. . ‘0 '- * . .0“......-...§J.. ”.0 ...‘-9".mh.‘"‘ -'.'o I'vottsu’o100010 v.0- sq » n V. . no! o (.0-I,< ’ .a... avionov-oo- ow pyA- .0 '0‘.-. '00 . O‘ItO-tn O V 0- 0‘. f..(—‘9.f.'.'. 'O ‘n13.5.'l m M mm m, m E M Nu. m. m m w W m m A mums: Thesis for the Degree of M A, WWW STATE UHNRSTFY I 1974 . c . _. a . .; . . . .. . 9 . .. . . ‘- . . z. . . ‘ . . . ... . . p . . . . . . .. ’ . . . . . . . u .. . .. .. .. . ... . . .. .V. I .. . o . . . . . . ~ . . . . . .... n . o . . ..4 n. . . . «. .c,. ..V._a..o.. _... ... ...... .o..¢u-a...4.._ - .m. . . u .oo.A .o .‘o... .. . If: .. . . . ........u........3. ran... m.:_n..u.xh..?...‘.. .255“: u... .3. “3.95M. ... ... 2.7 1.3.1.. J, n . 2.. . 2‘ . ...,. o.—.....Jb _ . .q.‘o ..L. I .?. . ... ‘ .u 1. .2 3 #rf. .29? - . . ‘ .V i... . o y . (‘5‘ q “3 _...q.§£s.h%3 Mtg“... .. LIBRARY Michigan State University THEblS DISPAR; LAB shlft dlff scores. ata was eight.day h an r ABSTRACT DISPARATE PERSONALITY SHIFTS AMONG LABORATORY TRAINING SUBGROUPS BY Fred Wise The present study investigated relationships between participants' initial prelab scores on a semantic differ- ential measure of interpersonal perception (Person Description Instrument--form X [PDIX]; Harrison, 1962, 1965, 1966) and their subsequent shifts coincident with a T-group experience. It was hypothesized that subgroups would shift differentially dependent on their prelab scores. Data was collected from 50 participants in an eight-day human relations training lab designed to enhance interpersonal competence. Additional PDIX ratings were obtained from one job colleague and one personal intimate, both chosen by and for the purpose of rating each partici- pant. Ratings were made at five weeks prelab, days 2 and 7 within the lab, and at six months postlab. Participants described themselves on the PDIX scales at all four occasions, while intimates and colleagues provided both -.- :23, -‘ Prelab and {>05th group participants 7. Participants W subgroups accordin separately for eac personal Warmth an Work, and Activity median percentage high (30%), middle all other sources associates) strong differences betwee each observation . Fred Wise prelab and postlab descriptions. Additionally, fellow T— group participants described participants on lab days 2 and 7. Participants were divided into high, middle, and low subgroups according to their prelab, self-descriptions separately for each of the three PDIX scales: Inter- personal Warmth and Acceptance, Power and Effectiveness in Work, and Activity and Expressiveness. The approximate median percentage of individuals in these subgroups was; high (30%), middle (40%), and low (30%). Descriptions from all other sources (intimates, colleagues, and T-group associates) strongly confirmed the substantial self-rating differences between the high, middle, and low subgroups at each observation. Results derived from complex analyses of variance indicated an overall change pattern across observations for the total group which closely resembeled the pattern of shifts exhibited by both high and middle subgroups. This included a decrement from prelab to day 2, a moderate increase within the lab, and a further increase from day 7 to postlab to a point not significantly different than their prelab scores. In sharp contrast, the low groups made increments at each subsequent observation time and were the only group to show significant prelab to postlab shifts. The ratings by intimages, Colleagues, and T-group associates provided some external confirmation of these changes. The hypothesis, that T-groups differentially influenced subgroups of individuals was clearly supported. It was suggested primarily of valr by themselves and measures of this lying processes r findings and sug. assessment of su. TheSiS Committee 302. ' urley Richter ' Thorr'n ard K. RUSS E Fred Wise It was suggested that perhaps such laboratories are primarily of value to persons who initially are described by themselves and others as relatively lower than peers on measures of this kind. lying processes were offered. Speculations about possible under— The implications of these findings and suggestions for further research on the assessment of such shifts were also discussed. Thesis Committee John R. Hurley, Chairman Dozier W. Thornton Richard K. Russell App OVE'Ad‘: x'; f I, / /// 1/ . - _ ,7 /',;,/,. x , 1".” ’1.“ li‘ ”7—4! I. (’4‘; 1 ff ’ // Date: {If—1 "if: (1/ "'J _ 41- .9 DISPA . h DISPARATE PERSONALITY SHIFTS AMONG LABORATORY TRAINING SUBGROUPS BY Fred Wise A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Psychology 1974 To Sigmund Freud, who, more than once, was known to have said ”This is Crazy" ACKNOWLEDGMENTS I would like to eXpress my deep appreciation to Dr- John Hurley, chairman of my thesis committee, for his continual support, encouragement, and guidance throughout the work of this thesis. I am also grateful to the members Of my committee, Dr. Dozier Thornton and Dr. Richard Russell, for their participation and continued help. I am also indebted to Tom Nicol and Mike Denny for their help in a seemingly unending battle of man against Ina-':31‘line, the computer, and their patient advice on the -statistical analyses of this study. Finally, I wish to thank Diane for her limitless s ource of energy and encouragement. iii TABLE OF CONTENTS LIST OF TABLES . . . . . . . . . . LIST OF FIGURES. . . . . . . . . . INTRODUCTION 0 O O O O O O O O O 0 Definitions and Purposes of T-groups . . REVIEW OF THE LITERATURE. . . . . . . Problems in Research . . . . . . . Factors Relating to Change . . . . . Effects of T-groups on Participants . . THEORETICAL JUSTIFICATION . . . . . . METHOD. 0 O O O O O O O O O O 0 Subjects 0 O O O O O O O O O 0 Instruments . . . . . . . . . . Des ign O O O O O O O O O O O 0 RESULTS............ Data. 0 o 0 Constitution of High, Middle, an Low Subgroups Measurement of Change . . . . . . . Main Effects of Time of Observation .. . Interaction of Observations With Subgroups Effect of Levels. . . . . . . . . DISCUSSION . . o o o o o o o I 0 Self-Report Ratings. . . . . . . . A S1_1rprise Finding . . . . . . . . Ratlngs by Others 0 o O o o O 0 0 Implications and Future Research . . . iv Page vi vii 12 20 33 44 44 45 47 49 49 50 51 52 62 64 71 71 77 80 83 Chapter BIBLIOGRAPHY . APPENDIX A Chapter Page BIBLIOGRAPHY O O O O O O O O O O O O O 86 APPENDIX A o o o o o o o o o o o o o 94 Means f. data ANOVA f ANOVA f ANOVA f F‘ratic THREy‘E for 5 LI ST OF TABLES Table Page 1. Means for all observations across sources of data . . . . . . . . . . . . . . 55 2. ANOVA for Scale IWA. . . . . . . . . . 56 3. ANOVA for Scale PEW. . . . . . . . . . 57 4. ANOVA for Scale A&E. . . . . . . . . . 58 5. F-ratios for mean differences across levels. . 63 6. Tukey's post-hoc analysis of mean differences for self-scores . . . . . . . . . . 66 Figure 1' TOtal grc and Agf N . subgroup warmth SubgrOUp Effect Subgroup EXpreS LIST OF FIGURES Efiigure Page 1. Total group mean scores on Scales IWA, PEW, and A&E O O O O O O O O O I O O O 54 2. Subgroup's scores on Scale IWA, Interpersonal Warmth and Acceptance . . . . . . . . 59 3. Subgroup's scores on Scale PEW, Power and Effectiveness in Work . . . . . . . . 60 4.. Subgroup's scores on Scale A&E, Activity and Expressiveness. . . . . . . . . . . 6l vii The pre groups and thei Purpose of this between a Persc atest of inte: and his Subseq of 18boratory methOdol-Ogy an that T'gl‘oups researchers: c “groups, (2) the methodo 10" he Shall then cerned with an result of T‘g~ D 4 £ Since ratory (NTL) INTRODUCTION The present study deals with encounter groups or T- groups and their effects on participants. The specific purpose of this study was to examine the relationship between a person's initial score (pre—lab, self-rating) on a test of interpersonal perception (PDIX; Harrison, 1964) and his subsequent scores on that same variable as a result of laboratory experience. Before proceeding into the methodology and results it comes to the author's attention that T-groups are no standardized phenomenon. It is, therefore, logical to proceed with a review of some researchers' opinions on (1) the nature of the purposes of T-groups, (2) the goals that they may have, and (3) some of the methodological problems involved in T-group research. We shall then review some of the specific studies con- cerned with actual assessed changes in individuals as a result of T-group experience. Definitions and Purposes of T-groups Since their inception at National Training Labo- ratory (NTL) in 1947, training groups or T-groups have enjoyed increasing myriad of purposes inproving relatior aids for universi and increasing ur Indeed, there is according to pur‘ Depending on the have been called groups, sensual. groups, or trai educational inn teaching goals Bradford, Gibk ature, One. bee ; .Ormulating a enjoyed increasing popularity. They have been used for a myriad of purposes including training management personnel, improving relationships within organizations, as teaching aids for university classes, stimulating personal growth, and increasing understanding of others, to mention a few. Indeed, there is no one standard type of T-group. They vary according to purpose, setting, structure, and practitioner. Depending on the emphasis on each of these variables, they have been called T-groups, encounter groups, sensitivity groups, sensual-awareness groups, growth groups, task- groups, or training labs. Some have called them a simple educational innovation while others attribute much farther reaching goals involving personal and social learning (Bradford, Gibb, Benne, 1964). In reviewing the many different types of groups reported on in research liter- ature, one becomes actuely aware of the difficulty of formulating a comprehensive definition of T-groups. It must also be kept in mind that research groups probably represent only a portion of the types of groups presently being conducted. Perhaps one of the best ways to define a T-group would be to examine (l) the goals and purposes of the group, and (2) the structure and processes that are used in attempting to achieve those goals. Campbell and Dunnette (1968) have composed a list Of Six goals for T-groups using resource material from AWinds. (1954), Bradford et al. (1964), Buchanan (1965). tiles (1960) , Shei IBM). It is as 1. Increased one's own context. 2. Increased Wthh in: understax includes 0! the a} to thee 3' InCrease 0f Proce 4‘ A Combi individ Skills Persona VEne a. 1eVels effect “Learn Crease goals inter} Altho enphasis Plac Miles (1960), Shein and Bennis (1965), and Tannenbaum et a1. (1961). 1. It is as follows: Increased self-awareness or self-insight concerning one's own behavior and its meaning in a social context. Increased sensitivity to the behavior of others which includes not only being able to recognize and understand another person's behavior, but also includes what would clinically be deemed as empathy or the ability to communicate that understanding to the other person. Increased awareness and understanding of the types of processes inherent in group functioning. A combination of the first three goals in which an individual would acquire diagnostic and conceptual skills or the ability to assess social, inter- personal, and intergroup situations. Increased ability to use #4 to effectively inter- vene at both the interpersonal and technological levels of group functioning so as to increase group effectiveness, satisfaction, and output. "Learning how to learn" where an individual in- creases his flexibility in using the previous five goals to establish more effective and satisfying interpersonal relationships. Although T-groups vary in regard to the amount of emphasis placed on each of the aforementioned goals, most investigators seem to agree that they are all necessary and integral parts of most T-groups. The degree of emphasis placed on these goals can often be seen in the structure or processes used to achieve these goals. For example, Shein and Bennis (1965), Benne (1964), and Harrison (1962), stress methods for organizational effectiveness improvement, Valiquet (1968) discusses individual work improvement, and Shutz discusses training designs aimed at increasing inter- personal effectiveness, to mention a few. Before designing structures and processes for achieving these goals, however, it is important to examine some basic assumptions of how an individual learns and specifically, how he can learn from a T-group. The liter- ature seems to suggest a common set of assumptions for what facilitates growth and learning in laboratory environments. Argyris (1962) presents seven factors that he feels contri- bute to an individual's learning within a T-group: l. The T-group emphasizes a participant's responsibi- lity for self-development. 2. Education in human-relations is a matter of re- learning or an unfreezing of old patterns of behavior and replacement with new, more adaptive patterns. 3. This re-learning process is much the same as the original learning process in which an individual learns in a social, interpersonal context. 4. The learning process is not only a matter of intellectual understanding, but also of emotional insight and learning. 5. The most effective development of an individual takes place when he becomes both more understanding and more accepting of himself. 6. As he becomes more understanding and accepting of himself, he will also become more understanding and accepting of others, thus reducing their defensiveness and facilitating their growth. 7. As his basic values change then so will his behavior. Also speaking to the issue of what group processes facilitate learning and the achievement of the previously mentioned goals, Gottschalk and Davidson (1972) state, A training create a 5. their own knowledge appraise t' ones. A l participan environmen to experim risks and artificial may safely old attitu and Practi ability to It can °f improving a Increasing awa behaVior' and 0‘ .hers, These A training lab is an educational procedure that aims to create a situation in which the participants, through their own initiative and control but with access to new knowledge and skilled professional leadership, can appraise their old behavior patterns and look for new ones. A lab recommends a temporary removal of the participants from their usual living and working environment, where attempts to re-evaluate attitudes or to experiment with new behavior patterns might involve risks and possible punishment. It provides a temporary artificial supportive culture in which the participants may safely confront the possible inadequacy of their old attitudes and behavior patterns and experiment with and practice new ones until they are confident in their ability to use them. It can be seen then that T-groups have primary goals of improving an individual's interpersonal competence, increasing awareness of his own attitudes, emotions, and behavior, and increasing his awareness and understanding of others. These goals are sought through providing the participants with a relatively unstructured experience in which they have the opportunity for examination of them- selves and others and for experimenting with new behaviors in a social context. The core of this self and other- examination usually rests on the process of feedback or more specifically, an individual is encouraged in a trusting atmosphere to both give and receive feedback about be- haviors and interactions in the group. He thus is afforded the opportunity to find out what others think of not only his old patterns of behavior, but also of any new behaviors he may wish to try out. Reviewi Purported Clair. the all-encomp; haPpen to an ir It seems as thc classified as 5 Some inVEStigag and have 90% a tions of T‘gror But althOugh me 1°°k at the lit variability in REVIEW OF THE LITERATURE Reviewing the literature on T-groups or reading the purported claims of laboratory trainers, one is taken by the all-encompassing nature of change that is supposed to happen to an individual as a result of group experience. It seems as though the experience could more correctly be classified as a cure-all for the social ills of mankind. Some investigators have also recognized this misconception and have gone as far as to directly point out the limita- tions of T-groups (Golembuski & Blumberg, 1970; Back, 1972). But although many claims have been made, one has only to look at the literature regarding T-groups to see tremendous variability in what changes in participants were reported, how long they lasted, and of what relevance to the indi- vidual's life-functioning they had. Added to this are the numerous methodological problems in the research. Indeed, articles quite often begin with an attack on previous methodology as justification for their study. We will review first some of these problems involved in T-group research, then some of the studies on factors related to an individual's functioning within a group, and finally some of the studies concerning actual change assessment in T- groups. Problems in Research Argyris (1964) has said that there has been more research conducted on the T-group method than on any other Specific management-development technique. A bibliography by Durham, Gibb, Knowles, and Harrison (1967) containing over one hundred abstracts of change studies in T-groups seems to attest to this. Indeed, at least one monthly journal, Journal of Applied Behavioral Science, is primarily dedicated to research on groups. But can we really say that these articles substantiate the claims made by their researchers? Perhaps, but it would do good to bear in mind some reservations concerning not only the methodologies, but also the theoretical justification behind some of these studies. A number of books and articles contain material directly addressing these issues (Napier, 1973; Shein & Bennis, 1964; Odione, 1963; Friedlander, 1967; Harrison, 1967; House, 1967; Luke & Seashore, 1966; Golembuski & Blumberg, 1970; Campbell & Dunnette, 1970; Stock, 1964; Bradford et al., 1964). The author sees six basic problems in T-group literature that should be considered when interpreting research results. First and extremely important is the matter of variability both in laboratory design and inplementatio before, group and practitic ammmmdate 1 is what Hart: T‘sroup trai: and prolifer Although Sta Purposes, it Settings for difficult tc learns fOrm justifiatbly Study with . differentia onpartiéip 1 ‘968‘ Peter implementation and in theoretical formulation. As mentioned before, groups vary in regard to goals, setting, structure, and practitioner and thus are designed differentially to accommodate their respective aims. Compounding this problem is what Harrison (1971) calls "a cult of originality among T—group trainers in which a dominant value is the invention and proliferation of new variations in training design." Although standardization would be desirable for research purposes, it would also reduce the number of suitable settings for laboratory application. It is therefore difficult to both ascertain specifically what a participant learns form T-groups, but more generally, what one could justifiably say about the comparability of results of one study with another. Also related here are the problems of differential effects of trainer style and group composition on participant learning (Stock, 1964; Bolman, 1968; Culbert, 1968; Peters, 1966; Lieberman, Yalom, & Miles, 1973). Second is the related problem of whether or not perceived or measured changes are in fact due to the laboratory or to some other intervening variable. This is perhaps best subsumed under the problem of controls or should I say, lack of such. Very few studies have made adequate use of control groups, probably because of the difficulty in obtaining them. Some investigators question the sound- ness of choosing random groups of people to serve as con- trols since they may not possess the same motivation as the people who sign up for T—groups (Massarik, 1965). To complicate mati motivation for usually report of their eXpeI Bach, 1968; Li (1965) has pr: desire T'grou muss and ha then t01d th a This has beer Change not a1 on a Waiting A th Assuming tha how does thi situation? that many 01 experienCQ complicate matters, while control groups may lack the same motivation for membership as participants, participants usually report only positive and possibly inflated accounts of their eXperiences in groups (Rogers, 1967; Mintz, 1969; Bach, 1968; Lieberman, Yalom, & Miles, 1973). Massarik (1965) has proposed one alternative, to use volunteers who desire T-group experience, half of which are placed in groups and half designated as controls. These controls are then told that they will have to wait to be in a group. This has been questioned, however, in that the controls may change not at random, but rather as a result of being put on a waiting list. A third problem is that of transfer of training. Assuming that individuals do change in a laboratory setting, how does this affect them in their work or everyday-life situation? Shein and Bennis (1965), for example conclude that many of the goals and processes of the laboratory experience (i.e.,--honesty and openness) are in fact antithetical to changes in effective job-functioning. Pattison (1965) has also reviewed research data and con- cluded that behavior change within groups is often not associated with behavior change outside of groups. It is easier to assume that learning is automatically transferred at the end of a group, that group processes will be hampered by attention to transfer issues, or that transfer 0f learning is not as important as the group experience itself. But unless practitioners, and some do, uniformly claim that th. experience in with the long ship. Even are still fac temPOrality . change a Pers last? The re Concerned Wit a grer‘it deal example, the} 0f hoUrS' ma] 10 claim that the laboratory is an intrinsically valuable experience in itself, there will still be those concerned with the long range and transfer effects of group member- ship. Even if we can assume a transfer of learning, we are still faced with a fourth and related problem, temporality. In other words, how long does it take to change a person and if he does change, how long will it last? The research most relevant to this problem has been concerned with the length of the group experience. There is a great deal of variability in laboratory length. For example, they may consist of weekly meetings for any number of hours, marathons, or a combination of the two. Some research supports the notion that participant change is related to length of meeting time (Bunker & Knowles, 1967). There is also literature supporting the idea that partici- pants vary in degree of change exhibited over time after the group (Harrison, 1966; Shutz & Allen, 1966; Lieberman et al., 1973). A fifth problem, to be eXplained in further detail in a later section, concerns dimension and direction of change. More specifically, do and if so, how and to what extent do individuals change differently? Stelle (1968) states that it is a faulty assumption that individuals change the same or have the same prerequisite skills for change in a laboratory situation. Harrison (1967) attacks the premise implicit in many articles, that is, individuals change : shown, : change : negative Yalom, 1 the Wldl in comp; in anotl amount 1 that he StUdY ma limit 11 Strictly is Obvic "ant to“: reseau:ck laboratc etc‘ (Ga Smith ’ 1 11 change in one direction only, for the better. It has been shown, in fact, that laboratory participants do not only change in differing degrees, but that they may make either negative or positive changes (Underwood, 1965; Lieberman, Yalom, & Miles, 1973). Again it is also important to note the wide range of variables studied and thus the difficulty in comparing the dimensions of change of one study to those in another study. An experimenter also limits the type and amount of change he will find by choosing the instruments that he will measure such change with. For example, one study may focus on changes in self-concept while another may limit its findings to changes in self-esteem. Sixth and finally, we must consider the more strictly methodlogical problems of T-group research. One is obviously instrumentation, or how can we measure what we want to? Campbell and Dunnette (1968) cite a number of researchers who discuss the problem of measuring what laboratory training purports to change; such things as interpersonal sensitivity, self-awareness, attitudes, values, etc. (Gage & Cronbach, 1955; Cline, 1964; Cronbach, 1965; Smith, 1966). For example, if we measure pre-post differ— .ences on questionnaire responses, are we measuring changes in attitudes, behavior, or merely familiarity with the questionnaire? More overriding may be the issue of exactly what is most important to measure? Related too are the statistical problems in measuring change. Harris (1963), Cronbach and Furby (1970), Tucker, Damarin, and Messick 12 (1966), and Lord (1968) have all discussed the problems of measuring any change. They state that one must consider the relationship between a person's initial score on a test and his change score. Another methodological problem is that of data collection. Harrison (1967) points out some of the problems with administering instruments on the first day such as initial anxiety and uncertainty about the laboratory experience. Yalom (1970) states that partici- pants almost always feel good at the end of lab and thus administering instruments at this time may also lead to erroneous conclusions. A number of investigators have discussed the merits of longitudinal research with more than two test administrations (Harrison, 1966; Bunker 8 Knowles, 1967; Shutz & Allen, 1966; Bare & Mitchell, 1972). Factors Relating to Change Bearing in mind the aforementioned cautions and problems in research, let us now turn to some of the literature concerning factors related to functioning and change within or as a result of a T-group. It is useful to consider two basic categories of factors; (1) those vari- ables that are part of the individual such as personality, desire for change, expectations, etc., and (2) those vari- ables related to group climate such as structure of the group, trainer orientation, group composition, etc. Let us first consider those variables within the individual that are related to behavior or change in 13 behavior in a T-group. Perhaps the most researched area has dealt with personality factors. Shutz and Allen (1966) found that the laboratory experience changed people selectively depending on initial personality character- istics, highly dominant members becoming less dominant and overly affectionate members also becoming less so. Harrison and Oshry (1966) found that members preferring high structure learned more than those perferring low structure situations. In a previous study they found that those individuals who were open to new ideas and expression of feelings were found to most effectively learn and apply their learnings from the group (Harrison & Oshry, 1964). Mathis (1958) found that passive, withdrawing people were least likely to change as they tended to prevent eXposure, confrontation, and exploration which are all seen as necessary prerequisites for personal development or growth. Steele (1968), using the Meyers Briggs as a prediction instrument, found a relationship between the sensation- intuition scale and those individuals who were rated as most effective in a T-group. This scale is designed to measure a person's preference for basic modes of conceptu- alizing the world. It was, however, less effective in predicting actual change as a result of the group. Greening and Coffey (1966) found that impersonals, those who rated others in terms of achievement, ambitions, and independence, gained more from laboratory training than did personals who rated others in terms of love and intimacy. Not (1960) found studied (ego directly aff Carlson (196 (CPI) was nc pants. Kerr sonality meg group, Liek is Easier t< encounter g: values than LiPPitt, Ka laboratory were more I than their elusive evi istics with Side the 9: 14 Not all research is positive, however. Miles (1960) found that none of the personality variables he studied (ego strength, flexibility, and need affiliation) directly affected laboratory outcomes. Massarik and Carlson (1962) found the California Psychological Inventory (CPI) was not predictive of laboratory change in partici- pants. Kernan (1964) also found no differences on a per- sonality measure (authoritarianism) before and after a T- group. Lieberman, Yalom, and Miles (1973) conclude that it is easier to predict a person's success or failure in an encounter group by his initial expectations, attitudes, and values than by measuring personality traits. Watson, Lippitt, Kallen, and Zipf (1961) analyzed extensive data on laboratory groups and concluded that personality factors were more related to how a person filled out a questionnaire than their actual behavior. In fact, they found no con- clusive evidence that behavior or personality character- istics within the group were predictive of any change out- side the group. Concerning more specific individual variables, Harrison and Lubin (1965) found that person-oriented people were more expressive, warm, and comfortable in a T-group setting than were work-oriented individuals. K01b: Winter, and Berlew (1968) report results indicating that those most likely to change in a T-group were those who initially were committed to the change goals of the group. Although their research was conducted on therapy groups which are somewhat 15 analogous to T-groups, Houts, Zimerberg, Rand, and Yalom (1965) also found that the only variables predictive of success were attraction to group and the person's general popularity. In summary, Stock (1964) concludes her review of laboratory training research by stating, "these findings converge on the idea that personality factors having to do with receptivity, involvement, lack of defensiveness and a certain kind of energy or openness may be the important facilitators of learning." The other set of variables that are seen to affect a participant's behavior is group climate. Included here are such things as group structure, trainer characteristics, and group composition. We shall examine some of the literature relevant to each of these factors. First, group structure including such things as length of group meetings, type of activities, and nature of communication is an extremely important variable affecting laboratory outcome. Bunker and Knowles (1967) compared the effects of laboratories of different length. Their assumption was that amount and kind of training outcome varies with respect to amount of input (length of lab). Thus by measuring outcome effects for different length laboratories, they could avoid some of the problems inherent in designs using control groups that had not had group eXperience. Using laboratories of two and three week lenghts, they found that those participants in the three- week group made more changes as viewed by both themselves 16 and others and that these changes were overt, active changes in behavior while the two-week sample changes were of less magnitude and more of a passive, attitudinal nature. They concluded that the longer labs provided more time for the participants to incorporate and test attitudinal changes and to then adopt them behaviorally. Other studies point to the fact that changes in both behavior and attitudes may not even be seen within the context of the laboratory situation but rather are only evident at a time quite a bit after the experience (Harrison, 1966; Shein & Bennis, 1965; Shutz & Allen, 1966). Related to laboratory length, Bare and Mitchell (1972) report a study in which they manipulated time patterns of the group meetings into what they called spaced (30 hours over 10 weeks), massed (24 hours of continuous meeting), and a combination of these. They found that the combination pattern was most conducive to behavior change of the participants, although the results were not signifi- cant when measured again at three months after the labo- ratory. In research concerning other group structure variables, Luke (1972) found that normative structures of T-groups often tend to be supportive of interpersonal encounter while norms of a conceptual nature are seen as less important. As a result, he concludes, it may be very difficult for members to make conceptual and behavioral changes. Stock and Luft (1960) conducted a study using high and low structure groups and found that low-structure 17 was more conducive to emotionality while the high structure groups were characterized by relatively superficial encounters, although were extremely active. Harrison (1965) interprets a number of studies in terms of confrontation theory and posits an optimum stress level for maximum participant learning. Stock (1964) also reviews a number of studies which reflect on the nature of group structure and its effects on participants. Egan (1970) discusses the varied effects of different group structures, examining such things as feedback, exercises, and norms in terms of "contract groups." Finally Lieberman, Yalom, and Miles (1973) explore many of the same variables in the context of encounter groups. Trainer behavior is the second major classification of group climate variables having an impact on partici- pantS' change. The trainer is probably the most important individual in the group in terms of effect on participant interaction (Hare, 1962). Luke (1972) also states that although many labs are purportedly a mutually accommodative learning environment in which trainers and participants have equal say in the normative structure of the group, .many T-groups do not even approximate this. He concludes that trainer style appears to determine which and how much members influence the normative structure of the group. Perhaps the most definitive and thorough examination of this phenomenon can be found in Lieberman, Yalom, and Miles' (1973) extensive study of encounter groups. Using eighteen 18 different groups with trainers representing most major styles of leadership, they conclude that trainer behaviors (often not related to their theoretical orientation) have a definite impact on whether or not a participant will be benefitted or harmed by the encounter group experience. More specifically, members who were seen to have benefitted most had trainers who were characterized by their caring and ability to provide a conceptual framework in which participants could interpret their experiences. Negative effects were associated with trainers who provided excessive emotional stimulation or were seen to have been extremely directive in terms of such things as rule setting and inter- action intervention. Two other excellent reviews of trainer role, function, and effect on a T-group may be found in Hare (1962) and Golembuski and Blumberg (1970). The con- clusion to be drawn appears to be that trainers do have a definite impact, but that it varies with regard to both his behavior and group composition. The third important factor of group climate contri- buting to an individual's change is group composition. Here we are interested in such variables as members' relationships, status in group, and whether or not the group is homogeneous or heterogeneous. We have already seen that an individual member's'personality, attitudes, values, or expectations may have an impact on his behavior or change in behavior within a group. It is logical then to assume that a combination of these, whether they be the sam' Shu and em in prq de co re fe 91 19 same or different across members, may also have an impact. Shutz (1961) used the FIRO-B to compose homogeneous groups and found that the group with individuals sharing a prefer- ence for receiving rather than initiating was relatively low in activity compared to a group whose members showed a preference for interaction. He also found that members who showed a high concern for affection were seen as primarily dealing with warm feelings in the group, while a group composed of people who expressed a preference for distant relationships were seen as being unable to deal with feelings. Harrison (1965) examined a series of studies on group composition and concluded; (1) compatible, homogeneous groups may depress conflict and inhibit the learning of the members (Harrison & Lubin, 1965; Harrison, 1965), (2) con- flict and incompatibility of personal styles may lead to exploration of alternatives and to learning (Harrison, 1965; Lieberman, 1958), (3) personal styles which depend on passivity and withdrawal for coping with interpersonal stress may prevent exposure, confrontation, and exploration which are all central to the learning process in T-groups (Mathis, 1958), and (4) poorly integrated, stress-vulnerable indi- »viduals have difficulty functioning in groups where ambiguity and emotionality are optimal for the learning of others (Powdermaker & Frank, 1953). Stock (1964) also looked at a number of studies on group composition. She came to the conclusion that group composition most cer- tainly affects the character of group interaction in that 20 the members' interaction, based on their initial orienta- tions, directly affects the normative structure of the group and thus the type and amount of interaction that will occur. It is clear then that there are many variables that are related to an individual's behavior or change in behavior in a T—group. It would seem logical to conclude that although studies have customarily focused on only one or a few of these at a time, there is probably a great deal of interaction among them. It is therefore very difficult to make any definitive "this always leads to this" infer- ences. Stock (1964) agrees stating, "It is difficult to separate out any single aspect of a laboratory and say this is what influenced learning." Effects of T-groups on Participants The previous sections have discussed various goals and purposes of T-groups, problems with research, and some of the specific factors relating to an individual's functioning or change in a group. We shall now briefly review some of the studies concerning actual assessed changes in an individual as a result of T—group experience. The literature can be divided into two main categories along the lines of whether it is concerned with changes in behavior or with changes in attitudes, values, and per- sonality characteristics. Further analysis could focus on whether the changes were observed within the group or c! ir. in 8a SE} of act 21 outside the group. Changes in behavior are most often measured by the observations of others, within the lab by other participants and outside the lab by colleagues, friends, or independent observers. Attitudes, values, and personality characteristics are usually measured, both inside and outside the lab, by participant responses to a questionnaire or some other kind of evaluative instrument. As mentioned earlier, T-groups have a varied range of goals. As well as behavioral change (to be discussed shortly) laboratories often have as goals, changes in self- perception, changes in interpersonal sensitivity or per- ception of others, and changes in attitude and personality characteristics. Luke (1972) states that since norms with interpersonal encounter as a focus take primary importance in T-groups, one would assume that these goals would be salient. We shall examine these three areas of change goals separately although they are obviously interrelated aspects of an individual's personality. A change in self-perception is commonly thought to be a precursor to an individual's growth or self— actualization. Zand, Steele, and Zalkind (1967) state, Generally, training laboratories attempt to help participants become more aware of their assumptions about interpersonal trust, openness, and expression of emotions. Laboratories usually attempt to increase a person's sensitivity to his feelings and the feelings of others. They may help a person to see the differ- ent styles of behavior that he and others use to cope with uncertainty, authority, and competition. It is assumed that the laboratory experience will become a foundation for future changes in behavior. cha lat mo: in P0 re as Th th th 56 SE er 22 Burke and Bennis (1961) found statistically significant changes in participants' self-perceptions during a two week laboratory. Eighty-four participants were seen to become more satisfied with their self-perceptions and to move more in the direction of their "ideal-self." An ancillary finding was that their perceptions of others seemed to converge with how those others saw themselves. Lieberman, Yalom, and Miles (1973) found that although self-reported changes are quite often biased, fifty to seventy percent of all participants of their encounter groups report some positive change in self-perception. The most frequently reported change was in the participants seeing themselves as being more honest and open in communication with others. They also reported an increased awareness and acceptance of themselves and seventy-two percent stated that they expected the change to be lasting. When interpreting results of studies dealing with self-perception, it should again be stressed that these self-ratings tend to be biased in the direction of positive change. Indeed, a number of studies dealing with self- perception report less than positive results. Gassner .(1964) used a control group and found no differences between them and his experimental group in terms of dis- crepancies between "actual" and "ideal" self-description. Stock (1964) found that individuals whose self-perceptions changed most were also most variable in their descriptions. She therefore concluded that they actually became less sure rea acc cal E101- shu. thej star nude thei 23 of the changes they made in the group. Campbell and Dunnette (1968) sum up much of this scepticism concerning self-perception studies; . . . it seems relatively well established that the way in which an individual sees himself may indeed change during the course of a T-group. However, there is no firm evidence indicating that such changes are produced by T-group training as compared to other types of training, merely by the passage of time, or even by the simple expedient retaking a self-description inventory after a period of thinking about one' 8 previous responses to the same inventory. Related to self-perception is perception of others. A number of investigators have studied and concluded that there is a positive relationship between self-acceptance and acceptance of others (Omwake, 1954; Sherrer, 1949; Rubin, 1967). It is assumed that as one becomes more willing and able to accept himself or see himself as he really is, he will also be better able to see others and accept them as they are. This is what Eric Berne (1966) called the "I'm OK, you're OK" position. Although many trainers claim that participants move more in this direction as a result of the laboratory experience, the evidence appears to be less conclusive. Shutz and Allen (1966) found that the large majority of their laboratory participants reported a greater under- standing and insight into others as well as an increased understanding of themselves. But they also caution that their findings are based entirely on self-reports and that participants are inclined to find something good about the experience regardless of what happened. This would probably 24 be especially true if they had to pay for attending the training, as is frequently the case. Miles (1960) found significant differences between participants and control subjects in analytical understanding of human behavior, insight into group processes, sensitivity to others' needs and feelings as well as an increased understanding of themselves. This study had the advantage of both control groups and independent observers. Argyris (1962) found that participants tended to describe others in more interpersonal terms after the laboratory. Using middle and top managers from an organization, Argyris reported a significant difference between them and a control group of managers selected from the same organization. His results do not address the issue of long-term change however, since they were only true when members were describing other members of the lab. He suggests counter-pressures from the organi- zation prevented them from transferring the learning out- side the lab. Similar results are reported by Harrison (1966) who conducted two studies designed to investigate changes in concept preference in interpersonal perception. He posited that concepts which an individual uses to structure his interpersonal relationships directly affect how he will respond to others. He further states that if it can be shown that a T—group affects these conceptual preferences, the participants will as a result change their interpersonal behavior. Harrison states that the direction of this change should be toward a greater abstractness, 25 complexity, and depth of interpersonal concepts (increased awareness of his feelings, perceptions, and attitudes toward others). His 1962 findings show a conceptual shift in the expected direction when describing other laboratory participants, but not when describing others outside the group. His second study found significant changes in concept preference, but only at three months post-lab. Finally, on the more negative side, Zand, Steele, and Zalkind (1967) found that participants (organizational managers) had statistically significant reductions in the extent to which they trusted work associates, the extent to which they saw work associates as trusting each other, the amount of help they saw others as willing to accept from others, and the extent to which they felt open to share ideas with their bosses or seek help from others. They explain their results as participants having changed from a less to more demanding definition of trust, open- ness, and acceptance of help. In summary it appears that laboratories do have the ability to change a person's interpersonal perceptions, but that the nature of this change may be dependent on (1) how it is measured (self-report or independent observer), (2) where and when it was measured (in lab, at work, or back home), (3) the nature of the laboratory experience (purposes and goals for change) and of course, (4) the individual's own ability to change. 26 We shall now turn to a third type of change, that affecting personality characteristics and attitudes. This area has had relatively little research. Until recently most research has focused on organizational change, or improving an individual personally and socially so that he may better function within a work organization. As a result, the variables studied are often defined only in terms of organizational efficiency such as ratings of job performance. Perhaps the paucity of research may also, in part, be due to the inherent problems with this type of study. Personality characteristics have always been diffi- cult to study in that there is no agreed upon method of measuring them and, like attitudes, it is often hard to decide upon what is relevant to measure. It is also diffi- cult to establish adequate control groups to aid in assessing such intervening variables as leader impact, length of lab, and laboratory techniques used. Finally there is the problem of specificity of variables measured. Quite often the variables are so global as to become meaningless. In contrast, when the variables measured become too ‘ specific, results tend to be inconclusive. For example, Stock (1964) studied changes in self-concept and found that participants who reported the most change were also those who become most unsure of what kinds of persons they were. Kassajarian (1965), using pre and post tests, found no significant changes in participants measured for 27 inner-directedness vs. outer-directedness. Campbell and Dunnette (1968) cite an unpublished study by Massarik and Carlson in which they administered the CPI to a group of college students and a study by Kernan (1964) in which the California F scale was administered to a group of engineering supervisors before and after a T-group. Neither study yielded significant results in terms of personality change, although the authors concluded that such basic personality variables may be unaffected by a relatively short laboratory experience. Perhaps the most positive results in the literature relevant to personality and attitude change have been found using the FIRO-B (Fundamental Interpersonal Relations Orientation-Behavior) personality test. This questionnaire includes a series of attitude items designed to measure what Shutz calls the three primary group behaviors; affection, inclusion, and control. Shutz and Allen (1966) found significant differences between experimental and control subjects in correlations of pre and post-test scores on the FIRO-B. They concluded that the lab experience (a Western Training Laboratories program) caused changes in the attitudes of the participants. Smith (1964) also reported significant attitude change in laboratory participants. They found that the disparity between an individual's own behavior or desire for a specific type of behavior and how that individual wanted others to behave decreased more for the laboratory sample than for the controls. Although 28 these studies seem to point to positive evidence that T- groups can change personality or attitudes of an individual, it must be remembered that only one measure was used and that other studies were not so conclusive. In summarizing personality and attitude change studies, Gottschalk and Davidson (1972) conclude . . . the evidence is meager that such participants undergo significant attitude changes or personality changes. . . . What is clearest is that these groups provide an intensive affective experience for many participants. In this sense, the groups may be described as potent. In reviewing behavioral change related to T-group experience, it becomes clear that most of the research is aimed towards measuring changes in job performance. A number of studies have been cited in support of the ability of a laboratory to change job behavior. Miles (1960) reports a well-controlled study designed to assess changes in job performance. Using three instruments to assess change, he found that 73 percent of the experimental group and only 29 percent and 17 percent, respectively, of the matched and random control groups showed statistically significant changes in job performance as measured by a "perceived change" rating made by job superiors, sub- ordinates, or peers. Bunker (1965) refined this same instrument and essentially replicated Miles' findings. Valiquet (1964) also obtained statistically significant results comparing experimental and control groups on the total number of changes observed by independent raters and 29 total number of changes reported by the subjects themselves. Buchanan (1964) and Blasfield (1962) analyzed T-group programs administered to entire organizational units. They found positive behavioral changes in delegation of authority, managerial performance, teamwork, and effectiveness of the unit. These results should be interpreted with caution however, since there was no control group and no units of change were reported. Other studies supporting changes in job performance include Boyd and Ellis (1962), Beer and Kleisath (1967), and Friedlander (1968). Although the aforementioned studies are cited in support of laboratory training for changing job performance, there are other studies which point to a more general behavioral change as a result of T-groups. Gottschalk and Davidson (1972) in a review of outcome research in encounter groups list a number of goals of T-groups. Among them are general behavioral changes such as improved group functioning, better socialization skills, and more effective interpersonal relationships. The literature seems to at least partly substantiate claims that T-groups are effective in meeting some of these goals. For example, Bunker (1965) found that 33 percent of the 229 members he studied showed favorable increases in openness, receptivity, toleration of differ- ences, understanding of self and others, and operational skills in interpersonal relationships while only 15 percent of his control group improved on these variables. Argyris (1965) reports a study using a set of categories he 30 developed to measure the interpersonal competence of members in a T-group. Using independent observers and with a relatively high interscorer reliability, he found significant positive behavioral changes in interpersonal competence within the group. These findings take on special meaning considering the rigor with which his studies were conducted. Bare and Mitchell (1972) showed that experimental subjects exhibited more behavioral change than did their control subjects, both at the end of a sensitivity training labo- ratory and three months after it. These behavioral changes were measured by individually constructed rating scales that were scored by self, group, leader, and colleagues. Shutz and Allen (1966) conducted a study in which they used the FIRO-B personality measure to assess participant per- ceptions of attitudinal and behavioral changes. After six months the participants reported feeling less tense and more honest and confident in their relationships with others. The authors note, however, that certain methodological difficulties such as self-report bias, inadequate control group, and self-selection of the sample pOpulation may in part contaminate their results. And finally, Underwood (1965) reports a field experiment in which changes in personal, interpersonal, and non—personal behavior as reported by work associates of lab members were measured. Their participants, fifteen supervisors drawn from several departments and organizational levels, were seen to have made both positive and negative changes as a result of the 31 laboratory training. Observers found a general increase in supervisory effectiveness, but a decrease in personal effectiveness. While their results must be interpreted in light of the structure and type of laboratory, the observers, and the different supervisory roles of the lab participants, they still suggest that laboratories have a potential for both a constructive and destructive impact on an indi- vidual's behavior. Not all research on behavioral change in T-groups is positive, however. Both the Underwood study just mentioned and a study by Yalom and Lieberman (1971) found negative as well as positive changes in behavior. Both authors suggest that negative behavioral change is probably a result of the "shake-up" type of eXperience that the laboratory provides. Some members may not be able to cope with this and thus the experience would not be constructive. That the laboratory provides this "shake-up" experience appears to be agreed upon by most T-group researchers. Some posit that it is necessary to "unfreeze" one's old values and attitudes in order to allow him to form new perceptions about himself and others, and in turn, change his behavior (Bennis, 1962; Bradford et al., 1964; Schein & Bennis, 1965). Argyris (1962) agrees, stating that although his findings point to the fact that laboratory training can change beliefs and values, whether or not this is translated into behavioral change outside the laboratory depends both on the participant's motivation and 32 the counterpressures from outside socializing influences. Zand, Stelle, and Zalkind (1967) conclude that The immediate effect of a laboratory experience, rather than producing changes in observable behavior, may be to change participants' perceptions. The laboratory seems to enable a participant to develop and inter- nalize new and probably more objective standards for evaluating his feelings and behavior and the feelings and behavior of others; this in turn may lead to a new perception of old events. Bunker and Knowles (1967) and Luke (1967) suggest that per- haps what is needed to translate learning from within a T- group to outside behavioral change is a program activity separate from the T-group which is aimed directly at facilitating a planning and action-practice model to help apply the participants' learning in back-home situations. At the present time, however, we are still left with the questions of whether or not negative changes result in positive changes later, what is the most effective way of producing behavior change, and what is the most effective way of insuring that changes within a laboratory will be translated to back-home settings? THEORETICAL JUSTIFICATION Much of the literature just reviewed focuses on change in an individual's attitudes, personality, or be- havior as a result of T-group experience. We have also examined research concerned with factors relating to change such as trainer effects, group composition, etc. But it appears that most of this research has proceeded from the tenous assumption that a group experience has the same effects on all participants. Or quite often, studies have been only concerned with the "average" effect on all parti- cipants (as in the case of organizational training labs that wish to increase the overall effectiveness of the organization). A number of investigators have questioned this assumption although, unfortunately, few studies report differential individual changes. Bunker (1965) states "there is strong evidence that groups, individuals, and even entire training programs have differential outcomes." He also stresses the idea that not only is there no standard outcome, but that no stereotyped ideal toward conformity exists within a group. Campbell and Dunnette (1968), in a review of effectiveness of T-groups, concur by stating 33 34 Such a generalized interpretation (T-groups have average effects on participants) may cover up important individual differences and training methods. Given a particular kind of outcome, certain kinds of people may benefit from T-group training while others may actually be harmed. The same reason may be applied to the interaction of differences in situational and organi- zational variables with the training experience. However, very few studies have investigated interactive effects. One way to view this problem is by examining the research designs that have previously been used. Basically, we find three standard research paradigms. The first and simplest design used to measure participant changes as a result of T-group experience is what may be termed the "Black-Box" method. Here a group is measured on some vari- able such as a personality, attitude, or behavior character- istic both before and after the laboratory. The basic assumption is that the laboratory is a type of "black-box" that everyone goes through and is affected by equally. The advantage to this method is obviously its simplicity, but it also carries several disadvantages. Among them are (1) problems of generalization because of lack of a control group, and (2) its inherent impreciseness both in terms of being able to pinpoint what variables caused changes and whether or not these changes were differential for the participants. The method is usually only used for pilot studies, when there are inadequate resources available, or when the investigator is not interested in generalizing his results to other populations. A definite improvement on this design is the use of a control group. 35 A second design is to manipulate some aspect of the group such as leader behavior, group size, exercises used, laboratory length, etc. and measure its effects on the participants according to some designated criterion vari- able such as personality, attitude, or behavior change. These results are usually in terms of pre-post difference scores and are then compared to scores of a control group in which the group variable was not manipulated. This method has definite advantages over the previous design in that it is possible to more accurately delineate what vari- ables were responsible for the change. It still has limitations, however, in terms of not being able to evaluate differential learnings of the individual participants. Quite often group averages are taken and can lead to erroneous conclusions. For example, if half the people change positively and half change negatively, the net change for the group is zero. This problem also arises when only unidirectional changes are measured (Harrison, 1971). This design is probably most widely employed, however, since it represents sufficient scientific rigor and is still within the fiscal and human resources of most .investigators. Also, as was stated before, it may be that the researcher is only interested in laboratory design and its ”average" effects on participants in which case this paradigm would be entirely appropriate. The same design has been used without a control group, but has obvious shortcomings. 36 The majority of research already reviewed has used this type of design. Included here would be those studies in which specific group variables were manipulated; length of laboratory experience (Bunker & Knowles, 1967; Bare & Mitchell, 1972), measurement of individuals at different times (Harrison, 1966; Shein & Bennis, 1965; Miles, 1960; Zand, Stelle, & Zalkind, 1967; Shutz & Allen, 1966), variations of trainer style (Greening & Coffey, 1966; Culbert, 1968; Cooper, 1969; Bolman, 1968), group composition (Powdermaker & Frank, 1953; Shutz, 1961; Harrison, 1965), and laboratory style (Byrd, 1967; Lieberman, Yalom, & Miles, 1973). The third type of research design is really a variation of the previous design, but with the notable addition of analysis of specific individual differences between laboratory participants. Design 2 is interested in correlations of variations in laboratory style and their effects on participants to predict what circumstances are most conducive to change the group as a whole. In contrast, Design 3 is interested in which individuals do best or worst under a certain condition and is more interested in -the differential effects of the lab on participants. This is not to say that Design 3 is not interested at all in laboratory style variables, but rather that its primary research focus is on the individual. Essentially we see two different methodologies for this assessment. 37 One is to divide the participants into categories according to some criterion variable such as scores on a personality test, ratings by independent observers, or situational variables like organizational status. Amount and type of change according to what the experimenter desires to examine are then measured for each of these categories. These measured changes need not necessarily be on the same variable as was the criteria for categorization of individuals, although that is frequently the case. Researchers using this type of methodology are usually interested in prediction of changes; namely, what charac- teristics of an individual are predictive of specific changes in that person. For example, a study might compare people who score high, medium, or low on a personality measure to see if they change differentially in behavior as a result of laboratory training. In fact, most research employing this methodology deals with personality factors as predictor variables (Stern, Stein, & Bloom, 1956; Vroom, 1960; Miles, 1960; Harrison & Lubin, 1965; Bennis, 1957; Stelle, 1968). Another method of analyzing individual changes is paccording to the relative changes made by the individual participants. This procedure takes a different approach, that is, a post hoc analysis to see who changed and in what degree. The investigator then goes back and attempts to discover what was correlated with or responsible for their change. Criteria used for categorization have most 38 commonly been amount of change exhibited on a specific variable. For example, individuals could be divided into groups of high, medium, and low changers or positive and negative changers. Their respective scores are then correlated with such variables as group process factors (group style, structure, etc.) or with other individual variables (scores on other personality tests, behavior in the group, relative changes on other measures). Examples of this type of research design can be found in studies by Argyris (1965) and Lieberman, Yalom, and Miles (1973). In an extremely thorough study, Argyris (1965) analyzed his data by categorizing individuals in terms of high medium, and low scores on a complex variable he called interpersonal competence. Lieberman, Yalom, and Miles (1973) also conducted an excellent study in which they used a composite score on a number of measures for each individual to categorize them into high, low, negative changers, and casualties. Both investigators, however, note the importance of assessing the learning environment in which these changes are made since a high changer in one laboratory may be a low or negative changer in another. Lieberman, Yalom, and Miles (1973) state "Differences among groups were more substantial both in number of people affected and in the type or area of change than were differences between those who participated in them and those who did not." HOpefully, in the future it will be possible to assess both the nature of laboratory experience 39 and the relative changes in individuals in order that laboratory training be made more effective in designing and implementing specific strategies to attain their goals, be they personal growth, organizational effectiveness, or whatever. The present study draws from the assumptions of this third type of research methodology. More specifically, it focuses on individuals and their differential changes in T-groups rather than on group structure variables as noted in the second type of design discussed. Essentially it is an attempt to show (1) that all individuals in a group do not change either in the same amount or direction, and (2) that an individual's initial position or how he sees him- self before the beginning of the group is predictive of both the amount and direction of change that he will exhibit during and after the laboratory. The first goal of this study seems well substantiated in the literature. Previously cited in this paper, research points fairly conclusively to the fact that individuals do change differentially, at least in regard to their per- sonality characteristics. It also seems reasonable to expect that individuals should be differentially affected by a laboratory experience. On one hand, varying labo- ratory structures and goals appeal to different types of people while it is also true that all individuals do not have the same prerequisite skills to learn from the laboratory environment. As Stelle (1968) states 40 This assumption (people learn best about behavioral concepts when they learn from immediate experience) seems to me to imply that all individuals tend to have an adequate ability to learn from the immediate data in the world around them, and I seriously question it. To me, the process of "learning from the data" requires a number of skills: to think thematically; to deal with the reality of multiple causation of behavior to use analogies to clarify a process; to make connections and to see correspondences which may be quite appro- priate but not one-to-one in their relationships to one another; and (tied in with the others) an ability to generate hypotheses and to understand the context in which the data occur. The second goal of this study is to evaluate the relationship between an individual's initial score on a variable (personality-type measure of interpersonal per- ception) and his subsequent change on that variable as a result of the laboratory experience. To the author's knowledge, this type of research has never been reported in the literature, although a number of investigators including statisticians have argued for the necessity of such a study (Harrison, 1971; Campbell & Dunnette, 1968; Cronbach & Furby, 1970; Tucker, Damarin, & Messick, 1966; Harris, 1962). There are, however, two types of studies which are closely related. These will be mentioned so as to present more clearly the different approach of this study. First are the correlational studies in which an initial score on some variable such as a personality characteristic is correlated with a postlab score on that same variable. Unfortunately, this does not allow for the investigator to compare, for example, the relative results of high initial scorers to low initial scorers. It only 41 allows him to make some statement about the general relationship between the two testing times for the group as a whole. He can make statements about the degree of changability of his sample from one test time to another by analyzing population variance, but this says nothing about the dimension and direction of specific individual's changes. The second type of research that may be found in the literature more closely approximates that of the present study. It utilizes some measure that when scored for an individual, yields differential categories. An example is a personality measure that classifies individuals into psychiatric types, i.e., obsessive-compulsive, paranoid, etc. Each categories' scores at pre and postlab are then correlated. This method gives the relationship of scores for each category, but has the same shortcomings as the previously mentioned research design, that of not being able to assess the relative relationship of an individual's initial score in a category with his subsequent change in the group on that variable. The present study departs from previous methodology by grouping individuals into high, middle, and low scorers based on their prelab, self-report on a personality instru- ment. It proceeds from the assumption that individuals do not change either in the same amount or in the same direction. It would therefore be inappropriate to analyze the data using correlational methods since the relationship 42 of an initial score of an individual to his subsequent scores is not linear. Proof for the nonelinearity of this relationship is meager, however. Although a number of investigators have hypothesized the possibility of differ- ential change of participants, few have sought to investigate the relationship of initial score to subsequent change. As a result, it remains largely theoretical speculation based on arm-chair observations of the T-group process. Let us examine some specific reasons to suspect a non-linear relationship. If, for example, an individual rates himself very high before the T-group on a desirable personality characteristic such as openness, it seems logical that it would be unlikely for him to improve since he already feels that he is well above the average. Con- versely, it may be that he holds a somewhat unrealistic view of himself and through the feedback processes of the group comes to see himself more as he really is and thus, rate himself lower at the end of the group. Clearly then it is easier for him to decrease or perhaps stay the same while an increase is seen to be much less probable. The obverse reasoning would say that those who rate themselves very low at the beginning of the group are more likely to increase than decrease further in their ratings of them- selves. The middle group is a different case. They probably hold fairly realistic views of themselves already, for instance, not rating themselves so high as to have no room for improvement, but also not so low that they feel 43 that they cannot improve. In their case, we would predict that they would report at least modest improvement if for no other reason than past studies tend to find that indi- viduals always report gains at the end of the laboratory, especially if they paid for the experience. As stated before, subjects were divided into groups of high, middle, and low scorers based on prelab, self- report data. This was done on the reasoning that such groupings should have some predictive value about subsequent changes. Although, data about the subjects was also collected from Intimates and Colleagues, this is not customary and to categorize the subjects accordingly would decrease the probability of replication. On the other hand, self-report data is usually readily available and would be extremely valuable if it could be shown that it is predictive of subsequent changes of participants in the laboratory. Data from Intimates and Colleagues was used, however, to provide some measure of validity or support for the accuracy of self—reports. METHOD The data for this study was drawn from the research done by Elizabeth J. Force for her Ph.D. dissertation. She collected data from 50 participants in an eight day training lab designed to increase their interpersonal competence. Particular emphasis was placed on communication skills and within that context, a greater awareness and understanding of self and others. Although an increase in job effective- ness was also a goal for the lab, it was not a primary focus. It was, however, measured under the assumption that as an individual becomes more facile in communication, he will in turn be better able both to establish more satisfying interpersonal relationships and to increase his personal effectiveness with work colleagues leading to greater job effectiveness. Subjects Subjects for this study were all participants in an eight day training lab conducted by Michigan Training Laboratories, Inc. from August 17 through August 25, 1968. The fee for participants was $200 tuition plus room and board costs. The fifty participants, thirty-three males 44 45 and seventeen females, came from varied occupations including five priests or pastors, two professors, one curriculum consultant, fourteen high and junior high school teachers, four principals, two school superintendents, one art coordinator, four counselors, two caseworkers, one psychiatrist, eleven psychology or social work students, two housewives, and one director of marketing. The subject sample also included a group of one hundred significant others, half of which were "Intimates" and half were "Colleagues" of the fifty participants in the laboratory. They were chosen so that whenever possible they knew the participants at least one year prior to the labo- ratory and would know them for an additional year after- wards. These significant others were chosen by the parti- cipants. Intimates were defined as family or very close friends while Colleagues were to be people who worked with and knew the participant well. Instruments Instruments for the original study were chosen for their predictive ability to reflect changes in interpersonal communication and perception as well as changes in intra- personal awareness and understanding. The instrument chosen for this study was the Person Description Instrument- form X (PDIX) because (1) it seemed to tap change variables that were thought to be congruent with the objectives of the laboratory, and (2) it was administered to the 46 participants at all four testing times; it was also given to Colleagues and Intimates at pre and postlab and to other members of the group at both intra-lab testing times. This instrument was developed by Harrison (1962, 1965, 1966) to study interpersonal perception. The scale is of semantic differential design and consists of three factorially independent subscales: Interpersonal Warmth and Acceptance (IWA), Power and Effectiveness in Work (PEW), and Activity and Expressiveness (AE). Each subscale con- sists of nine bipolar items separated by six numbers with dots between to permit decimal scoring (see Appendix A). Harrison has previously used this instrument and reports satisfactory test-retest reliability. Force (1969) also reported generally good stability for this measure with test-retest correlations for self-reports significant for all subscales across the four observation times. In general, Harrison assumes face validity for these subscales. He has, however, conducted a number of studies and found some evidence for external validity. In one study he found that organizational managers attending a training laboratory increased in self-reported interper- sonal skills (Harrison, 1962). In another study he found that person-oriented as opposed to work-oriented partici- pants were more expressive, warm, comfortable, and formed stronger interpersonal ties within the laboratory (Harrison & Lubin, 1965). And finally, he used the PDIX and found that participants increased in the number of 47 inferential-abstract concepts that they used after a laboratory (Harrison, 1966). Design The original study was constructed to obviate some of the previously mentioned problems with T-group research; in particular those problems dealing with bias of self? report data, temporality of changes, and dimensions of individual changes. Data was collected at five weeks pre- lab, on the second and seventh day of the lab, and six months postlab. It was assumed that this longitudinal approach to data collection would show the nature of changes over time (temporality). Bias in self-report data were somewhat controlled for by asking Intimates and Colleagues and other group members to provide ancillary data so as to substantiate self-reported claims. The present study is primarily of a statistical nature and was designed to explore the issue of dimensions of individual change by analyzing reSpective changes of high, middle, and low scorers based on prelab, self-scores. A packet containing ten personality-type measures (including PDIX) was sent out to all participants five weeks before the laboratory. Two identical packets were included with the instructions that they should be given to one Intimate and one Colleague. All data packets were then to be returned directly to the researcher, data not being available to T-group trainers or any other member of the 48 participant-Colleague—Intimate triad. The same procedure was followed for data collected at six months postlab. Within the lab, data was collected on the second and seventh day. It was assumed that, consistant with prior research, data collection at these times would reduce problems with initial day anxiety and spuriously high end of lab reports. Along with self-reports at these times, other group members were asked to rate the participants in their group on the same measures. The instructions for the PDIX, however, were to rate only the two members that you liked least and the two members you liked most. This was done for ease of administration and so as not to make data collection such a noxious event to the participants. Post-hoc analysis reveals that all but three subjects received at least two scores from other members of the group and it was thus assumed to be a valid indication of the group's perception. Groups consisted of ten members each, balanced for male-female ratio and minimizing prior acquaintances. Each group had two trainers which consisted of a senior, more experienced facilitator and a usually less eXperienced junior facilitator. Senior facilitators were all Ph.Ds. While junior facilitators had at least the equivalent of a master's degree. RESULTS Data A complete inventory of the original data may be found in the appendix of Force's (1969) dissertation. Data returns from all sources were high. Returns for prelab data were received from 48 of the participants (96%), 48 from Intimates (96%), and 46 from Colleagues (92%). Postlab data was received from 48 participants (96%), 41 Intimates (82%), and 38 Colleagues (76%). Since all participants were required to fill out instruments within the lab, almost complete data from this time period was received. Misread instructions or missing scores for one subject yielded incomplete self-report data on day 7 of the lab. As mentioned before, three subjects did not receive the minimum of two PDIX ratings by fellow T-group participants and thus were excluded from the group report data. Force examined those instances of missing data and concluded that nothing particularly characterized the other available data on those subjects. There were no trends for these participants to view themselves as significantly increasing or decreasing on any of the measures. She did note, however, that these participants had somewhat lower 49 50 prelab scores than did other lab members. In the present study, missing data was replaced using methods outlined in Winer (1962). More specifically, the row mean (S's scores across observations) and column mean (all scores for a certain observation time) were averaged to generate the missing score. Constitution of High, Middle, and Low Subgroups Theoretical justification for dividing the parti- cipants into three subgroups (high, middle, low) according to their initial self-report score has already been described. Individuals were divided into these subgroups according to self-reported, prelab scores. This was done separately for each of the three PDIX scales; Interpersonal Warmth and Acceptance (Scale IWA), Power and Effectiveness in Work (Scale PEW), and Activity and Expressiveness (Scale A&E). Thus, a person classified as a low scorer for one scale may have been a high scorer for another. An analysis of each individual's relative position on each scale revealed no Specific trends in the data. In other words, a person who was in a certain subgroup on one scale was not necessarily in the same subgroup on another scale. Three criteria were used to establish subgroups: (l) by dividing the range of scores on each scale into three equal parts; (2) by dividing the subjects into three equal size groups; and (3) by analyzing a scatterplot of all the scores on a specific scale and looking for breaks in the distribution of scores. Method #3 also included attention to differences in range between means of the three subgroups with an attempt at keeping the range between the low and middle groups' mean equal to the range between the middle and high groups' mean. A combination of these three methods was used for each scale and in most cases the distributions met the criteria for at least two of the three methods. In all cases though, particular emphasis was placed on method #3 since it was felt that breaks in the distribution constituted more evidence of high, middle, and low subjects than did the more arbitrary classifications based on sample size or range of scores. The resulting high, middle, and low groups on the three PDIX scales were of unequal membership (Scale IWA: 12 low, 24 middle, 12 high; Scale PEW: 10 low, 24 middle, 14 high; and Scale A&E: 18 low, 19 middle, 11 high). Because of missing and/or inadequate data for two persons, these categorizations were based on the 48 remaining participants. Measurement of Change A two—way analysis of variance was the primary statistical tool. A 3X4 factorial design, three subgroups at four observations, was used for self-reports while a 3X2 factorial design was used for Group, Intimate, and Colleague reports since they came from only two observa- tions (Group: Days 2 & 7; Intimate and Colleagues: prelab and postlab). A separate ANOVA was conducted on each 52 subgroup (high, middle, low) for each data source (Self, Group, Intimate, Colleague). Tables 2, 3, and 4 show the results of these analyses. Also of interest were the means for each subgroup, given in Table 5. Examination of the graph based on these means across the four observation times (Figures 2, 3, 4) indicated some further interesting trends and a Tukey's post-hoc analysis (Winer, 1962) was performed on the mean differences of self-scores at pre- post, day 2-day 7, pre-day 2, and day 7-post intervals. The results of this are shown in Table 6. Table 1 presents all subgroups' means for each scale across the four observation times. Total means are also presented graphically in Figure l. Consistant with the goals of this study, tables and figures are as fully as practical presented to diSplay comparisons between the three sub- groups. It should also be noted that comparisons with data analysis conducted on the total group will be made to show the difference in information obtained when the group is subdivided. In most instances these data were abstracted from Force's dissertation. Main Effects of Time of Observation Means for the total group across all observation times are presented graphically in Figure l and numerically in Table l. Efratio values from Tables 2, 3, and 4 (ANOVA for subscales IWA, PEW, and A&E) indicate signifi- cant mean differences for self-scores on all PDIX scales 53 (EIWA = 3.78, p<.025; EPEW = 4.90, p<.01; and £A&E = 11.08, p<.01--all dffs = 3/132). Figure 1 shows a drop on all scales from prelab to day 2 of the lab. From day 2 to day 7, there is a moderate increase which gradually rises to above their respective initial scores at six months postlab. Since §_tests were only significant for differences between means among the four series of observations, further analysis using Eftest values obtained by Force showed that the significant mean differences probably existed between prelab and postlab data. vaalues for scales IWA, PEW, and A&E were E_= 1.58, p<.10; £_= 2.22, p<.01; and 5‘: 4.66, p<.001. Group scores also appear to have similar significant increases from day 2 to day 7 on Scale IWA (2': 13.5, p<.01) and Scale PEW (§.= 12.5, p<.01)--both with g£,= 1/39. Scale A&E, however, did not significantly increase although the graph shows a similar trend for group members' scores to go up from day 2 to day 7. Intimate and Colleague scores for all scales showed no significant differences from prelab to postlab testing times on any scale. This is shown graphically where the mean differences do not exceed 1.5. We can conclude that the laboratory was effective in changing the participants' mean self-ratings on all scales over the four observations. The same conclusion can be drawn for Groups' ratings of participants except for the Activity and Expressiveness scale which showed 54 3e)? rou In‘fi' 1c colleague . ........ F“ Figure 1. Total Grou p Mean Scores on Scales TWA, and AGE mew, 55 Table 1 Means for all observations across sources of data Observations: Prelab Day 2 Day 7 Postlab Scale: IWA PEW AGE IWA PEW AGE IWA PEW AGE IWA PEW AaE L 30.8 33.6 30.5 33.0 37.2 31.6 35.9 37.1 31.8 37.3 38.7 34.8 M 38.4 39.4 37.6 37.1 38.6 34.9 38.1 39.3 35.6 38.6 41.4 40.4 Self H 44.3 44.6 44.1 38.4 38.8 37.8 39.0 40.3 39.1 41.5 42.7 44.1 Total i 37.8 39.2 37.4 36.2 38.2 34.7 37.7 38.9 35.5 39.1 40.9 39.7 L 31.6 30.6 29.5 31.6 34.0 30.2 M 33.4 38.4 31.1 31.0 40.0 32.4 Group H 35.0 35.4 35.2 38.1 37.9 37.4 Total i 33.3 34.8 31.9 35.2 37.3 33.3 L 36.8 40.8 34.0 38.5 41.9 36.5 M 39.4 42.2 37.7 38.4 44.3 40.4 Intimate H 43.9 44.5 43.4 41.8 43.6 40.8 Total i 40.0 42.5 38.4 39.6 43.3 39.2 L 37.3 41.9 38.1 40.9 43.8 36.6 M 39.3 42.2 39.9 38.9 42.8 39.7 Colleague H 42.2 45.6 43.2 41.4 44.7 41.5 36:81 i 39.6 43.2 40.4 40.4 43.8 39.3 56 Table 2 ANOVA for Scale IWA Source df MS F a Level of initial score 2 518.928 13.811 <.001 Error 45 3 g Test times 3 58.904 3.778 ’.025 Levels x tests 6 72.225 4.632 4.005 Error 132 Level of initial score 2 75.163 1.748 NS Error 45 2‘ 2 Test times 1 187.88 13.518 <.01 0 Levels x tests 2 1.531 .11 NS Error 39 Level of initial score 2 122.918 2.497 4.05 0 Error 45 ‘5 g Test times 1 16.918 .546 NS c *' Levels x tests 2 14.413 .465 NS Error 35 Level of initial score 2 1526.776 30.351 <.001 3 Error 45 S‘ .93. Test times 1 .650 .046 NS H o L) Levels x tests 2 43.878 3.099 NS Error 29 57 Table 3 ANOVA for Scale PEW Source df MS F a Level of initial score 2 292.022 6.444 <.Ol Error 45 ",3 g Test times 3 76.437 4.897 <.Ol Levels x tests 6 50.67 3.247 <.01 Error 132 Level of initial score 2 331,939 12,079 <.01 3' Error 45 g Test times 1 122.62 12.50 <.01 Levels x tests 2 6.05 .62 NS Error 39 Level of initial score 2 45.821 1.106 NS g Error 45 IE Test times 1 17.326 .854 NS H Levels x tests 2 18.636 .918 NS Error 35 Level of initial score 2 50.259 1.361 NS g Error 45 .78 lg Test times 1 4.167 3.692 NS (J Levels x tests 2 19.725 <.05 Error 29 58 Table 4 ANOVA for Scale A&E Source df MS F (1 Level of initial score 2 1140.543 26.031 ’.01 Error 45 3: 8 Test times 3 232.240 11.076 /.01 Levels x tests 6 36.905 1.76 NS Error 132 Level of initial score 2 272.796 4.289 /.05 Error 45 S“ 8 Test times 1 30.488 1.453 NS 0 Levels x tests 2 5.527 .263 NS Error 29 Level of initial score 2 328.121 4.009 /.05 3 Error 45 ‘8’ ‘3 Test times 1 26.67 1.358 NS c H Levels x tests 2 49.399 2.515 NS Error 35 Level of initial score 2 211.915 5.497 <.01 m g. Error 45 m 3 Test times 1 32.202 2.251 NS 0 0 Levels x tests 2 18.661 1.33 NS Error 29 60 calc PEW (Power and ' Scores on S Subgroups in Work) Figure 3. Effectiveness 61 Figure 4. Subgroups' Scores on Scale AaE (Activity and Expressiveness) 61 —-—-- --- -——----— ---— --_-- :h“— P54 lab d ' '_ty an . Act1v3 AGE ( ale ' Scores on SC ups Subgro 4. Figure Expressiveness} 62 non-significant within lab differences. Further analysis indicated that the significant self-rating changes were largely a result of differences between prelab verus postlab scores . Interaction of Observations With Subgroups As mentioned before, the subjects were divided into high, middle, and low subgroupings based on prelab, self- scores. The major hypothesis of this study was that subse- quent change over the four observations was dependent on these subgroupings. The relevant findings are graphically represented in Figures 2, 3, and 4, while their respective ANOVA results are given in Tables 2, 3, and 4. The overall interaction effect of observations and levels (high, middle, and low) will be reviewed before attending to the changes of these subgroups. The original hypothesis seems supported for self- reports on Scales IWA and PEW (E's = 4.63, p<.005 & 3.25, p<.01--b0th df's = 6/132). It also appears supported for Colleague's scores on Scale PEW (F = 19.3, p<.05, LEE = 2/29]) and approached significance on Scale IWA. There was no significant interaction between observations and level on Scale A&E for either of these groups. This was also true for both Intimates and Colleagues on all scales. The most significant conclusion to be drawn from this data is that a participant's prelab score importantly affects his subsequent self-scores, at least on Scale IWA and PEW. 63 Table 5 F—ratios for mean differences across levels Level: Low Middle High Self b IWA 6.88a (3/33) .93 (3/69) 4.20 (3/33) (’1 PEw 3.80b (3/27) .35 (3/69) 5'34 (3/39) b A&E 6.37a (3/51) .048 (3/54) 3'77 (3/30) Group b b b IWA 4.88 (1/11) .15 (1/23) 5.63 (1/11) paw 3.85 (1/9) .26b (1/23) 5.47b (1/13) A&E .09 (1/17) .527 (1/18) 3.17 (1/10) Intimate IWA .08 (1/11) .92 (1/23) .33 (1/11) pew .01 (1/9) .86b (1/23) .40 (1/13) A&E 4.36 (1/17) .04 (1/18) 1.99 (1/10) Colleague IWA 5.07b (1/11) .04 (1/23) .39 (1/4) PEw 7.87b (1/9) .96 (1/23) 2.22 (1/13) A&E 11.76a (1/17) .07 (1/18) 10.44a (1/10) Note: _E_‘_-Ratio @) Significance Levels: 3 C’ II p< p<.01 .05 64 This also held for Colleague's ratings of participants. Thus, a more detailed analysis of levels seems justified and warranted. Effect of Levels Initially we shall begin with a global approach to analysis of levels, that is, to discuss differences among the means of the high, middle, and low levels independent of observation time. Tables 2, 3, and 4 show that almost all F—ratios for level comparisons were statistically significant. Exceptions include Group scores on Scale IWA and the scores of both Intimates and Colleagues on Scale PEW. The import of this finding is attenuated, however, when we consider that the subgroups were classified according to initial self-score and thus significant differences between levels are not surprising, at least for self- reports. This bias was not inherent in the Group, Inti- mate, and Colleague data, however, since their levels were not dependent on prelab scores, but rather were determined by a matching procedure in which their scores were paired with their respective participants who were already sub- grouped. For instance, a Colleague would be placed in the low group if the participant he was rating had low prelab self-scores. Specific patterns of change for the subgrouped levels is the next issue. Table 5 compiles all Efratios for mean differences across levels for each subscale. The 65 relevant data are depicted in Figures 2, 3, and 4. We shall consider the Group, Intimate, and Colleague data and then proceed to a more detailed analysis of self-report data. The most notable trend in Group data is a signifi- cant increase from day 2 to day 7 (the only two observa- tions) on Scales IWA and PEW. This held for all subgroups (high, middle, low) except that the low group only approached significance on Scale PEW. No significant increase for any subgroup was found on Scale A&E. Inti- mates showed no statistically significant mean differences from day 2 to day 7 or any scale for any subgroup, except for the middle group on Scale PEW. In contrast, Colleagues exhibited a very clear trend toward significantly higher ratings of the low group on each scale (5| = 5.07, p<.05 IWA [93 = 1/11]; 311, = 7.88, p<.05 (g; = 1/9];_1=;ME = 11.76, EW p<.01 [fig = l/l7]). This is consistant with the original hypothesis that low groups have the greatest probability of changing positively over time. That neither high or middle groups showed any significant changes further supports the hypothesis that change is more difficult for them. The self-report data yield the most dramatic evidence for participant change. Since the Erratios are for across all four observations, a Tukey's post-hoc analysis for mean-difference comparisons (Winer, 1962) was conducted to identify the locale of significant mean differences. The results are shown in Table 6. Middle groups, as predicted, showed no significant changes over 66 Table 6 Tukey's post-hoe analysis of mean differences for self-scores Levels: Low Middle High IWA 6.5a .40 2.8 Pre-post PEW 5.0a 2 0 1.6 a A&E 4.7 2 9 0.0 IWA 2.9 .8 .6 Day 2-Day7 PEW ~.1 .7 1.5 A&E .1 .6 1.3 a IWA 2.2 -1.2 -5.9 Pre-day 2 PEW 3.5b —.8 -5.8al MB 1.1 -2.7 -6.2b IWA 1.5 .5 2.5 Day 7-post PEW 1.6 2.1 2.7 MB 3.2b 5.0a 4.981 Note: negative values indicate a decrease from first to second observation Significance level: a p <.01 b = p 5.05 67 the four testing times except on Scale A&E, Activity and Expressiveness ( = 5.04, p<.01 [df = 3/54]. Table 6 a... shows that this difference seems most attributable to a significant increase from day 7 to postlab. Graphically represented, they did, however, show the same trend for change as the total group; that is, a decrease from prelab to day 2 of the lab, a slight increase during the lab and a gradual increase from day 7 to six months postlab to a point slightly higher than their initial prelab score. Thus the middle group probably contributed much to the total change pattern or at least did not significantly alter it over the four observation times. A significant change over observations by the low groups on all scales is shown in Table 5. Further analyses, shown in Table 6, indicate this change is probably due to an increase between prelab and postlab testing times. All Erratios were significant at p<.01. Figures 2, 3, and 4 shows this relationship as increasing from prelab scores to each of the subsequent testing times. There were two other significant increases (Table 6) for the low group on Scale PEW (pre-day 2) and Scale A&E (day 7-post). The within-lab increases for low groups, shown by these graphs were not statistically significant, perhaps because of the conservative nature of Tukey's post-hoc analysis. This was also true for the middle and high groups. The high group also showed significant mean differ- ences over time on all scales (F __IWA = 4.20, p<.05 [513 = 68 3/33]; 2P = 5.34, p<. [§£_= 3/39]; and F = 3.77, p< EW -A&E .05 [fig = 3/30]. The Table 6 data suggests that this was probably due to a decrease, in contrast to the low group, from prelab to day 2. This is depicted in Figures 2, 3, and 4. These decreases are in the same direction as both middle groups' scores and the total group's scores as shown in Figure l, but are of a much greater magnitude. Perhaps the total group's scores were significantly affected by this loss from prelab to day 2 by the high group. Again it is important to note that the low group did not show this decrease from prelab to day 2, but contrarily showed sub- stantial increases especially on Scale PEW. Some other interesting data is noted in Table 6. All subgroups (high, middle, low) showed significant increases on Scale AsE (Activity and Expressiveness) from day 7 to postlab. This seems to support a longitudinal theory of change, or more specifically that individuals changed within the lab, but that these changes have to "incubate" and may only be fully evidenced at some time after the laboratory experience. Furthermore, the middle group, consistant with the original hypothesis, did not signifi- cantly change on any scale aside from the previously cited day 7 to postlab increase on Scale A&E. Finally, comparisons are drawn between the data for the total group and those of the low, middle, and high subgroups. The clearest finding is the significant variations in the self-reported changes made over time 69 between the three subgroups on all scales, as depicted in Figures 2, 3, and 4. The high group shows a marked decrease from prelab to day 2, a slight increase within the lab, and a gradual increase postlab to a level somewhat below their initial score. The medium groups have much the same pattern except that the initial decrease is of less magnitude and the day 7 to postlab data indicate an increase to somewhat above their initial score. This pattern is congruent with the total group pattern as seen in Figure 1. The low group, however, shows a distinctly different pattern, gaining on all scales at every observation point. Parti- cularly notable is the absence of a prelab to day 2 decre- ment in the low group. This seems to indicate that the total group pattern, at least between prelab and day 2, was more affected by high and middle groups' scores than by the low groups' scores. All scales show this same pattern for the three subgroups except that Scale A&E shows greater postlab gains for all participants. Differences between high, middle, and low group data and total group data are confirmed by the Colleague's reports. Again, the low group significantly increased on all scales while both middle and high groups show no statistically significant changes. Colleague data also revealed that there was a significant pre-post increase on Scale A&E for the total group while this was only supported for the low and high subgroups in this study. 70 Data from Intimates showed no significant changes on any scale by any subgroup. This was consistant with total group data. Another consistant finding was that there were no differences between total group and subgroup data for Groups. Both reported increases on Scales IWA and PEW with no change on Scale A&E. DISCUSSION The original hypothesis, that participants are differentially influenced by T-groups, appears well supported. The findings suggest that the laboratory was effective in changing participants on all three PDIX variables: Inter- personal Warmth and Acceptance, Power and Effectiveness in Work, and Activity and Expressiveness. However, a more complex pattern of change is clear when the data were analyzed in terms of high, middle, and low subgroups. There was obvious subgroup differences in the amount and direction of change. These changes provide insights into the ability of T-groups to differentially influence individuals. Major discrepancies between the findings of the total group versus the several subgroups will be examined. Within that context, a more detailed scruitney will be made of changes within the subgroups. Self-Report Ratings Self-ratings on the PDIX scales provide the strongest evidence of differential participant changes. The overall trend over observations for the total group's self-ratings (Figure 1, page 54) is a decrease from prelab 71 72 to day 2 of the lab, a moderate increase within the lab, and a gradual increase postlab to a point above the prelab score. This pattern held for all scales, although Scale A&E (Activity and Expressiveness) shows the most dramatic post- lab increment. This pattern might be "explained" by suggesting that participants undergo a type of "culture shock" when they enter the lab. More specifically, partici- pants have no secure relationships to hold on to as they had back home while at the same time are asked to let down their defenses. As a result, their self-perceptions were revised more negatively at day 2. This might be attributed to a desire to appear favorably before the lab, perhaps thinking this might enhance admission which may have inflated their self-descriptions. During the period of the laboratory they reevaluated themselves mindful of feedback from other group participants and also with awareness of their impact on others. This probably has the effect of modifying their self-perceptions toward a more "realistic" level, or at least to a point more congruent with how others see them. Within the lab, there were moderate increases on all scales, although none achieved statistical significance. These increases were only moderate probably because the participants' self-perceptions may have been in a state of flux and clear decisions to change were still in the process of being made. The five month postlab period shows parti— cipants increasing on all scales to a point just above their initial prelab scores. This postlab period might 73 then be viewed as an "incubation" period in which perhaps participants try out new behaviors that were discovered and seen only as possibilities within the lab. This pattern argues not only for the utility of laboratories to produce change in individuals, but also for the necessity of a longitudinal research design to adequately measure change. This overall pattern of change does not consistently hold, however, across the high, middle, and low subgroups. The sharpest deviation is apparent in the low group which made increments at each subsequent testing time on all scales, except for Scale PEW where the within-lab change was only -.1. This finding supports the hypothesized view that the low group has a greater probability of increasing than decreasing or remaining the same. Data from Table 6 (page 66) also shows that only the low groups changed significantly from prelab to postlab. This finding suggests that such laboratories may have greater utility for individuals who initially rate themselves rather negatively in comparison to the rest of the group. The author postulates two possible reasons for this pattern. First, the low group could perhaps be rating themselves low at prelab as a kind of "insurq‘be" that they would show positive changes during and after the labo- ratory. However, there are several strong arguments against this position. Both Intimates and Colleagues, although providing ratings higher than participants' self- ratings, rated participants in the same relative position 74 to the rest of the group as they do themselves; that is, lower than either the high or middle groups on all scales. Additionally, within-lab ratings by fellow group partici- pants confirmed this lower position. Thus, there appears some external validity to the subgroup's lower self-ratings. It is also unlikely that participants would plan this five weeks in advance of the lab. Any premedfiation would seem more likely to be in the direction of wanting to appear more favorable in order to be admitted to the lab. The second hypothesis to account for these changes seems both more plausible and is consistent with the study's original theoretical justification. It is that these participants may have held somewhat unrealistic negative views of them- selves before the group and after exposure to the group and feedback from its members, began to alter their self- perceptions positively toward a position more congruent with others' view of them. This positive self-perception continued postlab as shown by the statistically significant increments. The medium group evidenced a similar readjustment process from prelab to day 2, although it is manifested as a slight decrement. It is assumed that these individuals are relatively well-adjusted in that they rated themselves neither unreasonably high or low at prelab and in a manner congruent with how others rated them. As is typically the case, they tend to rate themselves somewhat higher than others rate them and subsequently readjust their 75 self-ratings slightly downward when they enter the lab, although this alteration was not statistically significant. Like the low and high groups, they make only moderate and statistically nonsignificant within-lab changes. Their postlab gains were also not statistically significant although they evidenced a general increase in self-ratings on Scales IWA and PEW (Interpersonal Warmth and Acceptance and Power and Effectiveness in Work). Only on Scale A&E (Activity and Expressiveness) did they show significant postlab gains. Perhaps Scale A&E is most congruent with the laboratory goals or at least what participants would tend to see as the more salient of lab goals. Quite often laboratories place a high priority on verbal activity and thus it is logical for participants, if they wanted to have accomplished something within the lab, to view themselves after the lab as having made gains on this scale. Even though their increment from day 7 to postlab was signifi- cant, it only brought the middle group slightly higher, but not significantly above their prelab score. Thus, although these middle groups made moderate prelab to postlab gains, generally the lab experience left them quite close to their original position. The high groups showed much the same pattern of changes as did the middle group with the exception of a greater loss from prelab to day 2 and a slightly greater increase from day 7 to postlab. Their prelab to day 2 loss was statistically significant on all scales. Again, no 76 significant changes occurred within-lab and the only signi- ficant postlab change was on Scale A&E, probably for the reasons previously cited. Most importantly, there were g2 statistically significant changes from prelab to postlab despite the wide fluctuation of their scores between these times. Are we to conclude that it was difficult for the high group to change upwardly? A closer look at the data reveals two possible explanations. First, assuming the same process of change with high groups as for low groups, it might be hypothesized that this group held unrealisti- cally high perceptions of themselves before the lab. It; could then be argued that upon entering the laboratory they shifted to a position more congruent with how others saw them. Like other subgroups, they made moderate increases within the lab and more gradual postlab increases to a point somewhat below their prelab scores. Perhaps they did not surpass their initial ratings because they began to see themselves more realistically. However, closer analysis suggests a second and perhaps more plausible explanation. The high groups differed from both the middle and low groups in their statistically significant decrement from prelab to day 2. This dramatic decrease in self-ratings on all scales seems to indicate not an actual readjustment in the face of external feedback, but rather an attempt to rate themselves as they thought others would rate them. This initial drop is so dramatic as to be suspect since not only 77 do none of the other groups show changes of such magnitude in that period of time, but it is difficult to believe that a group of individuals could alter their self-ratings to this degree after such a short exposure to group experi- ence. It may be speculated from this line of reasoning and a look at the high groups' subsequent return to a point not significantly different from their prelab rating, that the high groups were playing a sort of "game" in which it seems they had very little intention of changing. Instead, they fill out their instruments within the lab according to how they think others will see them, but outside the lab remain at much the same level. They do, however, show a slight downward trend in their ratings from prelab to postlab which may indicate that they see themselves a bit more congruently with others' perceptions of them, but these differences were not statistically significant. A Surprise Findigg Further speculation leads us to an unexpected, but important finding. Perhaps this same "moderating" process was functioning for all subgroups. It would point then to a conclusion that the laboratory had the effect of being a "moderator" variable. In other words, participants came into the lab from their respective high, middle, or low subgroups and rated themselves not as they had previously seen themselves, but more as they thought others would see them. The subgroups were still significantly different 78 from each other within lab, but it is hypothesized that the lab only moderates their scores more in the direction they assumed was expected rather than making everyone exactly the same. That all groups on all scales showed the same pattern of subsequent changes adds credence to this theory in that group members may be rating themselves according to what they feel is expected. This "moderating" theory has implications for T-group research. Perhaps laboratories are not places where participants are encouraged to see themselves more realistically, but rather are places where normative structure tends to "knock off the corners." Participants come to describe themselves less extremely, regardless of whether this extremity is positive or negative. An example of this can be found in the norms of a group which gives more attention to its low participants. Resultingly, the middle and high groups show very little overall change while the low groups make significant gains. Such differential learning could also be aided by a trainer who believes that everyone in the group should attain a certain level of functioning. He may then give more of his attention to the participants whom he feels need it most and in turn, set up a normative structure within the group for other members to do the same. Again, it should be mentioned that this is only speculation and that there is no direct supporting evidence that this occurred within this specific laboratory. 79 In summary, self-scores show different patterns of change for the total group than for the high, middle, and low subgroups. The total group exhibits a pattern of change on all scales in which there is a slight drop from prelab to day 2, a moderate within-lab increment, and a more gradual postlab increase to a point slightly above their initial score. Different change patterns are evident, however, within the high, middle and low subgroups. The most notable deviation is the low group which increases at all subsequent testing times. They are also the only group to exhibit a significant increase from prelab to postlab. The middle group almost exactly duplicates the total group's overall pattern of change. The high group, although having rather dramatic fluctuations, also somewhat duplicates the total group pattern. It is important to note that neither the high or middle group has statistically significant increments from prelab to postlab. A very interesting pattern of changes which bears directly on T-group methodology is evident when different time periods are considered. If we looked only at the change pattern from the beginning to the end of the lab, there would be moderate non-significant increases. When the postlab observations are added, significant increases are noted on all scales for the total group as well as for the high, middle, and low subgroups. The further addition of the prelab observations creates a totally different impression of change. Viewing the total group, we see in 80 general only slight non—significant prelab to postlab changes. But if we look at the high, middle, and low groups we see that only the low groups made significant prelab to postlab changes, the middle groups made non- significant changes throughout the four observation times, and the high groups made non-significant changes, but fluctuated a great deal. These patterns show that partici— pants do make differential changes and thus the original hypothesis appears to be well substantiated. These findings underline the necessity for longi- tudinal research with multiple measurement times. They also question the utility of the laboratory as a change agent or equal change agent for all participants. In this present study, at least, apparently only the low subgroups benefitted appreciably from the laboratory experience. Ratings by Others The preceeding findings were based solely on self- scores and, therefore, are subject to bias. For purposes of external validation, let us examine ratings made by other lab participants, Intimates, and Colleagues. As mentioned before, the within-lab self-reported changes for all subgroups on all scales showed increases, but were not significant. The group reports also showed increases, but on Scales IWA and PEW these changes were significant. There was also no differentiation between high, middle, and low subgroups, although their relative position to each 81 other is generally the same as was for self-scores. It does seem notable however, that all group scores were lower than self-reports. This seems to indicate that the group sees participants generally less well off then the partici- pants see themselves, but also rates them as having made more gains within the lab. As mentioned before, persons generally rate themselves more favorably than do others and thus these lower group ratings are logical. Group members, on the other hand, may attribute more change to participants due to a priori expectations. It is assumed that group members see more change in others because they hope for more change in themselves which could also explain the findings that their ratings did not discriminate between subgroups. It should also be noted that these results may have been in part due to the within-lab instructions to rate only the two highest and two lowest members of the group. In contrast, Intimates and Colleagues generally rated participants higher than participants rated them- selves. This might be explained by Intimates often being spouses while Colleagues were probably chosen by partici- pants with the eXpectation that they would provide high ratings. Again, there still exists the same relative relationship in ratings with high ratings by Intimates coinciding with high self-ratings, middle with middle, etc. This seems to support some external consensus as to a person's general position. However, few significant findings were observed for prelab versus postlab data, as 82 no statistically significant shifts were reported by Intimates for any of the subgroups. It was speculated that Intimates were probably not very discriminating, at least on the PDIX scales. In most cases they were spouses and probably tended to see participants in much the same manner as they had previously viewed them, regardless of any actual behavioral changes. In contrast, many Intimates reported quite positive behavioral changes in participants when asked to write general impressions. This may indicate that the PDIX was not sensitive to behavioral changes, but attended more to shifts in attitudes. The low and middle group Intimates did, however, tend to rate participants somewhat higher at postlab than prelab, although it was not statistically significant. Unlike Intimates, Colleagues appeared to have been somewhat more objective and we find that they report significant increases for the low groups on all scales. This strongly substantiates the self- reported prelab to postlab gains made by the low groups. Another notable finding is the changes made on Scale A&E (Activity and Expressiveness). As mentioned before, this scale is probably most congruent with laboratory goals or at least what participants may interpret as lab goals. It is assumed that since there is such a great importance placed on verbal activity, participants are likely to report increases on this scale. Looking at the data we find the most dramatic increases for self-reports on this scale (all subgroups reports significant changes). A closer look 83 reveals that these changes are greatest from day 7 to post- lab. Colleagues also report significant changes for high and low groups on this scale. The author hypothesizes that participants probably learn within the lab that what this scale taps are the most desirable changes. It is also true that the laboratory probably focuses on these things more than others and as a result, we would expect the most dramatic changes on this scale. Implications and Future Research These findings appear to indicate that the labo- ratory eXperience has the ability to affect participants on scales of Interpersonal Warmth and Acceptance, Power and Effectiveness in Work, and Activity and Expressiveness. The original hypothesis, that subgroups are differentially affected, also seems well supported. There are, however, cautions to be considered when interpreting these results. First is the problem of controls. It is tenuous, without a control group, to make definitive statements that this laboratory was solely responsible for participant changes, although adequate control groups are admittedly difficult if not impossible to obtain. Further research might employ more prelab observations so as to establish a baseline of participant responses to the measurement instrument. It may also be advantageous to use within-lab observers in conjunction with observers at prelab and postlab as further external validity checks for self-report data. 84 Another problem relates specifically to the measurement instrument. Although the results point fairly conclusively to the laboratories' ability to affect changes in the personality dimensions measured by the PDIX, other evidence (written reports by participants and intimates) brings attention to behavioral changes. It is desirable that future research investigate shifts in both these variables. Also, increased validity for the findings may be obtained using more than one instrument to measure change. Previously cited methods pertaining to statistical analyses which partial out the initial score from the change score correlations are also suggested. Finally, this study draws attention to the issue of actual change measurement. The results of this study indicate that the amount that participants are affected within the lab is about the same for all participants. The real differences between members' shifts lie in the prelab to day 2 losses, day 7 to postlab gains, and most importantly the prelab to postlab change. It would be easy to conclude that all participants were equally affected within-lab and postlab (both Group and self- reports show similar patterns of change for all members). But if change over all four observations is considered, only the low subgroups are shown to have made long-term changes, while the middle and high subgroups fluctuated but remained at much the same level as their prelab self- ratings. This finding emphasizes the need for longitudinal 85 research designs with multiple observations including both prelab and postlab measurement. Also underlined is the need for further examination of the differential effects of laboratory situations on participants. One possible method would be an in-group analysis of the interaction processes between individuals. As was speculated, the laboratory may have a "moderating" effect, much like peer-group pressure, that tends to make individuals less extreme, at least on self-reported per- ceptions. An in-group analysis of interaction may be able to identify any specific types or amounts of interaction that facilitate this. If it were then found that low individuals were the only ones to make long-term changes, it would have far-reaching implications on the utility of laboratories as catalysts for personal growth. It may be that laboratories serve more as normatizing agents rather than stimulating growth in all participants. These findings also bear on the issue of homogenity vs. heterogenity of group composition. Further research could examine differ- ential participants changes in laboratory groups that are composed entirely of individuals who rate themselves similarly. In summary, this study points to need for (1) multiple observers, (2) multiple measurement techniques, and (3) a longitudinal research design to adequately measure change. It also underlines the importance of assessing 86 differential effects of the laboratory experience on its participants. Further research seems definitely warranted to more specifically determine the interrelationships among these variables. BIBLIOGRAPHY BIBLIOGRAPHY Argyris, C. Interpersonal Competgpce and Organizational Effectiveness. Homewood, 111:: Irwin-Dorsey, I962. Argyris, C. T groups for organizational effectiveness. Harvard Business Review, 1964, 42 (2), 60-74. Argyris, C. Exploration in interpersonal competence--I. Journal of Applied Behavioral Science, 1965, 1, 58-830 Argyris, C. Explorations in interpersonal competence—-II. Journal of Applied Behavioral Science, 1965, l (3), 255-69. _ Argyris, C. On the future of laboratory education. Journal of Applied Behavioral Science, 1967, 2 (2), 153-83. Barl, C., and Mitchell, R. Experimental evaluation of densitivity training. Journal of Applied Behavioral Science, 1972, 8 (3), 263-76. Beer, M., and Kleisath, S. W. The effects of the managerial grid lab on organizational and leadership dimensions. In S. S. Zalkind (Chm.) Research on the impact of using different laboratory methods for interpersonal and organizational change. Symposium presented at the meeting of the American Psychological Assn., Washington, D.C., September, 1967. Benne, K. D. History of the T-group in the laboratory setting. In L. D. Bradford, J. R. Gibb, and K. D. Benne (eds.), T-Group Theory and Laboratopngethod. New York: Willy, 1964. Bennis, W. G. Goals and meta—goals of laboratory training. Human Relations Training News, 1962, 6 (3), 1-14. 86 87 Bennis, W. R., Burke, H., Cutter, H., Harrington, and Hoffman, J. A note on some problems of measurement and prediction in a training group. Group Psycho- therapy, 1957, 13, 328-41. Berne, E. Principlgs of Group Treatment. New York: Oxford University Press, 1966. Blake, R. R., Mouton, J. S., and Fruchter, B. A factor analysis of training group behavior. Journal of Social Psyghology, 1962, 28, 121-30. Blassfield, M. G. Depth analysis of organizational life. California Management Review, 1962, 5, 29-42. Bolman, L. G. The effects of variations in educator behavior on the learning process in laboratory human relations education. Unpublished doctoral dissertation, Yale University, 1968. Boyd, J. B., and Ellis, J. D. Findings of Research Into Senior ManagemepESeminars. Toronto: Hydro- Electric Power Comm. of Ontario, 1962. Bradford, L. P., Gibb, J. R., Benne, K. D. (eds.). T Group Theory and Laboratory Method Innovation in Re- educatlon. New York: John Wiley & Sons Inc., 1964. Buchanan, P. Evaluating the effectiveness of laboratory training in industry. In E. A. Fleishman (ed.), Studies in_Personnell and Industrial Psychology. Homewood, II1.: Dorsey, 1967. Bunker, D. R., and Knowles, E. S. Comparison of behavioral changes resulting from human relations laboratories of different lengths. Journal of Applied Behavioral Science, 1967, 3 (4), 505L23. Bunker, D. R. Individual applications of laboratory training. Journal of Applied Behavioral Science, 1965, l (2), 131-48. Burke, R. L., and Bennis, W. G. Changes in perception of self and others during human relations training. Human Relations, 1961, 11, 165-82. Byrd, R. E. Training in a non-group. Journal of Humanis- tic Psychology, 1967, 1 (1), 18-27. Campbell and Dunnette. Effectiveness of T. group experi- ences in managerial training and development Psyghological Bulletin, 1968, 19, 73-104. 88 Cline, U. B. Interpersonal perception. In B. A. Maker (ed.), Progress in Ex erimenEal Personality Research. New York: Academic Press, 196417 Cronbach, L. J. Processes affecting scores on "under- standing of others" and "assumed similarity." Psyghological Bulletin, 1955, 22, 177-93. Cronbach, L. J., and Furby, L. How should we measure "change" or should we? Psychological Bulletin, Culbert, S. A. Trainer self-disclosure and member growth in two t-groups. Journal of Applied Behavioral (Science, 1968, A (1), 47-73. Egan, J. Encounter: Group Processes for Interpersonal Growth. Belmont, Ca1if.: Brooks-COIe, 1970. Force, E. J. Personal changes attributed to human relations training by participants, intimates, and job colleagues. Unpublished doctoral dissertation, Michigan State University, 1969, 188 pp. Friedlander, F. The impact of organizational training laboratories upon the effectiveness and interaction of ongoing work groups. Personnel ngchology, 1967, 29 (3), 289-307. Gage, N. L., and Cronbach, L. J. Conceptual and methodo- logical problems in interpersonal perception. Psychological Review, 1955, 32, 411-22. Gassner, 8., Gold, J., and Snadowsky, A. M. Changes in the phenomenol field as a result of human relations training. Journal of Psychology, 1964, 28, 33-41. Golembuski, R., and Blumberg, A. SensitivityTraining and the Laboratory Appgoach. Itasca, I11.: Peacock, 1970. Gottschalk, L. A., and Davidson, R. S. Sensitivity groups, encounter groups, training groups, marathon groups, and the laboratory movement. In H. 1. Kaplan and B. J. Sadock (eds.), Sensitivity Through Encounter and Marathon. New York: Dutton, 1972. Greening, T. C. Sensitivity training: cult or contri- bution. Personnel, 1964, 11 (3), 18-25. Greening, T. C., and Coffey, H. S. Working with an impersonal T group. Journal of Applied Behavioral Science, 1966, 2 (4), 401-11. 89 Hare, P. Handbook of Small Group Research. New York: The Free Press, 1962. Harris, C. W. (ed.). Problems in Measuring Change. Madison: University of Wisconsin Press,’1963. Harrison, R. The person description instrument X. Un— published manuscript. Harrison, R. Cognitive change and participation in a sensitivity training laboratory. gournal of Con- sultant Psychology, 1966, 39 (6), 517-20. Harrison, R., and Lubin, B. Personal style, group compo- sition, and learning, part I. Journal of Applied Behavioral Science., 1965, l (3), 786-94. Harrison, R. Research on human relations training: design and interpretation. gournal of Applied Behavioral House, R. T group education and leadership effectiveness: a review of the empirical literature and a critical evaluation. Personnel Psychology, 1967, 22 (11), 1-300 Houts, P. S., Zimerberg, S. M., Rand, K. H., and Yalom, I. D. Prediction of improvement in group therapy. Archives of General Psyghiatry, 1967, 11, 159-68. Hurley, J. R., and Force, E. J. A multiobserver-—multi- method study of interpersonal competence and T- groups. Unpublished manuscript, Michigan State University, 1971, 61 pp. . T-group gains in acceptance of self and others. International Journal of Groungsychotherapy, 1973, '23, in press. Kassarjian, H. H. Social character and sensitivity training. Journal of Applied Behavioral Science, 1965, 1 14), 411—40. Kernan, J. Laboratory human relations training--its effects on the "personality" of supervisory engineers. Doctoral dissertation, Department of Psychology, New York University, 1963. Kolb, D. A., Winter, S. K., and Berlew, D. E. Self directed change: two studies. Journal of Applied Be- havioral Science, 1968, 4 (4), 453-71. 9O Lieberman, M. A. The influence of group composition on changes in affective approach. In D. Stock, and H. A. Thelen (eds.), Emotional Dynamics and Group Culture. Washington, D.C.: National Training Laboratories, 1958, Chapter 15, 131-39. Lieberman, M. A. The relation of diagnosed behavioral tendencies to member-perceptions of self and of the group. In D. Stock, and H. A. Thelen (eds.), Emptional Dynamics and Group Culture. Washington, D.C.: Nationa1 Training Laboratories, 1958, Chapter 5, 35-49. Lieberman, M., Yalom, I., and Miles, M. Encounter Groups: First Facts. New York: Basic Books,11973. Luke, R. The internal normative structure of sensitivity training groups. Journal of Applied Behavioral Science, 1972, 8 (41, 421-37. Mathis, A. G. "Trainability" as a function of individual valency pattern. In D. Stock and H. A. Thelen (eds.), Emotional Qynamics and Group Culture. Washington, D.C.: National Training Laboratories-- National Education Association, 1958. Massarik, F., and Carlson, G. The California Psychology Inventory as an indicator of personality change in sensitivity training. Masters thesis, University of California, Los Angeles, 1960. Mc Nemar, Q. Psychological Statistics. New York: John Wiley and Sons, 1972. Miles, M. B. Human relations training: processess and outcomes. Journal of Counsel Psychology, 1960, 1, Miles, M. B. Changes during and following laboratory training: a clinical-experimental study. Journal of Applied Behavioral Science, 1965, 1_(3), 215-42. Miles, M. B. Learning processess and outcomes in human relations training: a clinical-experimental study. In E. Schein and W. Bennis (eds.), Personal and Organizational Change Through Group Methogg. New York: Wiley, 1965, 244-54. Napier, R., and Gershenfeld, M. Groups: Theory and Experience. Boston : Houghton Mifflin Co., 1973. 91 Odiorne, G. The trouble with sensitivity training. In R. Golembuski and D. Blumberg (eds.), Sensitivit Training and Laboratory Approach. Itasca, 111.: Peacock, 1970. Omwake, K. T. The relation between acceptance of self and acceptance of others on three personality inventories. Journal of Consulting Psychology, 1954, 12' 443-46. Oshry, B. I., and Harrison, R. Transfer from here and now to there and then: changes in organizational problem diagnosis stemming from T-group training. Journal of Applied Behavioral Science, 1966, a (2), 185-98. Pattison, M. Evaluation studies of group therapy. Inter- national Journal of Group Psychotherapy, 1967, 15, 382‘9fi7r. — Peters, D. R. Identification and personal change in laboratory training groups. Unpublished doctoral dissertation, Massachusetts Institute of Technology, 1966. Powdermaker, F. B., and Frank, J. D. GroupPsychotherapy, Cambridge, Mass.: Harvard University Press, 1953, Chapter 4. Rogers, C. R. The process of the basic encounter group. In.J. F. Bugental (ed.), The Challenge of Humanistic Psychology. New York: McGraw-Hill, 1967. Rubin, I. Increased self-acceptance: a means of reducing prejudice. Journal of Perceived Social Psychology, 1967' 2' 233-380 Rubin, I. The reduction of prejudice through laboratory training. Journal of Applied Behavioral Science, 1967’ _3-’ 29-500 Shutz, W. C. On group composition. Journal of Abnormal Social Psychology, 1961, g, 275-81. Shutz, W. C., and Allen, V. L. The effects of a T-group laboratory on interpersonal behavior. Journal of Applied Behavioral Science, 1966, 3 (3), 265-86. Seashore, C. What is sensitivity training? In R. Golembuski and A. Blumberg (eds.), Sensitivipy Training and the Laboratory ApproagA. Itasca, 111.: Peacock, 1970. 92 Shein, E. H., and Bennis, W. G. Personal and Organizational Change Through Grouprethods: The Laboratogy Approach. New York: Wiiey, 1965. Sherrer, E. T. An analysis of the relationship between acceptance of and respect for self and acceptance of and respect for others in ten counseling cases. Journal of Consulting Psychology, 1949, $3, 169-75. Smith, H. C. Sensitivigy to Pegple. New York: McGraw- Hill, 1966. Solomon, A., Perry, 8., and Devine, R. Interpersonal Communication. Springfield, 111.: Thomas,'1970. Steele, F. I. Personality and "laboratory style." Journal of Applied Behavioral Science, 1968, A (1), 25-45. Stock, D., and Ben-Zeev, S. Changes in work and emotion- ality during group growth. In D. Stock and H. A. Thelen (eds.), Emotional Dynamics and Group Culture. Washington, D.C.: National Training Laboratories, 1958, Chapter 21, 192-206. Stock, D., and Luft, J. "The T-E-T Design." Typewritten manuscript, National Training Laboratories, 1960. Stock, D. A survey of research on t-groups. In L. D. Bradford, J. R. Gibb, and K. D. Benne (eds.), T- group Theory and Laboratory Method. New York:— Wiley, 1964. Tucky, L. R., Damarin, F., and Messick, J. A base-free measure of change. Psychometrika, 1966, 2$_(4), Underwood, W. J. Evaluation of laboratory method training. Training Directors Journal, 1965, l2}(5), 34-40. Valiquet, M. 1. Individual change in a management develop- ment program. Journal of Applied Behavioral Science, 1968, A (31, 313-25. Watson, J., Lippitt, R., Kallen, D., and Zipf, S. "Evalu- ation of a human relations laboratory program." Typewritten manuscript, Research Center for Group Dynamics, University of Michigan, 1961. 93 Yalom, I. D. The Theory and Practice of Group Psycho- therapy. New York: Basic Books, 1970. Zand, D. E., Stelle, F. I., and Zalkind, S. S. The impact of an organizational development program on per- ceptions of interpersonal, group, and organi- zational functioning. Journal of Applied Behavioral Science, 1969, 2 (3), 393-411. AP PEITDIX 94 THE PERSON DESCRIPTION INSTRUMENT--INTERNAL Coded to Identify Variables , The bipolar scales given below are to elicit your impressions of how . has behaved in recent weeks. Encircle the (write in name) -. point on each scale which best represents your impression of 2 '3 behavior. __0 o» o e o-'0' e D» 0 .0 D» o; .0 _o is» o. e c» o e at o e c-'o e 03' —-—— .. NONDEFENSIVE '0009000.00000000'00000000.os sssss 0.0000000. DP;F":IL‘$IVI'; WMILITY .e000000.00000.00.000.0000'0000so... 0000000 . I'I(,;“ ,"' SHOWS FEELINGS 'so.0.00.00.00.00'eoeeoeoe.eo ...... 'oo ..... ‘ “1111933918“: TACTLESS . '.......'........'........'........'.......' TACTFUL INFORMED ‘. 'oeoeooe.oeeeeooe'ooooeeeo.oecoco...o. ..... ' UNILK)BJ'ULII RESERVED I .0000....so.oooeo'oeeooeeo.eoseoooo.oases... OUTSIA'UKISII SYMPATHETIC 'so...oo'ooeeoooo'eoeoeoeo.eooeeooo'e.000... UNSYMl’h'Hiiu'oHf lxncoupETENT '.......'........'........'........'.......' COMPETHNT ENTHUSIASTIC .000...0.00.000000'00000000.00.00.00.0000000. UNENTIRJSIPJIW!“ TWENING .00.00100.ooeeeooe.ooogoeo00.0.00...0.0000000. NONTUIM'I'UUHL: 1m STams 'oo'eoeee.0000000..see-0000.00.00.09.oooeeee. HIGH STA'PUS MOTIONALLY ' .0000....OOOOOOOO.OOOOOOOO.IOOOOCOCO.0.00.... Ul'W‘IOIJAL sxpasssrvs_ CWSIDERATE ‘ .OOOOO...0.0.00.0.00000000.00000000.0000000. INCUNIS[HEP/JFK - uuxurnuaumxaL '.......'........'........'........'.......' [NFLUHHTIAL QUIET - .OOOOOOO'OOOOOOOO'00.......OOOOOOOO..OOOOOOO. '3')”l’ ACCEPTS OTHERS '.......'........'........'........' ....... 'Iuauarmac.nw' :W PRESTIGE .0000....oooeeooo'ooo050.100.0000so...0.00000. 3".(31‘ "’44'0. Dzuousrnnmzva '.......'........'........'........'.......' UNDEMONSTRATIVH Fm!“ .0.0......0.000000'OCOOOO...000......00.0.... I‘JE‘LEX“$III; Ianssponsxans '.......'........'........'........'.......' nmsnowsvuuw ACTIVE '0......'OOOOOOOO'OCOOOOO0.0.000.000.0000000. 9A3331VU REJECTS .eosoooo.00.00.00.00000000.oeoooooo'oooeooo. AL‘CEP'I‘S SUGGESTIONS Tmmmnn '.u.”.h.”.”.h.”.”.h.u.n.h.u.u'cmmums NONCOMMITTAL '.......'........'........'........'.......' COMMITS SELF mummwmwnw '.u.fi.h.u.n.5.g.u.8.”.H.H.n.§'Sflmmmn LAzy '.......'........'........'........'.......' "Aunwuuv1un PARTICIPATES '.......'........'........'........'.......' vnursn1pnwux ‘MUCH . . Iwrwsua: o-Internal warmth and Acceptance; A=Power and Lftvrtlvnt’. in Work; '_ "ACt1V1tY and Expressiveness.