Is positivism in sociology dead? While positivism may be dead in the sense that philosophers have found it to be wanting (Suppe, 1977), its ghost continues to haunt sociology in a variety of guises, ranging from sociological technicians' programs of sophisticated statistical manipulation of carefully quantified data to debates about whether sociology can be scientific without positivism (Halfpenny, 1982: 120-121; Turner, 1992). In theory circles, the term "positivist" has become one of derision (C. Bryant, 1985; J. Bryant, 1992; Gibbs, 1994). With the notable exception of Jonathan Turner, few sociological theorists seem intent on proclaiming their positivism, at least in public.(1) However, the question we ask is: Judging from their published work, do sociologists find positivism persuasive?
It would certainly not be surprising to find evidence that positivism is more in "bad odour" today than twenty years ago. By the late 1960s the "received view" of logical positivism was at its sociological zenith, firmly stamped in the burgeoning, and predominantly North American, sociological literature on theory construction (see, e.g., Zetterberg, 1965; Hage, 1972; Wallace, 1971; Dubin, 1969; Gibbs, 1972; Blalock, 1969; Mullins, 1971; Homans, 1967; Reynolds, 1971; Stinchcombe, 1968; Wilier, 1967). Positivism itself had been no stranger to controversy. Yet critical philosophical commentary about the "received view" (Suppe, 1977) was slower to make its way into the mainstream of sociological thinking, at least in North America. While positivistic sociology there was not without its critics (Blumer, 1969; Garfinkel, 1967; Glaser and Strauss, 1967), sociological skeptics were better established overseas (e.g., Adorno et al., 1976; Horkheimer, 1986; Habermas, 1972).
The last two decades have witnessed the working out in sociology of the philosophical criticisms of positivism (Suppe, 1977), particularly within relativistic approaches (Ritzer, 1975; Horan, 1978; Falk and Zhao, 1989) and realism (Sayer, 1992; Keat and Urry, 1982; Morrow, 1991). New voices have contributed to a growing scepticism about positivism's role in sociology. Feminist sociology has proposed a distinctive epistemology that confronts positivistic sociology on a number of grounds, particularly the difficulty of producing abstracted, impartial accounts given the embeddedness of knowledge within social relations (Acker, 1989; Haraway, 1988; Lengermann and Niebrugge-Brantley, 1992; Smith, 1989). Post-structuralists and post-modernists have challenged positivism for, among other things, its universalizing claims, its rationalism, and for the modern belief that theory mirrors reality (Best and Kellner, 1991). Proponents of historical-comparative methodology point to "intrinsic logical and substantive flaws" in the recent writings of positivist theorists (Bryant, 1992). Remarkably, some critics have even assembled evidence of the self-doubts expressed by leading proponents of positivistic sociology as evidence of "positivism's twilight" (Baldus, 1990).
Faced with this litany of criticism, the proponents of positivism in sociology have not remained silent. The assaults on positivism have clearly had an effect on this group. While some continue to be defiant (Lenski, 1988; Turner, 1992; Gibbs, 1994), others have advocated compromise (Collins, 1989) or reflection on failure with the hope of renewal (Hage, 1994). To what degree anti-positivist sentiments have diffused beyond theory circles remains an open question (Hage, 1994: 13; Collins and Waller, 1994: 29). For his part, Turner voices his suspicion that ". . . within the larger sociological community, my position is still in the majority. But social theorists have become decidedly anti-science" (1991: 590). Our research has been prompted by a similar suspicion, or curiosity, about what sociologists involved in the everyday (and every-night) activity of writing papers for academic journals find to be persuasive. How "positivistic" is their work?
We address this question through a content analysis of articles published between 1967-1970 and 1987-1990 in the official journals of the American, Canadian, British and Scandinavian sociological associations. We develop a 7-item scale that reflects what Bryant (1985) has characterized as "instrumental positivism" as it was developed in the "theory construction" literature of the late 1960s - as we will argue, the most representative of the many extant statements of positivism. The dimensions are assessed in each of 176 articles chosen at random to represent journals and time periods. These data reflect the extent of positivistic content in sociological work published in these journals in each period. We are particularly interested in changes over time and in cross-national differences. Our analysis is exploratory and inductive in the sense that we document change and then turn in our discussion to the question of how to make sense out of patterns in the data. In this spirit, we prefer to pose several orienting questions rather than stating explicit hypotheses. To what extent has the positivistic content of sociologists' written work changed in response to the stream of critical commentary levelled at positivism over the past two decades? Has Europe experienced greater change than North America? Are there distinctive patterns of change for each of the national association journals?
We begin our analysis by presenting and briefly justifying the characteristics of positivism that we looked for in our colleagues' published work.
Characteristics of Positivism
The controversy surrounding positivism concerns not only its relevance as a philosophy of science, but the very meaning of the term itself. Both the philosophical (Suppe, 1977) and sociological versions of positivism have changed over time (Bryant, 1985; Halfpenny, 1982; Giddens, 1977), so that specifying a single definition, or a unitary set of positivism's characteristics, is impractical. One sociological account of these developments stresses the importance of a differentiated view, with no less than twelve different understandings of the term "positivism" (Halfpenny, 1982).
Our tack is somewhat different. We focus on the characterization of positivism that is "received" in contemporary sociology among advocates of positivism and their critics. This version of positivism was articulated in the theory construction literature of the late 1960s (Zetterberg, 1965; Hage, 1972; Wallace, 1971; Dubin, 1969; Gibbs, 1972; Blalock, 1969; Mullins, 1971; Homans, 1967; Reynolds, 1971; Willer, 1967; Stinchcombe, 1968) and incorporates elements of several earlier forms of positivism (Halfpenny, 1982; Bryant, 1985). In particular, the theory construction movement adopted the Hempel-Oppenheim "covering-law" model of explanation and prediction (Hempel and Oppenheim, 1970; Halfpenny, 1982: 65) and the view of theories and the components of theoretical systems that has developed as part of the "received view" (Suppe, 1977; Keat and Urry, 1982: 9-26). The "instrumental" positivism (Bryant, 1985; Alexander, 1982) associated with developments in American sociology in the first half of this century was also given expression in the theory construction literature. Instrumentalism refers to a preoccupation with method - replicable, exact measurements, empirical relationships, and statistical techniques (Ross, 1991: 429).(2) We consider each of the elements of this view of positivism in turn (Table 1).
Concepts related in law-like statements
In the logical positivist view, theories incorporate sets of general, law-like statements whose truth or falsity is assessed by means of observation or experiment. Despite attempts in sociology to list propositions that might qualify as "laws" (Berelson and Steiner, 1964; Blain, 1971; Homans, 1974), it is generally recognized that few propositions meet the Hempelian criteria of simplicity, precision, economy, and universality (Halfpenny, 1982: 65). Hence, in our coding, the reference to the more relaxed term, "law-like" statements.
Laws (or "law-like statements") express relations among concepts, which are abstract, unobservable theoretical terms denoting sets of observable phenomena. Laws are thus distinguished from relations among observables, or variable relationships. One understanding of the term "positivism" is as a body of laws that express causal relations, an idea that has invoked considerable controversy both among positivists and their critics (Halfpenny, 1982: 68). In sociology, these linkages specify reasons why concepts should be linked in a particular way; such "theoretical rationales" are most often expressed in causal language (Hage, 1972: 85-99).
Our coding specifies that concepts be related in law-like statements to distinguish this use of concepts from applications in which concepts appear singly or in loosely aggregated "sensitizing schemes" (Turner, 1991: 10-11).
Nominal definition of concepts
The development of the received view is characterized by a relaxation of a strict operationalism in which non-observable terms in theoretical systems were taken to be meaningless (Hempel, 1958). Nominal or theoretical definitions thus outline a "meaning space" of concepts (Hage, 1972: 63-4) that is filled only in part by empirical indicators (instances of the concept in the social world). The meaning space is taken to be open to acknowledge the possibility of unanticipated empirical applications and novel conceptualizations. There is thus a close, dialectical connection between nominal and operational definitions: "without a theoretical definition, the indicators can remain too specific; without an operational definition, the meaning can remain too diffuse" (Hage, 1972: 67).
Operational definition / partial interpretation
Operational definitions specify empirical indicators of concepts; they thus link together the abstract, unobserved world of theory with the world of observation. The result of such definition is a "variable" signifying how the property varies over whatever social entities are to be measured. Logical difficulties of operationalism left the received view with the tenet that these measurement operations only partially interpret or define theoretical concepts (Suppe, 1977; Keat and Urry, 1982: 20). It is interesting that the popularized version of positivism in sociology continues to refer to the empirical realization of abstract concepts as "operational" definitions (Hage, 1972: 62-84).
Derivation of a hypothesis for empirical examination
Hypotheses specify "operational linkages" between variables, that is, the nature of the functional form of a variable relationship (Hage, 1972: 99-110). Thus they are particular empirical realizations of the relations between concepts in the unobserved, law-like theoretical statements. Under the covering-law model of explanation and prediction, descriptions of "lower-level laws" or generalizations can be derived logically as conclusions from the law-like statements and conditions that form the premises of an argument. An hypothesis, under this version of positivism, is an empirically defined, logically derived statement of relations between variables. The component parts of an hypothesis, variables, and "linkages," to use Hage's term, are implied in the concepts and relations specified in the more abstract statements of the theory.
In our experience, much hypothesizing in empirical sociological practice barely resembles this positivistic prescription. More often, sociologists present hypotheses based on implicitly deductive theoretical reasoning from which propositions, definitions, and the like could be reconstructed. Our coding allows for both types of hypotheses.
Formal language (logic or math) to express laws
"Logic" imparts its name to "logical positivism" with good reason. Under the received view, the terms of theoretical systems are formulated in a "first-order language," such as mathematical logic (Suppe, 1977: 50). The fundamental vision of the logical positivist movement involved a coupling of the dual strengths of empirical truth (building science on a bedrock of observation) and the analytical truth of logic. Because their derivations are usually informal, sociologists have only rarely translated verbal theories into formal languages (Freese, 1980) - this despite the exhortation to formalization present in much of the theory construction literature.
Variables related together empirically
The "instrumental positivism" described by Bryant (1985) has been less theoretically driven than the deductive-nomological model adopted by sociological theory construction would suggest. Indeed, this aspect of empirical research has been a sore point with the advocates of positivism (Willer and Willer, 1973; Warshay, 1975) as well as with its critics (Baldus, 1990:151), and has been a focus of content analysis of theorizing in the American Sociological Review (Wells and Picou, 1981). By itself, the presence of variable relationships is a sign of an empiricist, "instrumental" positivism. In combination with the other dimensions of positivism, this code refers to an empirical relationship that is related to a more abstract body of theoretical statements.
Use of statistical techniques / quantitative methodology
Quantitative methodology is, literally, a defining characteristic of instrumental positivism (Bryant, 1985: 139-140; Halfpenny, 1982: 27-45; Ross, 1991: 428-436). Although researchers with qualitative orientations have endorsed positivism (Halfpenny, 1982), contemporary critics invariably associate a preoccupation with statistical techniques with positivistic sociology (Baldus, 1990: 152; Bryant, 1992: 44).(3)
Two aspects of this characterization of positivism should be underscored. First, the positivism of the theory construction movement encompassed more than the Hempel-Oppenheim covering-law model. While the first five characteristics represent components of this model, the theory construction literature also emphasized empirical relationships and the use of statistical methods. Second, the seven characteristics in Table 1 do not constitute a set of necessary conditions in the sense that all seven must be present for an article to be considered to be "positivistic." These characteristics represent the methodological prescriptions of the theory construction literature, all of which were not necessarily met in sociological practice. To the extent that an article employs these characteristics, it is considered to be more or less "positivistic." Hence being "positivistic" is a matter of degree, rather than an all-or-none categorization. We would argue that this construal of positivism not only reflects sociological parlance in which research is referred to as being more or less "positivistic," but that it also permits us to assess the degree to which this occurs in practice.
Each characteristic of positivism was rated for each article using three different measures. The first coded whether a particular characteristic was absent (= 0), implicitly present (= 1) or explicitly present (= 2). In practice coding, we found that certain characteristics were sometimes present in an author's presentation without being stated directly; hence the "implicit" category. For example, "nominal definitions" might be presented discursively in the discussion of a concept, rather than as an explicit definition (see, e.g., the discussion of "class" in Hoff and Andersen, 1989). The raters were also asked to rate the difficulty of coding each characteristic, and their confidence in the coding, both on seven-point Likert scales. Scores on "presence" for the seven items were averaged to form a "positivism" index for each article.(4) Difficulty and Confidence indices were also calculated for each article.
We drew a stratified random sample of articles from the American Sociological Review (ASR), The Canadian Review of Sociology and Anthropology (CRSA), Acta Sociologica (Acta) and Sociology, during two time periods: 1967-1970 and 1987-1990. Sampling fractions were set to yield a minimum of approximately 20 articles for each journal in each of the two time periods. In total, 176 articles were coded. These journals were chosen in order to reflect sociological work in the official journals of the four largest English-speaking sociological associations. Presumably the published content of these journals reflects the policy of their associations' editorial boards concerning the nature of sociology and the mission of the journal in relation to such views.
This project was originally conceived as a way of providing students with first-hand introductions to the nature of positivism in sociological research and the method of content analysis. The articles were coded for class assignments by a group of 31 undergraduate students at the University of Victoria, who coded 85 of the 176 articles, and by 7 graduate students at the University of Alberta, who coded the other 91. The Alberta group also recoded 24 randomly selected articles coded by the Victoria students in order to assess test-retest reliability.
We began this exercise by having our students do background readings on positivism and content analysis in sociology. Both groups of students practised the coding scheme successively on two articles, and after each practice session we discussed the codes and the students' ideas for improving the coding. These discussions generated considerable interest among the students and some notable clarifications to the codes.
In our analysis we examined three different estimates of reliability: 1) proportion of variance attributable to differences between articles versus differences between coders; 2) test-retest reliability; and 3) item-to-total reliability. Inter-rater reliability was assessed by estimating "coder error" as the variance of ratings across raters coding the same article. Variances for the positivism index in three practice articles were consistently low, indicating that this index is substantially reliable.(5) We also estimated the proportion of reliable variance for the positivism index to be a very high 87% (Kerlinger, 1973: 447).(6) Internal consistency (Chronbach's Alpha) for the seven-item positivism index is .80, indicating that the scale is reliable.(7) Finally, test-retest reliability was estimated by having the second group of coders recode 24 articles. The resulting correlation (r = .64) is acceptable given our small sample.(8) Taken together, these different measures indicate that the positivism index is reliable. Factor analysis also indicated that these seven items all loaded substantially on a single factor.
Table 2 presents means and standard deviations on the positivism index for the four journals at two time periods. We begin by describing the patterns of change in the table using a measure of the standardized mean difference, or "effect size" (Cohen, 1977), and then report results for an ANOVA of these data.(9)
Means and Standard Deviations: Positivism Index
Period ASR CRSA ACTA SOCI Row Avg.
1967-1970 1.43(b) 1.22 1.29 1.17 1.29 (.358) (.537) (.564) (.567) (.507) 25 21 19 30 85
1987-1990 1.43 1.11 0.97 0.74 1.09 (.531) (.473) (.631) (.532) (.598) 28 19 22 21 90
Column Avg. 1.43 1.17 1.12 0.95 (.453) (.506) (.616) (.584) 53 40 41 41
a. ASR = American Sociological Review CRSA = The Canadian Review of Sociology and Anthropology ACTA = Acta Sociologica SOCI = Sociology
b. Cell elements are: mean, (standard deviation), and N.
The row marginals in Table 2 reveal a modest decline in positivistic content over the 20-year period (Cohen's d = .36). The ASR began, in the late 1960s, with a high index of positivistic content (averaging 1.43 out of 2.0). By the late 1980s this average remained the same. The other journals also began with averages in the range of 1.17 (Sociology) to 1.29 (Acta) and become less positivistic over time. The CRSA had the smallest decrease (d = .28), Acta had a moderate decrease (d = .56), and Sociology had a substantial decrease (d = .80).
Pooling the ASR and CRSA into a "North American" cluster and Acta and Sociology into a "European" cluster is also revealing. In the first period, the North American cluster was only slightly different from the European in terms of its positivistic content (d = .22). By the late 1980s, the gap had become substantial (d = .82).(10)
ANOVA results appear in Table 3. To control for the possible effects on variation in positivism of our raters' confidence and perceived difficulty, we included these indices as covariates in the analysis. The covariates account for a minuscule proportion of variation in positivism, and their effects are not statistically significant. Thus we can be reasonably sure that these rater characteristics have little to do with variation in the positivism index.
ANOVA: Positivism Index
Sum of Mean Source of Variation Squares DF Square F P
DIFFICULTY .065 1 .065 .241 .624 CONFIDENCE .037 1 .037 .137 .711
PERIOD 1.592 1 1.592 5.925 .016 JOURNAL 7.294 3 2.431 9.047 .000
PERIOD X JOURNAL 1.678 3 .559 2.082 .105
Residual 44.341 165 .269
Both the main effects of PERIOD and JOURNAL are statistically significant, and each factor accounts for a modest amount of variation in the positivism index (eta-squared for PERIOD = .032 and for JOURNAL = .096). The interaction between JOURNAL and PERIOD is not statistically significant, however. Post hoc t-tests suggest that many of the mean differences we described above can be inferred to the respective populations. None of the journals differed significantly in the late 1960s, but by the 1980s, the ASR is significantly different from each of the other journals and the CRSA differs significantly only from Sociology. None of the other inter-journal differences are significant at this period. The average positivism index for both North American journals taken together is not significantly different from the European journals in the 1960s, but becomes significantly different in the 1980s. Across time, only Acta and Sociology are significantly different.
It is also worth noting that while the level of positivistic content of the ASR remains high, the variation among its articles increased significantly between the two periods (F = 2.20, P = .05). Thus even the most positivistic journal has become more diverse, publishing across a wider range of the methodological spectrum.
As they leave the 1960s, all four journals have substantial positivistic content, and the ASR is the most positivistic of them all. By the time they enter the 1990s, each has changed. The European journals, Acta, and particularly Sociology, are much less positivistic. The CRSA is in a middle position, having declined enough to be distinguishable from the positivistic ASR, but not so much as to resemble the markedly less positivistic Sociology. The ASR continues to be the seat of positivism, although even its content has become much more diverse.
Critics of positivism will point out, as have our students perceptively noted, that our analysis is itself somewhat positivistic. Positivism looks at positivism! This sort of self-criticism is in the positivist spirit (Halfpenny, 1982: 119). We cheerfully concede the limitations of the work we present here. It does not address, in particular, the reasons for different degrees of change in positivistic content in each of the four journals. Such an issue is a matter for comparative-historical analysis of the nature of the sociological associations, their editorial boards and policies, and the socialization of sociologists. Some first steps in this direction have been taken by Abell (1994), who traces antipositivist developments in British sociology, and by Collins and Waller (1994), who attempt to explain antipositivist movements in the U.S. across a range of academic disciplines.
We can draw two major conclusions from our results. The first pertains to the question that prompted our study: Is positivism dead? To appropriate Sam Clemens' felicitous phrase, the rumours of positivism's death have been greatly exaggerated. If we assume that the ASR's editorial policy reflects the prevailing norms of scientific practice among members of the association, then it seems reasonable to infer that sociological practice among the association's membership is predominantly positivistic. With nearly 14,000 members, the American Sociological Association is the largest English-speaking sociological association in the world (personal communication, ASA Office). Besides, even though our data indicate a decline in the incidence of positivistic content in CRSA, Acta Sociologica, and Sociology, these journals continue to publish articles with positivistic content, to differing degrees.
These observations also have implications for the role that journals and their editors play in the development of sociology. If editors acted as "gatekeepers" in the sense that they maintained the status quo, we would expect continuity in the use of positivism. What we have observed, however, is a pattern of significant change in all of the journals. In some, levels of positivistic practice have decreased. In the one journal where the overall level of positivism has not changed, the degree of positivistic content has become much more varied. While these results indicate that these many editors have presided over considerable change, we cannot assess the degree to which they have initiated these changes or acted as representatives of their national sociologies. Interviews with editors, publication board members and others would provide useful leads in attempting to investigate these issues.
Three of the national association journals in our study have major competitors (the American, Canadian and British Journals), and there are a large number of other sociological journals in each of the four countries. Indeed, there has been a proliferation of sociological journals over the past two decades. Further research will examine the degree to which these journals have provided publication outlets for nonpositivistic approaches.
A second conclusion concerns the schism between the worlds of theory and research. This gap has been attributed to, among other things, the theorists' tendency to be more concerned with other theorists than with the problems generated by research (Menzies, 1982: 188-189; Abell, 1994: 107; Lieberson, 1992: 2-7; Collins, 1988: 494). Theorists have announced the demise of positivism in sociology in terms that give little voice to sociological practitioners. For instance, Alexander and Colomy assert that ". . . the discipline's stance toward orthodox positivism has changed fundamentally, that what once could be readily assumed about the nature of sociological inquiry has recently become an object of scepticism, if not downright derision. . ." (1992: 31). Perhaps Turner's sense of defeat, of representing a minority view among theorists, springs not from systematic evidence in the realm of sociological practice, but from experience in the realm of theory.
It has been argued that philosophies of science should reflect the actual practice of scientists. In fact, this was the case in the early logical positivist movement (Halfpenny, 1982: 118). If this is true, then "it is unlikely that the whole of scientific activity is characterised by those features that one philosophy of science identifies as central for example, the quest for laws in positivism, the formation of agreements among scientists in conventionalism, or the postulation of generative mechanisms in realism" (Halfpenny, 1982: 118). Debates over whether one or another theoretical or methodological orientation is in "twilight" or "dead" are misguided. Even a strong methodological prescription in feminist research (Smith, 1989) has flown in the face of a practice revealed to be multifaceted (Reinharz, 1992). A multifaceted sociology that includes positivism implies that the theorists' view that sociology is undergoing a radical paradigm shift is well off the mark.
Cole (1992: 238) suggests that while positivism is the dominant image of science held by most working scientists - sociologists included - their actual practice resembles "realism-constructivism," in which science is socially constructed yet constrained by input from the empirical world. Cole's position is based on studies of how social forces influence the content of scientific ideas as these ideas come to be accepted by scientific communities in a number of natural and social science disciplines. While this work covers a range of practice, it does not specifically assess the scientific content of journal articles. Our research suggests that, at least in this aspect of their practice, many sociologists continue to find positivism to be persuasive. At the same time, we find that positivism's influence is variable across journals and over time, much as Cole has stressed that social influence on the cognitive content of science is a variable that must be studied empirically (1992: x).
Critics of sociological positivism will find support for their views in our finding that at least parts of the sociological community have shifted their views. Our research does not answer the important question concerning the focus of the shift. To what do these sociologists look, if not positivism? Which critical alternatives do they find to be persuasive? We expect that these alternatives will include the perspectives of critics mentioned at the outset of this paper. Representing these alternatives more adequately in content analysis is a future goal, and this may inform a consideration of whether sociological practice is fundamentally eclectic, or whether there is a core set of elements from positivism that can be merged with elements from other philosophies of science.
1. "I'm a positivist and proud of it!" Jonathan Turner, speaking at the Pacific Sociological Association Annual Meeting, April 1986.
2. Certain of these elements can be traced to positivism's roots in the French sociological tradition of the 19th century (Bryant, 1985). Our interpretation of positivism as the methodological prescription of the theory construction movement is, admittedly, more restrictive than the political and world view of positivists such as Comte and Durkheim. We would argue, however, that this interpretation reflects most closely the understanding of "positivism" whose death is debated in theory circles, as well as the understanding of what being "positivistic" means in the broader sociological community.
3. Based on Bryant's characterization of "instrumental positivism" (1985), we initially included two other elements in this scale: "verification" and "value-free." These items were excluded from the present analysis because coders reported that operational definitions were ambiguous and because they both had relatively low reliability.
4. As we have defined the positivism index, being positivistic does not imply a set of necessary conditions, and articles are not "positivistic" or "not positivistic." It is readily apparent from our results that people do use bits and pieces of a positivist perspective.
5. These variances were .04, .06, and .02 for, respectively, one practice article done by students at the University of Victoria, and two practice articles done by students at the University of Alberta. The variance of the positivism index for all 176 articles was .32.
6. "Reliable variance" is variance associated with the variability of the property reflected in the item itself. The proportion of reliable variance is calculated as
var(t) - var(e) / var(t)
(see Kerlinger 1973: 447). We defined error variance as variance among raters coding the same item, in this case, the three practice articles, and we estimated the total variance as the variance of the 176 articles. For the full data set of 176 articles, we also estimated that most of the seven individual items making up the positivism index had substantial proportions of reliable variance (ranging from .62 for "formal language" to .91 for "derivation of hypotheses"). The two exceptions among the seven items used in the index were "concepts related in law-like statements" and "nominal definitions" (.31 and .41, respectively). The source of this ambiguity appeared to be in the practice articles themselves, in which both items were "implicitly" present.
7. The difficulty scale and the confidence scale were also internally consistent (a = .88 and .86, respectively).
8. This is probably a conservative estimate of test-retest reliability, since differences between undergraduate- and graduate-student coders would contribute to unreliability. As we would expect, the graduate-student coders tended to find somewhat less difficulty in performing the coding, and they expressed more confidence in the results.
9. Based on typical results across the social sciences, Cohen (1977: 24-26) suggests conventional effect size values for small (d = .20), medium (d = .50), and large effects (d = .80).
10. It could be argued that the CRSA should be differentiated from the ASR. The CRSA was moderately less positivistic than the ASR in the late 1960s (d = .42) and became even less positivistic by the late 1980s (d = .62).
Abell, P. 1994. "Sociological theory: What has gone wrong and how to put it right, a view from Britain." In Formal Theory in Sociology: Opportunity or Pitfall? J. Hage (ed.). Albany: SUNY Pr., pp. 105-120.
Acker, J. 1989. "Making gender visible." In Feminism and Sociological Theory. R. Wallace (ed.). Newbury Park: Sage, pp. 65-81.
Adorno, T.W., R. Dahrendorf, J. Habermas, H. Pilot and K.R. Popper. 1976 . The Positivist Dispute in German Sociology. London: Heinemann.
Alexander, J. 1982. Theoretical Logic in Sociology. Vol. 1: Positivism, Presuppositions, and Current Controversies. London: RKP.
Alexander. J. and P. Colomy. 1992. "Traditions and competition: Preface to a postpositivist approach to knowledge cumulation." In Metatheorizing. G. Ritzer (ed.). Newbury Park: Sage, pp. 27-52.
Baldus, B. 1990. "Positivism's twilight." Canadian Journal of Sociology, Vol. 15, pp. 149-163.
Berelson, B. and G.A. Steiner. 1964. Human Behaviour: An Inventory of Scientific Findings. New York: Harcourt, Brace and World.
Best, D. and S. Kellner. 1991. Postmodern Theory. New York: Guilford.
Blain, R. 1971. "On Homans' psychological reductionism." Sociological Inquiry, Vol. 41, pp. 3-25.
Blalock, H.M., Jr. 1969. Theory Construction: From Verbal to Mathematical Formulations. Englewood Cliffs, N.J.: Prentice-Hall.
Blalock, H.M., Jr. 1984. Basic Dilemmas in the Social Sciences. Beverly Hills: Sage.
Blumer, H. 1969. Symbolic Interactionism. Englewood Cliffs, N.J.: Prentice- Hall.
Bryant, C.G. 1985. Positivism in Social Theory and Research. London: Macmillan.
Bryant, J. 1992. "Positivism redivivus? A critique of recent uncritical proposals for reforming sociological theory (and related foibles)." Canadian Journal of Sociology, Vol. 17, pp. 29-53.
Cohen, J. 1977. Statistical Power Analysis for the Behavioral Sciences. New York: Academic.
Cole, S. 1992. Making Science. Cambridge, Mass.: Harvard Univ. Pr.
Collins, R. 1988. Theoretical Sociology. San Diego: Harcourt, Brace, Jovanovich.
Collins, R. 1989. "Sociology: Proscience or antiscience?" American Sociological Review, Vol. 54, pp. 124-139.
Collins, R. and D. Waller. 1994. "Did social science break down in the 1970s?" In Formal Theory in Sociology: Opportunity or Pitfall? J. Hage (ed.). Albany: SUNY Pr., pp. 15-40.
Dubin, R. 1969. Theory Building. New York: Free Press.
Falk, W. and S. Zhao. 1989. "Paradigms, theories and methods in contemporary rural sociology: A partial replication." Rural Sociology, Vol. 54, pp. 587-600.
Freese, L. 1980. "Formal theorizing." Annual Review of Sociology, Vol. 6, pp. 187-212.
Fuller, S. 1992. "A plague on both your houses: Beyond recidivism in the sociological theory debates." Canadian Journal of Sociology, Vol. 17, No. 1, pp. 62-68.
Garfinkel, H. 1967. Studies in Ethnomethodology. Englewood Cliffs, N.J.: Prentice-Hall.
Gibbs, J. 1972. Sociological Theory Construction. Hinsdale, Ill.: Dryden Press.
Gibbs, J. 1994. "Resistance in sociology to formal theory construction." In Formal Theory in Sociology: Opportunity or Pitfall? J. Hage (ed.). Albany: SUNY Pr., pp. 90-104.
Giddens, A. 1977. Studies in Social and Political Theory. London: Hutchison.
Glaser, B. and A. Strauss. 1967. The Discovery of Grounded Theory. Chicago: Aldine.
Habermas, J. 1972 . Knowledge and Human Interests. London: Heinemann.
Hage, J. 1972. Techniques and Problems of Theory Construction in Sociology. New York: Wiley.
Hage, J. (ed.). 1994. Formal Theory in Sociology: Opportunity or Pitfall? Albany: SUNY Pr.
Halfpenny, P. 1982. Positivism and Sociology: Explaining Social Life. London: Allen and Unwin.
Haraway, D. 1988. "Situated knowledge: The science question in feminism and the privilege of partial perspective." Feminist Studies, Vol. 14, pp. 575-600.
Hempel, C.G. 1965 . Aspects of Scientific Explanation and Other Essays in the Philosophy of Science. New York: Free Press.
Hempel, C.G. and P. Oppenheim. 1970 . "Studies in the logic of explanation." In Readings in the Philosophy of Science, B. Brody (ed.). Englewood Cliffs, N.J.: Prentice-Hall, pp. 8-27.
Hoff, J. and J.G. Andersen. 1989. "The Danish class structure." Acta Sociologica, Vol. 32, pp. 23-52.
Homans, G.C. 1967. The Nature of Social Science. New York: Harcourt, Brace and World.
Homans, G.C. 1974. Social Behavior: Its Elementary Forms. New York: Harcourt, Brace and World.
Horan, P. 1978. "Is status attainment atheoretical?" American Sociological Review, Vol. 43, pp. 334-341.
Horkheimer, M. 1986. Critical Theory. New York: Continuum.
Keat, R. and J. Urry. 1982. Social Theory as Science. 2d ed. London: RKP.
Kerlinger, F.N. 1973. Foundations of Behavioral Research. 2d ed. New York: Holt, Rinehart and Winston.
Lengermann, P.M. and J. Niebrugge-Brantley. 1992. "Contemporary feminist theory." In Contemporary Sociological Theory. G. Ritzer (ed.). New York: McGraw-Hill, pp. 308-357.
Lenski, G. 1988. "Rethinking macrosociological theory." American Sociological Review, Vol. 53, pp. 163-171.
Lenski, G. 1991. "Positivism's future - and sociology's." Canadian Journal of Sociology, Vol. 16, No. 2, pp. 187-195.
Lieberson, S. 1992. "Einstein, Renoir and Greeley: Evidence in sociology." American Sociological Review, Vol. 57, pp. 1-15.
Menzies, K. 1982. Sociological Theory in Use. London: RKP.
Morrow, R. 1991. "Critical theory, poststructuralism and critical realism: Reassessing the critique(s) of positivism." Paper presented at the Annual Meeting of the Canadian Sociology and Anthropology Association, Kingston, Ontario.
Mullins, N.C. 1971. The Art of Theory: Construction and Use. New York: Harper and Row.
Reinharz, S. 1992. Feminist Methods in Social Research. New York: Oxford.
Reynolds, P.D. 1971. A Primer in Theory Construction. Indianapolis: Bobbs-Merrill.
Ritzer, G. 1975. Sociology: A Multiple Paradigm Science. Boston: Allyn and Bacon.
Ritzer, G. 1992. Metatheorizing. Newbury Park: Sage.
Ross, D. 1991. The Origins of American Social Science. New York: Cambridge.
Sayer, A. 1992. Method in Social Science. 2d ed. London: Routledge.
Scheuch, E. 1992. "German Sociology." In The Encyclopedia of Sociology. E. Borgatta and M. Borgatta (eds.). New York: Macmillan, pp. 762-772.
Smith, D. 1989. "Methods of writing patriarchy." In Feminism and Sociological Theory. R. Wallace (ed.). Newbury Park: Sage, pp. 34-64.
Stinchcombe, A.L. 1968. Constructing Social Theories. New York: Harcourt, Brace and World.
Suppe, P. 1977. The Structure of Scientific Theories. Chicago: Univ. of Illinois Pr.
Turner, J. 1991. The Structure of Sociological Theory. 5th ed. Belmont, Calif.: Wadsworth.
Turner, J. 1992. "If not positivism, then why is sociology important?" Canadian Journal of Sociology, Vol. 17, pp. 54-61.
Wallace, W. 1971. The Logic of Science in Sociology. Chicago: Aldine.
Warshay, L.H. 1975. The Current State of Sociological Theory: A Critical Interpretation. New York: McKay.
Wells, R.H. and J.S. Picou. 1981. American Sociology: Theoretical and Methodological Structure. Washington: Univ. Pr. of America.
Willer, D. 1967. Scientific Sociology: Theory and Method. Englewood Cliffs, N.J.: Prentice-Hall.
Willer, D. and J. Willer. 1973. Systematic Empiricism: A Pseudoscience. Englewood Cliffs, N.J.: Prentice-Hall.
Zetterberg, H. 1965. On Theory and Verification in Sociology. 3d ed. Totowa, N.J.: Bedminster.
RELATED ARTICLE: Table 1
Characteristics of Positivism
1. Concepts related in law-like statements
2. Nominal definition of concepts
3. Operational definition / partial interpretation (empirical measurement of concepts)
4. Derivation of a hypothesis (or hypotheses) for empirical examination
5. Formal language (logic or math) to express laws
6. Variables related together empirically
7. Use of statistical techniques / quantitative methodology…