참고문헌 (47)

1. Constructing Social Theories Stinchcombe A. L. 1968 

2. Professionalization Vollmer H. M. 1966 

3. The Structure of Scientific Revolutions Kuhn T. S. 1970 

4. Theory of Collective Behavior Smelser N. J. 1967 

5. 10.7202/001299ar Smelser . “Theory,” 115 - 18 . Mayhew, Gray, and Mayhew, op. cit., n2., Mayhew and Levinger, op. cit., n2., R. Girod, “Typologie Sequentielle de la Mobilite et Analyse Causale,” Sociologie et Societes 8 (1976): 2 

   상세보기

6. The Organizational Life Cycle Kimberly J. R. 1981 

7. Theoretical Methods in Social History Stinchcombe A. L. 1978 

8. Earlier versions of this paper have been criticized for shortchanging the contributions of the comparativist sociologists (and historians) to sequence analysis. It is correct that the comparativists have discussed contingent processes and conditional events in a wide variety of settings, thereby questioning evolutionary models such as modernization theory. Yet, traditional narrative historians have discussed contingent processes and conditional events for many years. At their most abstract, comparativists come under the “careers” heading; they give ideal typical alternative models for revolutions, or organizations, or whatever. They derive these models by contrasting the “conditions under which” things occur in one case, but not in another. Their work does not raise and consider the forms of theorizing about sequences, and these are my concerns here 

9. History and Theory Hull D. L. 253 14 1975 10.2307/2504863 

   상세보기

10. History and Theory Dray W. H. 149 17 1978 10.2307/2504843 

    상세보기

11. The Analytical Philosophy of History Danto A. C. 1965 

12. Story and Discourse Chatman S. 1978 

13. Anatomy of Criticism Frye N. 1966 

14. The Structure of Professionalism Cullen J. B. 1978 

15. For an excellent example, see Dray, op. cit., n10 

16. The Social Construction of Reality Berger P. L. 1966 

17. In this discussion of order, I have assumed linear sequences. This rules out sets of sequences that proliferate from a common origin or that concentrate onto a common termination. Thus, one might argue that all professions start out developing in the same order, say with education preceding associations, but that they then take a variety of different paths. These cases, which represent important historical genres, seem to be best considered under the heading of multiple sequences 

18. American Quarterly Moore R. L. 200 21 1975 10.2307/2712342 

    상세보기

19. In historical writing, the arrival of a steady state normally leads to sharp shifts in focus. It often signals the end of a story. At other times, it leads to a sudden speeding of narrative time, in which the duration of the steady state is treated as a single, whole period to be analyzed at once. Such a separation operates only within the historical discourse, not in the social process itself 

20. One Hundred Years of American Psychiatry Zilboorg G. 507 1944 

21. Hannan, M T, Tuma, N B. Methods for Temporal Analysis. Annual review of sociology, vol.5, 303-328.

    상세보기

22. Blau , P. and Duncan , O. D. 1967 . The American Occupational Structure 167 - 68 . New York Early path analytic studies were particularly cautious about this assumption, a caution that has vanished with time and use. For an excellent discussion 

23. It should be emphasized that these hypotheses concern sequences of particular values of variables, that is, sequences of events. One can also raise in this context questions about the sequences of variables , that is, about the proper structure of paths between the four variables at the three time periods. It is this order that must be presumed knowable, fixed across cases, and invulnerable to variation in the time horizons of the variables 

24. Allison, Paul D.. Testing for Interaction in Multiple Regression. The American journal of sociology, vol.83, no.1, 144-153.

    상세보기

25. Time Series Analysis Box G. E. D. 1970 

26. Stewman, Shelby. Markov models of occupational mobility: Theoretical development and empirical support. Part 1: Careers†. The Journal of mathematical sociology, vol.4, no.2, 201-245.

    상세보기

27. Panel Analysis Wiggins L. M. 1972 

28. Tuma, Nancy Brandon, Hannan, Michael T., Groeneveld, Lyle P.. Dynamic Analysis of Event Histories. The American journal of sociology, vol.84, no.4, 820-854.

    상세보기

29. Scientific Discovery Wimsatt W. 213 1980 10.1007/978-94-009-9015-9_13 

    

30. The Origins of the World War Fay S. B. 1966 

31. I have omitted from consideration here the important class of models for sequences found in operations research under the heading of dynamic programming. The general problem of this literature is to develop sequences of actions that fulfill a criterion, usually optimality, over the path of or at the termination of an alternating sequence of states and responses (actions) that is generated by the interaction of system structure, actor response, and random exogenous disturbance. In its normal form, dynamic programming is done under Markovian assumptions in discrete time with a fairly limited set of possible events. While the dynamic programming problem may be expressed much more generally, encompassing the general problem of sequential contingency, the standard solution strategy of backward induction from a known terminal state depends on the assumption of Markovicity. The technique has a number of diverse applications and has found specific use in decision theory. In general, dynamic programming has the strengths and weaknesses of other Markovian approaches. First, the one-step dependence implies the assumption that all of the causal past acts only through its shaping of the most recent past. Second, the actual observed sequences are treated as model-generated, as observables to be reduced to the succession of one-step transitions predicted by the model. On the general theory of dynamic programming, see K. Hinderer, Foundations of Non-stationary Dynamic Programming with Discrete Time Parameter (New York, 1970) and B. GIuss, An Elementary Introduction to Dynamic Programming (Boston, 1972). An interesting group of applications is found in Puterman M. L. Dynamic Programming and Its Applications (New York, 1978), while examples in decision theory are D. M. Kreps and E. L. Porteus, “Dynamic Choice Theory and Dynamic Programming,” Econometrica 47 (1979) : 91-100, and R. M. Cyert and M. H. DeGroot, “Sequential Strategies in Dual Control Problems,” Theory and Decision 8 (1977): 173-92 

32. A useful analogy to the distinction between these two conceptions of the social process is the distinction between harmonic and contrapuntal conceptions of polyphonic music. Classical harmonic theory treats music as a series of snapshot ensembles of sound (chords), whose successive relation is determined by formal rules analogous to causal laws in sociology. The first importance of a given note is synchronic; it helps determine the chord of which it is a part, just as causal variables in the general linear model are first understood in synchronic ensemble. As in that model, too, certain notes of the chord are more important than others, given the type of chord; and the focus of this importance may switch from one line to another as chords progress. The earlier, contrapuntal conception of music is more like the sequence approach. Each line is its own melody, and each note's first importance lies in furthering that melody. Lines are free to move at their own speed, rather than being subordinate to an underlying harmonic rhythm. Time horizons vary. As a result, there are no chords and no overall rules for their succession. The importance of different lines is a function of their inherent melodic interest and of their relation to other lines moving in parallel, contrary, or other relative motion. The intense conflicts and resolutions that make up the overall texture of a contrapuntal piece are not planned by harmonic design, but appear to arise out of the accidental confluences inherent in the relative motion of the voices themselves. The historical relation of the harmonic and contrapuntal conceptions is also similar to the relation of standard and sequential approaches to the social process. Counterpoint is the older style, but it was overwhelmed by the tremendous musical power of harmony after the seventeenth century. Only in the present century, when the possibilities of classical harmony seemed exhausted, did composers like Hindemith, Bartok, and Webern begin a conscious return to contrapuntal principles. To many modern listeners, their music seems, like the sequential approach to the social process, to deny itself many of the benefits of its alternative, the harmonic style 

33. I am, of course, ignoring all the theoretical issues involved in saying that two events are distinguishable. In empirical practice, these will be central issues of analysis. Is registration of a profession the same event as state-sanctioned monopoly of service? Not in England. Yet, functionally, the two are equivalent in the United States. Any cross-national study of professional development would have to deal with this problem directly, and more generally, it constitutes a major problem in this type of approach 

34. Psychological Bulletin Hubert L. J. 1098 86 1979 10.1037/0033-2909.86.5.1098 

    상세보기

35. Philosophical Transactions of the Royal Society of London Kendall D. G. 125 269 1970 10.1098/rsta.1970.0091 

    상세보기

36. A cycle is a “loop” within the permutation. In a cycle the i th element of the original sequence is replaced by the j th , the j th by the p th , and so on until some element is replaced by the i th This closes the loop, which is said to have period (or length) r , where r is the number of replacements. Any permutation may be uniquely factored into cycles that have no elements in common (disjoint cycles). Cycles may be of any length, from one (an element stays where it is) to n , where n is the total sequence length. While the set of cycles generates the full symmetry group, which has n! elements, it is itself considerably smaller, since many permutations are not cycles themselves, but products of cycles. Nonetheless, the number of possible cycles is large. It follows the sequence of the so-called logarithmic numbers, there being 3, 8, 24, 89, 415 possible cycles for sequences of lengths two through six. In datasets where any kind of overall regularity is present, observed cycle distributions will involve only a few of these. The problem with cycle decomposition as a sequence analysis technique is that regularity is inferred not by observed cycles but by unobserved ones. Thus, a regular, necessary subsequence of length r , whose elements always follow in immediate succession, will be evident because it will rule out all cycles of length greater than n - r . To be sure, this is a large exclusion, since there are many more long than short cycles. But immediate succession is a strict, and unusual, condition 

37. Hogan, Dennis P.. The Variable Order of Events in the Life Course. American sociological review, vol.43, no.4, 573-586.

    상세보기

38. Philosophy and the Historical Understanding Gallie's W. B. 1964 

39. One such measure is Mayhew's coefficient of sequence inequality, a standardized sum of pairwise differences between event likelihoods. See Mayhew et al., op. cit., nl. Other measures might be based on modal events or on direct fitting of multinomial distributions using moments 

40. I am omitting any discussion of some other rescue strategies. One might, for example, solve the problem of repeated events by so refining the state space that events would not be repeatable. This would permit the techniques of unique event analysis, but at the price of distinguishing events (state registration and associational certification of professions, for example) that we might not really like to separate, or about whose order we have no hypotheses, or which haven't occurred in all cases. I also omit discussion of the approaches to this problem that base measures of sequence dissimilarity on “distance” in terms of elementary operations (insertion, deletion, substitution) required to transform one sequence into another. There is an important book in press on this subject (see note 7 above). I am indebted to Dr. J. B. Kruskal for providing me with draft material from this work 

41. Finite Markov Chains Kemeny J. G. 1976 

42. Stochastic Processes Doob J. L. 1953 

43. The inertial case offers a useful illustration of the Markov model's weak formulation of a sequence hypothesis. Here the Markov model includes predictions about off-diagonal transitions that are not really part of the theory. If a jurisdiction is increasingly stamped as belonging to a particular dominant profession throughout its career, then if a new profession takes it over, we expect the erstwhile dominant profession to take that jurisdiction back more than we expect it to take over another jurisdiction of that new profession. Yet, under the Markov model, these two events have the same likelihood 

44. Abbott, Andrew. Professional Ethics. The American journal of sociology, vol.88, no.5, 855-885.

    상세보기

45. I have omitted two other issues concerning aggregation. One is the loss of information where there are several sets of cases, each group following its own typical sequence. Here aggregation hides the sequence effects. This problem may be handled by scaling or clustering the cases, before other analysis, to distinguish such groups. Another issue in sequence aggregation is the relative complexity and numbers of the sequences to be aggregated. Sequence datasets can be very small or very large. In the small datasets characteristic of historical sociology, such as in the study of professions or revolutions, there is considerable particular information about each case, of which much or most must be sacrificed for the kinds of aggregation suggested here. Large sequence datasets are often generated by survey methods that themselves perform this screening function. It is this difference in size and in relative importance of particular detail that has kept the two areas methodologically distinct despite their common theoretical interest in sequences, forcing historical sociologists to choose, in general, comparative methods (see, e.g., T. Skocpol and M. Somers, “The Uses of Comparative History in Macrosociological Inquiry,” Comparative Studies in Society and History 22 [1980]: 176-97) 

46. Granovetter, Mark. Threshold Models of Collective Behavior. The American journal of sociology, vol.83, no.6, 1420-1443.

    상세보기

47. Chains of Opportunity White H. C. 1970 10.4159/harvard.9780674437203