 to the Present (New York: Pantheon, 1977), pp. 108–123Google Scholar; 174–86.

17. The boundaries we have drawn between types of content analysis are made to highlight the different possible approaches to quantitative newspaper research. Some researchers clearly straddle these boundaries. Paul Hiniker, in the examples cited above, uses thematic analysis to draw conclusions about changing leadership styles in Chinese organizations: in other words, thematic content can be one kind of indicator of a limited range of social processes (notably leadership and mass communication). This extension of the basic thematic approach takes on the added burden of those using this third analysis strategy: to think through the problems of using a specific kind of newspaper content to gain an understanding of the actual social process of interest.

18. Fogel, Robert W., “The new economic history: its findings and methods,” Economic History Review, Vol. 19 (12 1966), pp. 642–56Google Scholar; 652–53. The controversy surrounding Fogel's own work on the economics of American slavery should be sufficient to disabuse anyone of the notion that statistical technique can be anything more than an aid to the logical thought and argumentation that is the staple of all good scholarship, historical and otherwise.

19. Examples of such scholarship are readily found in the interdisciplinary journals devoted to social, economic or urban history. The supplementary volume to Fogel, Robert W. and Engerman's, Stanley study of American negro slavery, Time on the Cross: A Supplement (New York: Little, Brown, 1974)Google Scholar, contains a general essay on statistical developments in history and also describes the coding of slave plantation journals and newspaper advertisements for runaway slaves. Coded newspaper accounts, further, form the core of our knowledge about the history of collective violence in Europe. See Tilly, Charles, Tilly, Louise and Tilly, Richard, The Rebellious Century, 1830–1930 (Cambridge: Harvard University Press, 1975)CrossRefGoogle Scholar, for a careful recounting of this research strategy and an example of its potential.

20. See, for example, the following articles, all of which use data on social and political conflict that has been enumerated, classified, and coded from newspaper accounts: Feierabend, Ivo K. and Feierabend, Rosalind, “Social change and political violence: cross-national patterns” in Graham, H. D. and Gurr, T. R. (eds.), Violence in America: Historical and Comparative Perspectives (New York: Praeger, 1969), pp. 632–87Google Scholar; Gurr, Ted, “A causal model of civil strife: a comparative analysis using new indices,” American Political Science Review, Vol. 62 (12 1968), pp. 1104–124CrossRefGoogle Scholar; Inverarity, James, “Populism and lynching in Louisiana, 1889–1896,” American Sociological Review, Vol. 41 (04 1976), pp. 262–80CrossRefGoogle Scholar; Jenkins, J. Craig and Perrow, Charles, “Insurgency of the powerless: farm worker movements (1946–1972),” American Sociological Review, Vol. 42 (04 1977), pp. 249–68CrossRefGoogle Scholar; Russet, Bruce, “Inequality and instability: the relation of land tenure to politics,” World Politics, Vol. 16 (04 1964), pp. 442–54CrossRefGoogle Scholar.

21. See Paige, Jeffery M., Agrarian Revolution: Social Movements and Export Agriculture in the Underdeveloped World (New York: The Free Press, 1975). Pages 86–103 and 388–402Google Scholar explain the coding procedures and research strategy. Paige relies heavily on the New York Times Index and on regional press summaries for Africa, Asia and Latin America. The approach parallels that used in the chapter on “Political protest and executive change” in Taylor, Charles Lewis and Hudson, Michael C., World Handbook of Political and Social Indicators (New Haven: Yale University Press, 1972), pp. 59–199Google Scholar.

22. See, for example, Snyder, David and Tilly, Charles, “Hardship and collective violence in France, 1830–1960,” American Sociological Review, Vol. 37 (10 1972), pp. 520–32CrossRefGoogle Scholar; Tilly, Charles, “How protest modernized in France, 1845–1855” in Aydelotte, W., Bogue, A., and Fogel, R. W. (eds.), The Dimensions of Quantitative Research in History (Princeton: Princeton University Press, 1972), pp. 192–255Google Scholar. Tilly's, CharlesFrom Mobilization to Revolution (Reading, Massachusetts: Addison-Wesley, 1978)Google Scholar, contains a comprehensive summary and bibliography of, among other things, the research that has resulted from these coding projects. The appendix, pages 245–306, contains a comprehensive discussion of the coding procedures both for the French study and the one currently in progress for England during the tumultuous years of 1828–32.

23. Tilly, Charles and Rule, James, Measuring Political Upheaval (Princeton: Center of International Studies, Woodrow Wilson School of Public and International Affairs, Princeton University, Research Monograph No. 19, 1965), p. 65Google Scholar.

24. Tilly and Rule, ibid. pp. 61–100, explains the strategy for detecting relevant press biases in historical studies of collective action. Some have elaborated the art of checking media biases into an exacting science. See, for example, Snyder, David and Kelley, William, “Conflict intensity, media sensitivity and the validity of newspaper data,” American Sociological Review, Vol. 42 (02 1977), pp. 105–123CrossRefGoogle Scholar; and Danzger, Herbert, “Validating conflict data,” American Sociological Review, No. 40 (10 1975), pp. 570–84CrossRefGoogle Scholar.

25. Note that thematic and elite analyses do not face identical problems. When thematic analysis is applied properly to the limited range of questions for which it is appropriate, the problem is only that of moving from A to B, since the manifest content of the media is of immediate interest. For elite analysis, researchers can confidently move directly from A to C, since they can make a convincing case for the validity of most printed information about individual elite careers.

26. For an introductory discussion of the assumption of independence and the kinds of sampling problems discussed here, see Blalock, Hubert M., Social Statistics (New York: McGraw-Hill, Second Edition, 1972), pp. 142–45Google Scholar.

27. This is the approach used by Gamson, William A. in his The Strategy of Social Protest (Homewood, Illinois: Dorsey, 1975)Google Scholar. Gamson, employing historical monographs as his source of information, wanted a representative sample of challenging groups in the American polity for which he could code characteristics and form an argument. From a bibliography of 75 books covering each type of political group, he enumerated over 4,000 index entries for the period 1800–1945. From this comprehensive list he selected 467 at random for closer examination.

28. Experienced researchers are painfully aware of the drastic reduction in the number of provincial and local newspapers available outside China after the early 1960s. Research projects which straddle this great divide will be unable to adopt this “fixed proportion” sampling approach. Systematic between-periodical comparisons and, if called for, adjustments in findings, will be a standard tool of the trade for any study which spans a time period of decades.

29. Visitors' reports range from the standard guided tour of short duration to longer stays by foreigners in a specific setting for the express purpose of study. Recent village studies by American social scientists are an example of the recently expanded role of this latter type of visitors' report.

30. See Andors, , “Factory management,” and China's Industrial Revolution, pp. 97–134Google Scholar, which presents a slightly revised version of the earlier article.

31. Andors, , “Factory management,” p. 451Google Scholar.

32. Andors, , it should be noted, was quite frank in stressing this possible drawback: “The judgments about degree of continuity are of necessity preliminary, based on somewhat fragmentary evidence, subject to personal interpretation” (China's Industrial Revolution, p. 119)Google Scholar. The point here is that the level of personal interpretation could be greatly reduced by formulating careful rules for recognizing that certain “something” that sets apart different leveb of continuity. This is a difficult exercise, but crucial to the success of the entire research method.

33. There are two discrepancies between Andors' Tables 1 and 2 in the original article. The totals recorded in the nine right-hand columns of Table 2 do not completely tally with the raw data presented in Table 1 for two enterprises – Tsingtao Iron and Steel, and Tientsin Vehicles. We have corrected the totals in Table 2 to coincide With the information in Table 1. All subsequent tables reported here are based on these corrections.

34. Andors assigned the codes 1, 2 and 3 to “high,” “some” and “little or no” continuity, respectively. We changed the codes so that 2 was applied to “high,” 1 to “some” and 0 to “little or no” continuity. The higher the coded value, the higher the subjective assessment of continuity. This makes possible easily interpretable mean values. One should note, however, that these mean values are based on a questionable assumption on our part. These three continuity categories are ranks, and we cannot strictly assume that the distance between “high” and “some” is the same as that between “some” and “no” continuity. We could just as easily have assigned scores of 9, 7, and 1 to our three categories. We use these mean values solely for the purposes of illustration, with this caveat.

35. Note Andors', observation (China's Industrial Revolution, pp. 104 and 120)Google Scholar that the triple combination committee had become more an organ for communication than an arena for co-operative decision-making.

36. These significance tests are described in Blalock, , Social Statistics, pp. 220–23 and 226–28Google Scholar. A careful reader will note that, strictly speaking, the overall means for the rows (mode of participation) still violate the assumption of independence. Most of the enterprises sampled contribute examples of more than one mode of participation, so the means for each row are derived from samples that are not completely independent of one another. The bias this introduces tends to make the row means look more alike, since a firm scoring high on one mode of participation will be more likely to score high on another. When applying a difference of means test to the total row means for “mode,” therefore, we must recognize that we are looking not at differences in the continuity of these modes of participation reported for China generally but at differences within single enterprises. Any differences we find, therefore, will probably tend to understate differences existing in the country as a whole. Note that the total column means for sector do meet the criterion of independence, since these are distinct enterprises.