University Micrdfilms International

IN F O R M A T I O N T O U S E R S

This reproduction was made from a copy of a document sent to us for microfilming. While the most advanced technology has been used to photograph and reproduce this document, the quality of the reproduction is heavily dependent upon the quality of the material submitted. The following explanation of techniques is provided to help clarify markings or notations which may appear on this reproduction. 1.The sign or “target” for pages apparently lacking from the document photographed is “Missing Page(s)” . I f it was possible to obtain the missing page(s) or section, they are spliced into the film along with adjacent pages. This may have necessitated cutting through an image and duplicating adjacent pages to assure complete continuity. 2. When an image on the film is obliterated with a round black mark, it is an indication of either blurred copy because o f movement during exposure, duplicate copy, or copyrighted materials that should not have been filmed. For blurred pages, a good image o f the page can be found in the adjacent frame. I f copyrighted materials were deleted, a target note will appear listing the pages in the adjacent frame. 3. When a map, drawing or chart, etc., is part of the material being photographed, a definite method o f “sectioning” the material has been followed. It is customary to begin filming at the upper left hand comer of a large sheet and to continue from left to right in equal sections with small overlaps. I f necessary, sectioning is continued again-beginning below the first row and continuing on until complete. 4. For illustrations that cannot be satisfactorily reproduced by xerographic means, photographic prints can be purchased at additional cost and inserted into your xerographic copy. These prints are available upon request from the Dissertations Customer Services Department. 5. Some pages in any document may have indistinct print. In all cases the best available copy has been filmed.

University Micrdfilms International 300 N. Zeeb Road Ann Arbor, Ml 48106

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

R eproduced with permission of the copyright owner. Further reproduction prohibited without permission.

8503003

W in k e lm a n , M ic h a e l J a m e s

A CROSS-CULTURAL STU D Y OF M AGICO-RELIGIOUS PRACTITIONERS

University of California, Irvine

University Microfilms International

Ph.D.

1984

300 N. Zeeb Road, Ann Arbor, Ml 48106

Copyright 1984 by W inkelman, Michael James All Rights Reserved



PLEASE NOTE:

In all cases this material has been filmed in the best possible way from the available copy. Problems encountered with this d ocum en t have been identified here with a check m ark V

1.

Glossy photographs or p a g e s _______

2.

Colored illustrations, paper or p rin t______

3.

Photographs with dark b ac k g ro u n d ______

4.

Illustrations are poor c o p y _______

5.

Pages with black marks, not original

6.

Print shows through as there is text on both sides of p ag e______

7.

Indistinct, broken or small print on several p a g e s \ /

8.

Print exceeds margin re q u irem en ts______

9.

Tightly bound copy with print lost in spine____

10.

C om puter printout pages with indistinct p rin t______

copy______

11 .

P a g e (s )_____________ lacking when material received, and not available from school or author.

12.

P a g e (s )_____________ seem to be missing in numbering only as text follows.

13.

T w o pages n u m b e re d ______________. Text follows.

14 .

Curling and w rinkled p a g e s _______

15 .

.

O th er______________________________________________________________________________

University M icrofilm s international



UNIVERSITY OF CALIFORNIA Irvine

A Cross-Cultural Study of Magico-Religious Practitioners

A dissertation submitted in partial satisfaction of the requirements for the degree Doctor of Philosophy in Social Sciences

by MICHAEL JAMES WINKELMAN

Committee in Charge: Professor Duane G. Metzger, Co-Chair Professor Douglas R. White, Co-Chair Professor Robert L. Morris

1984


The dissertation of Michael Winkelman is approved, and is acceptable in quality and form for publication on microfilm:

'ildvdy? jva/H-. yyk^/af/x i

Committee CQpCjrair

Committee Co-Chair

University of California, Irvine

1984

ii


MICHAEL JAMES WINKELMAN

ALL RIGHTS RESERVED


DEDICATION

This dissertation is dedicated to the sociocultural milieu which fostered this work.

iii


CONTENTS List of Tables ..................................................... vi List of Figures .................................................... vii Acknowledgments .................................................... viii Curriculum Vitae ................................................... x Abstract ............................................... Chapter 1: Introduction ............................................. 1 Research Goals....................................................4 1.1 Types and Distributions of Magico-religious Practitioners. 4 1.2 Trance States and Magico-religious Practices .............11 1.3 Theoretical Perspectives.................................. 15 1.4 Universals of Magico-religious Practices ................. 19 1.5 Advantages and Limitations of this Study................. 22 Chapter 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9

2: Methodology .............................................28 Introduction ......................................... .....28 Determining Magico-religious Practitioner Roles ......... 31 The Sample ................................................. 35 Development of the Coding Format ......................... 37 Variables .................................................. 42 Sampling Unit Interdependence and Autocorrelation ....... 50 Measure of Similarity- Gower's Coefficient ............... 53 Other Statistical Procedures ..............................54 Data Reliability .......................................... 58

Chapter 3: Cluster Analysis of Practitioners ..................... 60 3.1 Validation of Cluster Analysis Solution................... 66 3.2 Discussion of Cluster Analysis Solutions ................. 69 Chapter 4: Practitioner Characteristics .......................... 79 4.1 Practitioner Type Characteristics ........................ 80 4.1.1 Priests .................................................. 80 4.1.2 Malevolent Practitioners ................................ 83 4.1.3 Mediums .................................................. 84 4.1.4 Healer Complex ........................................... 85 4.2 Variable Distributions......................................89 4.3 Sex Restrictions on Access................................. 97 4.4 Selection Procedures and Major Role Activities........... 101 Chapter 5: Practitioner Types and Configurations and their Relationship to Social Complexity Variables ................. 106 5.1 Entailment Analysis of Practitioner Types ............... 108 5.2 Configurations of Practitioners .......................... 110 5.3 Practitioner Types and Social Variables ................. 113 5.4 Practitioner Configurations and Social Variables ........ 125 iv

with permission of the copyright owner. Further reproduction prohibited without permission.

xii

Chapter 6: Trance States........................................... 134 6.1 A Psychophysiological Model of Trance States............. 137 6.2 Psychophysiological Aspects Trance Induction Procedures..145 6.3 Temporal Lobe Syndromes and Trance States................ 155 6.4 Trance Training Data Analysis............................. 163 6.4.1 Hypothesis: No Distinct Trance Types ................... 164 6.4.2 MDS Representation of Trance Variables .................168 6.4.3 Practitioner Types and Trance Induction Training ...... 174 6.4.4 Trance Variables Entailment Analysis ...................178 6.4.5 Possession Trance and Temporal Lobe Conditions ........ 181 6.4.6 Possession Trance and Social Complexity V a r i a b l e s 192 Chapter 7: Divination...............................................199 Chapter 8: Malevolent Practitioners........................... ....214 8.1 Hypothesis: Malevolent Practitioner Types ............... 220 8.2 Malevolent Practitioners- Origins and Associations....... 231 Chapter 9.1 9.2 9.3 9.4 9.5

9: Shamanism........ 247 Shamans- What, Where and W h y .............................. 248 Transformation of the Shamanic Role ....................... 254 Healers .................................................... 260 Mediums .................................................... 265 Priests .................................................... 272

Chapter 10: Conclusions............................................ 276 References.. ........................................................ 281 Appendix #1 Magico-Religious Practitioners ........................297 Appendix #2 Variables ............................................. 313 Appendix #3 Coding Instructions .................................. 317

v


L IS T OF TABLES

TABLE #

Page

Table #1-

Societies in Sample ......................................36

Table #2-

Practitioner Type Classification ........................63

Table #3-

K-Means Classification Differences ..................... 69

Table #4-

Core Shaman with Social Variables....................... 115

Table #5-

Shaman/Healers with Social Variables.................... 117

Table #6-

Healers with Social Variables........................... 119

Table #7-

Mediums with Social Variables........................... 119

Table #8-

Malevolent Practitioners with Social Variables......... 122

Table #9 - Priests with Social Variables........................... 124 Table #10- Practitioner Type Configurations and SocialConditions.129 Table #11- Comparison of MDS and Cluster Analysis with

Data...... 166

Table #12- Comparison of Data with Residuals....................... 167 Table #13- Practitioner Type and Trance Induction...................174 Table #14- Practitioner Type and Composite Trance Induction........ 177 Table #15- Trance Type and Amnesia (1)............................. 184 Table #16- Trance Type and Amnesia (2)............................. 188 Table #17- Trance Type and Tremors/Convulsions..................... 189 Table #18- Trance Type and Spontaneous Onset....................... 190 Table #19- Trance Type and Compulsive Motor Behavior.............. 191 Table #20- Anova- Possesion and Social Variables................... 195 Table #21- Anova- Possession Variables with Political Integration and TLD Variables........................................ 196 Table #22- Comparison of Data with Residuals....................... 224

vi


LIST OF FIGURES

FIGURE #

Page

Figure #1

cluster Analysis- All Magico-religious Codes (ave)....

71

Figure #2

Cluster Analysis- All Magico-religious Codes (comp)___

73

Figure #3

Cluster Analysis- Magico-religious Practitioners (comp) 75

Figure #4

Cluster Analysis- Magico-religious Practitioners

Figure #5

Entailment of Selection and Activities...........

Figure #6

Entailment of Practitioner Types .....................

109

Figure #7

Practitioner Type Configurations .....................

112

Figure #8

MDS Practitioner Trance Characteristics (1&2)........

169

Figure #9

MDS Practitioner Trance Characteristics (1&3)........

170

Figure #10 Entailment of Trance Variables .......................

180

Figure #11

MDS Malevolent Practitioners (1&2).....................

226

Figure #12 MDS Malevolent Practitioners (1&3).....................

227

(ave) 103

Figure #13

MDS Non-Malevolent Practitioners.................. 241

Figure #14

MDS

Figure #15

MDS All Practitioners with Malevolent Activities (1&3)

All Practitioners with Malevolent

77

Activities

vii


(1&2)242 243

ACKNOWLEDGEMENTS I wish to thank some of the people who have helped me bring this research project to its present form.

In particular I thank Elizabeth

for helping me through the most intense period by accepting such a major share of the care of Christopher.

I was very fortunate to have

DuLne Metzger, Doug White and Robert Morris as my committee.

Their

ability to wisely guide without trying to determine the course or form of these investigations has allowed for the development of a research project which could not have been envisioned in advance. I am indebted to a number of people for support and assistance.

I

wish to thank Roger Blashfield, Michael Burton and Mike Slagley for computer programs essential to the analyses presented here, and Helen Wildman, Cheryl Larsson, Kathy Alberti, Jim Ashurst, and Jerry Keys, for other forms of technical assistance.

I thank Drs. William

Elmendorf, William Simmons and James Watson, who provided me with information about particular societies in this sample with which they had firsthand knowledge, and Glen McAlpine, Peter King, Lori Tomchak, and Toshio Wantanabe for translation and coding assistance.

I also

thank David Jacobs, Carmella Moore, Stephan Schwartz, Erika Bourguignon, Joe Jorgensen and Joseph Long for comments on portions of this work. I gratefully acknowledge several forms of institutional support which were provided for me in the course of this project.

A National

Science Foundation Dissertation Improvement Grant was provided for coding reliability checks on the data collected here.

The UCI Patent

Fund provided resources for review and coding of some of the

viii


foreign language materials necessary for this project.

The American

Society for Psychical Research awarded the Arthur C. Twitchell 1982 Scholarship, which provided me with crucial support during this work. Indispensable assistance was provided by the UCI Interlibrary Loan staff, who obtained much of the material essential to this research. I also thank the University of California, and indirectly the people of California, for the essential institutional support for this project; without it, this work would not have been possible.

ix


VITA Michael James Winkelman 4612 Verano Place Irvine, Ca. 92715 714-856-0328 PERSONAL Born in San Marcos, Texas Married, one child.

School of Social Sciences University of California, Irvine Irvine, Ca. 92717 714-856-6691 November 27, 1954

EDUCATION B.A. Psychology, Behavioral Science, Rice University

1976.

Ph.D. Social sciences, emphasis Anthropology. University of California, Irvine. 1984. Dissertation: "A Cross-cultural Study of Magico-religious Practitioners". TEACHING EXPERIENCE Teaching evaluations on file in School of social Sciences, UCI with Mary Lou Eckerman. April-June 1984. Teaching Assistant, Introduction to Archeology. School of Social Sciences, UCI. January-March 1983. Teaching Assistant, Ethnographic Journals. School of Social Sciences, UCI. Worked with Dr. Duane Metzger in developing and teaching ethnographic observations course integrated with use of word processors. August-September 1982, 1984. Instructor, Introduction to Parapsychology. Summer Session, School of Social Sciences, UCI. April-June 1982. Teaching Associate, Ethnography. School of Social Sciences, UCI. September 1980-December 1981. Teaching Assistant, Ethnography. School of Social Sciences, UCI. Managed UCI Field Station, Tijuana, B.C., Mexico. Organized and provided transportation for students, supervised field station, assisted in development of student research projects. April-June 1980. Co-lecturer for credit, ESP and Anthropology. School of Social Sciences, UCI. Developed and taught course with Dr. Joseph Long.

x


PUBLICATIONS 1980 Science and parapsychology: an ideological revolution. Revision 3(1):59-64. 1981a The effects of formal education upon extrasensory abilities: the Ozolco study. Journal of Parapsychology 45:321-336. 1981b "The effects of schooling and education upon extrasensory abilities" in Research in Parapsychology 1980. Edited by W. Roll and J. Beloff. Methuen, N.J.:Scarecrow. 1982a Magic: a theoretical reassessment. Current Anthropology 23:37-44, 59-66. 1982b Cross-cultural research as methodology. Journal of Parapsychology 46:17-27. 1983 The anthropology of magic and parapsychological research. Parapsychology Review 14(2):13-19.

GRANTS 1981-82. UCI Chicano/Mexico Program. $1000 grant for study of development procedures and techniques of group of spiritualist healers in Tijuana, B.C. 1982. Patent Fund, UCI. $1400 grant for coding of foreign language materials in the Standard Cross-cultural Sample. 1983. National Science Foundation Dissertation Improvement Grant. $7263 for coding reliability checks on "A Cross-cultural Study of Magico-Religious Practitioners".

PROFESSIONAL AFFILIATIONS Associate and Referee, Current Antrhopology. Member, American Anthropological Association Member, Association for Transpersonal Anthropology Associate Member, Parapsychological Association


ABSTRACT OF THE DISSERTATION A CROSS-CULTURAL STUDY OF MAGICO-RELIGIOUS PRACTITIONERS by Michael James Winkelman Doctor of Philosophy in Social Sciences, emphasis Anthropology University of California, Irvine, 1984 Professor Duane G. Metzger, Co-Chair Professor Douglas R. White, Co-Chair This study provides a cross-cultural assessment of magicoreligious practitioner roles, using societies in a subsample of the Standard Cross-Culture Sample.

Cluster analysis based upon more than

one hundred variables provides a typology of four major types of magico-religious prctitioners, labeled here as: Priests, Mediums, Malevolent Practitioners and members of the Healer Complex, subdivided into Shamans, Shaman/Healers and Healers.

There are four principal

configurations of magico-religious practitioners found cross-cultura1ly: 1) Shaman (or other practitioners of the Healer Complex); 2) Healer Complex and Priest; 3) Healer Complex, Priest and Malevolent Practitioner or Medium; 4) Healer, Priest, Malevolent Practitioner and Medium. An evolutionary model strongly implicates agriculture, political integration and the presence of classes as determining the transition from one configuration of magico-religious practitioners to the next. Multiple regression with autocorrelation controls was used to establish these relationships with controls for diffusion.

xii


Psychophysiological research is reviewed to establish that trance states associated with the training of magico-religious practitioners share common features in establishing an altered state based in hippocampal-septal synchronous slow wave dominance of the frontal cortex.

The relationship of these psychophysiological changes

associated with trances to those precipitated by temporal lobe dysinhibitions is reviewed.

These types of trance states are found

associated with a least one type of magico-religious practitioners in all societies of the sample.

The incidence of trance types labeled

here as possession is strongly and significantly related to both social variables (political integration) and psychophysiological variables (characteristics indicative of temporal lobe dysinhibitions). This study concludes that magico-religious practitioners have derived from three bases.

The Shaman, Shaman/Healer, Medium (and to

some extent Healer) have a psychophysiological basis related to the induction of trance states.

The Priests have their roles largely

deriving from their positions as leaders of sedentary agriculture societies with hierarchical political structures.

The Malevolent

Practitioners derive from the political integration of societies and the persecution of others,

in particular local members of the Healer

Complex.

x iii


CHAPTER 1 INTRODUCTION

This thesis is the initial phase of a project directed towards a cross-cultural investigation of magical and religious practices.

This

research is based upon analysis of coded data for a wide range of variables characterizing magico-religious practitioners found in a cross-cultural sample.

Reported here is the preliminary analysis of

the structure of this data and its relationship to selected variables assessing social conditions.

This research was directed towards

determining: 1) the types, characteristics and distributions of magico-religious practitioners found cross-culturally; 2) the relationship of magico-religious practitioner types to the social conditions of the societies in which they were present; 3) the relationship among magico-religious practitioner types; and 4) the relationship of trance states to magico-religious practices. This study has provisionally employed the label "magico-religious" to refer to the phenomena which are the subject of inquiry.

The use

of the term magico-religious is not to assert a fundamental similarity of the varied phenomena subsummed under the label.

Rather, it

reflects a cultural and disciplinary bias which tends to see magical and religious phenomena as substantially, if not fundamentally


2 similar, combined with an uncertainty as to how the ideal type distinctions between magic and religion relate to the broader characteristics of the phenomena labeled magical and religious. In the abstract, Frazer's distinction between magic and religion is rather simple and clear.

Magic is used to refer to the presumed

manipulation and control of supernatural forces while religion is used to refer to the propitiation and supplication of what is conceived of as supernatural beings.

However, in the context of actual practices,

magical and religious acts are frequently intermingled (see Wax and Wax 1963).

This indicates that the distinction between magic and

religion is of limited use in classifying activities, and Dundes (1982) suggests that the effort to separate magic and religion is fruitless.

Others (e.g., Durkheim 1915, Hammond 1970) consider magic

to be a part of religion, but this hardly obviates the need to distinguish between the phenomena to which the terms magic and religion refer or to determine the relationship of these ideal types to the ethnographic data.

If religion is defined as activities

involved with belief in supernatural beings, then most of magic is religion, but it leaves a residual aspect of magic involved in the manipulation of impersonal supernatural power (mana). Anthropologists have been concerned with the nature of these ideal/polar types which, when compared with ethnographic data, seldom prove to be useful or illuminating.

It is apparent that the labels

magic and religion refer to an overlapping domain of phenomena which are not clearly specified or differentiated.

Given this uncertainty


3 with respect to the nature of magic and religion and their differences and relationships (e.g., see Wax and Wax 1963 and Rosengren 1976 with subsequent comments), it was decided "magico-religious".

to refer to the phenomena as

It is intented that the analysis of the data

collected here shed light upon the nature of the relationship of the concepts of magic and religion to the phenomena of the domain. Anthropologists consider magic and religion to be similar in their assumption of an order of the universe which is generally glossed as the supernatural.

This supernatural order generally involves a set of

ontological beliefs which presume the functioning of non-human beings (spirits), or the ability to use special powers occurring outside of the normal experience, knowledge and access of humans.

From a

"scientific point of view", these phenomena do not objectively exist as believed by the participants of these systems.

Most would agree

that if supernatural phenomena or powers as conventionally understood do exist, they cannot be explained by currently known forces or laws of nature as recognized by Western science. The considerations here do not address the appropriateness of the concept of the supernatural in unifying the phenomena referred to as magic and religion, nor does it provide a consideration of the ontological status of the supernatural, or the ultimate reality of supernatural phenomena.

The term supernatural is used here to refer

to those beliefs and practices which involve beliefs in spiritual beings or special sources of power which allow humans to influence the course of nature in ways not accessible to ordinary human functioning and abilities.

The use of the term "supernatural" is not to imply any


4 acceptance of the ultimate nature of supernatural reality.

The term

is used as a convenient gloss to refer to a domain of belief which is conventionally understood by the use of the term.

RESEARCH GOALS AND FINDINGS The following sections integrate a revised version of the original research plan with the findings presented in the subsequent chapters. The intent of this review is to orient the reader to the general plan of research, to present the overall structure of the considerations which follow, and to provide an overview of the major findings of this research project.

1.1 Types and Distribution of Magico-religious Practitioners A primary goal of this research was to determine an etic typology of the different types of magico-religious practitioners encountered cross-culturally.

This assumption that magico-religious practitioners

can be classified into groups of distinct types is a hueristic utilized in the initial organization of the investigation into the domain of inquiry.

In the later chapters (8 and 9), the data analyzed

is reviewed to argue that some of the different types found here are best understood as the transformation of some practitioner types into others as a consequence of social changes.

Whether or not the entire

domain of magico-religious practitioners is best represented as discrete types or continuous gradations between ideal types is not formally addressed in this phase of the research, largely because the


5 number of cases in this study exceeds program limitations for the statistical procedures necessary to formally address the problem. The determination of a cross-cultural typology of magico-religious practitioners will help suggest etic labels for magico-religious practitioners found in different societies.

The lack of formal

cross-cultural approaches in assessing magico-religious practices has resulted in the lack of generalizable terminology and inconsistency and disagreements in the applications of presumed etic terminology. For instance, some writers (e.g., Siikala 1978) argue that the term "shaman" be restricted to Siberian and sub-artic practitioners, while others (e.g., Lewis 1971; Peters and Price-Williaras 1981) use it more broadly to refer to any magico-religious practitioner entering trance states.

Hultkrantz (1978) provides a detailed set of definitions and

considerations as to what constitutes shamanism, apparently including modern spiritualists within the shamanic tradition.

This inves

tigation will help address the issues of the universality, distribution and characteristics of shamans by empirically identifying the shamanic like practitioners which are classified together. A similar problem is encountered with respect to magical practitioners called sorcerers and witches.

Evans-Pritchard (1937)

introduced anthropologists to the Azande distinction between sorcerers and witches.

Simply, sorcerers are thought to engage in conscious,

deliberate manipulative acts causing harm, while witches are thought to engage in unconscious psychic acts which cause harm (see Chapter 8).

However, Evans-Pritchard's distinctions have not been widely

investigated.

Thus far there has been a lack of research determining


6 the appropriateness of these distinctions for other societies, or determination of the relationships of these different ideal types to social variables.

This study addresses the issue of differences in

magical practitioners engaged primarily in malevolent acts, with the intent of exploring the validity of the witch/sorcerer distinction and its relationship to social complexity variables. The development of an empirical and cross-culturally valid typology of magico-religious practitioners and the determination of the appropriate extension of etic terminology was accomplished by this research since it was based in a detailed assessment of a wide range of descriptive characteristics of the practitioners.

The

characteristics of magico-religious practitioners were assessed in several major areas (see Methods, Section 2.5 and Appendix #2). variable areas include:

The

political powers, social and economic status

and roles, mythological, and psychological characteristics of the practitioners, selection and training procedures, types of magico-religious activities performed, magical techniques employed, and trance induction procedures and the physical characteristics of the trances associated with training.

The final coding of data used

here included nearly 300 variables which were combined and reduced to 98 variables for the purpose of the analyses reported here.

This

approach overcomes the problem of comparability of practitioners types from different cultures by basing analysis and determination of shared similarities in a wide range of observed characteristics rather than upon the labels used to refer to the roles.

It is hoped that the wide

range of variables used in deriving this typology will ensure that the


typology is valid, and not likely to substantially change with the addition of other variables characterizing the practitioners. The typology derived was used as the focus for additional investigations of magico-religious practitioners.

The core

characteristics of each type of practitioner have been determined (see Chapter 4) and the similarities and differences between types have been subject to preliminary exploration (see Chapter 9).

Another

primary concern was determination of the types of configurations of practitioner types, that is, which types of practitioners occur together, and an investigation of the social and geographic determinants of types, both of which have been addressed in Chapter 5. Some of the research questions related to the practitioner types and their characteristics which were formulated prior to the analysis of data and which have been addressed below are: 1) The distribution of the shaman type practitioners and the characteristics of practitioners undertaking similar roles in cultures where shamans are not present (Chapters 4 and 9); 2) Determination of the relationship between selection procedures and the functions of practitioners (Chapter 4). 3) A determination of the cross-cultural similarities and differences in malevolent practitioners and an investigation of the usefulness of the witch vs. sorcerer distinction offered by Evans-Pritchard (1937) and Murdock (1980) (Chapter 8); 4) Cross-cultural differences and similarities in practitioners with primary healing functions and the distinguishing characteristics of different types of healers (Chapter 9).


The essential findings with respect to these major questions are presented here.

The results of Chapter 4 suggests that there are four

major types of magico-religious practitioners which have been provisionally labeled as follows: members of the Healer Complex (Shaman, Shaman/Healers and Healers); Mediums; Malevolent Practitioners; and Priests. Shamans occur universally in hunting and gathering societies, without agriculture or political integration beyond the local level. As a result of the changes concomitant with the development of agriculture and sedentary lifestyles, the shaman's role and activities apparently undergoes changes to the practitioner type labeled a Shaman/Healer.

Although the Healer role shows considerable

similarities with the Shaman role, those practitioners classified as Healers occur predominately in societies with political integration beyond the local level, suggesting that these social conditions contribute to a further transformation of the shamanic role.

The

finding that Shamans, Shaman/Healers and Healers have strong exclusion relationships (Chapter 5) further supports a transformation model. Differences among the subtypes of the Healer complex are discussed in Chapter 9.

Briefly, the hypothesized transformation of the Shaman

into the Shaman/Healer and Healer subtypes occurs as a consequence of the transformation of societies from hunting and gathering to agricultural and politically stratified systems.

The concomitant

changes in role activities include a decrease in the use of trance states, and increase in the formal organization of the practitioner


9 group, an increase in involvement with activities such as propitiation, sacrifice, agriculture rituals, and the development of elaborated material systems for divination (e.g., like the I Ching and Tarot; see Chapter 7). The practitioners labeled as Mediums utilize possession trance states as the basis of their interaction with the supernatural.

In

the use of trance states, Mediums are similar to those practitioners labeled as Shamans.

However, the Mediums occur almost exclusively in

societies with political integration beyond the local level, again indicating that the social changes concomitant with the development of politically stratified societies brings about the development of this practitioner type. Malevolent practitioners refer to those practitioners generally labeled as sorcerers and witches; these practitioners are found almost exclusively in societies with political integration beyond the local level, and most (but not all) such societies have practitioners classified as Malevolent Practitioners present.

The sorcerer/witch

distinction offered by Evans-Pritchard is substantiated by the data of this study.

There are sorcerer like practitioners thought to engage

in deliberative manipulative acts, and witch like practitioners thought to engage in unconscious psychic acts.

However, the

differences between the sorcerer-like and witch-like practitioners are best represented as continuous, not discrete, rejecting the hypothesis of discrete types in favor of one which sees the differences as forming a continuum.

Political integration is shown to be the central

social variable contrasting sorcerer-like and witch-like


practitioners, suggesting that social forces related to the process of political integration are responsible for the hypothetical transformation from sorcerer to witch.

Further analyses (see Chapter

8) are used to argue that the roles of the Healer Complex (Shamans, Shaman/Healers and Healers) provide the social and magico-religious roles which are transformed into Malevolent Practitioners.

It is

suggested that the process of integration of local communities into politically stratified systems either systematically persecutes these practitioners, transforming them into the Malevolent Practitioner role, or continues to utilize these attributed beliefs and practices in the persecution of lower status or class individuals within the society. The practitioners labeled as Priests are found only in societies with an agricultural (or pastoral) economy, and predominate in societies with political integration beyond the local level, of which they are political leaders.

These social and religious leaders

apparently have their origin in kinship based hierarchies and selection for the role is primarily on the basis of senority within certain kinship groups, membership in privileged classes, or the exercise of political power. Examination of the relationship between magico-religious activities and role selection and training has indicated three main structures (see Chapter 4).

Briefly, trance training leads to

professional roles involving healing and divination; selection through social succession or political processes leads to a range of social and political powers and magico-religious activities involving


11 agriculture, propitiation and sacrifice; selection through persecution or ascriptive labeling leads to professional roles based largely or exclusively upon performing acts which are harmful to others.

These

three structures largely correspond to the Healer Complex and Diviner roles, the Priest role, and the Malevolent Practitioner role, respectively.

1.2 Trance States and Magico-Religious Practices The relationship between trance training and the types of magico-religious activities confirms the importance of trance with respect to magico-religious practices.

Determination of the

relationship of trance states to magico-religious practices was another principal goal of this research project.

As Norbeck (1961),

Mauss (1950), and de Vesme (1931) pointed out, trance and possession are central aspects of magico-religious practices.

However, although

some form of trance induction plays a central part of many magico-religious practices, not all magical practices involve trance induction, and those which do show a quite varied set of induction procedures, observable outcomes, and explanations for trance states (i.e., soul journey or possession). This research intended to determine the different types of trance states associated with magico-religious practices in a manner similar to the procedures outlined above with respect to determination of magico-religious practitioner types.

The suggestions of a number of

previous researchers investigating trance states (e.g.,Ludwig (1965), Bourguignon (1968), Prince (1966), Lex (1976), Kelly and Locke 1981))


12 were followed in developing a range of variables to use in assessing the nature of the trances in magico-religious rituals (see Appendix 2. #82-98).

The variables used assessed a wide range of induction

factors and observable conditions, including: sleep conditions, drug use, motor behavior, food intake, auditory driving, physical stimulation, convulsions, collapse, unconsciousness, and cultural interpretation of the trance. When this project began, there were only a few explicitly developed hypotheses with respect to altered states and magicoreligious practices.

The intention had been to determine the

distribution of different types of trance states and their relationship to: 1) cultural beliefs about trance states (possession versus non-possession) and 2) the complexity of the society.

It was

hypothesized that there would be psychophysiological differences between possession and non-possession trance states, and a reduction in the incidence and use of trance states in magico-religious practices in more complex societies. In the initial stages of data analysis it became apparent that the assessment of trance states could not be based solely on kinds of induction procedures or observable consequences, but had to be based in the psychophysiological effects of these procedures since a variety of procedures could in fact induce similar if not identical alterations in state of consciousness (e.g., mescaline and psilocybin).

This assumption led to a review of research on the

psychophysiology of altered states of consciousness and the psychophysiological effects of a wide variety of behaviors and agents.


The central idea which emerged from this review on the psychophysiology of altered states of consciousness was that widely varying trance induction techniques were directed towards establishing a relatively homogenous change in state of consciousness.

Lex (1976,

1979) has suggested a similar idea, namely that procedures frequently utilized to induce trance states involve a manipulation of the balance between the sympathetic and parasympathetic nervous systems with the effect of establishing the preeminence of the right hemisphere processes.

Mandell (1980), whose ideas have strongly influenced the

perspective developed here, has suggested a more fundamental mechanism for the similarity in varied altered state induction techniques.

He

suggests that a number of different techniques frequently associated with trance induction have common effects in effecting a drive arrest release sequence of the biogenic amine inhibitory system which triggers temporal lobe limbic and hippocampal-septal hypersynchronous discharges which impose a synchronized slow wave pattern upon the frontal cortex (see Chapter 6).

Mandel's review provided a basis for

establishing the necessary psychophysiological effects of a range of induction techniques and for arguing the essential similarity of a number of trance states arrived at by diverse procedures (see Chapter 6, Trance States).

Analysis of the trances associated with the

training of the magico-religious practitioners indicates that the differences among different trance induction procedures are continuous rather than discrete, supporting at the empirical level the contention that a variety of different trance induction procedures are directed towards a common set of psychophysiological changes.


The assumption that a single type of trance state underlies a variety of manifestations is consistent with several anthropologists' considerations, for instance Peters and Price-Williaras (1980), who explicitly decided not to distinguish between shamans and spirit mediums in assessment of the experiential characteristics of trance practitioners.

However the types combined by Peters and Price-

Williams correspond closely to the distinction between trance and possession trance suggested by Bourguignon (1968, 1976).

Since these

states are generally associated with significantly different social and psychological conditions (locus of control, amnesia), it was hypothesized that they would be associated with neurophysiologically distinct states.

The relationship of these different types of trance

states (possession versus non-possession) to induction procedures, the characteristics of the trance, and the social conditions of the societies in which they exist remained an open issue until later stages of the data analysis. The analyses presented in Chapter 6 argue that although possession and non-possession trances share a common set of psychophysiological changes, they differ in some fundamental aspects.

Whereas

non-possession trances are generally deliberately sought, possession trances generally show spontaneous initial onset, characteristics such as siezures, convulsions, tremors and other behaviors which suggest that the practitioners have evidence of temporal lobe syndromes. Mandel has related the psychophysiology of temporal lobe syndromes and related brain conditions to the psychophysiology of trance states induced by a variety of other procedures.


15 This study also confirms and extends Bourguignon’s (1968, 1976b, Bourguignon and Evascu 1977) findings of a relationship between possession trance states and measures of social complexity.

The

presence of possession trances states associated with magico-religious practitioner training was assessed for each society in the sample and compared with a number of measures of cultural complexity.

It is

demonstrated that political integration significantly predicts the incidence of possession, while other social complexity variables such as social stratification and fixity of residence did not provide significant explanation of the variance in possession beyond that provided by political integration.

However, it was also shown that

variables assessing temporal lobe syndromes (tremors, convulsions, amnesia, etc.) significantly predict the incidence of possession trance, and that independently significant prediction of variance in possession trance is provided by both social variables (political integration) and psychophysiological variables (indicative of a temporal lobe syndrome).

1.3 Theoretical Perspectives The research questions about types and characteristics of magico-religious practitioners, their relationships to social conditions, and the role of trance states in magico-religious practices were shaped by a broader perspective on magico-religious practices.

This section: 1) reviews some of the principal research on

the relationship between magico-religious practices and social conditions which contributed to the theoretical perspective guiding


16 this research; 2) outlines the theoretical perspective developed; and 3)

specifies a few specific hypotheses which were to be tested in an

exploration of the perspective.

The general perspective developed

here is that there are systematic differences in the magico-religious practices of different societies which is a function of factors related to sociel complexity.

The least complex societies are

typified by the hunting and gathering societies without hierarchical political integration; the more complex societies are typified by agriculture, social stratification and hierarchical political integration.

Although this study was directed towards discovering

unknown patterns in magico-religious data, previous research suggested systematic patterns which were expected. Norbeck notes a distinction among magico-religious practitioners, contrasting shamans and priests, associated with simpler and more complex societies respectively.

The shaman is considered to be in

direct personal contact with the spiritual world, grappling with spiritual entities, while the priest is seen as removed from direct contact with spiritual beings, dealing with them through offerings and propitiations but generally lacking experiental (altered state) contact (Norbeck 1961).

D'Andrade (1961) and Textor (1967) examined

the relationship of the use of dreams to seek supernatural power with social variables, finding relationships which Bourguignon (1972) points out indicate that the presence of vision quests and dreams as sources of power are negatively related to social complexity. Similarly, Jorgensen's (1980) analysis of Western North American Indian societies indicated that sedentary societies with centralized


political authority and sodalities indicative of priesthood had no vision quest on the part of lay people and limited use on the part of shamans, while less centralized societies showed widespread access to vision quests for both individuals and shamans and a greater incidence of shamans. Bourguignon's (1968, 1976b) findings on the relationship of trance and possession trance to indices of social complexity suggest a related pattern.

Trance, defined as an ego oriented altered state of

consciousness without loss of memory or displacement of personality is significantly associated with indices of simpler societies. Possession trance, a trance state in which it is believed that the practitioner's normal personality is displaced by a spirit entity, is significantly associated with indices of greater social complexity. Greenbaum (1973a,b) extended Bourguignon's findings in a regional study of Sub-Sahara Africa, suggesting that possession trance was associated with rigid societies-characterized as being non egalitarian, with ascriptive status, autocratic hierarchical political systems, centrally controlled with fixed residence, fixed religious rites and fixed group membership.

Lewis's (1971) discussion of

ecstatic religious forms suggests similar relationships; possession cults and religions arise from oppression, pressure or constraints to which subordinate members of the society or subjected communities are exposed. Murdock (1980) suggests a differential relationship between supernatural theories of illness and indices of social complexity. Sorcery, defined as involving conscious techniques and manipulation of


materials which can generally be employed by anyone, was contrasted with witchcraft, defined as innate and generally operating unconsciously.

Beliefs in sorcery and witchcraft as causes of illness

are negatively and positively associated respectively with several indices related to social complexity.

Similar patterns are revealed

in Whiting (1950), who found the presence of sorcery negatively associated with jurisdictional hierarchies, and Roberts (1976) who found a strong positive relationship between indices of social complexity and the presence of beliefs in the evil eye, in which an individual is believed to exercise a harmful influence upon others, although generally inadvertently. A general perspective extracted from this previous research is that systematic differences in magico-religious practices can be organized around the principle that beliefs and practices in more complex societies indicate a displacement of responsibility, repression of awareness, and reduction of direct ego control over information revealed or actions taken; practices and beliefs in simpler societies indicate a more direct ego contact with supernatural power and experiences.

This general perspective was to be explored by

testing a number of specific hypotheses about the relationship of magico-religious variables and the social variables. The specific hypotheses which were to be tested were: 1) The use of trance states in magico-religious practices is strongest in simplest societies, and more complex societies have a lesser incidence of trance states and a reduction in the intensity of trance experiences;


2) The presence of trance states labeled as possession (Bourguignon 1976) are more prevalent in more complex societies; 3) Malevolent magico-religious practitioner roles characterized by repression or projection of responsibility (as in the unconscious use of power like witchcraft), will be associated with more complex societal conditions than those practices characterized by direct responsibility for actions (e.g., the sorcerer, who is thought to carry out consciously directed manipulative techniques); and 4) Divination procedures in which the practitioners appear to have the most direct control over outcome are predicted to be more prevalent in simple societies; divination procedures which minimize direct control over the outcome through mechanical and randomizing procedures, or which indicate a displacement of responsibility for outcome through possession trance beliefs will be more prevalent in complex societies.

1.4 Universals of Magico-religious Practices. Cross-cultural similarities in magico-religious practices have led anthropologists to hypothesize the universality of particular beliefs, practices and principles.

These apparently include but are not

limited to the following: the presence and action of spirits as fundamental aspects of magico-religious practices (Tylor 1871); the presence of the principles of sympathy and contagion as basic characteristics of magical action and belief (Frazer 1929); a concept of supernatural power (mana) (Marett 1909); the enactment and visualization of desired ends (Malinowski 1954); the use of supernatural power to heal and to cause illness and death; the ability


20 to influence natural systems, crops, animals and humans; abilities such as clairvoyance, telepathy and precognition; the use of trance states as an integral aspect of magical practice (Mauss 1950, Norbeck 1961); and the presence of divination procedures (Wallace 1966). However, there has been no systematic effort to verify that these presumed universals are in fact present in all cultural groups or in a representative sample of cultures.

In fact, Bourguignon (1982) has

suggested that the supposed universals of magic are not found in all societies.

This study cannot conclusively address the issue of the

universality of these traits, since it focuses upon the practitioner role and does not include assessment of the magico-religious beliefs and practices present among the general population and not associated with a professional magico-religious role.

However, some of these

hypothesized universals (e.g., divination) are substantiated in the course of this research. The primary concern in this study with the issue of universals of magico-religious activity is at the level of the institutional structure of magic and religion and their origins, such as the theoretical positions proposed by Frazer (1929), Mauss (1954), Durkheim (1915), Malinowski (1954) and O'Keefe (1982).

In the initial

phases of this project it was presumed that the data collected here could not bear directly upon the issues of origin of magico-religious practices.

However, it has since become apparent that the data

collected here do bear upon this issue in a fundamental way.

The

implications of this data for an understanding of the origins of magico-religious practices is addressed indirectly in the context of


21 the relationship of selection and training procedures to magico-religious activities (Chapter 4), the relationship of magico-religious practitioner types to social conditions (Chapter 5), the origin of Malevolent Practitioners (Chapter 8), and the transformation of the shamanic role (chapter 9). The concluding chapter directly addresses the issue of the origin of magico-religious practices, suggesting the bases for the various types of magico-religious practitioners found in this study and the social forces which lead to their development and transformation.

The

theory of magico-religious practices derived from this research suggests three main bases for these practices.

The trance state

capacities of humans provide the basis for what can be referred to as magic in the strict sense (direct manipulation).

The Shaman,

Shaman/Healers and Mediums derive from this trance basis, and the Healer type is seen as a specialization and evolution of this capacity and the traditions which it has spawned.

The forces associated with

political and social power in agriculture societies provide the basis for what might be seen as religious in the stricter sense of involving worship, propitiation and supplication. Priest role.

This is the basis for the

The conflict between political power and charismatic

trance based power provides the basis for the third main aspect of magico-religious practices.

The Malevolent Practitioner role results

from the persecution of the practitioners of the Healer Complex or other non-magical individuals by those holding social and magical power.


22 1.5 Advantages and Limitations of this Study. Ethnographic descriptions have provided a basis for theoretical generalization about magico-religious practices and beliefs (e.g., Tylor 1871, Lang 1898, Marett 1909, Frazer 1929, Durkheim 1915, Mauss 1950, Malinowski 1954, Evans-Pritchard 1957, Wallace 1966, Lewis 1971, O'Keefe 1982).

However, this research is not based upon systematic

cross-cultural analysis of the actual elements of magico-religious practices or their relationship to social and cultural factors.

There

have been other studies utilizing cross-cultural samples which have related magico-religious variables to social variables (e.g., Whiting 1950, Whiting and Child 1953, Swanson 1960, D ’Andrade 1961, Textor 1967, Davis 1971, Roberts 1976, Murdock 1980 and Peters and Price-Williams 1980).

However, most of these studies are based upon

samples which are not well representative of world societies, or else have unassessed interdependencies within the population.

There is

some overlap in the magico-religious practices and beliefs assessed in these studies which suggest some consistent relationships between social conditions and magico-religious practices across the various studies.

However, these findings cannot be systematically integrated

since these studies are based on different samples and different variables.

Since the generalizations offered about magico-religious

practices have been based upon data derived from samples of convenience, they are not likely representative of the variation in human societies, and are therefore subject to serious uncontrolled biases.


23 This study overcomes these limitations on comparison, integration and generalization of some of these apparently related findings by assessing a wide range of descriptive characteristics of each culturally recognized type of magico-religious practitioner found in a subsample of the Standard Cross-Cultural Sample (SCCS) (Murdock and White 1969).

The SCCS is a stratified representative sample of the

world's societies, providing a basis for a more reliable assessment of the worldwide variability in magico-religious beliefs and practices, and their relations to social conditions.

The SCCS was carefully on

the basis of statistical assessment of the interrelationships among societies in the Ethnographic Atlas, and is therefore representative of the broadest possible sample of the world's societies.

Steps were

taken to select cultures with minimal extent of historical influences, diffusion, or common derivation.

The interdependencies of the

societies of the sample with respect to language, economy, political institutions, descent and geographical distance have been assessed (Murdock and White 1969) and analyses have been developed to control for these biases (see Methods 2.6).

Other investigators have already

coded over 700 social and cultural variables for the societies in the SCCS (e.g., see Barry and Schlegel 1980).

Utilizing a subsample of

the SCCS for this study provides the advantage of utilizing existing data on the social and cultural conditions of the societies used in this study; in the subsequent analyses the social complexity variables reported by Murdock and Provost (1973) have been utilized.


24 The sample utilized does have its limitations as well.

A

subsample of the SCCS has been utilized in this study; this smaller sample is not adequate to represent the range and variation in language groups, and the limited regional representation prevents analysis of regional patterns.

It is hoped that this study will be

extended to a larger sample of societies, something approaching Murdock's (1981) 150 cultural provinces.

Another major limitation of

the SCCS is that it is a synchronic sample, focusing upon each society at a given point in time, rather than assessing a given social system across time.

In the long run it would be desirable to employ a

diachronic sample to substantiate the principal findings of this study which indicate that there is a transformation and development of practitioner types as a function of social changes.

A more

substantial sample, as well as more research into ancient cultures is necessary to determine regional patterns and to separate hypotheses of independent invention from those of diffusion. The focuses of the SCCS has limited the scope of the findings. The SCCS generally focuses upon societies at a particular time in the historic past when some amount of information was present about the societies, but when the societies were presumably minimally influenced by the colonization from foreigners, particularly Western Europeans. This focus upon the historic past precludes a more detailed consideration of the wide variety of magico-religious practitioners found in modern Indo-European and other international cosmopolitan societies.

The practitioners in the sample do not include cases of

practitioners primarily involved in millenarian, nativistic, messianic


25 or revitalization movements and other religious changes and phenomena associate with industrialization and modernization of traditional societies (e.g., see Wilson 1973; Linton 1943; Wallace 1956). The analyses presented here are the initial phase of an ongoing project.

The findings presented here are offered tenatively,

recognizing some methodological limitations and the incomplete evaluation of the data collected in this study.

All of the results

presented here should be subjected to replication on a different sample of societies. The data used in this study were not subject to formal reliability checks, which are presently in progress.

Future publications will

provide data and reliability checks for the magico-religious practitioners considered in this study.

However, the consequences of

using the data with reliability checks are likely to be minimal for reasons discussed in the methodology section (2.9).

Briefly, the

coding is largely based upon the presence of characteristics, with their absence being a default category.

I feel that the major problem

with the reliability of the data is the issue of missing data, not miscoding of data.

Missing data has been adjusted for here by using

analyses which minimized the effects of missing data (see Methods 2.7). Another aspect of reliability has to do with the determination of magico-religious practitioner roles within each society.

This issue

is only peripherally addressed by the coding reliabilty checks being employed.

Although this has not been conclusively addressed in this

research, the considerations taken and the roles recognized are presented here (Appendix 1).

It is hoped that this will provide the


26 basis for scrutiny by specialists in the societies concerned, and result in the correction of any errors in recognition of raagicoreligions practitioner roles.

Several such errors have been

recognized in the recoding process and have been noted in Chapter 5. This research has addressed a limited range of questions about magico-religious practices.

The data collected on magico-religious

beliefs and practices of the people in the societies considered here is limited to that pertaining to magico-religious practitioner roles, and for the most part does not deal with magico-religious beliefs and practices accessible to the population as a whole, or the wider belief system within which magico-religious practices are explained,

while

this allows some specific research questions to be addressed, it is a limited focus, one of many different approaches which are necessary for an understanding of magico-religious phenomena.

Summary This thesis is directed towards several major research goals, including: 1) Determination of an etic typology of magico-religious practitioners, the characteristics of the different types, and the relationship of practitioner types to social variables; 2) Determination of the relationship of practitioner role selection to

the major magico-religious activities and the differences among

different types of magico-religious practitioners;


27 3) Determination of the types of trance states, the relationship of trance states to magico- religious practitioners, the psychophysiological characteristics of trance states, and the relationship of trance types to social variables; and 4) Exploration of a general perspective which suggests systematic changes in magico-religious practices as a function of social complexity.


CHAPTER 2 METHODOLOGY

A number of issues relevant to the construction of the research project and the analysis of the data are addressed in this chapter. The reader is provided with a general orientation to the processes involved in arriving at the data used for this project and the procedures used for selecting the SCCS subsample used in this study. The variables used in this study are briefly considered. This chapter also introduces some of the less common statistical procedures used in the analyses presented in the subsequent chapters.

2.1 Introduction. The process by which meaningful patterns are discovered and the process by which those findings are justified are generally quite different processes.

Generally the formal presentation of the data

focuses upon the explanation of the operationalisms and the justification of claims; most of the processes taken in the construction of data are frequently overlooked.

These considerations

are of essential importance because they constitute a generally unconscious foundation for the data.

This is not as problematic in

studies where only a few variables are being considered, as it is in this study where hundreds of variables are analyzed to arrive at a

28


29 final presentation of a few sets of variables.

The facts eventually

revealed are in a fundamental sense formed when the data or variables are created.

The variables chosen, or in this case the coding

categories employed, were central to the creation of the data and the paterns which are latter "discovered" in the analyses.

From this

perspective, the process of forming the data is more important than the analyses which are eventually employed to present structures already created. Since the processes undertaken in construction of the data are therefore of fundamental importance, records were kept of the process of creation of the variables so that this aspect of the investigation could also be presented.

However, the initial effort made to maintain

records of this process of creation of the variables quickly revealed that the process of observing, recording and reconstructing the discovery process was a task requiring demands at least as great as that of the primary investigation.

If a record were to be kept of

every step of the process of creating variables, deciding between alternative presentations of variables, determining what variables to delete and combine, and all of the reasons and justifications for the decisions, the documentation of that process could easily exceed hundreds of pages.

This is particularly true of this study where

hundreds of descriptive characteristics have been screened, coded, modified, combined, discarded, etc. in the process of arriving at a final set of variables.

Furthermore, the self-reflective process

could lead to an infinite regression of considerations.


Although a complete consideration of the process of creating the data was beyond the scope of this project, aspects of the discovery process have been presented which will hopefully serve to clarify the overall processes involved in arriving at these findings.

Rather than

providing a single section focusing upon documentation of the discovery process, several aspects of this process have been presented as a part of the overall presentation of the research.

The overall

process of construction of the research project is presented below, providing the general outlines of the considerations taken in constructing the data and a general overview of the major aspects of the process of developing and employing the coding schema.

The

culturally recognized magico-religious practitioner types which have been coded here have been specified in Appendix #1, along with some information about decisions made in arriving at a determination of culturally recognized practitioner types.

The order of presentation

of analyses roughly parallels the actual analyses performed on the data, allowing for a presentation of the data which more closely parallels the discovery process.

In order to provide a background for

the context of the investigations, some of the author's general theoretical orientations relevant to this research and the goals of the research were presented in the introduction.

This is to orient

the reader to the investigator's general expectations and orientation to the field of inquiry.

These expectations included that there would

be systematic cross-cultural similarities and differences in magicoreligious practitioners and the other explicit hypotheses presented in the introduction.


31 Another theoretical influence upon the author which is relevant to the area of inquiry but not considered here is addressed in other publications (Vinkelraan 1982, 1983).

These articles explore the

implications of parapsychological research for understandings of magico-religious practices.

The author has no reason to expect that

these perspectives have impinged upon the coding process in ways which have had a determinant effect upon the data and analyses presented here.

2.2 Determining Magico-Religious Practitioner Roles. The development of the coding schema used here (see Appendix #2) was guided in particular by the intention of determining cross-cultural similarities and differences in magico-religious practitioners.

Since the intent of this research was to discover

types of magico-religious practitioners, the focus of the investigation was upon magico-religious roles present in the societies in the sample, and the characteristics associated with or attributed to those occupying those roles.

A role is understood in the

conventional sense as that which defines the rites, duties and obligations of an individual performing a specialized function within the social group. Determining what constituted a magico-religious role in a given society was generally not problematic.

The society's, the

ethnographer's and the present investigator's recognition of a magico-religious role were generally the same.

The magico-religious

practitioner roles are apparently distinct in ways which were easily


32 recognized by both the ethnographers and the society under consideration since the recognition of a role appeared to have been unproblematic for the ethnographers.

The present investigators

recognition of magico-religious practitioner roles was generally only those (and all of those) implicitly or explicitly recognized as magico-religious practitioners by the ethnographers.

The few

exceptions (e.g., the Pentecosts) are discussed in Appendix #1. Each culturally recognized magico-religious practitioner role was individually assessed.

These roles included leaders of communal

rituals which were thought to involve interaction with supernatural beings, supernatural roles which were acquired on the basis of senority within the social group, and those roles which were generally attributed to an individual but not admitted to by the individual (e.g.,

witches and sorcerers).

However, those individuals who

function solely in the role of an assistant to other practitioners have not been coded separately, but their activities have been coded as those of the practitioner which they assist.

No speculation is

offered as to the overall characteristics of these assistants and how they might relate to the practitioner types arrived at below. Those magico-religious practices not enjoying an association with a distinct professional role, that is, the magico-religious abilities assessible to the general populace, have generally not been assessed. The healing practitioners whose knowledge and techniques are based in what is considered by the culture as unrelated to supernatural power (e.g., the Babylon physicians, called asu or ria-zu, and the Amhara surgeon-herbalist, called the woggesha) were not considered either.


33 However, there were two societies, the Mbuti and Siriono, which had no recognized magico-religious practitioners roles.

These societies were

assessed for the magical abilities and characteristics assessible to all members of the culture.

The Mbuti were also coded for the

characteristics associated with a men's society (molirao) which carried out magico-religious activities.

These non-practitioner cases have

been excluded from all of the analyses except the initial exploratory cluster analyses. The determination and identification of culturally recognized magico-religious roles was on occasion difficult.

Frequently sources

would fail to provide indigenous terms identifying a practitioner, but would refer to what was eventually presumed to be a single culturally recognized practitioner role by a variety of English terms (e.g., medicineman, witchdoctor, magician, sorcerer, etc.).

Even when there

were indigenous terms provided they occasionally varied between sources.

Frequently cultures/languages had a variety of names for

what was presumed to be a single role.

The process of deciding what

were culturally recognized practitioner roles frequently involved deciding which descriptions to group together as pertaining to a single practitioner role.

Appendix #1 provides the information used

to decide on culturally recognized roles.

Included in Appendix #1

are: 1) the practitioner type labels derived from the classification procedure findings of Chapter 3; 2) other labels used to refer to the practitioner in the text; and 3) indigenous terms for the magico-religious practitioner role.

The appendix also includes

material of general interest on some of the societies and


34 practitioners, particularly when there were coding problems or other peculiarities in the process. A number of cultures had practitioners with highly specialized traditions, where some practitioners may engage in activities not engaged in by other practitioners who are in most other respects identical.

For example, among the Trukese there are practitioners

which may engage in any or all of the specializations of magico-religious activities involving agricultural rites, divination, healing and malevolent activities.

This raised the issue of whether

to consider the different specializations as one magico-religious practitioner role or several.

These ambiguous situations were

resolved on the basis of selection and training characteristics.

If

selection and training conditions were the same, the different specializations were considered a single role.

The consequence was

that where there appeared to be highly specialized roles, they were recognized as a single role with specialization. The coding of a highly specialized practitioner roles as a single practitiioner role rather than several should have had minimal impact upon the classification of practitioner types.

This is because the

differences between the different specializations of a given practitioner role are limited to a few variables of the many (98) which are used to classify the practitioners.

Coding each

specialization separately would have merely proliferated instances of the same magico-religious practitioner type; all of the different specializations would most likely have been classified as the same type since they share most characteristics in common.


However, if the different specializations had an exclusive differentiation in the selection processes and training procedures for the practitioner role, the specializations were coded as separate practitioners (e.g., see Tanala).

Also, when there was no way to

unambiguously discriminate whether a particular individual practitioner pertains to one specialization of a role/type or another (e.g., whether a sorcerer or a shaman) and was generally thought to engage in both types of activities, the practitioner was coded as a single type and the role specialization was assessed as a variable. The discussion of malevolent practitioners in the introduction suggests that a single society may have more than one culturally recognized magico-religious practitioner role which engages in largely or exclusively malevolent practices (e.g., the Azande/Evans-Pritchard sorcerer vs. witch distinction).

However, no evidence was encountered

for two distinct practitioner types engaged largely or exclusively in malevolent practices within a single society in the sample used here. The practitioners in a given society which were characterized by members of their own cultures as engaging in largely malevolent acts and nothing else were coded as a single practitioner role.

2.3 The Sample. This study is based upon the data from societies in a subsample (26%) of the Standard Cross-Cultural Sample (SCCS) (Murdock and White 1969).

The sample used in this study was arrived at by first

selecting every other odd number society in the SCCS (every fourth society) as the initial sample.

Since the societies are ordered on


36 the basis of geographical proximity, this sample provides a group of societies which is presumably well representative of the SCCS.

Of

this initial sample of 47 societies information on magico-religious practitioners was unavailable or inadequate for 2 of the sample societies the

Hadza and the Tuehueleche.

The Hazda have been

replaced by the !Kung Bushmen, who are in the same cultural province (Murdock 1981).

None of the indigenous South American societies in

the sample had an economy based on extensive agriculture.

The Mapuche

were substituted for the Tuehuelche, since the Mapuche are a indigenous South American society with extensive reliance upon agriculture as a mode of subsistence.

The 47 societies used as a

basis for this study are shown by geographical region in Table #1.

TABLE #1 SOCIETIES IN SAMPLE AFRICA Nama Hottentot Kung Bushmen Ovimbundu Mbuti Ibo

EURASIA Samoyed Toda Kazak Garo Vietnamese Semang Tanala Japanese Chukchee

NORTH AMERICA Montagnias Kaska Twana Paiute Hidatsa Creek Zuni Aztec

CIRCUM-MEDITERR Wolof Fulani Fur Kafa Amhara Tuareg Babylonians Romans Kurd

INSULAR PACIFIC Iban Alor Kimam Lesu Pentecost Marquesans Trukese Atayal

SOUTH AMERICA Bribri Callinago Saramacca Jivaro Slriono Tupinamba Cayua Mapuche


Two of the societies in this sample had no practitioners (Mbuti and Siriono.

These groups were among the simplest in terms of

measurements of social complexity, and show evidence of deculturation.

The Mbuti, descendants of the pygmies, had adopted a

Sudanic language and developed a symbiotic relationship with their Negro neighbors.

The Siriono were a starving hunting and gathering

group, who lacked magico-religious practitioners typical of the other societies in the Equatorial language group (see Stearman 1984 on Siriono deculturation).

Since these cases appear to be special

exceptions rather than the rule, and since there was not a sufficient basis to evaluate the hypothesis of deculturation effects upon magicoreligious practices, they were excluded from the analyses here.

2.4 Development of the Coding Format Prior to formulating a coding format, ethnographic materials covering magico-religious practices of approximately 30 societies chosen on the basis of convenience and representative of the major regions of the world were reviewed and notes taken on the central characteristics of the practitioners and their activities.

The

characteristics of the magico-religious practitioners found in these societies were reviewed and organized into common categories (e.g., selection procedures, magico-religious activities, political/social roles, trance induction techniques, types and sources of power, etc.).

This initial schema was then applied to the coding of several

societies.

The societies reviewed in the formation of the coding

categories and those initially coded did not serve as a basis for the sample used in this study.


After this initial phase of development of the coding format, the coding schema was then applied to the assessment of the practitioners in the societies in the sample.

However, the process of modification

and development of the coding format continued.

Variables were added

as other salient characteristics were noted; some characteristics were grouped in large variables constituting "checklists" of characteristics (personality characteristics, types of techniques, types of trance induction procedures).

This led to an enormously

cumbersome set of overlapping variables, which required that the codes be synthesized and reduced.

Codes which proved to be rare or not

consistently reported (e.g., certain descriptions of practitioner during trance states) were removed, and many overlapping variables (e.g, sources of power, acquisition of power, types of power) were condensed into multiple category variables. The coding schema continued underwent continued revision in an effort to reduce the coding format to a smaller number of central variables and to provide clarification of the variables.

One of the

continued efforts was to derive a smaller set of variables which as adequately assessed the characteristics of the magico-religious practitioners.

This effort was seen as attempting to develop

variables which represented repeatedly observed regularities in the practitioners, in some sense, the categories inherent to the practices.

This process of revision occurred 12 times in the course

of coding the 47 societies of the sample, requiring that the entire sample be recoded at the termination of the first complete coding. During this process of coding a set of coding instructions (see


39 Appendix 3) were developed to clarify the variables which were being coded.

In the process of coding an extensive search was made for

sources providing information on magico-religious practices.

Sources

utilized included the Peabody Museum Catalogue, bibliographic searches, HRAF files and bibliography, as well as the SCCS bibliography and the citations present in the already completed studies utilizing the SCCS (see Barry and Schelgel 1980).

The newly

acquired sources were frequently the most important sources on magico-religious practices, and are cited in Appendix #1. When the coding had been completed, the coding format included approximately 360 variables, including the multiple codings for repetitive variables such as the altered state induction procedures associated with the different magico-religious activities.

The data

was subjected to a series of analyses to help reduce the extremely large set of variables prior to beginning code reliability checks. The data focusing upon altered states of

consciousness was then

analyzed with cluster analysis and entailraent analysis procedures. This initial analysis indicated that two major psychophysiological variables, convulsions and unconsciousness, stood in an exclusion relationship to one another.

This strong psychophysiological

relationship prompted a review of psychophsyiological research to develop a psychophysiological model which would help organize the trance data.

The model presented in the chapter on Training and

Trance States was eventually formulated. The remaining data was then subjected to a wide variety of exploratory analyses (cluster analyses, factor analysis, entailraent


40 analysis) designed to reveal the general structure and interrelations of the data, to generally confirm the hypotheses which had generated the research, and to assist in the consolidation of redundant variables.

These analyses were directed by the research goals

specified above, and proceeded in much the same manner as the analyses presented in the subsequent chapters.

The implicitly recognized

groupings of the magico-religious practitioners were first ascertained by cluster analysis of different subsets of the variables.

Because of

program limitation on number of cases by variables, a subset of 70 variables was selected and used in BMDP2M cluster analysis of cases (Dixon and Brown 1979).

This revealed a structure of the data similar

to that revealed in the cluster analysis of cases presented below. However, this analysis was based upon the use of a measurement procedure which was inappropriate for the data, and did not as clearly reveal the results presented here.

The relationship of this

practitioner type data to social complexity variables was briefly explored, but not in the extensive manner reported in Chapter 5. A number of other exploratory analyses were carried out.

These

included an examination of the relationship between divination procedures and social variables.

The 16 variables originally used to

characterize divination procedures were synthesized into 5 basic procedures, and analyses were performed revealing relationships between divination and social variables similar to those outlined in chapter 7.

The association of psychophysiological states and

induction procedures with social complexity variables was explored as


41 was the interrelations of the groups of variables assessing political powers, magical abilities, and

magical techniques.

These analyses served to clarify and consolidate variables, to confirm and modify the general assumptions about the nature of the data, and to indicate areas in which the assessment of practitioners was weak.

About 10 months after the completion of the coding the

process of recoding the variables was begun, implementing the variable clarification suggested by the initial analyses.

At this time code

checkers were trained to assess the reliability of the data coded here.

However, self imposed deadlines for completion of the

dissertation could not be met if the entire sample were to first be recoded and then checked for reliability prior to analysis.

A reduced

set of variables was computed from the data originally coded, implementing to the extent possible the categorical changes in coding and syntheses suggested by these initial exploratory analyses.

The

societies which had missing data in the initial phases of analyses were coded and included at this time.

The original data was reduced

to a set of 98 variables through the combination of variables and the elimination of those variables very low in frequency, of little theoretical intrest, or lacking clarity because of ambiguous coding instructions.

Reduction to this number of variables was determined by

the limitations (n=100) of the program used to calculate similarity. This new set of data was computed from the original data with the SPSS programs (Nie et al. 1975).

This output was then compared variable by

variable against the original coding sheets in order to eliminate errors which resulted from earlier data punching errors or from


42 inappropriate combinations of variables into the new set of variables.

Since the original coding had included considerable notes,

this provided a basis for ascertaining that the new codes were in fact accurate in their characterization of the original data.

These

variables (see Appendix 2) provide the basis for the analyses presented here.

2.5 Variables. The variables used as the basis for this research were chosen with the intention of assessing all of the major areas which could be considered to include principal characterizations of magico-religious practitioners.

Mythological explanations are the only major areas of

characterization not extensively considered in the codings.

The only

variable areas considered in this coding process and not integrated into the set used in the analysis here were those characterizing the trance induction procedures and trance characteristics associated with magico-religious activities other than training.

These were excluded

from the reduced set of variables used here on pragmatic groundsprogram limitations of the number of variables which could be considered simultaneously in the calculation of similarity (see 2 .1 ). The trance variables associated with magico-religious activities were eliminated since they constituted an easily identifiable and well circumscribed large group of variables (n > 30) which were frequently repeated since there were different trance procedures associated with different activities.

This was not considered to be a complete loss

of relevant data on the practitioners since the trance states


43 associated with training were assessed, and were generally similar to the trance states used by the practitioners in their magico-religious activities. Since the variables and the coding instructions are considered extensively in appendices #2 and #3, they are not presented extensively here.

However, a brief overview of the major variables

areas is provided at the end of this section so as to advise the reader to the range of characteristics assessed.

The reader should

refer to the appendices for a more complete description of the variables.

These variables do sample the entire range of variables

relevant to a characterization of similarities and differences in magico-religious practitioners.

Although it is not a complete

assessment of potentially relevant variables, it is expected that the wide range of characteristics assessed has provided a sufficient basis for an accurate and stable classification of magico-religious types.

The finding that the empirically derived typology of Chapter 3

corresponds so strongly to the social complexity variable (see Chapter 5) strengthens this argument.

However, it is certain that the

classification would change if certain subgroups of the data were used rather than the complete data employed here. In the major analyses of the data there has been no distinction made between different types or categories of variables, for instance those measuring social status and those assessing beliefs that a practitioner transforms him/herself into an animal.

Presumably most

anthropologists would want to consider variables assessing social status and those assessing a belief in a transformation into an animal


44 to be of different ontological statuses.

The ethnographer’s

assessment that a practitioner was of high social status would likely be considered as representing an "objective physical and social reality", while the statement that "the practitioner transforms himself into an animal" would be considered a mythological belief. The transformation into an animal is considered a cultural belief lacking an actual process in physical reality; such an assessment would likely persist even if the ethnographer reported seeing or undergoing such a transformation. Concepts/variables such as social or economic status or political power seem objective, and verification or consensual validation of the accurateness, appropriateness and objectivity of the assignment of particular values along those variables to a particular individual appears to be relatively unproblematic.

But the concepts of social

status or political power are not denotative.

The terms form part of

systems of conventional understanding which we have tacitly agreed upon.

Status is not present in physical reality.

It is a construct

of mind and culture, conventionally agreed upon as referring to differences we tend to presume that we commonly perceive.

As with

beliefs in animal transformation, they are cultural constructs we have learned to perceive in order to express our collective experiences of the world. Whether or not one wants to assume the ontological equivelence of such variables is immaterial to this work.

However, no

differentiation is made between variables which are presumed to be accepted as physical facts or pertaining to objective reality and


45 those variables which are thought to reflect mythological beliefs rather than physical realities.

Although separation into types of

variables would likely be a productive area of inquiry, it is not addressed here. The lack of differentiation between such variables should not be seen as an implicit or explicit assertion that the present investigator intends that variables such as those assessing a culture's belief in humans transforming into animals be treated as representing an objective physical reality.

Such variables, like all

others, have to be treated as reflecting a culture's beliefs, and may not have analogues in the physical world.

References to spirits,

gods, supernatural power, animal transformation, etc. should be interpreted as referring to cultural beliefs held by the members of the societies under consideration, not as assertions by the author as to the objectivity of the entities or phenomena to which such beliefs subscribe.

The same qualification is extended to the use of terms

referring to magico-religious activities.

Reference to healing,

divination, or a malevolent magical act against another does not imply that healing, prediction of future events or harm to another human has actually resulted from the acts; it means that members of the society considered the activities in which the practitioner engaged to have that particular goal. in all cases, variables represent: 1) the present investigators perception that 2) the ethnographer(s) reported that certain beliefs, practices or characteristics were a) attributed to the magico-religious practitioner by the culture, or b) inferred by the


46 ethnographer as present and accurately characterizing the practitioner.

In all cases, the information about magico-religious

practitioners is filtered through the ethnographer's and present investigator's cognitive systems.

2.51 Variable Areas. Socio-Political Assessments:

Presence of political/legislative,

judicial, military and economic power, economic and social status, degree/type of role specialization, presence of practitioner group, extent of joint practitioner activities, participation in life cycle activities, sex restriction on access to role, moral assessment of practitioner by culture, psychological characteristics.

Context and Motive of Activities: These variables assess aspects central to Durkheira's distinction between magic and religionindividualistic versus collective; and Titiev's (1952) distinction between critical and calendrlcal rituals- whether activities occur in response to unfavorable situations or as a part of a seasonal annual ritual determined by an exact or approximate period in the solar cycle.

Context of activities: a) public, b) client group, or c) private.

Practitioner's motivation for activities: a) client request, b) social/calendrical function, c) personal reasons


47 Magico-Religious activities Intending: Healing- response to immediate illness Protection- activity to preclude future illness or calamity Divination, food acquisition, increase/protect agriculture productivity, propitiation/worship of spirit entities, malevolent acts- causing harm to others, weather control. Techniques Employed; Rubbing/massage, physical care (washing, simple surgury), herbal remedies, spells, discharge of power, manipulative, imitative and exuvial techniques, charms, exorcisms, spirit control, sacrifice. Power Relationships: Animal relationships, spirit relationships, awareness and control of power, types of power, mode of acquisition of power. Practitioner Selection: The selection characteristics are not exclusive; a given practitioner role in a society may have several different possible selection procedures for a role or may require several criterion be met [e.g., an ancestor who was a practitioner (social inheritance) and vision quest or involuntary visions]. Types of Selection Procedures: Social labeling- role is attributed to individual and generally denied by the accused. Biological inheritance- social labeling in which all genetic descendants of practitioner are thought to acquire role through model similar to genetic inheritance. Illness- individual selected because of illness or in order to cure Involuntary visions indicate selection


48 Spirit selection- experiences interpreted as intervention by spirits who require that individual seek role Vision quests Voluntary self selection Social inheritance- tendency for descendants of practitioner to acquire role Social succession- a single descendant of practitioner acquires role Political appointment/action involved in acquiring role. Training: Spirit training, training by individual practitioners, training by formal group. Altered State Induction in Training:

Alcohol, sensory isolation,

sleep, sleep deprivation, hallucinogens, fasting, dancing, behavioral inhibition, auditory driving (e.g., drumming, chanting, singing). Altered State Characteristics:

Tremors, convulsions, collapse,

unconsciousness, motor behavior, spontaneous seizures, amnesia.

2.52 Social Complexity Variables. The assessment of the relationship between magico-religious practitioners and the social variables which have been considered in this study have utilized the data on social variables provided by Murdock and Provost (1973).

The variables of major concern in this

study are briefly presented below. been used in the text.

The following conventions have

When the social complexity variables have been

used exactly as coded by Murdock and Provost (1973), they have been referred to in the text with capitalized letters (e.g., Political Integration).

When recodes of the variables provided by Murdock and


49 Provost have been used (e.g., low versus high levels of agriculture), the variables have been referred without capitalization (e.g., "the binary recode for agriculture").

Fixity of Residence: Settlements are sedentary and permanent Settlements are sedentary but impermanent Settlement pattern is semisedentary Settlement pattern is seminomadic Settlement pattern is fully nomadic Agriculture: Major source of food supply and conducted by intensive techniques Major source of food supply but not conducted by intensive techniques Contributes more than 10% of food but less than other sources Present but contributes less than 10% of food supply Absent or confined to non-food crops Technological Specialization: Craft specialists including smiths, weavers and potters Metalworkers or smiths but no loom weaving and/or pottery Loom weaving but no metalworking Pottery but no metalworking or loom weaving Metalwork, loom weaving and pottery absent or unreported Density of Population in territory controlled by society: Mean density exceeds 100 persons per square mile Mean density averages between 26 and 100 persons per square mile Mean density averages between 5.1 and 26 persons per square mile Mean density averages between 1 and 5 persons per square mile Mean density averages less than 1 person per square mile Level of Political Integration: Three or more administrative levels above the local community, e.g., state organized into provinces and subdivided into districts Two administrative levels above that of local comunity One administrative level above the local community Stateless society with politically organized local communities Stateless society without centralized political authority Social Stratification: Three or more distinct classes or castes Two classes and hereditary slavery or castes Two classes but no hereditary slavery or castes No Formal classes but hereditary slavery or important status differences based upon wealth Egalitarian, lacking classes, castes, hereditary slavery and important wealth distinctions


50 2.6 Sampling Unit Interdependence and the Autocorrelation Method If one is concerned with obtaining general laws about human societies, the population is all human societies which have existed. We are obviously unable to obtain data on most human societies which have existed, and are therefore restricted to those societies which have been observed and recorded.

Since the samples available are

necessarily not representative of all human societies which have existed, the inferences which can be drawn are technically limited to the sample studied or the actual population from which they were drawn.

One has to recognize the inherent limitations present in the

available data and make efforts to compensate for the shortcomings. One important reason for using a random sampling is to obtain what is presumed to be a representative sample.

However, a stratified

probability sample is a much more efficient manner of obtaining such representativeness.

Another more important reason for selecting a

random sample is its usefulness in obtaining independent observations.

The independence of observations, or the lack of

correlation in the error residuals, is necessary for the validity of estimates in correlational analysis (White, Burton and Dow 1981). However, random sampling does not guarantee such independence. The interdependence of sampling units is a longstanding problem facing formal cross-cultural research.

The importance of independence of

observations lies in the fact that if observations are associated, measures of association can be spuriously inflated by reduction in variance which results from what amounts to repetition of the same case.

If tests of probability are to be used to evaluate the


51 significance of joint occurrences of variables as a basis for inference about causal processes, the sampling units need to be indepedent or the variance of the correlations will be underestimated, leading to spurious levels of significance (White, Burton and Dow 1981). Murdock and White's (1969) research shows that there is considerable interdependence even among samples selected with the intention of reducing such interdependence.

Furthermore, their

assessment of the interdependence indicates that a worldwide sample which would not have significant interdependence with respect to such basic variables as language, economy, political integration and rules of descent would be so small (n=15) that it would prohibit most statistical analyses.

The problem of how to control for sample

interdependence has been addressed with several different methods (e.g., see Naroll 1961, 1964; Naroll and D 1Andrade 1963), but all have serious shortcomings.

More recently White, Burton and Dow (1981) have

utilized a solution, known as the autocorrelation method, which overcomes the limitations of the previous solutions (see also Dow, White and Burton 1983). The autocorrelation method assesses the statistical inter relatedness of societies on the basis of some known measure such as language similarity or geographic distance.

The known

patterns of

relatedness can then be used to predict a societies' score on some particular variable of interest on the basis of the society's autocorrelation within the sample of societies considered.

This

predicted value is then subtracted from the societies actual score on the variable.

This provides a set of residual scores which can be


52 considered independent of the effects of diffusion as assessed by the particular measure of relatedness used (language or geographical distance).

Correlations between the residuals and the independent

variables of interest (or their residuals from the autocorrelation procedure) are independent of the effects of diffusion, and overcome the major objections raised to the validity of cross-cultural sampling and research. The autocorrelation controls established here utilized procedures developed by Dow, Burton, White and Reitz (1984), and provided in an interactive program made available by Dr. Michael Burton, School of Social Sciences, University of California, Irvine.

This program

provides controls based upon language affinity and geographical distance.

The distance measurement utilized an exponential decay

function of .05, which maximizes the assessment of diffusion and relatedness within about 300 miles.

These autocorrelation procedures

allow for prediction of both the unbiased estimates of the effects of the explanatory variables, and the effect of the historical or diffusion model in the context of a linear regression model (see also White, Burton and Dow 1981; Dow, White and Burton 1983). Jorgenson (1979) points out that causal inference from correlations with synchronic data are not warranted without autocorrelation controls for geographical propinquity, environmental similarity, language relatedness and time.

The autocorrelation

methods used here substantially met these criterion with controls for language similarity and geographical propinquity.


53 2.7 Measurement of Similarity- Gower's Coefficient. Since the variables used here were binary, qualitative and ordinal, the usual measures of similarity were inappropriate, since they assume continuous data.

Similarities among cases were therefore

computed with a fortran program implementing Gower's Coeficient* (Gower 1971), which was provided by Dr. Roger Blashfield, Department of Psychiatry, University of Florida.

This measurement program allows

for the determination of similarity of binary and qualitative data on the basis of shared characteristics rather than upon the distributional properties of the data.

This program also assesses the

similarities of cases on the basis of quantitative or ordinal data by basing the assessment on the difference between the two values divided by the range of the variable.

This program combines the measures of

similarity as computed for the different types of variables and combines them into a single measure. Another advantage of this program was that the assessment of binary data does not contribute to the measure of similarity between two cases on the basis of their conjoint absence on a variable.

When

variables are treated as binary, conjoint presence contributes to the

* "Two individuals i and j may be compared on character k and assigned a score sijk, zero when i and j are considered different and a positive fraction, or unity, when they have some degree of agreement or similarity. The similarity between i and j is defined as the average score taken over all possible comparisons: v v Sij = £ sijk/ X k=l k=l

£ijk"(Gower 1971:858-9).


54 measure of similarity, and absence for one case and presence for the other contributes to a measure of (dis)similarity.

However, if the

value is absent for both cases, the variable does not enter into the measure.

In the major assessments of similarity, the binary variable

option was used with variables which had shown significant correlations with data quality assessments in the initial analyses and variables which had very high or very low frequencies; this allowed for a minimization of the effect of missing data upon the measurement of similarity. 2.8

Other Statistical Procedures. In this section several of the statistical tools used in the

subsequent chapters which are not very common are briefly presented. 2.81 Entailment Analysis Entailraent analysis involves procedures for determining implicative relationships among variables, or tendencies towards such set-subset relationships among binary variables.

The purpose of

entailment analysis is to determine if the presence of some attribute logically implies the presence of another attribute.

Entailment

relationships may be pure logical relationships without exceptions, or they may be statistical relationships with exceptions; the latter are referred to as material entailment relationships (see Vhite and McCann 1981).

A material entailment requires a positive statistical

correlation between the variables, and allows a limited number of exceptions. Entailment analysis takes the contingency tables for all pairs of variables being considered, selecting those which are conditional (A


implies B), or conditional with a specified level of exceptions. These conditional relationships are assembled into into chains of conditional relationships (entailment chains), which are then checked for transitivity (If A implies B and B implies C, then A implies C within an acceptable level of exceptions).

The weak entailment

relationships are then removed to make the entire entailraent structure transitive.

The overall significance of the structure can then be

assessed by comparing the entailments with those expected by chance. The final implicative structure can then be diagramed in an entailogram (see White, Burton and Brunder 1977).

In the analyses

reported here, the default parameters of the entailraent analysis program (SOC file, UCI Computing

Center) were used with, maximum

exceptions lowered to .2, the signal/noise ratio increased to 1.5 and the minimum relevance increased to .2.

2.82 Quadratic Assignment Program In several cases there was the interest in determining which of two statistical representations of data was most efficient in accounting for the data in order to assess hypotheses.

In this study

the comparisons were between cluster analysis representations and multidimensional scaling (MDS) representations.

If the MDS solutions

are significantly better than the cluster analysis solution in reprsenting the data, this indicates that the differences in the data are not those which are best represented as distinct clusters or types.

The greater success of the MDS solutions indicate that the

differences be seen as continuous gradations rather than discrete.

R eproduced with permission o f the copyright owner. Further reproduction prohibited without permission.

56 In order to determine how well the representations accounted for the data, the cluster analysis representations and the multi dimensional scaling representations were transformed into a matrix of differences.

To obtain these matrices, the coordinates from the MDS

solution were run through a program which computed a matrix from the interpoint distances.

The amalgamation order of the cluster analysis

solutions were used in a similar manner to construct a matrix of dissimilarities.

These matrices representing the MDS and cluster

analysis solutions were compared with the original data matrix, using gamma as a measure of association. If the gamma values suggest differences in the success of the different solutions in representing the original data, the solutions can be compared to determine if the differences are significant.

The

residuals of the solutions can also be compared with the original data to determine if there is a significant portion of data which is not accounted for.

In order to make these comparisons, the Quadratic

Assignment Program (QAP) (Hubert and Schultz 1976; Hubert and Golledge 1981) was employed.

QAP is a set of data analytic techniques which

compares two matrices and provides an index similar to a correlation coefficient between the elements of two matrices. In order to make these comparisons, the matrices constructed from the cluster analysis and MDS models were run through a program which standardizes two matrices and substracts them from the standardized form of the original data matrix.

The residual output of this program

is then entered into the Quadratic Assignment Program in order to compare it with the original data matrix and to determine if there is


a significant amount of the original data that is explained by the residuals.

The comparisons provide a Z score for the association of

the residuals, with the original data.

If there are significant

differences between different solutions, or if a significant ammount of the data are not explained by the solutions, this will be indicated by a Z score of significant magnitude.

In order to determine if

representations leave significant portions of the original data unaccounted for, the matrices representing the solutions can be subtracted from the original data and the residuals compared with the original data with QAP.

Again a Z score provides the measure of

association; if there is no significant association between the residuals and the original data, one can assert that the solution has accounted for all of the data. 2.83 Profit-Identifying MDS Dimensions. When multidimensional scaling (MDS) representations of objects are used, it is desirable to determine how the organization of the objects in the space is related to the variables which characterize the objects.

In this study, in order to determine the relationship

between the practitioners and the variables used to construct the similarities among practitioners, the Profit program of the MDS(X) series (Chang and Carrol 1968) was used.

Profit uses as input the

coordinate points from the MDS representation and a set of measures which are properties associated with the entities represented by the coordinates.

Profit performs a "regression on optimization of linear

(or non-linear) fit" for the properties in the coordinate space by calculating a property vector in the coordinate space such that the


projections on the stimulus points (objects/practitioners) are minimized.

This optimized projection is expressed as a measure of

association rho, the value of which is similar to the multiple linear regression correlation coefficient from the regression of the property vector on the coordinates which represent the stimulus points (practitioners).

This measure of correlation of the properties with

the coordinate space can be used to determine which of the variables are most strongly associated with the coordinate space’s representation of the differences and similarities among practitioners.

The projection point of the property vector onto the

coordinate space allows for an identification of the major dimensions of the representation of the practitioners.

Variables with high

correlations and aligning closely with the axes of the coordinate space can be said to be the central characteristics underlying the differences and similarities characterizing the practitioners.

2.9 Data Reliability. The findings presented here should be accepted as tentative for several reasons.

One is that there have been no checks for coding

reliability on the data used here.

However, given the nature of most

of the data, it is expected that data reliability checks will have minimal effects upon the overall findings reported here.

This is

because most of the original coding of data involved coding for the presence/absence of a certain characteristic, not deciding between values on a scale or distinguishing between the most appropriate characterization, where more discriminating judgement is called for.


59 The issue of whether an item was reported as present was seldom problematic, and the author has no reason to expect that there are any particular kind of motivated coding errors.

However, such biases are

not necessarily readily assessible to consciousness. What was occasionally problematic was whether the data was truly absent or if there was a failure to report the phenomena.

In the

initial set of cluster analyses the data was analysed in two different ways, one which minimized the effect of missing data on the assessment of similarities among practitioners.

An informal comparison of the

analyses utilizing the assessment which minimizes the effect of missing data with another assessment not designed to compensate for missing data shows that the former provides a clearer resolution in the cluster analysis solution of practitioner types.

Although there

was missing data due to the low data quality of some sources, it is expected that improvement in the coverage will not substantially change the overall pattern of the results since the analyses sensitive to missing data have provided satisfactory solutions to the classification problem and apparently overcome the distortion which could have been caused by the missing data.


CHAPTER 3 CLUSTER ANALYSIS OF PRACTITIONERS

Cluster analysis procedures were used to determine the different types of magico-religious practitioners.

All of the variables

reported in Appendix #2 were used in the analyses.

The similarity

among cases was determined through the implementation of the Gower's Coefficient procedure described in the methodology section.

The lower

triangular matrix output from the Gower's Coefficient program was transformed into a full matrix with ones on the diagonal and read as input data into the BMDP1M Cluster Analysis program (Dixon and Brown 1979).

The similarity among practitioners was explored under several

different measurement conditions in order to determine if stable clusterings would be found under different conditions.

Analyses were

performed under single, average and complete link amalgamation rules. Several cases not present in the classification presented below were included in the initial analyses.

These cases included the general

access to magico-religious abilities present in the two societies which did not have magico-religious practitioners present (Mbuti and Siriono) as well as the codes labeled the Mbuti Male Society, the Paiute Sorcerer and the Toda Malevolent Practitioner.

The latter two

cases were excluded from the final classification because it was

60

R eproduced with permission o f the copyright owner. Further reproduction prohibited without permission.

61

determined that these codings represented specializations of the roles of the Paiute Shaman and Toda Healer, respectively (see Appendix #1). In the initial set of cluster analyses, the similaritiy among practitioners was determined under two different measurement conditions.

One computation of similarity considered all of the

variables as continuous (quantitative measurement condition), while another divided the variables into continuous, qualitative and binary variables (mixed variables condition).

In the mixed variables

condition, the variables with more than two values (n=42) were treated as quantitative.

Twenty variables which had correlated significantly

with data quality assessments in the exploratory analyses or which were very low or very high in frequency were treated as binary variables in order to reduce the influence of missing values on the measure of similarity.

This reduced the effect of missing data

because conjoint absence did not contribute to the measure of (dis)similarity.

The remaining

variables (n=36) were treated as

qualitative measures, equally weighting presence and absence. The solutions provided under the quantitative measurement condition provided a larger number of small groups which were not clearly organized into hierarchical structures or clearly separated into larger groups.

However, the solutions based upon the mixed

variables condition, which was designed to reduce the effect of missing data, generated a few larger clusters which were more distinctly separated.

Since this consistently occurred under

different amalgamation rules (complete and average), the mixed variable measurement was used for all subsequent analyses.


62

The different solutions considered suggest that there are four major types of magico-religious practitioners, with one type divided into three subtypes.

The solutions which were relied upon in arriving

at this classification are the average and complete link solutions based only upon the practitioner cases (Mbuti and Siriono codes excluded; see Figures #3 and #4 below).

The classification derived

from these cluster analysis solutions is presented in Table #2.

The

labels used to identify the different magico-religious practitioners types were suggested by the names frequently employed by the ethnographers to refer to the practitioners which were included in the type. The labels applied to the magico-religious practitioner types arrived at in the classification are: Priests Mediums Malevolent Practitioners Healer Complex:

(Core) Shamans, Shaman/Healers, and (Core) Healers.

The material which immediately follows briefly discusses the resolution of the differences across different cluster analysis solutions and the procedures employed in the validation of the classification.

The subsequent section provides a more detailed

consideration of the different solutions examined. Although there are some differences between the various solutions provided by cluster analysis, similar and stable grouping are present across different amalgamation rules, with few cases shifting between different clusters under different solutions.

The differences between


63

TABLE #2 PRACTITIONER TYPE CLASSIFICATION CORE SHAMANS Kung Num Master Samoyed Semang Chukchee Montagnias Kaska Twana Paiute Callinago Creek Shaman Creek Priest Jivaro Cayua Tupinamba SHAMAN/HEALER Nama Shaman/Healer Fur Magician Roman Eastern Cult Roman Sorcerer/Witch Kurd Dervish Japanese Ascetic Dyak Shaman/Healer Kimam Shaman/Healer Pentecost Bwari Hidatsa Shaman/Healer Zuni Shaman/Healer Aztec Shaman/Healer Bribri Shaman/Healer

MEDIUMS Ibo Curer-Medium Fulani Medium Kafa Ekko Amhara Zar Wolof Medium Babylonia Oracle Kazak Shaman-Medium Vietnamese Dong Japanese Miko Japanese Founder Toda Medium Tanala Medium Trukese Medium Marquesa Medium Ataya1 Shaman-Medium Alor Seer-Medium Saramacca Medium MALEVOLENT Ovimbundu WoloE Fulani Fur Kafa Amhara Tuareg Babylonia Kurd Garo Japanese Tanala Alor Marquesa Creek Zuni Aztec Saramacca Mapuche

HEALER Ovimbundu Healer Ibo Oracle Fulani Healer Fur Puggee Amhara Scribe Tuareg Marabout Babylon Exorcist Vietnamese Thay Garo Healer Tanala Healer Lesu Healer Truk Healer Marquesa Ceremonial Priest Saramacca Obia Toda Healer Jivaro Priest Mapuche Healer

PRIESTS Ovimbundu Chief Ibo Sac. Priest Ibo Rain Priest Wolof Shiek Fulani Guard Fur Rain Priest Fur Sultan Amhara Priest Tuareg Chief Babylonia King Babylonia Soothsayer Roman Pontiff Roman Augur Roman Paterfamilias Kafa King Toda Dairy Priest Kazak Headman Garo Headman Kurd Mulla Vietnamese Ancestor Vietnamese Cult Vietnamese Priest Japanese Headman Dyak Priest-Diviner Alor Priest Pentecost Priest Marquesa Insp. Priest Atayal Priest Tanala Headman Creek Rain Priest Zuni Katchina Zuni Bow Priest Zuni Rain Priest Aztec Priest Bribri Priest Mapuche Priest Saramacca Headman


64

solutions involved: 1) a different order of amalgamation between groups; 2) a shifting of a few problematic cases between groups; and 3) the formation of small anomalous clusters.

These differences

occurred as a result of different amalgamation rules, or the inclusion or exclusion of certain cases, particularly the non-practitioner cases (Mbuti and Siriono). The practitioners which were associated with different clusters in the different solutions were classified with the application of a few simple rules.

The practitioners which formed the small anomalous

groups in some solutions were classified with the larger groups with which they were clustered in other solutions.

Most other differences

were resolved by placing the practitioners in the group with which they were most consistently classified across solutions.

The

inconsistencies with respect to the practitioners classified in each group were dealt with as follows:

Priests-

The Marquesan Inspirational Priest, the Creek Rain Priest,

and the Ibo Rain Priest were the only cases classified in the Priest type which were not clustered in the Priest cluster in all of the cluster analyses.

The four practitioners clustered with both the

Priests and the types of the Healer Complex (Toda Healer, Ibo Oracle, Jivaro Priest, Roman Cult) were classified with the Healer Complex type with which they were clustered.


65

Mediums-

The only practitioner in the Medium group not consistently

clustered there is the Japanese Miko.

When the Japanese Ascetic is

excluded from the analysis, the Japanese Miko is clustered within the Medium group rather than as an outlier to the Medium or Shaman/Healer clusters.

Since the the Japanese Miko and Japanese Ascetic were more

similar to each other than they were to any other practitioners, they amalgamated first, and therefore distorted each others1 relationship to the practitioner type in which they were finally classified (Medium and Shaman/Healer, respectively).

Malevolent Practitioners- The Garo Malevolent Practitioner is the only practitioner classified as Malevolent Practitioner which was not clustered in that group under all cluster analysis solutions.

The

Creek Rain Priest and the Roman Sorcerer/ Witch were clustered with the Malevolent Practitioners under one solution, but were classified with the practitioners with which they were clustered in the other solutions (Priest and Shaman/Healer).

The Ibo Rain Priest was also

clustered with the Malevolent Practitioners in some soultions and with the Priests in others; it is classified in the Priest group on functional grounds.

Healer Complex-

The clustering of some of the cases into the Healer

or Shaman cluster varied under the different amalgamation rules, or as a function of the inclusion of certain non-practitioner cases.

The

differences between solutions with respect to the Healer Complex were resolved by creating a transitional group which included those cases


66

which were occasionally clustered with the Shamans and occasionally with the Healers.

Those practitioners consistently clustered with the

practitioners primarily labeled as shaman across the different solutions are labeled Core Shamans; those practitioners in the Healer Complex which are never found in the cluster with the Core Shamans are labeled Core Healers; those cases which varied between the Core Shaman group and the Core Healer group were classified as Shaman/Healers.

3.1 Validation of Cluster Analysis Classification Since the bulk of analyses which follow depend upon the classification arrived at here, consideration of the validity of the solution is important, especially given the few anomalous cases in the cluster analysis groupings.

Gordon (1981) points out that

classification analysis has been largely considered as a form of exploratory data analysis, and as a result there has been a tendency to approach classification in a less rigorous manner, especially with respect to hypothesis testing.

Gordon suggests that one "rule of

thumb" for determining numbers of groups is to plot the numbers of groups against the amalgamation value for the groups and to decide the number of groups on the basis a discontinuity in the slope.

Plotting

the average link amalgamations suggest a discontinuity after 4 groups, while the complete link solution, which has a small anomolous group, suggests a discontinuity after 6 groups.

However, as Gordon

indicates, this procedure can be unreliable. Although the significance of a cluster analysis solution can be estimated through the comparison of the data with data sets generated


67

under Monte Carlo procedures, this approach is apparently not common. Rather, the tendency is to evaluate the validity of the solution through other procedures.

Aldenderfer and Blashfield (1984) consider

five other techniques which have been used to validate cluster analysis solutions, including:

cophenetic correlation, significance

tests on the variables used to create the clusters, replication on an independent sample, significance tests on independent variables and software validation indices.

The cophenetic correlation and the

significant tests on variables used to generate the clusters have serious methodological problems.

Aldenderfer and Blashfield point out

that the cophenetic correlation method assumes normal distribution of the values in the matrices being compared, an assumption that is generally violated.

They indicate that Monte Carlo studies of the

cophenetic correlation indicated it was a misleading indicator of quality of the cluster solution.

Efforts at validation through the

use of significance tests comparing the different cluster groups on the variables used to generate the solutions is unacceptable since the formation of the clusters is in fact based upon significant differences in the values of the variables in different clusters. Replication on an independent sample is a highly desirable validation procedure, but is beyond the scope of this project.

Validation of the

classification presented in Table #2 was sought through the prediction of the classification with independent measures, and with a software validation procedure, the BMDP K-Means program. One attempt at validation of this solution with independent measures utilized the SPSS Discriminate Analysis (Nie et al. 1975)


68

program and the social complexity and regional variables reported by Murdock and Provost (1973) as predictors.

The initial analysis

attempted to predict the six groups: Priests, Mediums, Malevolent Practitioners, Core Healers, Shaman/Healers, and Core Shamans.

The

classification was relatively unsuccessful, predicting only 42% of the cases correctly.

Redistributing the Shaman/Healer group to the Core

Shaman and Core Healer groups through BMDP K-Means analysis (Dixon and Brown

1979) and entering the new Shaman and Healer group with the

Priests, Mediums and Malevolent Practitioners did not improve the prediction success (45%).

In both cases, the program predicted the

Shaman and Priests well, but consistently misclassified the other groups, generally as

Shamans or Priests.

The reason for the failure

of the social complexity variables to predict this classification becomes clear when the relationship of the practitioner groups to social complexity variables is considered (see Chapter 5).

Briefly,

the Priests, Healers, Mediums and Malevolent practitioners all occur together in the most complex societies, preventing their distinction on the basis of social complexity variables. The BMDP K-Means procedure (Dixon and Brown 1979) was used to validate the cluster analysis classification as specified in Table #2.

The K-Means program takes that assignment of cases and then

redistributes the cases among groups, seeking to maximize the Euclidean distance between clusters using the variables employed to generate the clusters.

This procedure was employed for the six

groups, with no data standardization since the variables were all binary or three step ordinal.

This procedure reassigned 14 cases, or


69

only 11% of the sample.

The cases which were reassigned to other

groups are presented in Table #3.

Those cases which were reassigned

were primarily those which were inconsistently classified in the earlier analysis, plus a few shifts from the Shamans, Healers and Mediums to the Shaman/Healer group.

Since the classification

differences with the K-Means validation were minimal, the classification presented here was accepted without modification.

TABLE #3 K-MEANS CLASSIFICATION DIFFERENCES CORE SHAMANS Hidatsa Shaman/Healer Mapuche Healer

HEALERS MALEVOLENT PRACTITIONERS Zuni Katchina Creek Rain Priest Nama Shaman/Healer Dyak Shaman/Healer

SHAMAN/HEALERS Semang Shaman Toda Curer Jivaro Priest Atayal Shaman-Medium Alor Seer

MEDIUMS Marq. Insp. Priest

PRIESTS Ibo Oracle Fur Healer

3.2 Discussion of Cluster Analysis Solutions This section considers some of the cluster analysis solutions examined in the development of the classification arrived at here. The cluster analyses solutions based upon the continuous variable assessment are not presented since an informal comparison of these solutions with those based on the mixed variables assessment, which adjusted for the effects of missing data, showed the latter to provide much clearer clusters.

Since the large number of cases analyzed here

has resulted in very large dendograms, they are not presented. However, modified and truncated versions of the dendograms are presented for four of the principal solutions.


70

The single link solution with all cases (not shown here) exhibited the phenomena of chaining, which frequently occurs with single link amalgamation (Dunn and Everitt 1982).

Chaining refers to an

amalgamation phenomenon in which a cluster or a few small clusters are formed, but the majority of the cases link to this initial cluster one after the other without the formation of individual clusters.

This

solution was therefore unacceptable, but did reveal some limited clustering reflecting the classification presented above.

This

included individual clusters representing the core of the Priest group, most of the Malevolent Practitioners, the Core Shamans, and the core of the Medium group.

This was the only analysis in which the

Roman Sorcerer/Witch was clustered with the Malevolent Practitioners. In the single link solution, the group of Priests and Malevolent practitioners amalgamated first, followed by the non-practitioner cases labeled Siriono General and the Mbuti Male, illustrating the pivotal, intermediate or transitional role of these non-practitioner cases in the overall structure of the data. The average link solution (Figure #1) suggests five groups of practitioners, with five outlier practitioners.

Clusters

corresponding to the Core Shamans, Malevolent Practitioners and Mediums are clearly separated.

The Priest group and a group

containing the Healers and most of the Shaman/Healers merged together in a manner which suggested a continuity between the types.

However,

there was a group of five practitioners which merged with the Mediums at distances similar to the distances found in the amalgamation between major clusters.

Although this solution is appealing and


71

*x-CZ tw e

S

< u »

sswcc ■vs

xao •• • Itii

-E

.2

“5 4)

-E

x

X