What counts as an experiment?:
A transdisciplinary analysis of textbooks, 1930-1970

of Guelph

Ontario Institute for Studies in Education

American Journal of Psychology
Winter 1996, Vol. 109, No. 4, Pp. 599-616

©1996 by the Board of Trustees of the
University of Illinois


The textbook definition of experiment as manipulation of an independent variable while holding all other variables constant is generally treated as transdisciplinary and transhistorical. We examined the rise of this definition in psychology and other disciplines by comparing 236 introductory texts from psychology, sociology, biology, and physics published during the 1930s, 1950s, and 1970s. The definition of experiment in psychology texts did not approach uniformity until the 1970s and was not borrowed from texts of other disciplines. The standard definition is relatively absent from physics, infrequent in biology, and appears in sociology after its development in psychology. We discuss the enshrinement of experimentation as the sole method for the discovery of causes.

Despite widespread disagreement on fundamental issues in psychology, there is a remarkable social consensus on the definition and role of experiment. In nearly all modern texts, experiment is defined as manipulating an independent variable, holding all other events constant, and observing the effect on a dependent variable (Winston, 1988). This strategy is distinguished from correlational, observational, and other kinds of studies and is presented as, generally, the best and frequently the only way to discover causes. These views are typically described as the scientific method, which, according to the dominant mythology, psychology acquired from more established sciences, particularly physics. No source or citation is provided for these ideas, yet they form a cornerstone of psychological pedagogy.

How did this view spread throughout psychology textbooks? Was this conception borrowed from other sciences, or is it unique to psychology? An informal search suggested that in other disciplines, the terms independent variable and dependent variable are rarely found. If the standard definition developed within psychology rather than within established science, then the usual description of experiment may involve an "origin myth" (Samelson, 1974) for a cherished methodological dictum. In this paper, we document the development of the textbook view of experimental method from the 1930s to the 1970s and examine the changing status of experiment. Second, we compare the construction[1] of experiment in introductory psychology texts to texts in sociology, biology, and physics in order to explore the universality of the psychologists' version of experimental method. Finally, we argue that the shift in psychology textbook rhetoric, regarding the nature of experiment, paralleled important shifts in research practice and helped psychologists build a distinct epistemological place on which to stand.

As noted by Morawski (1992), introductory textbooks are a "denigrated literature," presumed to reflect a distorted and commercialized view of the discipline. As such, textbooks have rarely been the subject of serious historiographic inquiry, with the exception of the intense interest in James's Principles (e.g., Evans, 1990). Recently there has been increased documentation of the changing content of texts (Weiten & Wight, 1992) and the changing relations of author to audience (Morawski, 1992). Moreover, recent histories of specific works (e.g., Fuchs, 1992; Winston, 1988) have highlighted the role of the introductory text in the history of psychology. In this paper, our interest is in the role of textbooks in socializing new recruits to the discipline. For this aim, the textbook treatment of method is critical in discursively locating psychology among the sciences and for demarcating psychology from everyday or folk knowledge.

The meaning of experiment[2]

Since the Renaissance, the term experiment has been used in diverse ways to describe a variety of procedures such as a trial, a diagnosis, or a dissection (see Hacking, 1975). The modern distinction between experimentation and observation was drawn most clearly by John Frederick William Herschel (1830/1987), who contrasted "noticing facts as they occur" with "putting in action causes and agents over which we have control, and purposely varying their combinations" (p. 76). John Stuart Mill (1843/1974) repeated this distinction and maintained that only with "active" experimentation (i.e., manipulation) could the cause or causes of a phenomenon be determined.[3]

In psychology, some textbook writers such as J. M. Baldwin (1889) relied explicitly on Mill for the idea that experimentation was the only route to the discovery of causes. Wundt (1902) also made a clear distinction between experimentation and observation, but his views of causality and the place of experiment differed substantially from those of Mill. Modern textbook conceptions of experiment are more closely tied to the positivist traditions of Mach, with his emphasis on the search for functional relations, than to the traditions of Wundt (see Danziger, 1979). Ebbinghaus (1885/1913) omitted observation as a method and described experimentation as the one "method of science." In these formulations, the term variable was not widely used. As Dzinas and Danziger (1992) have shown, Galtonian statisticians also played a major role in introducing the concept of variables as the subject of psychological inquiry.[4]

With the rise of Galtonian and neo-Galtonian models for research (see Danziger, 1990), the question of what constituted an experiment became considerably more complex. From as early as 1890 onward, J. M. Cattell referred to his studies of mental tests as "experimental." These anthropometric studies (e.g,. Cattell, 1890) did involve some manipulation; for tests of "pressure causing pain," "least noticeable difference in weight," and others, the stimuli are varied. However, the purpose of variation in Cattell's approach was to identify individual differences, rather than to describe a general relationship. Thus Cattell began a tradition in which the studies of differences among individuals and naturally occurring groups were termed experimental. As outlined by Winston (1990), the question of what was to be included as experimental continued to be unsettled into the 1930s, and many writers continued to use the term experimental as synonymous with empirical.

In North American psychology, the modern textbook definition of experiment[5], including the mathematical terms independent and dependent variable, was introduced by E. G. Boring (1933).[6] Tolman, Skinner, and others began to use these terms in the early 1930s, but it was Robert S. Woodworth (1934, 1938) who popularized this terminology through his widely read introductory text, Psychology, and his "Bible," Experimental Psychology (see Winston, 1988, 1990). Thus it is in the mid-1930s that a clear shift in the linguistic conventions regarding experimentation begins.

Toward textbook uniformity

To examine changes in the textbook construction of experimental method, introductory texts in psychology, sociology, biology, and physics were surveyed during three time periods: 1930-39, 1950-59, and 1970-79. In the history of psychology, these three decades encompass the introduction of the modern definition of experiment, the dominance of experimental method, and the beginnings of doubt about experimental method. Twenty texts from each discipline at each time period were randomly selected from those available among four university library collections. For physics during the 1930s, only 16 texts were available. Before inclusion in the sample, the preface, table of contents, and first chapter of each potential text were examined to insure that the text was college or university level, introductory, and general in scope.

The first question examined was whether the texts included a general discussion of research methods, in which experiment is typically defined and contrasted with other research strategies. As shown in Figure 1, the percentage of texts with discussions of research methods increased from 50-90% in psychology, from 25-70% in sociology, from 20-45% in biology, and from 16-30% in physics. Even in the 1970s, such discussions were absent from the majority of biology and physics texts. The inclusion of a research methods section in sociology texts appears to have followed the change in psychology texts.

Figure 1. Prevalence of research methods sections in introductory textbooks

The absence of research methods sections in textbooks of biology and physics might be explained by assuming that such a discussion would appear in laboratory manuals rather than in textbooks. However, this is not the case. Laboratory manuals in biology begin with a discussion of the microscope, a tradition dating back at least to the late 1800s (e.g., Pillsbury, 1893) and continuing through the 1970s. Physics laboratory manuals provide instructions on proper deportment in the lab and focus on problems of measurement error and graphing of data. In only one recent physics manual (Bernard & Epp, 1980) have we found a discussion of experimentation using the terms independent and dependent variables, but the discussion is in the context of how to graph data rather than how to set up an experiment.[7]

The use of independent and dependent variables to define experimentation increased dramatically from the 1930s to the 1970s in psychology. As shown in Figure 2, the use of these terms increased from 5% in the 1930s (i.e., 1 text, Woodworth's Psychology) to 95% of the sample from the 1970s. In sociology texts, the use of these terms has risen from 0-40%. In contrast, the terms independent and dependent variables were used once in 1970s biology texts and never used in physics texts of any period. In sum, the use of these terms in texts of the other disciplines was rare.

Figure 2. Use of the terms independent and dependent variable in introductory texts.

The most important question for this analysis was how definitions of experiment differed by decade and by discipline. The following classification was developed to categorize the definition of experiment in each text: (a) no explicit definition; (b) an empirical or systematic study, or data collection, with no mention of control or manipulation; (c) observations or repeated observations under controlled or standardized conditions, with no mention of manipulation; and (d) manipulation of a factor or variable while controlling or holding all others constant. These categories were found to be exhaustive for this sample of texts.

The distribution of definitions by discipline is shown in Figures 3-6. In psychology (Fig. 3), the shift from heterogeneity to homogeneity is clear. In the 1930s, 40% of texts had no definition, 30% broadly defined experiment as an empirical study, 10% used the notion of controlled observation, and only 20% defined an experiment in terms of manipulation of one variable while holding other variables constant. This diversity was still present in the 1950s texts, but by the 1970s, 80% of the texts defined experiment in terms of active manipulation. The pattern for sociology texts (Fig. 4) shows some similarity to the trend for psychology but remains more diverse. That is, there is a decline in texts with no definition of experiment, and a rise in the

Figure 3. Psychology texts: Definitions of experiment

use of manipulation of a variable, but only to 50%. When sociology texts of the 1970s do define experiment in terms similar to psychology texts, the choice of examples is noteworthy: such texts sometimes use the Milgram obedience work or the Zimbardo prison experiment as illustrations of sociological experiments. Thus sociology text writers may expropriate the treasured work of experimental social psychologists in order to present sociology as a discipline that includes experimentation.[8]

Figure 4. Sociology texts: Definitions of experiment

Biology texts (Fig. 5) also show an increase in the proportion of texts with an explicit definition of experiment, but by the 1970s, experiment is more likely to be defined as controlled observation than active manipulation. In contrast, physics texts (Fig. 6) show no substantial increase in explicit definitions. When there is a definition, no particular version is favored. Of particular interest is that in the 1970s sample, no physics text defined experiment in terms of active manipulation of a variable. In the 25% of texts with a definition, 15% defined experiment as an empirical study, and 10% defined experiment in terms of controlled observation. Physics texts often describe the precise measurement of a quality or measurement to test a theoretical prediction as examples of experiment.

Figure 5. Biology texts: Definitions of experiment


Figure 6. Physics texts: Definitions of experiment

The construction of method in psychology is further revealed by examining the categories each text contrasts with experiment (Fig. 7). Five categories were identified in these texts: (a) correlational studies; (b) comparison of naturally occurring groups or "quasi-experiments"; (c) case study, clinical, or biographical method; (d) naturalistic observation; and (e) surveys or questionnaires. Other methods that were not contrasted with experiment, such as introspection, were not included in this categorization.[9, 10]

Figure 7. Differentiation of experimentation from other methods in psychology texts.

In the 1930s, psychology texts frequently differentiated between naturalistic observation and experiment (45%) but rarely differentiated experiment from any other categories. The 1950s and 1970s showed increasing differentiation, and by the 1970s, experiment was frequently contrasted with naturalistic observation (80%), case studies (70%), and surveys (60%) but less frequently with correlational studies (25%) and comparison of natural groups (20%). Given Woodworth's influence, it was somewhat surprising that his 1938 distinction between correlational and experimental studies did not become universal. One likely explanation is that textbook writers feared introductory students would be confused if the same term were used for both a statistical technique and a research strategy, a concern mentioned more recently by Cook and Campbell (1979). The distinction between experimental and correlational research appears to be more common in textbooks for courses in experimental psychology. In a preliminary examination of 10 texts from the 1970s, we found that 6 contrasted experimental and correlational studies. In other respects, the construction of method in the experimental texts was similar to the introductory texts: 9 used the terms independent and dependent variables, and all 10 defined experiment as manipulation of a variable while holding all others constant.

The way in which psychology textbooks treat the issue of causality is more difficult to explore and does not lend itself to simple categorization. The presentations range from those in which causality is never mentioned to those in which the cause is identified as the independent variable. Nevertheless, there is a detectable shift in tone from the 1930s to the 1970s. In the earlier texts, the treatment of diverse methods is more evenhanded, and clear statements about the superiority of experiment are rare. By the 1970s, texts almost universally warn of the dangers of inferring causality from correlation and describe experiment as the "most accurate" or "most reliable" or "most powerful" method. Other methods were to be employed only when experimentation was impossible or unethical. Texts for experimental psychology were even more extreme in their position. According to Kling and Riggs (1971) in their revision of Woodworth and Schlosberg (1954), "the great majority of the facts of modern psychology are derived from formal experimentation" (p. 2). An analogous stance would be difficult, if not unthinkable, for the writer of a biology text, in that it would deny the contributions of nonexperimental evolutionary biologists and ethologists.

Disagreement over the role of experiment is not unique to the history of psychology. In biology, the tension between naturalists and experimentalists at the turn of the century was a major issue for the discipline (Allen, 1979). Jacques Loeb's (1912/1964) position clearly implied an epistemologically superior role for experiment when he declared that "all 'explanation consists solely in the presentation of a phenomenon as an unequivocal function of the variables by which it is determined" (p. 26). To follow this Machian program and discover functional relations in biology would require experimental physiology, rather than the paleontologist's naturalistic observations, which Loeb sometimes derided (Allen, 1979, Note 16; see also Pauly, 1981). The rather incomplete resolution of this conflict (see Mayr, 1982) allowed for the appearance of a tolerant pluralism in the texts surveyed here.

The meaning of uniformity

In sum, introductory psychology textbooks gradually adopted a highly uniform view of experiment as defined by manipulation of an independent variable. This uniformity was achieved relatively recently, (i.e., between the 1950s and 1970s). By the 1970s, psychology texts imply that experiment is epistemologically superior to other methods. This view was not borrowed from the textbooks of other disciplines, although other disciplines may have recently begun to borrow the construction of method used by psychology texts. In physics, which psychologists traditionally take to be the model science, discussions of research method and definitions of experiment are generally absent. When physics texts define experiment, they generally do so in a much broader manner than psychology texts and with a different meaning accorded to manipulation of a variable.

It is tempting to view the uniformity of psychology texts as a simple function of the marketplace. Although textbook content involves business as well as academic considerations, market forces can explain homogeneity, per se, but not the adoption of any particular view. A second potential factor in the development of uniformity may be the commitment of textbook authors to experimental or nonexperimental work. A careful analysis of the training and research history of textbook authors would reveal whether experimentalists are overrepresented among these authors, and have therefore been able to control the presentation of method. An overrepresentation of experimentalists among authors would then, in turn, require explanation.

Perhaps the most important factor in textbook homogeneity of experimental method is an issue identified by Danziger (1990): psychology's anxiety regarding investigative practice. Anxiety about method and the desire to "paper over" any disagreements about method in order to appear legitimately scientific may have been a powerful force for uniformity in texts. To admit to the introductory psychology audience that basic methodological issues are a matter of social negotiation and uncertainty would threaten psychology's claim to scientific status. Instead, it was important for psychology, relatively low in prestige, to borrow features of more prestigious fields (see Whitely, 1984). Unfortunately, what was borrowed was not, as Danziger (1990) put it, "the essence of successful scientific practice" but frequently "certain unexamined and decontextualized features of scientific practice" (p. 120).

In physics, with established status, such anxiety was not at work. Consequently, discussions of method and a narrow definition of experimentation were unnecessary. If this analysis is correct, we would expect texts from early in the development of physics as an independent discipline to provide more explicit discussions of experimental method than modern texts. In one such early text by Parker (1853), experiment and observation are carefully distinguished using language very similar to Herschel's (1830/1987) Preliminary Discourse on the Study of Natural Philosophy. A full analysis of the changes in physics texts, as well as those in biology and sociology, is beyond the scope of this paper. However, further interdisciplinary comparisons may help us explore how the social construction of scientific method is tied to the collective sense of identity of a discipline. In the analysis of texts, as in the analysis of all discourse, what is left out is just as revealing as what is put in. That is, what is omitted, as in the omission of a definition of experiment in physics texts, may reveal shared assumptions that are taken for granted by a scientific community (see Tracey, 1988).

Moreover, it is necessary to look beyond the obvious features of the definition to the meaning that experimental procedures have in other disciplines. In modern physics experiments, variables are certainly manipulated, but the focus is on intricately derived theoretical predictions, not the independent variable itself. Although causality is still implicit in some aspects of physics (see Suppes, 1970), experiments in particle physics typically do not look for the way in which the manipulated variable functions as a cause (see Franklin, 1990 for examples). In psychology, identification and focus on the independent variable as cause, the tradition developed by Boring and Woodworth, remains common though not universal. Thus experiments may share surface features across disciplines, supporting the appearance of unity in method, while remaining fundamentally divergent in deeper structure.

The homogeneous, mythologized view of scientific method that came to dominate psychology texts was not a minor semantic shift in terminology. During the period from 1930 to 1970, empirical studies in psychology journals increasingly made use of treatment group designs with explicit manipulation (see Danziger 1990). In the Journal of Experimental Psychology, treatment group designs increased steadily between 1930 and 1960, while other research designs declined sharply (Winston & Blais, 1990). A later and more sudden change occurred in the Journal of Abnormal and Social Psychology. The proportion of studies using an experimental manipulation jumped from 27% in 1949 to 82% in 1959 (Christie, 1965).

This uniformity in texts and research practice was not universally shared. A number of psychologists continued the tradition of J. M. Cattell, in which nonmanipulative investigation of individual differences was considered experimental. For example, Anne Anastasi (personal communication, May, 1989), author of the most widely known text on psychological testing, reported that she considered all of her work on the interrelationship of abilities to be experimental. Anastasi was trained at Columbia in the late 1920s, and the vestiges of J. M. Cattell's influence were still present in the curriculum.

R. B. Cattell (1990) and other multivariate researchers have always insisted that factor analytic studies are "experimental," to the point of calling their organization the Society of Multivariate Experimental Psychology. Cattell (1990) recalled:

In my first visit to the United States in 1937...I was surprised to note a course labelled as "experimental psychology." Was not all psychology experimental?... An experiment is an analysis of carefully observed data, controlled or uncontrolled (p. 48).... I once submitted a multivariate experimental article to the Journal of Experimental Psychology when Arthur Melton was editor and had it returned with "This is not an experiment." (p. 56)

Cattell's resistance is understandable; by defining multivariate correlational studies as nonexperimental, other psychologists were effectively denying factor analysts the right to make statements about the causes of behavior. Insofar as experimental psychologists viewed the identification of causes as coextensive with scientific explanation, factor analysts were thus unable to provide explanation and were relegated to the epistemologically inferior role of mere description. But Cattell never gave up insisting that manipulation was NOT the only route to causality:

Causal connections can be established when it is shown that an invariable time sequence of A followed by B exists, and to show this it is not necessary to manipulate A: it suffices to know that B follows A, not A follows B. To recognize that a rain cloud is one necessary causal condition for a rainbow I do not have to create a rain cloud. It is true that manipulation usually also means a time sequence, but not that establishing a time sequence requires manipulation. Consequently, a factor analytic or correlational experiment in which the data gathering established the existence of a time sequence permits a factor analytic experiment to establish causal connections (1978, p. 9)

The requirement of manipulation is so deeply entrenched that to most experimental psychologists, Cattell's position is simply wrong and the term "correlational experiment" is self-contradictory. A general discussion of the problem of causality and causal inference is beyond the scope of this paper (see Bunge, 1959; Muliak, 1987; Suppes, 1970; Wallace, 1972). However, it is obvious that astronomy, paleontology, ethology, and other disciplines have successfully investigated the causes of phenomena without much recourse to manipulations. So great was the need to deny this possibility that one popular experimental psychology text argued "even astronomy owes much of its viability to the improvement in observational techniques and apparatus constantly emerging from experimentation in allied sciences" (D'Amato, 1970, p. 7).


The standard view of methodology in psychology textbooks, in which a particular form of experimentation is elevated to favored status, can be understood as an attempt to legitimize psychology as a fundamentally practical science. By defining experimentation in this way, psychology presented itself as a science that would uncover not just "relationships" in the form of correlations, but manipulable conditions that brought about change. Thus experiments would do more than reveal the nature of the world; they would provide socially useful information. It is no accident that the modern treatment group design had its origins in applied, rather than basic, research (see Danziger, 1990). The notion that causal information should be of practical, rather than purely theoretical, value has its apotheosis in Cooke and Campbell (1979), who extolled the value of a "recipe" approach to causality and argued that in any science, a good explanation provides causes that can be manipulated. From this perspective, the justification for experiment lies primarily in a philosophical position about the simple interrelationship of explanation, cause, and manipulation. The success of physics and the role of experiment in physics is not particularly relevant to this argument, and, in this respect, the model for experimentation in psychology was not physics but agriculture, education, and engineering.

Thus the privileging of experiment in psychology may be more than a desire to ape established natural science. Additionally, it is understandable as the logical extension of a rather crude, but widely held, philosophy of science based on, in Smith's (1992) terms, "the technological ideal." This view emphasizes the Baconian traditions of manipulation, control, and production of effects for shaping and reshaping nature and thus supports technological above contemplative aims. As shown in Ross's (1991) detailed analysis, the technocratic belief that positivist science would provide rational control over human life was a major concern in all social sciences during the late 19th and early 20th centuries, as the disciplines were organized and professionalized. Thus the change in status accorded to experiment in psychology is intertwined with the general history of positivism and scientism in modern Western thought.

The nearly universal promotion of the standard view of cause and experiment had at least four important consequences for psychology during the 1930s through the 1970s. First, it helped delegitimize alternative, nonexperimental models for scientific psychological inquiry, such as the natural history approach proposed by Gardner and Beatrice Murphy in the 1930s (Pandora, 1991). Second, it helped psychology distinguish itself from other social science disciplines as having special (i.e., causal) information not available through the uncontrolled studies of other disciplines. Third, it helped to justify the choice of laboratory animals as subjects for a widening range of problems, in that animals would permit the degree of manipulation and control necessary for causal inference (Winston, 1990).

Most important of all, the enshrinement of the manipulated independent variable helped to justify the laboratory study of attitude change, aggression, competition, moral development, and other socially embedded phenomena in the ahistorical, acultural, and decontextualized approach that was so common in the 1950s and 1960s. As Danziger (1990) has shown, the isolation of environmental variables in this way both presupposed and supported a highly individualistic model of human existence. Given that experimentation was seen as the only route to causality, this strategy seemed appropriate. But the persistence of this faith made it difficult for experimental psychologists to appreciate the developments taking place in statistics, sociology, economics, and political science during the 1950s, 1960s, and 1970s. In these fields, sophisticated strategies for drawing causal inferences from correlational data, most notably structural equation modeling, were emerging (see Blalock, 1964; 1971).[11] Whether the increasing use of these strategies in psychology will challenge the special epistemological place accorded to experiment remains to be seen.


This research was supported by a grant from the Research Board of the University of Guelph to the first author. Portions of this paper were presented at the 1991 annual meeting of CHEIRON, the International Society for History of the Behavioral and Social Sciences, Slippery Rock, PA. We thank Kurt Danziger, Rolf Kroger, and Judi Winston for criticism and comments. Correspondence concerning this article should be addressed to Andrew S. Winston, Department of Psychology, University of Guelph, Guelph, Ontario N1G 2W1 Canada (E-mail: awinston@uoguelph.ca). Received for publication July 22, 1994; revision received June 26, 1995.

[1] This paper deals only with basic definitions of experiment and differentiation of experiment from other methods. Obviously, the social construction of method involves many more dimensions not dealt with here, such as the social roles of experimenter and subject, quantification, statistical inference, etc. See Danziger (1990) for a general discussion.

[2] For useful discussions of experimentation and the natural sciences, see Franklin (1986), Gooding, Pinch & Schaffer (1989), and Hacking (1983).

[3] For a general discussion of scientific method in Galileo, Bacon, Newton, Herschel, Mill, and others, see Blake, Ducase, & Madden (1960).

[4] There were other attempts to label the components of experiment. The philosopher, W. S. Jevons, (1874) introduced the terms "variable" for the factor that was varied and "variant" for the factor that was affected by the change. However, Jevons thought that this approach was suitable only for phenomena subject to continuous quantitative variation. E. C. Tolman (1932) used the terms "independent causes" and "causal variables"; however, Tolman's use was in the context of defining his theoretical position, rather than in the context of defining experimental procedures.

[5] The modern definition of experiment generally implies some notion of randomization, although this concept is rarely mentioned explicitly in textbook definitions. See Hacking (1988) for an excellent history of randomization.

[6] I am indebted to Rand Evans for pointing out to me the appropriate passage in Boring (1933) pp. 8-9.

[7] That is, independent and dependent variable were used as in their original mathematical sense in which x and y are interchangeable.

[8] The role of experimentation in sociology was the subject of early 20th century discussions, as outlined by Ross (1991), pp. 370-371.

[9] Generally, introspection was treated in texts as a variety of data, rather than a research strategy, and was usually contrasted with "objective" data (e.g., see Woodworth, 1934).

[10] The names of the categories do not necessarily correspond to those used by text authors. For example, Woodworth's (1934) "genetic" method would be classified as a case study.

[11] Ironically, Herbert A. Simon, so admired by experimental psychologists and often counted as an experimental psychologist, provided much of the groundwork for the development of causal inference from nonexperimental data (see Blalock, 1964).


Allen, C. E. (1979). Naturalists and experimentalists: The genotype and the phenotype. In W. Coleman & C. Limoges (Eds.), Studies in history of biology, (Vol. 3, pp. 179-209). Baltimore: Johns Hopkins University Press.

Baldwin, J. M. (1889). Handbook of psychology: Senses and intellect. New York: Holt.

Bernard, C. H., & Epp, C. D. (1980). Laboratory experiments in college physics. (5th ed.). New York: Wiley.

Blake, R. M., Ducase, C.J., & Madden, E. H. (1960). Theories of scientific method. Seattle: University of Washington Press.

Blalock, H. M. (1964). Causal inferences in nonexperimental research. Chapel Hill: University of North Carolina Press.

Blalock, H. M. (Ed.). (1971). Causal models in the social sciences. Chicago: Aldine, Atherton.

Boring, E. C. (1933). The physical dimensions of consciousness. New York: Century Co.

Bunge, M. (1959). Causality: the place of the causal principle in modern science. Cambridge, MA: Harvard University Press.

Cattell, J. M. (1890). Mental tests and measurements. Mind, 15, 373-381.

Cattell, R. B. (1978). The scientific use of factor analysis in behavioral and life sciences. New York: Plenum Press.

Cattell, R. B. (1990). The birth of the society of multivariate experimental psychology. Journal of the History of the Behavioral Sciences, 26, 48-57

Christie R. (1965). Some implications of research trends in Social Psychology. In 0. Klineberg & R. Christie (Eds.), Perspectives in social psychology (pp. 141-152). New York: Holt, Rhinehart and Winston.

Cook, T. D., & Campbell, D. T. (1979). Quasi-experimentation: Design and analysis issues for field settings. Boston: Houghton Mifflin.

D'Amato, M. R. (1970). Experimental psychology: Methodology, psychophysics, and learning. New York: McGraw-Hill.

Danziger, K. (1979). The positivist repudiation of Wundt. Journal of the History of the Behavioral Sciences, 15, 205-230.

Danziger, K. (1990). Constructing the subject: Historical origins of psychological research. New York: Cambridge University Press.

Dzinas, K., & Danziger, K. (1992, June). How psychology got its variables. Paper presented at the annual meeting of the Canadian Psychological Association. Quebec City, Canada.

Ebbinghaus, H. (1913). Memory: A contribution to experimental psychology (H. A. Ruger & C. E. Bussenius, Trans.). New York: Teachers College. (Original work published 1885)

Evans, R. B. (1990). William James and his Principles. In M. C. Johnson & T. B. Henley (Eds.), Reflections on The Principles of Psychology: William James after a century (pp. 11-31). Hillsdale, NJ: Erlbaum.

Franklin, A. (1986). The neglect of experiment. New York: Cambridge University Press.

Franklin, A. (1990). Experiment, right or wrong. Cambridge: Cambridge University Press.

Fuchs, A. (1992, June). The origin and evolution of the first American textbook in psychology. Paper presented at annual meeting of CHEIRON, Windsor, Ontario, Canada.

Gooding, D., Pinch, T., & Schaffer, S. (Eds.). (1989). The uses of experiment: Studies in the natural sciences. New York: Cambridge University Press.

Hacking, I. (1975). The emergence of probability. New York: Cambridge University Press.

Hacking, I. (1983). Representing and intervening. New York: Cambridge University Press.

Hacking, I. (1988). Telepathy: Origins of randomization in experimental design. Isis, 79, 427-451.

Herschel, J. F. W. (1987). A preliminary discourse on the study of natural philosophy. Facsimile edition. Chicago: University of Chicago Press. (Original work published 1830)

Jevons, W. S. (1874). The principles of science: A treatise on logic and scientific method. London: MacMillan.

Kling, J. W. & Riggs, L. A. (1971). Woodworth and Schlosbergs Experimental Psychology. (3d ed.). New York: Holt.

Loeb, J. (1964). The mechanistic conception of life (D. Fleming, Ed.). Cambridge, MA: Harvard University Press (Original work published 1912)

Mayr, E. (1982). The growth of biological thought: Diversity, evolution, and inheritance. Cambridge, MA: Harvard University Press.

Mill, J. S. (1974). A system of logic, ratiocinative and inductive. Toronto: University of Toronto Press. (Original work published 1843)

Morawski, J. (1992). There is more to our history of giving: The place of introductory textbooks in American psychology. American Psychologist, 47, 161-169.

Maalox, S. A. (1987). Toward a conception of causality applicable to experimentation and causal modeling. Child Development, 58, 18-32.

Pandora, K. (1991, June). Dissenting from experimentalism: The natural history model and psychology during the 1930s. Paper presented at annual meeting of CHEIRON, Slippery Rock State University, Slippery Rock, PA.

Parker, R. C. (1853). A school compendium of natural and expert mental philosophy. New York: A. S. Barnes.

Pauly, P. J. (1981). The Loeb-Jennings debate and the science of animal behavior. Journal of the History of the Behavioral Sciences, 17, 504-581.

Pillsbury, J. H. (1893). A laboratory guide for an elementary course in biology. New York: Silver, Burdett.

Ross, D. (1991). The origins of American social science. New York: Cambridge University Press.

Samelson, F. (1974). History, origin myth, and ideology: "Discovery" of social psychology. Journal of Theory in Social Behavior, 4, 217-231.

Smith, L. D. (1992). On prediction and control: B. F. Skinner and the technological idea of science. American Psychologist, 42, 2 16-223.

Suppes, P. (1970). A probabilistic theory of causality. Amsterdam: North-Holland Publishing.

Tolman, E. C. (1932). Purposive behavior in animals and men. New York: Appleton.

Tracey, K. (1988). A discourse analysis of four discourse studies. Discourse Processes, 11, 243--259.

Wallace, W. A. (1972). Causality and scientific explanation. Ann Arbor: University of Michigan Press.

Weiten, W., & Wight, R. D. (1992). Portraits of a discipline: An examination of introductory textbooks in America. In A. E. Puente, J. R. Matthews, & C. L. Brewer (Eds.), Teaching psychology in America: A history (pp. 453-502). Washington, DC: American Psychological Association.

Whitely, R. (1984). The intellectual and social organization of the sciences. Oxford: Oxford University Press.

Winston, A. 5. (1988). Cause and experiment in introductory psychology: an analysis of R. S. Woodworth's textbooks. Teaching of Psychology, 15, 79-83.

Winston, A. S. (1990). R. S. Woodworth and the "Columbia Bible": How the psychological experiment was redefined. American Journal of Psychology, 103, 391-401.

Winston, A. S., & Blais, D. (1990, June). Achieving methodological consensus: An analysis of the Journal of Experimental Psychology, 1930-1960. Paper presented at the annual meeting of CHEIRON, Westfield State University, Westfield, MA.

Woodworth, R. S. (1934). Psychology. (3d ed). New York: Holt

Woodworth, R. S. (1938). Experimental psychology. New York: Holt.

Woodworth, R. S., & Schlosberg, H. (1954). Experimental psychology (2d ed.). New York: Holt.

Wundt, W. (1902). Outlines of psychology, 4th ed. (C. H. Judd, Trans.). Leipzig: Engelmann.