Structural Models of the American Psychological Society at Birth

The American Psychological Society (APS) was formed August 12, 1988, in response to events and actions occurring within the American Psychological Association (APA) and the general community of psychologists before that date. Some consider the APS formation to be "revolutionary," fomented and effected as a revolt against the APA power structure; which has been increasingly oriented toward private-practice psychology. Others consider its formation to be a natural product of the changing nature of the field of psychology. Whether the APS is an example of revolution or evolution, this new organization has even within its first year had important influences on psychology (e.g., VandenBos, 1989). My purpose here is to present empirical data analyses that show structural portrayals of the APS early in its existence, and that reflect the relationship of the APS to both the APA and the field of psychology. The portrayals serve several purposes. First, they demonstrate some of the dynamics that drove the development of the APS. Second, they give perspective to the current structure of the field of psychology. And third, they show some possible models for the structural organization of the APS.

awarded in health-service provider areas doubled. By 1980, notably higher percentages of health-service providers were joining the APA, while higher percentages of academic psychologists were resigning or not joining. In the early 1960s the APA was split 50-50 between academic and professional affiliation; by 1981, 64.7% was composed of clinical and health service psychologists.
Several efforts to reorganize the APA in response to such demographic and political shifts have occurred during its history. Recently, a plan developed by the 1985 Task Force on the Structure of APA-allowing up to five fairly autonomous assemblies-was voted down by APA Council in February 1987. Following this event a new APA task force called Group on Restructuring wasappointed, which developed a more streamlined threeassembly model. This plan passed APA Council in February, 1988, and was sent to the body of APA voters in Summer 1988, where it was overwhelmingly defeated. 1 To lobby for reorganization and develop political cohesion, the Assembly for Scientific and Applied Psychology (ASAP) was formed in Summer 1987. When ASAP's major goal-reorganization of APA-was voted down, ASAP took advantage of the 1988 annual APA meetings in Atlanta to further its organizational goals. On August 12, 1988, by a vote of 419 to 12, ASAP was transformed by its members into a free-standing American Psychological Society (APS), with specified goals to "advance the discipline of Psychology and preserve its scientific base; to promote public understanding of psychological science and its applications; and to encourage the 'giving away' of psychology in the public interest" (quoted from the lead article of the first APS Newsletter, later renamed the APS Observer).
1. Requiring 66.7% majority for passage, the re-organization plan actually received 42.8% of the votes cast. The'intensity of feeling on both sides was illustrated by the voting percentage: Nearly 40% of the ballots were returned, compared to "typical" return rates of 25% or lower in previous APA elections. I will not try to objectively document the positions or the heated debates on APA reorganization. Some of the flavor of these issues can be found in ongoing commentary within APA's journal, The American Psychologist (e.g., Altman, 1987;Howard et aI., 1987;Matarazzo, 1987;Rodgers, 1988;Spence, 1987;VandenBos, 1989), and early comments, letters, and articles by leaders of the APS in the first several issues of the APS Observer. Rather, I will portray the structure of the APS around its birth. Defining the nature of a nascent organization is rather difficult. First, it takes some time for an organization to collect data on itself. Second, new organizations can grow so rapidly that defining "organizational structure" is like shooting at a moving target. Third, until the APS reaches some critical mass of membership and ideology, attempts to "define the organization" can be both frustrating and irrelevant.
The structure of the APS rises above some of these problems, however. The APS came about because of structural problems within the APA, an already existing, complex, and carefully studied organization. Unlike many new organizations, the APS already has a formal relationship to the field of psychology, defined by its goal to "advance the discipline of psychology and preserve its scientific base." Thus, existing data can be used to shed light on the structure of the APS early in its development, even before explicit organizational efforts have occurred.
In an earlier article (Rodgers, 1988), I used data from the APA Central Office to present a structural portrayal of the APA. This portrayal is particularly interesting within the context of the recent events noted above. The same data will be used to develop similar structural portrayals of the APS at its inception.

THE DATA AND THE METHOD
The data in Rodgers (1988) and for the current study came from APA records of the 60,106 members and the 41 APA divisions in 1986. A "membership overlap matrix" was constructed indicating the overlapping membership' between all pairs of the 41 divisions. For example, there was a higher membership overlap for Division 12 (Clinical) with Division 42 (Independent Practice) than Y!'ithDivision 3 (Experimental). Several different measures-including correlation-type measures and overlap statistics-are discussed in detail in Rodgers (1988) and Adkins (1954). Rodgers (1988) drew on work by Adkins (1954Adkins ( , 1973, who used similar data earlier in APA's history. Their studies applied factor analysis, cluster analysis, and multidimensional scalin"g models to divisional overlap data. The three datasets can, in fact, be combined to produce a longitudinal dataset with information about the APA's 82 structure in 1951,1970, and 1986. 2 The current article will take advantage of data and methods from this earlier work to study the earliest stages of the APS. Different subsets of the 1986 APA data will be used, each showing the APS in a different way. Since the APS was formed in response to perceived structural problems within the APA, it is of interest to inspect such data. 3 Figure I shows an enhanced version of the cluster analysis from Rodgers (1988, p. 380). This figure demonstrates that, even prior to the formation of APS, there was a clean, empirically-based split between "Health Care Practitioners" and" AcademiclResearch Psychologists." Clearly, the way psychologists joined divisions in 1986 separated them into one of "two types of psychology." Subsets of the APA overlap data explicitly tied to the APS can be defined using information indicating the reaction of APA divisions to the formation of the APS. At its inception, a $250 APS membership category was defined called "organizational affiliate." Because the APS was chartered during the APA meetings, its formation was quite salient to the APA divisions, which were holding executive and business meetings during the convention. The APS encouraged APA divisions who wished to affiliate. Early in Fall 1988, an APS Task Force chaired by Ann Howard polled divisional officers to determine divisional reaction to the APS: 13 of the 41 divisions from 1986 had immediately voted to pay the $250 affiliation dues to APS; 15 were uncertain or were still considering affiliation; 13 were clearly not going to support the APS through dues or otherwise.
A second indicator of divisional interest in the APS also exists. In January 1989, the APS sponsored a "Summit Meeting of Scientific Psychological Societies" on the University of Oklahoma campus. Fifty different organi-2. Using an individual differences multidimensional scaling model (lNDSCAL) the differences between the underlying structure of the 17 divisions which existed in 1951 across these three different time periods have been examined. In fact, a larger shift is shown by these analyses between 1951 and 1970 than in the more recent period. Of course part of the recent shifts in the APA are reflected in the presence of new divisions, and not in the structure underlying the original divisions.
3. Obviously, the APA data are less than perfect data for studying the structure of the APS. For example, many members who are joining the APS are those who have not been members of the APA in recent years. Logan Wright; the APS Logistics Officer; reported (personal communication) that after one year of APS existence, approximately ' 13 of the 6000 APS members are psychologists who have not recently been members of the APA. Of the remaining ¥3, approximately half ('13 of the total) resigned from the APA after joining APS, and the other half (the remaining ' 13 of the total) are simultaneously members of both the APA and the APS. Thus, the data to be used in the present analyses only account for around ¥3 of the APS membership. The other '/3 are undoubtedly different in important and systematic ways from the ¥3 for whom information is available; further study of this subset will have t~" await more complete information from the APS itself, however. Fig. 1. The structure of the APA in 1986, cluster analysis of divisional overlap (from Rodgers, 1988), with reactions by each division to the formation of APS in August, 1988. The "tree diagram" produced by the two cluster analyses (from SAS's PROC TREE) is constructed so that similar divisions appear to be clustered within the diagram; thus, the visual clusters correspond to empirical clusters. There is a "distance scale" (which is not included in the figures) running along the X-axis and indicating at what distance the various divisional clusters are joined. In Figure 1, for example, Division 1 and 2 are joined in a bar that extends far to the left, indicating low distance between them. The bar joining the Academic and Health Care clusters (on the plot between Divisions 21 and 12), on the other hand, is by far the shortest bar, indicating a large distance between these clusters. Within Figures I and 2, the divisions are listed on the right as they were empirically ordered by the cluster analysis; the labels on the left are the author's interpretation of the clusters. zations chose to send representatives, including 19 APA divisions. The organizations represented at this summit meeting were published in the March 1989 APS Observer. The 13 divisions that immediately affiliated with APS attended the summit meetings, and an additional 6 also sent representatives.
Two analyses will be presented. The first will compare the behavior of the 41 divisions to the structure defined in Figure 1. This is, in a sense, a validity study testing Whether the behavior of divisions was predictable from the 1986 APA data. The second analysis will present substructure underlying divisions that supported the APS. _VOL. I, NO.2, MARCH 1990 The cluster and scaling analyses used previously (Rodgers, 1988) were run on various subsets of APA divisional data defined above. Because of the interpretability of the cluster models, results will be based on these (which used PROC CLUSTER in SAS with Ward's procedure). These analyses can help guide the APS in developing a reasonable and empirically-based organizational structure. Figure 1 has been enhanced by indicating the reaction of the 41 divisions to the formation of the APS. Based on Neuropsychology, and Health-clustered together within the Developmental/Health-related cluster. Fig. 2. Cluster analysis of the APA divisional overlap data for the 19 divisions that attended the APS summit meeting in February, 1989. the original figure, Rodgers (1988, p. 382) concluded that •'The present analyses . . . suggest . . . a two-assembly model, in which Health Care/Clinical Psychology and Academic/General Psychology would further subdivide their interests." The behavior of the APA divisions did strongly reflect this structure. The right-hand side of Figure 1 indicates whether each of the 41 divisions affiliated with APS, was uncertain, or was opposed: 11 of the 13 active joiners were from the "Academic Cluster"; 11 of the 13 who had no intention of joining were from the "Health Care Cluster." Furthermore, 15 of the 19 divisions who attended the APS Summit Meeting five months after its formation came from the "Academic Cluster." Clearly, the APS is an outgrowth of substructure already present within the APA at an earlier time.

RESULTS
The next analysis considered the structure underlying two subsets of APA divisions supporting the APS. The "affiliators" analysis suggested three basic substructures: Experimental, Applied, and Social/Developmental. The "Summit attenders" analysis showed the same three clusters, except that developmental split off to combine with several Health-care divisions. Figure 2 shows results of this cluster analysis. Four divisions that attended the APS summit but were not in the" Academic Cluster" from Figure I-Counseling, Women, Clinical 84 CONCLUSION The basis of the eventual formation of the APS is apparent in 1986 data from the APA divisions (Figure 1). The APS has more than ideological and conceptual legitimacy; it has a clear empirical basis within the structure of the profession of psychology.
The structure underlying two-thirds of the early APS membership can be derived from APA divisional overlap data ( Figure 2). Four distinct clusters emerge: Experimental, Social, Applied, and DevelopmentallHealthrelated. Any organizational structure of the APS that is eventually developed should account for these areas within scientific psychology that cohere on objective empirical grounds.