III' II“ WWII m I II III“ II I I +4 I I I IIIIIIII'III IIIIIIJ M I III III) III I'IIII [II IIIII'II I a .ulI I'I‘III HII'I-I'Il, “‘1 IIIII IIILII' III'IIIIIII13f I. [I “I” W! I III I ”I“; )1! HI I 'II . I,” I. 'II II‘IW . I _‘.I " |IIIII IIII IIIIIII. IIIIIIIIIIIII I I II "IIIIII‘IIIII'III:IIII‘II'III‘I” ‘ " 'III III I- ' I III ‘ _- -t $— _ A Ar‘ 1 - -Ai — __ - - 23-51 . _ F .- __ 1 W m ‘ W‘— -__- ' ' 1 .4 W I . A ~ ‘.__ III I II‘II II II II III .' "I .I ,. IIIIIII‘IIIIIIIIIII IIIIII IIIIIII .,.I . II, IIIIIII IIIIIII I .' | I IIIIIIIII «III M . =. . I ., I."'IIIJIIIIII III" III III IIIIIIII III" III IIIII (HI 1.. . ’II III" IIIIIIIIII III IIIIIIIIIII II:IIIIIIHII‘I [M II” II ttttt ‘ ~‘-—. :- In NI?” I. III! ' :. III? II I II‘ III I" IIII‘I' III: II III , I , I» III II I I III IIIIIIII IIII IQIII ' .I .. I I III II III II 'I “:I III I .I II III W: ,4 vegan: ‘fi' ll mu; IllljljlllllflllllllIlflllllslllll LIBRARY Michigan $5“ UnivchIW This is to certify that the thesis entitled EFFECTS OF SITE VISITS ON INNOVATION ADOPTION presented by Mitchell Fleischer has been accepted towards fulfillment of the requirements for Ph . D. degree in PSYChOIOQL w/f - Major professo/ Date July 12, 1979 ‘ «4 \c ‘ “WW ,3“; flu“ Mu; (I E; 39' .Mzsifit EFFECTS OF SITE VISITS ON INNOVATION ADOPTION By Mitchell Fleischer A DISSERTATION Submitted to Mighigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Psychology l979 ABSTRACT EFFECTS OF SITE VISITS ON INNOVATION ADOPTION By Mitchell Fleischer While a number of solutions to various social problems have been developed and tested, there still remains the difficulty of finding ways to increase the adoption and implementation of such solutions. This difficulty is shown to have two central components, individual and organizational resistance to change. One technique that might reduce resistance from both of these sources is a visit to the site of an innovation. An experiment was performed to test the effects of such a visit on innovation adoption. The context of the experiment was an effort to disseminate the Community Lodge, an innovative residential treatment program for chronic mental patients. The target of the dissemination effort was a national sample of state hospitals. TWenty-four state hospitals were randomly assigned to either exper- imental or control conditions. Hospitals in the control condition received a workshop on the Lodge and further consultation assistance. Hospitals in the experimental condition sent one staff member on a site visit to an exemplary Lodge in addition to receiving the work- shop and consultation assistance. Mitchell Fleischer The results of the experiment indicated that the Site Visit intervention had significant, but rather weak effects in increasing advocacy for the Lodge, decreasing uncertainty about the Lodge, and increasing boundary spanning. There was also an indication that the Site Visit had a weak impact on adoption of the program. A cluster analysis of the data indicated that two clusters of variables, dis- cussion about the program and attitude-certainty about the program, were both related to change, and that high scores on both were necessary, but not sufficient conditions for change to occur. It was concluded that a Site Visit as provided in the con- text of this experiment has only minimal effect. Some suggestions for increasing the impact of a site visit are to include additional interventions and to strengthen the components of the visit itself. Tailoring the intervention based on a number of organizational factors is also suggested. For all of those peOple still in the institutions and boarding homes ii ACKNOWLEDGEMENTS As is often the case, a very large number of people helped in a lot of ways to make this research possible. Naturally some pepple helped more than others. First and foremost among these was my chairman, advisor, and friend, Lou Tornatzky. Lou made this happen in so many ways from beginning to end. This simply wouldn't be here had it not been for his encouragement and advise. Thanks, Lou! Particular thanks must go to Dorothy Berger and the staff and patients of the Fairweather program at Anoka State Hospital, and the staff and Lodge members of Tasks Unlimited, where the site visits took place. They were kind enough to chauffeur site visitors around, to have their privacy invaded, and to put up with numbers of site visitors for days at a time. That all of the site visits went well is a testament to their skill and organizational ability, not to mention good humor. The members of my dissertation committee, Bill Fairweather, Bill Davidson, and Charlie Johnson assisted in many ways. Bill Fairweather in particular helped me a great deal to understand the concepts and the data I was dealing with. The staff of the MSU-NIMH Innovation Diffusion Project helped in many ways as well. Esther Fergus and Joe Avellar provided much emotional and intellectual support throughout the project. Dennis LaFave kept me laughing and in good spirits. JoAnn Ohm sent out a lot of questionnaires and saved me a lot of work. The National Institute of Mental Health supplied financial support for this research, as part of a grant to Lou Tornatzky and Bill Fairweather. Lou and Bill's decision to provide me with part of this funding is greatly appreciated. Last, but not least I want to thank my wife, Carol, and my son, Ethan, for putting up with what they did while this research was in progress. The project involved much travel and time spent away from home, particularly during Ethan's first year in the world. I hope you think it was worth it. iv TABLE OF CONTENTS LIST OF TABLES . LIST OF FIGURES LIST OF APPENDICES INTRODUCTION Problems of Individual Resistance Uncertainty . Reduction of Uncertainty: Communicator Credibility. Problems of Organizational Resistance . Organizational Structures Linkage and Boundary Spanning Site Visits as a Technique for Overcoming Resistance to Change . . The Present Study Hypotheses . . Effects on Social Process . Effects on Intervening Variables . Effect on Adoption of the Lodge METHOD The Innovation to be Disseminated Sample . . Hospitals . . Individual Respondents Research Design Procedure . . Approach Phase . Persuasion Phase Site Visit Phase . . . Follow- -Up and Consultation Phase . Measurement . . . . Communication Network Item . Certainty Scale Attitude Scale . Page vii ix Knowledge Test . . . . . . . . . . . . . . . 36 Change Scale . . . . . . . . 36 Internal- External Locus of Control Scale . . . . . . 37 Similarity Scale . . . . . . . . . . . . . . 37 RESULTS . . . . . . . . . . . . . . . . . . . 38 Equivalence of Treatment Groups . . . . . . . . . . 38 Effects on Social Process . . . . . . . . . . . . 4l Effect on Discussion . . . . . . . . . . . . . 4l Effect on Advocacy . . . . . . . . . . . . 43 Effects on Intervening Variables . . . . . . . . . . 45 Effect on Uncertainty. . . . . . . . . . . . . 45 Effects on Boundary Spanning . . . . . . . . . . 47 Effect on Attitude . . . . . . . . . . . . . . 5l Effect on Knowledge . . . . . . . . . . . Sl Effect on Movement Toward Adoption . . . . . . . . . 5l Post- Hoc Tests . . . . . . . 53 Post- Hoc Test of Effect on Locus of Control . . . . . 53 Post- Hoc Test on Attitude and Uncertainty . . . . . 56 Effects of Similarity and Site Visitor Characteristics . . 59 Similarity . . . . . . . . . . . 59 Site Visitor Characteristics . . . . . . . . . . 60 Cluster Analysis . . . . . . . . . . . . . . . 60 DISCUSSION . . . . . . . . . . . . . . . . . . 67 Social Process . . . . . . . . . . . . . . 67 Movement Toward Adoption . . . . . . . . . . . . 69 Factors Influencing Adoption . . . . . . . . . . . 72 Hospital Factors . . . . . . . . . . . . . . . 73 Summary . . . . . . . . . . . . . . . . . . 74 APPENDICES . . . . . . . . . . . . . . . . . . 76 REFERENCES . . . . . . . . . . . . . . . . . . l6l vi Table 11. 12. 13. 14. 15. 16. 17. LIST OF TABLES Summary of ANOVA of Pre-Test Differences Between Conditions . Effect of Position and Condition on Dropping Out Effect of Site Visit on Perceived Discussion Effect of Site Visit Effect of Site Visit Effect of Site Visit Effect of Site Visit Knowledge Effect of Site Visit Effect of Site Visit Effect of Site Visit Effect of Site Visit Effect of Site Visit Consultant Effect of Site Visit Programs . Effect of Site Visit Effect of Site Visit on on on on on on on on on on on on Personal Involvement Mean Discussion Strength Star's Mean Discussion . Uncertainty of "How-To" Uncertainty of Effectiveness . Uncertainty of Feasibility Overall Uncertainty . Calls to Consultant . Sending Letters to Visits to Other Lodge Attitude Toward Lodge Knowledge About Lodge Movement Toward Adaption by Condition . Effect of Site Visit on Locus of Control . "How-To" Knowledge vii Page 39 41 42 44 44 44 46 46 48 48 50 50 50 52 52 55 55 57 Table Page 18. Effectiveness . . . . . . . . . . . . . . . 57 19. Overall Uncertainty . . . . . . . . . . . . . 58 20. Attitude Toward Lodge . . . . . . . . . . . . 58 21. Cluster Analysis - Cluster Loadings . . . . . . . 62 22. Correlations of Clusters to Change and Each Other . . . 62 22a. Correlations of Variables in the Cluster Analysis With Change . . . . . . . . . . . . . . . 64 viii LIST OF FIGURES Figure Page l. Design of the Site Visit Experiment . . . . . . . . 26 2. Flow Chart of Activities and Data Collection . . . . . 27 3. Contents of Questionnaires . . . . . . . . . . . 33 4. Measurement of Hypotheses . . . . . . . . . . . 34 5. Movement Toward Adoption by Condition Over Time . . . . 54 6. Cluster Scores of O-Types . . . . . . . . . . . 63 7. Movement Toward Adoption by O-Type Over Time . . . . . 65 ix Appendix A. Sample of Hospitals . 8. Cover Letter to Superintendents . C. Telephone Protocol - First Call to Superintendent . D. Reminder Letter to Superintendent E. Workshop Decision Follow-Up Letter . F. Workshop Confirmation Letter . G. Schedule for Hospital—Community Treatment Program Workshop . . . . . . H. Workshop Questionnaire . I. Site Visit Letter J. Sample Site Visit Schedule K. Site Visitor Questionnaire L. Mail Out Questionnaire and Cover Letter M. Follow-Up Letter . N. Phone Follow-Up Questions . 0. Communication Network Questions . P. Overall Certainty Scale Q. Certainty of How-To Knowledge Subscale R. Certainty of Effectiveness Subscale S. Certainty of Feasibility Subscale LIST OF APPENDICES Page 77 79 82 86 88 9O 92 94 . 102 . 105 . 108 .112 .122 .124 .135 .137 .141 .144 .146 Appendix T. Z< Attitude Scale . . Knowledge Test Change Scale Internal-External Locus of Control Scale . Similarity Scale . xi Page 148 151 153 155 158 P\ v Hfrc CHAPTER I INTRODUCTION It is a truism that our present-day society is beset by more and more problems. Paradoxically at the same time, more and more solutions are being develOped by scientists and others, but rela- tively little has been done in a systematic way to promote the use of these solutions among the persons who could most benefit by them. Havelock (1973) has written about a new "science" of knowledge utilization that might provide systematic techniques for the diffu- sion of innovations. Havelock and Rogers and Shoemaker (1971) have reviewed recent developments in this area. The path of an innovation or new program from its initial conception to mass adoption is long and arduous. People seem to inherently resist change, whether that change might benefit them or not. As LaPiere (1965) has pointed out, almost all innovations, from the wheel to the computer, have been feared, ridiculed, and resisted. Given the relative success of the human race, it may be said that, in the past, society has managed to adopt those innova- tions it needed, while discarding those it deemed useless. This argument states that change takes place slowly because the "natural processes“ in society need time to determine an innovation's suita- bility. Although that point is moot, it is clear that while the need for change is accelerating as society grows more complex, the 1 capacity of individuals and organizations to cope with change is becoming increasingly strained (Fairweather& Tornatzky,l977: Toffler, 1971). Thus, it seems necessary to make the change process more rational and efficient. As evidenced by Havelock's (1973) and Rogers and Shoemaker's (1971) reviews, and some more recent work (Emrick, 1977) innovation diffusion research has been steadily increasing in recent years. However, surprisingly little of that work has involved experimenta- tion to test the effects that various interventions can have on the diffusion process. Fairweather, Sanders, and Tornatzky (1974) have provided an example of what can be done toward this end by performing an experiment to diffuse an exemplary mental health innovation, the Community Lodge. The present study is an extension of that work, and attempted to diffuse the same innovation. Frequently (and particularly in the area of human services) an innovation must not only be accepted by an individual, but must be introduced into an organization. Although organizations are indeed composed of individuals, the whole in this case is very different from the sum of its parts. While organizations often resist change for many of the same reasons as do individuals, due to the complexity of most organizational structures additional problems present themselves when an organization is the target of a change effort. As a consequence, in order to adequately discuss ways of overcoming resistance to change, it will be necessary to divide the discussion into two parts: 1) problems of individual resistance; and 2) problems of organizational resistance. Problems of Individual Resistance Among the primary causes of resistance to change is simple fear of the unknown. As LaPiere (1965) has pointed out: Men always and everywhere accept with considerable complai- sence what is familiar to them, whatever it may be and however disagreeable it may seem to members of another, different society, apparently because almost anything familiar is less disturbing emotionally than is something unknown. It is an oversimplification to say that men fear the unknown; it would be better, perhaps, to say that what is designated as fear (or apprehension, dread, or the like) are those emotional disturbances that are induced by the contemplation of or exposure to what is unknown or unfamiliar and hence unpredictable (p. 177). Havelock (1973) says that this fear, or threat, occurs if the new behavior required by an innovation represents unfamiliar elements. or if the change threatens an individual's status. LaPiere agrees, calling the latter a "rational basis for resistance" in that an innovation may be contrary to an individual's self-interest. Fear, and consequently resistance, occurs because the individual is uncertain as to the outcome of his adopting the innovation. If the change agent can clarify the ambiguity surrounding the innovation the potential adopter will be more certain about the outcomes of adopting it. Uncertainty. Rogers and Shoemaker (1971) have described the change process in a different fashion. They ascribe five characteristics to an innovation. These are (1) observability; (2) trialability; (3) relative advantage; (4) compatibility; and (5) complexity. Four of these five characteristics (observability, trialability, complexity and compatibility) are directly related to uncertainty_about the innovation. If the potential adopter can observe the effects of the innovation,lw3will be more certain about the results if he adopts it. Similarly, if he can try the innova- tion (or a imart of it) before committing himself to it, he will be 'less likely to fear it. Finally, if the innovation is compatible with his present technology or belief system, he will have less reason to be uncertain. Schon (1967) has also discussed resistance to change in terms of uncertainty. In a manner similar to that of March and Simon (1958), Schon discusses the differences between risk and uncertainty. Risk is the likelihood that an action will produce an unwanted result. Note that in risk there is a known probability involved. In uncertainty, the probabilities involved are unknown. Uncertainty can be the result of two things, either not enough information or too much information. While it is probably obvious how insufficient information could result in uncertainty, the problem of too much information may be less clear. The difficulty with a surfiet of information is in the individual's or organization's inability to process it. An overwhelming amount of information simply overcomes one's ability to organize and make sense of what one has. Schon feels, as does Galbraith (1967) that this inability to process the information available is one of the major causes of uncertainty leading to resistance to change. Reduction of Uncertainty: Communicator Credibility. It would seem then that the best way to reduce the uncertainty involved in resistance to change is to provide better means of understanding and processing the information that is available. This leads to two courses of action. One Option would be to alter the organizational systems involved so that information can be processed more effec- tively and uncertainty thereby reduced. This clearly has greater implications for organizational resistance and will be discussed at length in that section. A second alternative would be to alter the quality of the information itself so that it is more comprehensible and believeable. This later course would have direct impact on individuals. There are a number of ways to improve the quality of infor- mation that is being provided. Clearly, if someone is more certain that the informationlmais receiving is accurate, his uncertainty about the effects of an innovation will be reduced. The quality of the information, as perceived by the potential adopter, is mediated by certain characteristics of the communicator. For the most part these relate to how credible the change agent is perceived as being. Some of the factors involved in credibility are the legitimacy of the change agent's role (Havelock, 1973), the trustworthiness, prestige, and perceived expertise of the change agent (Hovland & Weiss, 1951; Hovland, Janis, & Kelley, 1953), and whether the po- tential adopter believes the change agent has an ulterior motive for his activities (Walster, Aronson, & Abrahams, 1966). Another crucial factor in communicator credibility is how similar the commun- icator and recipient are. Brock (1965) showed that “the extent that the recipient perceives that he and the communicator share an attribute, that is have a similar relationship to an Object, to that extent is the recipient's behavior with respect to that object likely to be modified by the communicator's influence attempt" (p. 650). Berscheid (1966) also showed that amount Of opinion change is related to communicator-recipient similarity, as long as that similarity was relevant to the Opinion being changed. In a more recent study Tuppen (1974), using cluster analysis, found that there were five dimensions having to do with communicator credibility. These were trustworthiness, expertise, dynamism, charisma, and co-orientation. This last related to how similar the recipient of the communication thinks the communicator's ideas are to his own. Rogers and Shoemaker discuss characteristics of the change agent that make adoption Of an innovation more likely. One of the primary characteristics is homOphyly (or similarity) between the change agent and the potential adopter. Roger's rationale for this principle is that communication is facilitated when two individuals are similar. One of the reasons behind the rather consistent finding that similarity makes a communicator more credible could have to do with the reduction in uncertainty accompanied by what Festinger (1954) called the Social Comparison Process. Under conditions Of anxiety people tend to evaluate themselves (and their abilities, Opinions, and emotions) by comparison with others. Schacter (1951) referred tO this process in another way by saying "on any issue for which there is no empirical referent, the reality of one's own Opinion is established by the fact that other people hold similar Opinions" (p. 191). He later qualified this (Schacter, 1959) by saying that comparison could only be effectively made if the others were similar in some relevant manner. Radloff (1961) stated the main hypothesis to be derived from Social Comparison Theory: "A person who is uncertain about the correctness Of one Of hisopinions,fbr which he finds no Objective criteria available by which to evaluate its correctness, should seek affiliation with other people in order to evaluate his Opinion via social comparison." Radloff provided subjects with information about the Opinions Of a) no others, b) irrelevant others, c) peers, d) experts. The subjects need for social comparison decreased from a) to d). This showed that ten- dencies toward comparison varied according to strength Of evaluative need. These studies add weight to the above discussion concerning information. The potential adopter of an innovationis;1ikely to perceive information as being of higher ggality (with the conse- quent reduction in uncertainty) if he perceives the change agent as being more credible and similar. Social comparison theory seems to provide an underlying explanation for this process. Problems of Organizational Resistance It might appear that organizational resistance to change would have no relationship to individual resistance, however organi- zations resist change for many Of the same reasons as individuals. One Of the most important of these is the degree of uncertainty associated with the innovation. However, the process involved in reducing this uncertainty by individuals in the aggregate may be very different from that used by a single individual. The reasons for this relate to the very nature and purpose Of organizations. Organizational structures. One Of the primary functions of an organization is tO reduce uncertainty about the task it is tO perform (Schon, 1967; Zaltman, Duncan, & Holbek, 1973). A variety Of organizational structures have been developed to enable organi- zations to deal with uncertainty. In a classic bureaucracy, such as that described by Weber (1947), a rule is available to answer every question, a routine available for every task. Thus, uncer- tainty is theoretically eliminated by the provision Of rules and a chain Of authority to respond to all situations. Unfortunately, except for the most routine tasks, it is impossible to derive a set of rules to cover every possible circumstance. In addition, those rules, once developed are highly resistant to change. As a consequence, "bureaucracy, like other forms Of organization, dis- courages the emergence Of changes from within and resists the impact of changes imposed from without" (LaPiere, 1965; p. 409). The potential introduction of an innovation threatens the entire struc- ture of the organization, since (if the innovation is of any conse- quence) a large number of rule changes and unorthodox decisions will have to be made. However, some types of organizations are better equipped to deal with change than others. A number Of organizational theorists have develOped what are known as "contingency theories" Of organizational structure. Two of these, Litwak (1961, 1968) and Perrow (1970) contend that organizational structure should depend on the nature of the task being performed by the organization. In Litwak's model, an organi- zation that needs to perform a highly routinized, uniform task should utilize a formal, bureaucratic structure. On the other hand, primary groups are better suited to perform tasks which are non- uniform or where expert technical knowledge is not required. An example Of such an instance would be when there is no knowledge about the task (so that experts cannot be trained) or when the task is so complex and non-routine that rules about it cannot be made. Such a task requires a relatively informal "human relations" type of organization. An organization that must perform both types Of tasks must include components that are either bureaucratic or informal, with each having its own area of concern. Linkage roles are necessary to keep the two components on the same track. Cer- tainly a considerable amount of resistance would be expected when an organization that was set up to perform one type Of task is called to perform another. A bureaucratic organization called upon to perform a non-uniform task might be simply unable to perform for lack of guidelines. A human relations type of organization that was required to take on a series of routine tasks would be Operating at a very inefficient level. In either case resistance would be expected due to the uncertainty arising from an unusual situation. Transformation Of the organization into another format might 10 actually be necessary for the organizationix>be able to accomplish the task at all. A typical example would be the formation of what Litwak calls a "professional" organization. This is a type in- volving division into components with links between them. Hall (1962) has provided some evidence to support Litwak's model. Thus, there would seem to be two "basic" organizational structures, which for the sake of convenience will be called bureaucratic and human relations. One, the bureaucratic, is designed for stability, while the other, human relations, is designed for change. A considerable body of research has provided evidence for this assertion (e.g. Burns & Stalker, 1961; Hall, 1962; Lawrence & Lorsch, 1967; Hage & Aiken, 1970; Baldridge & Burnham, 1975). In the middle is Litwak's "professional“ organization, containing some elements that are bureaucratic, some that are human relations. The existence of these differing structures can have great impact on the innovation process. Zaltman and Duncan (1977) con- sider the innovation process to include two phases, initiation and implementation, each Of which requires different structures. The initiation stage requires flexibility and decentralization, in order to encourage a flexibility Of thought and action that is needed to initiate change. The implementation stage requires a more formal, structured, task oriented structure to enable the organiza- tion to accomplish the specific steps necessary to actually set up a new program or implement some technological innovation. Thus not only is it possible for an organization to have differing structures 11 in different units, but that may actually be necessary for complete adoption Of an innovation to take place. Linkage and Boundary Spanning One point that has not been discussed is the connection between the two types Of organizations. Communication between ini- tiating and implementing units is Obviously essential, however the structural differences are Often a source Of friction which interferes with communication. In addition, some connections need to be made with outsiders in order to gain necessary input in the form Of new ideas, suggestions for change, and Objective feedback. Havelock (1973) has provided a detailed discussion of this concept, which he calls linkage. In his discussion Havelock des- cribes ten characteristics Of organizations that inhibit knowledge flow and change. The focus here will be on three of these. The first is the presence of a "coding scheme barrier.“ By this he means that members Of an organization, because of their common experiences and interests, develop a unique vocabulary and means of communicating. This enhances stability, but also prevents effective communication with outsiders. The second characteristic is related tO the coding scheme barrier, that is the stable social structure and social relationships within the organization. Because any change is likely to disrupt the stability Of these relationships it is likely tO be resisted. The third characteristic is related to the first two and that is fear of the "malevolence of outsiders." "The boundaries which separate the 12 organization from its environment (e.g., buildings, dress, rules) encourage the formation Of organizational myths which help members to deal with the uncertainty and ambiguity Of change brought on by outside forces. Thus knowledge from the outside can be seen as a threat to the organization, not only in terms Of upsetting the orderliness as a consequence Of deliberate change, but also as a direct maligning Of the organization and its members" (Havelock, 1973, p. 6-8). One Of the most important ways of overcoming these organization barriers is through the use of linking roles between the develOper Of an innovation and the potential adopting agency. One type Of linkage role is that Of change agent. Frequently the actual innovator is neither willing nor able to advocate change himself (LaPiere, 1965). Therefore an advocate takes the role of creating the linkage between the innovation and the potential adopter. Havelock divides this particular role into three subtypes (conveyor, consultant, and trainer); however, the common element in all of them is that they are outside the organization. Consequently, they are liable to encounter all Of the problems involved in attempt- ing to bridge the organization barriers to change. Another role is necessary, what Havelock calls the Innovator role. This is not necessarily the inventor Of the new process or product. This person may only be an innovator in the context Of his social system, in that he is the first person in that system to adopt or advocate the innovation. While this type of innovator may not be an actual opinion leader he may act as "a demonstrator and quasi-Opinion leader for the real gpinion leader" (p. 7-14). Thus, this type Of 13 linkage role may be said to to provide the first true link into. the organization. A link of this sort may provide the possibility for social comparison processes to take place within the organiza- kw1primarily because the "innovating" individual is likely to be much more similar to peOple within the organization lzhan is an outside change agent. Thus, individual uncertainty may be reduced in a manner favoring the innovation. Thompson (1967) has discussed the linkage concept in another fashion. Thompson stresses the need for organizations to have "boundary-spanning structures." These are units within the organi- zation that have the responsibility of linking the organization with its environment, and dealing with the differing demands and struc- tures that exist there. Thompson emphasizes the need for such boundary-spanning units to be structured differently from the rest Of the organization, and the problems these units may have as a result. He also claims that organizations that face a greater need to change will have.greater numbers Of such units. From the point Of view of an outside change agent then, one key to overcoming organizational resistance to change is the creationcnione or more linkage roles, both within the organization and with its environment. In addition, it may be necessary to encourage the formation of a variety of organizational structures, depending on the existing structure and the type of innovation being introduced. In the case of a mental hospital adopting the Community lodge,it could be necessary to encourage the formation of a small, informal group that would initiate the new program. It might also 14 be necessary to create linkage between the potential users Of the innovation (at the target hospital) and the originators. Linkage would also be needed between the small innovative group and the unit that might be implementing the program, and with the hospital at large. Site Visits as a Technique for Overcoming Resistance to Change From the preceding discussion it can be seen that the fol- lowing are some of the key problems in disseminating innovations to organizations: 1) Reducing individual uncertainty; 2) altering organizational relationships; and 3) creating boundary spanning mechanisms. While a variety Of techniques have been used to over- come these problems (Rogers & Shoemaker, 1971; Havelock, 1973; Emrick, 1977) some have received less attention than others. One technique that has been the object Of relatively little research is the use of a visit to the site of an innovation. Theoretically, a site visit should resolve a number of difficulties described above that may arise due to uncertainty. At the actual site of an innovation the visitor is provided with information of much higher quality than he could receive through other means such as brochures or workshops. The visitor is shown as much information as he is capable of interpreting with his own eyes and ears. He is in a position to direct questions about the innovation1x>the people most qualified to answer them, the people who use the innovation. Thus, uncertainty about the innovation will be reduced because the potential adopters have become more familiar with it. In the same 15 way, the Observability Of the innovation, as described by Rogers and Shoemaker (1971) will be improved. Assuming the visit is sponsored by an outside agency the Operators Of the innovation would be the most credible people to talk to about it. Certainly they know more about how it operates than anyone, and since they are not going out Of their way to make a persuasive communication they should be perceived as having little to gain from having others adopt. As many researchers have shown (Hovaland & Weiss, 1951; Hovland, Janis, & Kelly, 1953; Walster, Aronson, & Abrahams, 1966; among others), these characteristics of the communicator improve his credibility, resulting in increased opinion change. A site visit provides potential adopters with peers with whom they can compare themselves (by way of Festinger's Social Comparison Process) in order to reduce uncertainty. This would especially be the case if the visit took place shortly after the visitor had received information about the innovation, thereby making the comparison more salient. In general these peers would be much more similar to the potential adopter than would the typical outside change agent, who is Often a professional consultant. With a peer acting as a change agent the probability of acceptance should increase, as Brock (1965), Berscheid (1966) and Tuppen (1974) have shown. A site visit can also help in an organizational sense. Assuming that the individual site visitor has been convinced that his organization should adopt the innovation he would then become an 16 internal advocate. In a sense he would be playing the Innovator role as described by Havelock (1973). While the concept of site visits seems to be widely accepted (Havelock, 1973; Rogers & Shoemaker, 1971), there have been rela- tively few experimental tests Of their effectiveness. It has been widely used by the agricultural extension specialist and county agents with their demonstration farms. When one farmer in an area has been convinced to try a new product or procedure (such as hybrid corn, or a new'1ilowing technique) other farmers are then given the Opportunity to visit the demonstration farm or talk with that farmer socially. Clark (1962) related this technique to the field of edu- cational innovation. Education today may have roughly the same relationship to its practitioners that existed in the field Of agriculture in the latter part of the 19th century. At that time, the primary vehicle of communicationto the practitioner was the printed word from research to practitioner. The impact on agricultural practice was slight. Interposed now between the researcher and practitioner are two levels Of translation. The extension specialist can read the research and translate it into some- thing the county agent can understand. The county agent, however, does not typically pass this information directly on to the practitioner. Instead he provides an Opportunity for the farmer to visit another farm in his neighborhood where the new practice is being employed. The situation is a real one. The farmer using the new method is risking his own money on his own farm. The visiting farmer has a chance to see what is going on and talk tO the experimental farmer about it (p. 111). Certainly a similar situation exists today as regards researcher- practitioner communication in many fields, such as mental health, public health, or social services. It would seem likely that such fields could benefit greatly from the concept used in agriculture. 17 Richland (1965), in an attempt to test the effectiveness Of the agricultural field extension service concept as applied to education, tested what he called "Traveling Seminars." These were groups of thirty "educators" who traveled around visiting school sites that had exemplary innovations. At the conclusion of the site visits the four seminar groups gathered in one location to discuss what they had seen and what they were going to do. Unfortunately, the seminar groups included no teachers, although school adminis— trators were included, in addition to representatives of state departments Of education and colleges of education. During their five day tours each group visited from three tO five school dis- tricts and Observedifltm1seven to eleven different innovations. The school districts from which the administrators were chosen were selected because Of "a known interest in research," among other criteria. In a follow-up one year later the school administrators were asked questions about how many innovations they had adopted or were considering adopting. This was compiled into an "innovation index." There was a significant difference between the tour groups and a no treatment control group both on innovation gain scores and on an Analysis Of Covariance of the post-test scores. Unfor- tunately, there may have been some bias in the control group due to the inclusion Of some non-volunteers from the experimental groups. While showing that a site visit plus a conference could help to make schools more innovative, Richland did not attempt to influence educators to adopt a specific innovation. In addition, no attempt was made to isolate the effect of the site visits nor was an attempt 18 made to compare the effectiveness of site visits with that Of other persuasion techniques. Glaser and Coffey (1967) attempted to diffuse a more speci- fic group Of innovations and did test a number of specific techniques. Their ”innovation" was an organization for the mentally retarded that was using a number Of specific (although unrelated) practices. Their measure Of adoption was the number Of these practices that their target organizations adopted. They used three persuasion techniques: (1) an attractive booklet that described the various practices used by the innovating organization; (2) the same booklet, plus a combination visit and conference at the site of the organi- zation; and (3) the booklet, site visit-conference, plus a consul- tation visit from the founder of the exemplary organization. When asked if they had adopted one or more practices, the organizations in the booklet alone condition were significantly less adoptive (p < .10) than the organizations in the other two conditions. How- ever, using an overall change score there was no difference between experimental conditions, although all three groups showed signifi- cantly more change than did a no treatment control group. Glaser and Ross (1971) attempted a much more specific and controlled experiment. Their innovation this time was a specific type of therapy to be used in a variety Of mental health organiza- tions (e.g., Mental hospitals, Community Mental Health Centers, etc.) called Saturation Group Therapy (SGT). This study used the same conditions as Glaser and Coffey did. This time however there 19 was no effect at all. Not one organization adopted the innovation after a six-month follow-up, although at least two seemed as if they probably would. Glaser and Ross believe the problem lies in the nature Of their innovation. While it was ideal from their research standpoint, it did not have sufficient relative advantage, nor was it sufficiently compatible with the potential user systems. In a more recent study Larsen, Artunian, and Finley (1974) used site visits in an attempt to make Community Mental Health Centers (CMHCs) more innovative, in general. They had three experi- mental conditions, plus a control group. These conditions were: (1) Written materials were sent to the CMHC that described a number of innovations and where more information about them could be Ob- tained; (2) Written materials were sent, plus the CMHC was provided with a set amount Of money which they could use to send staff to visit the site of any innovation described in the materials; and (3) In addition'UJthe written materials and the money for the site visits a consultant visited the CMHC prior to the site visits to discuss innovation and any Of the specific innovations in the materials. The results Of this experiment showed no significant differences between conditions as to innovativeness. In general the CMHC staffs liked all Of thetreatments and thought they were very useful. With the exception of the Richland study it would seem that most Of the experimental work to date has not shown site visits to be a very effective technique for the diffusion Of innovations. However, each of the studies has had some methodological flaw which 20 hampered its effectiveness. Probably the major weakness Of all of the studies was the lack of direct assistance to help the potential adopters to implement the innovation. Fairweather, Sanders, and Tornatzky (1974) have shown that this is crucial to get a signifi- cant rate of adoption. What may have happened, (and certainly did happen to Glaser and Ross) is that so few adoptions took place that there was no chance for the site visit to take effect. In addition none Of the studies actually tested for the effectiveness of the site visits alone, it was usually part of a conference. Often no attempt was made to involve the whole organization. In other words, information was passed to only one or two individuals in an organi- zation, one Of whom made the site visit. Such a situation is likely to lead to an authoritarian type Of decision. Rogers and Shoemaker (1971) and Fairweather et a1. (1974) have shown that such decisions are less likely to lead to change. Thus it would seem that site visits, a concept with much theoretical backing but little direct evidence in its favor, is still in need Of empirical testing as to its effectiveness. The Present Study This study experimentally tested the effects of a site visit on adoption Of a specific innovation (the Community Lodge) by a na- tional sample Of mental hospitals. A number of mental hospitals were provided with a workshOp about the Lodge program, after which half Of the sample was Offered the Opportunity to send a staff 21 member to visit the site of a hOSpital which was operating a number of Lodges. Hypotheses A number Of hypotheses were made concerning the effects that such a visit ought to have on anental hospital and its staff. These were divided into three categories. Effects on social process. The first can be called direct effects Of the visit on social process variables. If the site visit is to have any effect at all, the site visitor must talk to others about the visit and act as an advocate for the program. This results in two hypotheses: l. A site visit will increase the amount of discussion about the Lodge among hospital staff. 2. A site visit will result in increased advocacy for the program at the target hospital. Effects on intervening variables. It was suggested that uncertainty about the innovation might be an intervening variable between discussion about the program and actual adOption Of it. Uncertainty can have many components, three Of which appear to be of importance in this instance: 3. A site visit will reduce the amount of uncertainty hospital staff feel concerning their knowledge about how-to set up a Lodge. 4. A site visit will reduce the amount Of uncertainty staff feel concerning the effectiveness of the Lodge program. 5. A site visit will reduce the amount of uncertainty staff feel concerning the feasibility Of the program. 6. A site visit will reduce overall uncertainty about the program. 22 Boundary spanning was noted above as an important variable in the innovation process that might be affected by a site visit. 7. The extent of external Boundary Spanning will be increased as a result of the site visit. Other variables that might be of importance as interveners in the adoption process are attitude toward the program, and know- ledge about the program. This resulted in two more hypotheses: 8. A site visit should result in an improved attitude toward the Lodge program. 9. A site visit will result in hospital staff having more knowledge about the program. Effect on adoption of the lodge. Finally, as a result Of these intervening processes: 10. A site visit should result in increased movement toward adoption Of the Lodge program. CHAPTER II METHOD The Innovation to be Disseminated The innovation that was to be disseminated in this experi- ment was the Community Lodge Program, which was developed by George Fairweather in the 1960's as an effective method Of treatment for chronic mental patients. The Lodge program consists Of two parts, a hospital phase, known as the Small Group Ward (Fairweather, 1964), and a community phase, known as the Lodge (Fairweather, Sanders, Maynard, & Cressler, 1969). A central idea Of the program is the develOpment Of intense group cohesion and peer dependence through the use of a variety Of group reinforcement techniques. A key element in maintaining a strong group identity is a reduction in staff influence over individual members of the group. This is in direct conflict with the traditional role of the mental health professional as a “helping" person. The program also has a strong community orientation, which runs counter to the very insular role that mostmental hospitals have assumed. The Lodge in particular is very complex and difficult to implement, involving a variety of tasks that are not typically included in the training Of most psychiatric hospital personnel. 23 24 Sample Hospitals. The sample was twenty-four (N = 24) state hospi- tals (Appendix A) that (1) had not been contacted about the Lodge innovation in the seven years previous; (2) were within 1000 miles of E. Lansing, MI; (3) agreed to have a free workshop on the Lodge program; and (4) agreed to permit one staff member to make an all expense paid site visit, should that Opportunity be provided. These hospitals had all been contacted by Fairweather, Sanders, and Tornatzky (1974) in their diffusion study, and had been assigned a change score by them, depending on what steps they had taken toward adoption Of the Lodge. Hospitals were matched on that change score and then randomly assigned to either experimental or control conditions. A total of 44 hospitals were contacted in order to Obtain the 24 that met all Of the above criteria. Twenty hospitals that were contacted were either unwilling or unable to have the workshop. All hospitals that agreed to have the workshop also agreed to permit staff to make the site visit. Individual reppondents. A total of 606 individuals attended the 24 workshops and filled out a Workshop Questionnaire. More than 90% Of these respondents were employees Of the hopsital where the workshop was given. The rest were representatives Of various community agencies. Of those respondents, 386 responded to a follow-up questionnaire that was mailed to them. 25 Research Desigp The design Of the experiment was a simple, two cell design (Figure 1). Hospitals were matched on their change score from the Fairweather, Sanders and Tornatzky (1974) study and then were assigned to either experimental or control conditions. Thus, there were twelve hospitals in each condition. Procedure The study consisted of four phases (Figure 2), as follows: Approach Phase. TwO brochures describing the HOSpital- Community Treatment Program (the named used in the project to repre- sent the combination Of Small Group Ward and Lodge programs) was sent to all hospital superintendents along with a cover letter (Appendix B) describing the kinds of assistance that could be pro- vided to the hospital in setting the program up. After approximately ten days the superintendent was called. A brief description Of the Lodge and the assistance that could be provided was presented (Appendix C), and the superintendent was asked if he would allow a workshop to be presented on the Lodge at his hospital. A decision was not demanded at this point, although the superintendent was encouraged to make one. One week after the telephone call a reminder letter (Appendix D) was sent. Ten days after that a follow-up call (Appendix E) was made to Obtain a decision, if one had not been reached. Workshop volunteers then moved on to the Persuasion Phase. 26 m4 mcwm 8:» to cmwmao _ mesmPJ 2m; 2.» + 2.5 359.33 3:333 m4apu< to “Logo zo_a N mg:m_m mewmccowammao aauzoppod mcosampmh mgwmccowummzo uzolpwmz Auwcmmc cmczv cowboy—:mcou mcwm: unevumwao pmmz ucmm nogmxcoz cocoa mgucoe m_-m Ampmwpqmom nmv :uzoppou Axpco mpmuwamo; wcwm:cowumm=o u_mw> muwm moxme nocmxcoz poucmswgmgmeFV coammv> mumm gmaeme compm F coupe mace mm1m— uwmw> mumm auaucou Pawpacw Am_aomamoglamv mgwmccowummao aocmxcoz quwamo; pm aogmxcoz Loumm mAmu c¢_-~e cowmmsmgma nogmxsoz so» aomucoo meawcm Am—muwamo; we“ mpcmewmcmcgm mxmz gmuwm mxmu mmuo some; < umuomp_ou mama mmwu_>wuu< cowgma meme omega 28 Persuasion phase. The persuasion phase consisted Of two parts, a brochure mailing and a workshop presentation. Brochures were mailed along with a letter confirming the workshop date (Appendix F) to the hospital superintendent or other designated contact person. These were to be distributed to all personnel that were expected to attend the workshop. Distribution of these brochures is uncertain, although 42% Of the staff who attended the workshops said that they had read the brochure. A one day workshop at the hospital constituted the bulk of the persuasion phase activity. This included a two hour lecture, complete with slides and a movie, a question period, and a three hour discussion of the steps necessary to set up a Small Group Ward. A schedule for the workshops can be found in Appendix G. At the conclusion Of the lecture and question period the workshop participants were informed about the kinds of assistance that could be provided to them, should they request it. This assistance included: (1) eight copies Of a manual describing in detail the steps necessary to set up a Small Group Ward (Tornatzky, Fleischer, Avellar, Fergus, 8 Fairweather, 1976); (2) two capies Of a manual describing in detail the steps necessary to set up a Lodge program (Avellar, Dittmar, Fergus, Tornatzky, Fleischer, & Fairweather, 1976); (3) free telephone consultation concerning the program; (4) the possibility of a free two day consultation visit to the hospital, if a Small Group Ward had been set up and Operated for at least 60 days; and (5) for those hospitals in the Site Visit (experimental)condition only, the Opportunity for one member Of the 29 staff to make an all expense paid visit tO a state hospital in Minnesota that had a functioning Small Group Ward and several lodges. In the afternoon, at the conclusion of the workshop, parti- cipants filled out the Workshop Questionnaire (Appendix H). Participants who left early were asked to fill this questionnaire out before they left. In the Site Visit condition, after the questionnaires were completed the participants were asked to decide who would be the one to make the Site Visit. In about half of the hospitals a decision was made immediately. In the rest Of the hospitals the decision was made within two weeks. Site visit phase (experimental group_on1y). Each hospital in the Site Visit condition was invited to send one staff member on an expense paid visit to an exemplary Lodge program. All twelve hospitals in the Site Visit condition participated in this phase of the experiment. Two of those hospitals chose to send an additional staff member on the visit, at hospital expense. It should be noted that there were no activities in this phase for hospitals in the Control condition. Those individuals who were selected as site visitors were informed by mail and telephone about the details of their trip to Minnesota (Appendix I). The visitors were asked to make their own arrangements for transportation (they were later reimbursed) and to make their own way to a designated hotel, where they were met by a member Of the hospital staff in the morning. 30 It should be noted that from three to six individuals were involved on each of three site visit dates. On the first date three peOple from different hospitals were involved. On the second date, about a month later, five people from four hospitals partici- pated, and on the third date, another month later, there were six individuals from five hospitals. The hospital that hosted the Site Visits has one of the best examples Of a Community Lodge program in the nation. Their initial Lodge was set up eight years ago, as a result of the Fairweather, Sanders, and Tornatzky (1974) diffusion study. The staff had re- mained in contact with the Michigan State research group on a regular basis. Since their initial Lodge they had set up five addi- tional ones (as of this writing). They were a very enthusiastic group, and when presented with the idea Of hosting the site visits, were quick to accept. It was agreed to pay $350 for each Of the three visits, as compensation for staff time and expenses during the visits. This money was donated to the Lodges. A sample schedule for the Site Visits can be found in Appendix J. During their two day stay the visitors had the Oppor- tunity to meet with staff and residents of the Small Group Ward and Lodge programs, to Observe the various meetings that are a part Of the program taking place, to visit Lodge residences, and to visit with Lodge members while they were working in their place of business. Ample opportunity was provided for the visitors to talk with the residents and staff on an informal basis. Before the 31 visitors left for home they each filled out a Site Visitor Questionnaire (Appendix K). Follow-pp and consultation phase. Three months after the workshop all workshop participants (in both conditions) were sent a Mail-Out Questionnaire (Appendix L). Those individuals who did not return the questionnaire were sent a Follow-Up letter with an additional questionnaire (Appendix M). All questionnaires were sent with a business reply (postage-paid) envelope for the respon- dents'convenience. Of the 606 workshop participants who were sent the Mail-Out Questionnaire, 386 returned them, for a response rate Of 63.7%. Follow-Up telephone calls were made to the designated contact at the hospital at the following intervals. 3 months after the workshop 6 months after the workshOp 9 months after the workshop 13 months after the workshOp DOOM-4 The calls had two objectives. The primary objective was to collect data concerning the extent to which the hospital was making progress toward setting up the Small Group Ward or Lodge (Appendix M). The secondary purpose was to provide telephone consultation assistance or to answer questions concerning the program. There was also additional assistance available. Staff at the workshop were informed that they could receive direct consulta- tion assistance in setting up the Small Group Ward or Lodge if they had Operated a Small Group Ward for at least 60 days. This consul- tation assistance consisted of a two day visit by a consultant. 32 The first day involved a trouble-shooting procedure to find and solve problems with the Small Group Ward. The second day consisted Of discussion about how to take the first steps toward setting up a Lodge. Additional copies of the Lodge manual were also provided at this time. One hospital began a Small Group Ward three months after the workshop, but did not request consultation. Three hospitals began Small Group Wards in the interval between six and nine months and received consultation in the interval between nine and thirteen months. Thus, there were no consultation visits made until after the nine month follow-up period had passed. Measurement Measurement for this study involved the use of four question- naires, which were composed Of a number of different scales and individual items (see Figure 3). The four questionnaires were: the Workshpp Questionnaire (Appendix H), the Site Visitor Question- naire (Appendix K), the Mail-Out Questionnaire (Appendix N). Compo- sition of these questionnaires is summarized in Figure 3. A number Of questionnaire items and scales were developed for the purpose Of testing the specific hypotheses discussed in the previous chapter. The various scales are discussed in detail below, with a summary Of which measure tests which hypothesis provided in Figure 4. Communication network item (Appendix 0). This was an item on the Mail-Out Questionnaire that was designed to determine the 33 mcmmccowummzc to mucmpcou N mesmwu mpmpwamo; gwsuo op muwmm> ozone mco_amm=o .2 2053.6 33m mmcmcu .m cowummzc xcozpmz :owuouwcassou .m mcowpmmac covenawowucma .m amok mmumpzocx .n mpmom ulH .o mpmom muaumup< .n mpmomnam xuwpwnmmmmu upmomnam mmmcm>wuomwwu upcomaam mmcm_3oc¥ ouuzoz mpmom zucwmucmu .m mcowummzc uwmm> mo mmmc—zcmmz .n m_mom xuwgmpmewm .m cowumozum .zcoumwg xgoz .ma< mcowpmmaa cowumamowpgmm “mop mmvm_zocx mpmom m1H mpmum enzymau< mpmumnzm xuwpwnwmmmu m—wumnam mmmco>wuumw$m m—momnsm mmuo—Iocg calzo: mpmom aucwmugmu .w o mcosa o Lepwmw> mowm attaccoaumaao oao-pwaz mcwmccowummao nosmxcoz 34 Hypothesis How Measured l. Site Visit (SV) should increase discussion about program. 2. SV will increase advocacy 3. SV will reduce How-To Uncertainty. 4. SV will reduce Uncertainty about eEffectiveness. 5. SV will reduce Uncertainty about Feasibility d . a. Q1 & Q2 on MOO b. Communication Network question on MOQ . 05 on MOQ Network Question on MOQ . How-TO Certainty subscale on MOQ . Effectiveness Certainty subscale on MOQ . Feasibility Certainty subscale on MOO 6. SV will reduce Overall 6. Overall Certainty Scale on MOQ Uncertainty 7. SV will increase Boundary 7. a. Visits Question on Phone Spanning. Follow-Up O. 6. #Letters sent to Consultant c. #Calls made to Consultant 8. SV will improve Attitude 8. Attitude Scale on MOO toward program. 9. SV will increase knowledge 9. Knowledge Test on MOQ about program. 10. SV will increase movement 10. Change Scale derived from toward adoption Of program. Telephone Follow-Up Calls Figure 4 Measurement Of Hypotheses 35 extent to which each respondent had communication concerning the Lodge with each other respondent. Data from this item was used to determine whether the Site Visit intervention increased discussion and advocacy about the program. Certainty_scale (Appendix P). This scale was designed to measure overall uncertainty about the program. It was composed Of three subscales: (1) Certainty of How-TO Knowledge Subscale (Appendix Q). This subscale was composed Of 10 items that had to do with the respondents' certainty Of the extent to which they knew how to implement both the Small Group Ward and Lodge. Internal consistency reliability, or alpha (Guilford, 1954), for this sub- scale was .82; (2) Certainty of Effectiveness Subscale (Appendix R). This subscale was composed of five items having to do with the respondents' certainty that the Samll Group Ward and Lodge were effective programs. Internal consistency reliability for this sub- scale was .77; (3) Certainty of Feasibility Subscale (Appendix S). This consisted Of three items having to do with the respondents' certainty that setting these programs up was feasible within the context Of their hospital and community. The realiability of this subscale was only .56. The combination of the three subscale scores resulted in the score for the overall Certainty Scale, which had 18 items and a reliability Of .82. 36 Attitude scale (Appendix T). The Attitude Scale was composed of eight items, dealing with the respondents' agreement with the program and its components and with their belief that the program ought to be adopted by their institution. The internal consistency reliability of this scale was .88. Knowledge test (Appendix U). This consisted Of four multi- ple choice items designed to determine if the respondent had basic knowledge about the program. A respondent's score for this test was the number Of correct answers out Of the four. Change scale (Appendix V). This was a scale composed by Tornatzky, et al., (1978) to determine the extent to which a hOSpi- tal has moved toward adoption Of the Small Group Ward or Lodge. There are seven steps involved in Small Group Ward adoption and 26 steps in Lodge adoption. A hospital's change score was calcu— lated in the following manner: each step toward adoption Of the Small Group Ward was given one point. Each step toward adoption of the Lodge was given .337 point. Thus, the maximum score a hospi- tal could receive was (7 x l) + (26x.337) = 15.75. It was necessary to weight the Lodge steps much less due to the number of steps involved as compared to the Small Group Ward. The assignment of seven points to Small Group Ward implementation and 8.75 points to Lodge implementation was based on previous archival data from several hospitals on the relative amount Of time and effort involved in implementing the two respective subcomponents Of the total program. [(‘7 37 A hospital received a change score at each Of the Follow-Up periods at 3, 6, 9, and 13 months after the workshop. The number Of steps achieved toward adoption was derived from the Telephone Follow-Up questionnaire. A number of additional measures were used in this study that were not involved in specific hypothesis testing. Their primary purposes were as measures that seemed to be important indicators Of the process taking place during adoption. These included: Internal-external locus of control scale (Appendix W). This was a modification Of the Rotter (1966) I-E Scale, developed by Tornatzky, et al., (1978), following previous work by Bond and Tornatzky (1972). The modifications were designed to make the scale directly applicable to staff employed in a mental institution. This scale had eight items and a reliability Of .78. Similarity scale (Appendix M). The Similarity Scale was composed of seven items relating to how similar the Site Visitor felt his own institution was to the one he was visiting. It was used only on the Site Visitor Questionnaire. Its reliability was .76, with a sample of only 14. In addition to data derived from the four questionnaires, some information was gathered about the hospital itself from the City_and County Data Book (U.S. Census, 1974) and from the Hospital Data Book, published by the American Hospital Association (1977). CHAPTER III RESULTS qujvalence Of Treatment Groups One possible difficulty that could have arisen with the study that was just described is the problem Of "experimenter effects" (Rosenthal, 1966). Experimenter effects as bias in an experiment may occur when the experimenter is not "blind'I to the treatment conditions and has the opportunity to affect his subjects' outcomes. This could be a problem in the present study, since there was only one experimenter/consultant (namely the author), and he was aware of which hospitals were in each condition. Since the bulk of the consultant's time with each hOSpital's staff was during the workshop, it may be that he presented his material differentially or said something different for hospitals in different conditions. The best way to test for the possibility of experimenter effects would be to take measures of the outcome variables after the completion Of the workshop, but prior to the time that the Site Visit took place. The WorkshOp Questionnaire meets the qualifi- cations for such a test. The absence of differences on the measures contained in that questionnaire would Offer strong support for a lack of bias on the part of the consultant. Table 1 is a summary of the Analyses of Variance (ANOVA) comparing the two conditions on the various scales and items on the pre-test (Workshop) 38 39 Table 1 Summary of ANOVA Of Pre-Test Differences Between Conditions Variable F (df) p 1. Years staff worked at hospital .530 (l, 22) ns 2. Level of staff education .245 (1,22) ns 3. Staff age .521 (1,22) ns 4. Number journals read 3.381 (1,22) .1 5. Satisfaction with present .114 (1.22) ns programs 6. Personally involved in .089 (1.22) ns decision making 7. Staff discussion about .257 (1.22) ns workshop 8. Certainty Of "How-to“ knowledge .171 (1,22) ns 9. Certainty of feasibility .108 (1,22) ns 10. Certainty of effectiveness 1.000 (1,556) ns 11. Attitude .804 (1,345) ns 12. Locus of control .075 (1,22) ns 13. Overall certainty .014 (1,22) ns 14. Knowledge .237 (1,582) ns 15. Position of staff 2.168 (1,22) ns 16. Area of staff training .001 (1,22) ns 4O questionnaire. The table shows that there were no significant (.05 level) differences between the two conditions on their responses to the pre-test. Another potential problem with this study could have arisen as a result of a differential return rate of the post-test (Mail-Out) questionnaire. If this occured it might indicate that different kinds of people were responding in different conditions, possibly biasing the results. Of 386 questionnaires returned, 202 were from the Site Visit condition, while 184 were returned from the Control condition. This difference is not significant (t = 1.00, 22df). In addition, several ANOVAs were computed to test for the effect that a number of personal variables had on dropping out. For these analyses, whether or not an individual dropped out Of the sample was treated as a dichotomous variable. Thus, it was possible to test for the effect that demographic variables and the Site Visit condition had on it. More important, it becomes possible to see if there was an interaction between the demographic variables and the condition in their effect on dropping out. Table la is the ANOVA table showing the effect of both the respondents' status position in the hospital and condition on dropping out. It can be seen that while position does have a significant effect (F = 5.96, df = 4,586, p < .001), condition does not, nor is there an interaction between position and condition. This lack Of an interaction can be interpreted to mean that the intervention did not differentially effect individuals in the status positions as far as dropping out of the sample was concerned. 41 Similar results were found for variables such as area of training, education, age, and work experience Of the respondents. TO summarize, respondents in the two conditions appear equivalent at the time Of post-testing. Table la Effect Of Position and Condition On Dropping Out Source df MS F p Position (P) 4 1.438 5.956 < .001 Condition (C) 1 .403 1.671 ns P x C 4 .106 .440 ns Subjects 586 .242 Total 595 Effects on Social Process Effect on discussion. The first hypothesis (see page 21) stated that the Site Visit should result in greater discussion about the Lodge taking place within the hospital. This hypothesis was tested in two ways. The first involved staff perceptions about the amount of discussion. This was measured by two questions on the Mail-Out Questionnaire (MOQ). One asked the respondent to indicate how much discussion about the program took place among hospital staff (Appendix 42 K, Question 1). Table 2 indicates that the Site Visit did not significantly affect response to this question (f = .447, df = l, 22, ns). Table 2 and many Of the following tables may appear somewhat unusual to some readers. In this experiment hospitals are nested within conditions (Winer, 1976), and thus, the variance due to hospitals must be considered separately. If the differences between hospitals (the nested factor) are ppt significant, then the error terms may be pooled. Thus, in a situation where the hospital factor is significant it is used as the error term in computing the F ratio for condition. When the hospital factor is not significant, a pooled error term is used. In this case, there was a significant difference between hospitals (F = 3.22, df = 22, 344, p < .001) although not between conditions. This means that hOSpitals vary considerably on this measure, based on some organizational variables. The implications of this will be discussed in the next chapter. Table 2 Effect of Site Visit on Perceived Discussion Mean Discussion Site Visit Control 5°urce 2.73 2.62 df MS F p Conditions 1 .992 .447 ns *Hospitals 22 2.218 3.222 .001 Subjects 344 .688 Total 367 * = error term used to compute F ratio 43 The other question having to do with perceived discussion asked the respondent to indicate how much discussion concerning the program he or she was personnally involved in (Mail-Out Question- naire, question 2). The results shown in Table 3 indicate that there were no differences between conditions (F = .203, df = 1, 377, ns) or hospitals. The second way discussion was measured was more direct. This involved the Communication Network Question (Appendix N). The difference between the two conditions on the mean discussion based on this question was not significant (Table 4; F = 2.30, df = l, 22). Thus, discussion was not increased as a result of the Site Visit. Effect on advocagy. Perceived advocacy was measured by a question on the Mail-Out Questionnaire (Question 5) that asked whether the respondent thought that there was an advocate for the Lodge in the hospital, and if so who that was. The differences between the two conditions on this variable were not significant (F = .152, df = l, 300). Advocacy was more directly measured by means of the Communi- cation Network Question. In this case the presence of a sociometric "star" was determined in each hospital. This was defined as the individual who had the greatest number Of incoming links in the net- work, that is, the person whom the most respondents named as someone they spoke with concerning the Lodge. Each hospital's "star" then received a communication score which was the mean of all Of his links with other respondents. As Table 5 shows, 44 Table 3 Effect Of Site Visit on Personal Involvement Mean Discussion Site Visit Control 2.71 2.77 Source df MS F J; Condition 1 .323 .203 ns Hospitals 22 1.673 1.053 ns Subjects 355 1.590 *POOled error 377 Total 378 * = error term used to compute F ratio Table 4 Effect Of Site Visit on Mean Discussion Strength Mean Discussion Experimental Control 2.80 2.67 Source df MS F p Condition 1 .103 2.303 ns *Hospital 22 .045 Total 23 * = error term used to compute F ratio Table 5 Effect of Site Visit on Star's Mean Discussion Mean Discussion Experimental Control 3.56 3.18 Source df MS F ,p Condition 1 .851 3.494 p < .10 *Hospital 22 .244 Total 23 eta2 = .137 * = error term used to compute F ratio 45 the Site Visit hospitals' stars had marginally greater communication than did the stars from the Control hospitals (F = 3.49, df = l, 22, p < .10). A measure of association, eta-squared, indicates that a fairly substantial portion of the variance (13.7%) in this type Of discussion can be explained by the Site Visit condition. It should be noted that in seven Of the twelve Site Visit hospitals, the site visitor was the communication "star." In eleven Of the twelve Site Visit hospitals the site visitor was the person named most Often as the advocate on Question 5 of the M00. To summarize the results, described so far, there is some evidence to indicate that the Site Visit did have some effect on communication in the hospital, at least in terms Of concentration Of communication around one "star." This can be interpreted as the existence Of greater advocacy taking place. Effects on InterveningyVariables Effect on uncertainty. This was measured by use Of the Certainty Scale, which was composed Of three subscales. l. Certainty of "How-TO" Knowledge subscale - this subscale consisted Of items asking the respondent to indicate how certain he was about his knowledge Of how to set up the program. Table 6 shows that the Site Visit had a significant effect on Certainty Of "How-TO" knowledge (F = 6.20, df = l, 344, p < .05). Unfortunately, only a small proportion of the variance is accounted for in this case by the condition. 46 Table 6 Effect Of Site Visit on Uncertainty Of "How-to" Knowledge Mean Uncertainty Experimental' Control 2.96 3.15 Source Of MS F p Condition 1 , 2.555 6.200 p < .05 Hospital 22 .454 1.110 ns Subject 322 .409 *POOled error 344 .412 Total 345 * = error term used to compute F ratio 2. Certainty Of Effectiveness subscale - respondents answered these questions by indicating how certain they were that the Lodge was an effective program. The Site Visit had no effect on this (Table 7, F = .004, df = l, 22, ns). although there was a significant difference between hospitals (F = 1.86, df = 22, 339, p < .05). Table 7 Effect Of Site Visit on Uncertainty of Effectiveness Mean Uncertainty Experimental Control 2.70 2.71 Source df MS F p Condition 1 .004 .004 ns *Hospital 22 1.100 1.860 p < .05 Subject 339 .592 Total 362 * = error term used to compute F ratio 47 3. Certainty of Feasibility subscale - this subscale consisted Of items asking about certainty that the program was feasible within the context Of the respondent's hospital and community. There was no difference between either conditions (F = 1.72, df l, 360, ns, Table 8) or hospitals on this subscale (F = 1.24, df 22, 338, ns). 4. Overall Certainty Scale - the Overall Certainty Scale was composed of the sum Of the three subscales just described. As indicated by Table 9, the Site Visit did marginally reduce uncer- tainty about the program (F = 3.23, df = 1, 22, p < .10). There were also significant hospital differences (F = 1.60, df = 22, 303, p < .05). As a summarycfi the results about uncertainty, it can be said that the Site Visit did have an effect on uncertainty, although more so for "How-TO" uncertainty. This may give some indications Of some of the strengths and weaknesses of the site visit technique. Effects on boundary spanning. As noted in the first chapter, boundary-spanning or linkage with the external environment may be critical in the innovation process. The extent Of boundary spanning was measured in three ways. 1. Telephone Calls to consultant - These were spontaneous calls to the consultant requesting some kind of information. Excluded from this count were calls having to do with the workshop, the site visit, or any other administrative matter. The actual measure used here is the number Of hospitals that made such calls 48 Table 8 Effect of Site Visit on Uncertainty of Feasibility Mean Uncertainty Experimental Control 3.37 3.45 Source df MS F p Condition 1 .4802 1.751 ns Hospital 22 .3397 1.238 ns Subject 338 .2743 *POOled error 360 .2783 Total 361 * = error term used to compute F ratio Table 9 Effect Of Site Visit on Overall Uncertainty Mean Uncertainty Experimental Control 2.94 3.08 Source df MS F tp Condition 1 1.238 3.230 p < .10 *Hospital 22 .383 1.597 p < .05 Subject 303 .240 Total 326 eta2 = .015 * = error term used to compute F ratio 49 in each condition. In other words, a hospital received a score Of one (made one or more calls) or zero (made no calls) on this measure. As indicated by the Chi-Square table (Table 10), six hOSpitals in the Site Visit condition made calls to the consultant, while none did in the Control condition. This difference is signifi- cant (X2 = 5.55, df = l, p < .02). 2. Letters to consultant - This was similar to the telephone call measure. Again only letters to the consultant requesting information were used, as Opposed tO letters concerning such "admin- istrative" matters as setting dates for site visits or consultation. Table 11 shows that seven site visit hospitals sent letters, while only three control hospitals did so. This difference is not 51901- ficant (x2 = 1.54, df = 1, ns). 3. Visits to other Institutions - Only included here are spontaneous visits to institutions that had a Lodge program, but excluding any visits to the original site visit hospital in Minnesota, or to another hospital that was included in the Site Visit condition. Table 12 shows that four site visit hospitals made such visits, while none of the control hospitals did so. This difference was not significant (X2 = 2.70, df = 1, ns). It should be noted however, that one hospital in the site visit condition sent personnel to visit the hospital in Minnesota on a separate occasion. Also, one of the site visitors made a visit to another site visit hospital that was beginning to implement its own Lodge program. Another point Of interest is that several Of the site 50 Table 10 Effect of Site Visit on Calls to Consultant Site Visit Control Total Made Calls 6 O 0 Made no Calls 6 12 18 Total 12 12 2- X — 5.55 p < .05 Table 11 Effect of Site Visit on Sending Letters to Consultant Site Visit Control Total Sent Letters 7 3 10 Sent no Letters 5 9 14 Total 12 12 x2 = 1.543 ns Table 12 Effect of Site Visit on Visits to Other Lodge Programs Site Visit Control Total Made Visits 4 O 4 Made no Visits 8 12 20 Total 12 12 x2 = 2.70 ns 51 visitors made efforts on their own to disseminate information about the Lodge to local Community Mental Health Centers and other mental health agencies, while no such efforts were reported at the control hospitals. Given the evidence that has just been presented, it can be stated that to some extent, the Site Visit seems to have increased the amount of boundary spanning or linkage that the hospitals had with their environment. Effect on attitude. The Attitude Scale measured the respondents' agreement with the Lodge program'scomponents and belief that it should be adopted by their hospitals. As Table 13 shows, the Site Visit had no effect on attitude toward the Lodge (F = .310, df = 1, 22), although there were significant hospital differences on this variable (F = 2.05, df = 22, 273, p < .01). Effect on knowledgg, The knowledge test consisted of items testing knowledge about a variety Of aspects of the Lodge. Table 14 shows that there were no significant differences between either conditions (F = .265, df = 1, 384) or hospitals. Effect on Movement Toward Adoption The measure used to determine movement toward adoption was the Change Scale. Data for this was collected during the Telephone Follow-up calls. Thus there are measures for extent of adoption at 3, 6, 9, and 13 months after the workshop. The mean adoption level for each condition at each time period is shown in the graph 52 Table 13 Effect Of Site Visit on Attitude Toward Lodge Mean Attitude Score Experimental Control 3.95 3.90 Source df' MS F p Condition 1 .157 .633 ns *Hospital 22 .509 2.054 p < .005 Subjects 273 .248 Total 296 * = error term used to compute F ratio Table 14 Effect Of Site Visit on Knowledge About Lodge Mean Knowledge Score Site Visit Control 3.00 3.06 Source Of MS F p Conditions 1 .279 .265 ns Hospitals 22 .735 .685 ns Subjects 362 1.074 *POOled Error 384 1.054 Total 385 * = error term used to compute F ratio 53 in Figure 5. Table 15 shows that the differences apparent in the graph approach significance (F = 2.86, df = l, 22, p < .11). This test was a repeated measures Analysis of Variance (Winter, 1976). A small, but substantial, portion Of the variance in change score (6.3%) does seem to be accounted for by the Site Visit. Thus, there are indications that the Site Visit may have had a small impact on adoption of the Lodge over all four follow—up periods. The lack of significant effect for Time indicates that between three and thirteen months the level of adoption did not increase for the entire sample. This completes the tests Of the hypotheses described in Chapter 1. The Site Visit was shown to have significant effects on staff uncer- tainty concerning their knowledge about how to set up the Lodge program and on the extent of linkage the hospital had with its environment. There were also marginal effects on advocacy and adoption of the pro- gram. In addition tO hypothesis testing there were a number Of addi- tional, post-hoc, tests that were performed on some additional variables. Post-Hoc Tests Post-hoc test of effect on locus Of control. The Locus Of Control Scale measured respondents' feelings about whether they or some external forces (e.g. hospital administrators) had major influence on their activities within the hospital and on hospital affairs. There were no differences between conditions on this scale (Table 16; F = .59, df = l, 22), although there were signi- ficant hospital differences (F = 1.78, df = 22, 333, p < .05). 54 week Lm>o cowuwccou ma cowuqou< newzo» p:mEm>oz m mgamwd mIHzo—z 2. m2: m. m m m a _ _ — \\\. llllll I. .I ''''''''' ‘\\ .250 _E.:ooullll .28 32> 25.111 m. NOLLdOOV OHVMOJ. .LN3W3AOW 55 Table 15 Movement Toward Adoption by Condition Source df MS F p Conditions 1 30.083 2.858 p < .11 *Hospital 22 10.527 Time 3 2.401 .767 ns C x T 3 .801 .256 ns *H x T 66 3.129 Total 95 eta2 = .063 * = error term used to compute F ratio Table 16 Effect of Site Visit on Locus of Control ‘_Mean Degree of Internality (Higher is More Internal)_ Site Visit Control 3.21 3.13 Source df MS F p Conditions 1 .483 .590 ns *Hospitals 22 .819 1.780 < .05 Subjects 333 .460 Total 356 * = error term used to compute F ratio 56 Post-hoc test on attitude and uncertainty. If in fact the reduction in uncertainty noted above was due tO discussion and interaction with the Site Visitor (either directly or indirectly) then it would seem that the effect should be stronger for those people who interacted more with the visitor, or at least discussed the program more. Since no one in the control group had the opportunity to speak with a site visitor, the only comparison that can be made is between people who were more involved in discussion about the program. Thus, both groups were divided at the median on Question 2 Of the Mail-Out Questionnaire, personal involvement in discussions about the program. A comparison was made between conditions for respondents above the median on that question. Table 17 reveals that the Site Visit did have a highly significant effect for this group of people on their certainty Of "How-To" knowledge (F = 11.76, df = l, 175, p < .001). Note that this is a much stronger effect than was Obtained for the entire sample (Table 6). Certainty Of Effectiveness Of the program was marginally enhanced by the Site Visit (Table 18; F = 2.84, df = 1, 163, p < .10), something that did not occur for the entire sample (Table 7). There was no effect on Certainty of Feasibility. The overall Certainty Scale shows a highly significant effect for the Site Visit (Table 19; F = 11.05, df = 1, 148, p < .001) similar to that for How-TO Certainty. Finally, there are indications that the Site Visit may have affected attitudes that these people held about the program (Table 20; F = 2.97, df = l, 133, p < .10), a result not found for the entire sample. 57 Table 17 "How-To" Knowledge (above median on discussion) gtperimental Control 2.72 3.07 Source df MS F p Condition 1 4.012 11.76 p < .001 Hospitals 22 .211 .59 ns Subjects 154 .360 *POOled Error 176 .341 Total 177 eta2 = .053 * = error term used to compute F ratio Table 18 Effectiveness (above median on discussion) Experimental Control 2.38 2.59 Source df MS' F _p Condition 1 1.514 2.783 p < .10 Hospitals 22 .619 1.159 ns Subjects 163 .534 *POOled Error 185 .544 2 _ Total 186 eta - .015 * = error term used to compute F ratio 58 Table 19 Overall Uncertainty (above median on discussion) Experimental Control 2.72 2.99 Source df MS F p Condition 1 2.403 11.17 p < .001 Hospitals 22 .192 .88 ns Subjects 148 .218 *POlled Error 170 .215 Total 171 eta2 = .062 * = error term used to compute F ratio Table 20 Attitude Toward Lodge (above median on discussion) Experimental Control 4.13 3.99 Source df MS F p Condition 1 .643 2.937 p < .10 Hospitals 22 .239 1.103 ns Subjects 133 .216 *POOled Error 155 .219 Total 156 eta.2 = .019 * = error term used to compute F ratio 59 These data indicate that those peOple who were more involved in discussion concerning the Lodge program (and who were, presum- ably more interested in it) were more influenced by the Site Visit than were their less interested peers. The evidence that has been provided has shown that the Site Visit condition reduced uncertainty, increased advocacy and boundary spanning, and resulted in a trend toward greater movement toward adoption Of the Lodge. In order to better understand the processes which took place that resulted in these findings some additional analyses were performed. Effects Of Similarity_and Site Visitor Characteristics Similarity. The Similarity Scale (Appendix W) was given to the Site Visitor as part Of the questionnaire filled out at the conclusion Of the Site Visit. It measured the extent to which the visitor perceived the Site Visit hospital as being similar to his or her own. One way to examine the similarity scores is to compare adopting versus non-adopting hospitals. In this case adopting hospitals were defined as those hospitals in the Site Visit condi- tion that had made some movement toward adoption, and had not regressed, as Of 13 months after the workshop. There were four "adopting" hospitals and eight "non-adopting" hospitals. A t-test comparing the two groups showed that the adopting hospitals per- ceived the visited hospital as significantly more similar than did the non-adopting hospitals (t = 3.37, df = 12, p < .01). This is confirmed by the correlation found between similarity and 60 adoption at 13 months, which was .697. Thus, perceived similarity would seem to be an important factor in the success Of a Site Visit in increasing adoptionCHran innovation. Site visitor characteristics. Correlations were computed to determine whether certain characteristics Of the Site Visitors were related to adoption. The visitor's occupational area Of training seems particularly important. Area Of training (i.e. nursing, psychology, social work, etc.) can be considered to be a scale depending on the relative status of the discipline. Thus areas of training were ordered from one tO five in the following order: aide, nurse, social worker or business, psychology, psychiatry (based on Tornatzky, et al., 1978). Area of training had a correlation Of -.53 with adoption, thereby indicating that the less prestigeous the visitors' area of training, the more effective was the visit. Questions were also asked Of the Site Visitor asking whether he or she thought the Site Visit would be useful for him/ herself or the hospital. Usefulness for self correlated .55 with adoption and usefulness for hospital correlated .62 with adoption. This would seem to indicate that the visitor left the visit with a fairly accurate idea about whether or not the innovation could be useful for his/her hospital. Cluster Analysis One technique that can be used to enhance understanding of the data that has been presented, particularly in terms of the 61 organization involved, is Cluster Analysis. The particular method of Cluster Analysis that was used in this study involved the BCTRY computer package as developed by Tryon and Bailey (1970). In addition to the usual breakdown of variables into clusters (cluster analysis of variables or "V" analysis), the BCTRY system allows the user to find respondent typg§_based on standardized cluster scores. This can be Of particular value in the present situation since the method, known as Cluster Analysis Of Objects ("0" Analysis) could be used to find organizational types in such a manner as to possibly predict change. The first step in the Cluster Analysis is the analysis of variables ("V" Analysis).. The results Of the V Analysis can be found in Table 21. The V Analysis was performed using the scales and individual items from the Mail-Out Questionnaire. The results indicate that there are two clusters of variables, an Attitude- Certainty cluster (which might be considered to be tapping some cognitive dimension concerning the Lodge) and a discussion cluster. The relationships between the clusters themselves and with the four change scores are shown in Table 22. The two clusters are clearly related. The discussion cluster is shown to be rather highly related to change, while the cognitive cluster is shown to be related, but less so. The Cluster Analysis Of Objects ("0" Analysis) revealed four "0" Types Of hospitals. The standard scores Of the O'Types on each of the clusters is presented in Figure 6. Basically the 62 Table 21 Cluster Analysis - Cluster Loadings Cluster 1 - Discussion (Reliability = .862) Variable Loading 1. Perceived Discussion (MOQ, 01) -1.000 2. Mean Discussion (MOQ, Q 33) .767 3. Highest % Of Vote for Advocate (M00, 05) .742 4. Mean Link Strength Of Star (MOQ, Q 33) .612 5. Personal Involvement in Discussion .449 Cluster 2 - Certainty-Attitude (Reliability = .840) Variable Loading 1. "How-To" Certainty Subscale .838 2. Effectiveness Certainty Subscale .837 3. Attitude Scale .758 4. Feasibility Certainty Subscale .507 Table 22 Correlations of Clusters to Change and Each Other Cluster 1 Cluster 2 Cluster 2 (Attitude-Certainty) .406 Change 3 months .335 .176 Change 6 months .430 .220 Change 9 months .588 .487 Change 13 months .397 .285 63 mmaxhio Lo mmgoom emumzpu o weaned mmwhmDJQ mantra 2063.520 i>._.z_<._.mmo N _ - d ¢w&>.—..Ou ............. .\.\.\. n ECO. ii... .11.... N we; .11.... ii... . EEO "111 on 0v On 00 Oh 8313013 OEZIOEJVONVIS $38038 64 Table 22a Correlations of Variables in the Cluster Analysis with Change Variables Change at 13 Months Knowledge .113 How—TO Certainty .202 Effectiveness Certainty .298 Feasibility Certainty .163 Attitude .176 Perceived Discussion .328 % Naming Advocate .177 Actual Mean Discussion .335 Mean of Advocate .493 four O-Types are: l) O-Type 1 (6 hospitals) - average attitude- certainty, with very low discussion; 2) O-Type 2 (11 hospitals) - moderate attitude-certainty, with moderate discussion; 3) O-Type 3 (4 hospitals) - moderate to high attitude-certainty, with high discussion; and 4) O-Type 4 (3 hospitals) - very low attitude-certainty, with low discussion. Figure 7 depicts the mean movement toward adoption by each O-Type at each of the four follow-up periods. 0- Type 3 is seen tO have the greatest amount of change, O-Type 2 a somewhat lesser amount, and O-Types 1 and 4 essentially none. This would seem to confirm the notion, discussed earlier, that reduced uncertainty and increased discussion would lead to greater change. What is of particular interest is that pptn_reduced uncertainty and increased discussion seem to be necessary for change to take place. However, while those are necessary for change, they are by no means sufficient. There were hOSpitals in both O-Types 2 and 3 that produced no change at all. 65 week Lm>o maxk-o An :owunou< uemzop pcmsm>o= N «gnaw; $320.2 2. m2; n. m n o I _ \\\v 0 JD \\\ \\\\\ .‘l ......... I. .o \ K x 1 N .\ N .\ .x .s x x 1 n v E>PI°I .......... \u n wm>hion 1.1.! s. N ghéiui \iliit — wghn°Il .|.‘.‘. I? ¢ NOlidOCN OHVMOJ. lNBWBAOW 66 To conclude this chapter, one final test should be noted An overall test Of significant was performed comparing experimental and control conditions on multiple dependent measures. These measures were a) the two cluster scores, b) the locus of control measure, and c) the knowledge test. The last two were included separately due tO their being dropped from the cluster analysis. The multivariate F did not reach the .05 level Of significance (F = 2.46, df = 4, 19, ns). CHAPTER IV DISCUSSION Social Process Perhaps the clearest findings Of this study havetx>do with the way in which social processes within the hospitals were affected by the Site Visit. The uncertainty that hospital staff felt con- cerning the Community Lodge was reduced as a result of the Site Visit. This effect was particularly pronounced for the uncertainty the staff felt concerning their knowledge about how to set up the program. This may have indicated that the visit was perceived more as a “how-tO-do it" lesson than as a means Of providing evidence for the effectiveness or feasibility of the Lodge. That there was no effect on uncertainty of feasibility is of considerable interest. On the one hand this seems to indicate that a Site Visit has no effect on this variable. On the other hand, combined with the finding that similarity increased the effec- tiveness