STRATEGIES UNDER NON-TRANSFERABLE uTILIm ; g
AN EXPERIMENTAL STUDY OF THE EEEEOTSOF : -
DIVISIBILITY OF 'PAYOFF, COGNITIVE COMPLEXITY.
AND MACHIAVELLIANJSM 0N STRATEGY SELECTION
IN A MIXED MOTIVE GAME ’ ’

Thesis for the Degree of Ph. ‘D.
MIOIIIOAN STATE UNIVERSITY
LAWRENCE H. NITZ'
1959.

 

‘5' _.fl.... TI." .___ 5

LED 1: AK)
Michigan State
University

 

‘fld

This is to certify that the

thesis entitled

Strategies Under Non-Transferable Utility: An
EXperimental Study of the Effects of Divisibility
of Payoff, COgnitive Complexity, and Hachiavellianism
on Strategy Selection in a Mixed Motive Game
presented by

Lawrence H. Nitz

has been accepted towards fulfillment
of the requirements for

Ph.D. Political Science

degree in

T” a r
flfiM/K '74 .441.qu (

Major professor

 

Date Hay 11+, 1969

 

0-169

ABSTRACT
STRATEGIES UNDER NON-TRANSFERABLE UTILITY: AN
EXPERIMENTAL STUDY OF THE EFFECTS OF
DIVISIBILITY OF PAYOFF, COGNITIVE COMPLEXITY,
AND MACHIAVELLIANISM ON STRATEGY
SELECTION IN A MIXED MOTIVE GAME
Game-theoretic treatments of political decision processes typically
assume that the payoff of a decision is homogeneous and infinitely
divisible. This assumption permits the transfer of goods to which the
analytical concept "utility" may be applied. This study explores stra-
tegies that may be selected in a mixeddmotive situation constructed with
either easily divisible payoffs or relatively indivisible payoffs.
Recent studies of strategy selection (e.g., Riker, 1967) have con-

tended that personality or individual differences do not have a significant
effect on processes relevant to political strategy selection. A more
general question, however, has been raised (Greenstein, 1967); Under
what conditions do individual differences account for political behavior?
The mixeddmotive situation used in this study systematically controls
conditions in which differential strategy patterns might be predicted
from individual difference measures of attitudes or skills supposedly
related to strategic behavior.

Studies by Cole (1969), Phillips and Nitz (1968) and Nitz and Phillips

(submitted for publication, Journal of Conflict Resolution) indicate that

 

the probability of executing a strategy choice and the ease of divisibility
of the payoff elicit differential strategies in mixed—motive situations.
In this study a mixeddmotive coalition game known as the "political con-

vention paradigm" presented subjects with a finite set of strategy options

as they played the roles of faction leaders in two mock political party
conventions. Each subject played the role of a faction leader faced with
one opponent who could muster an equal number of convention votes and one
opponent who could control a greater number of votes. The subject also
played the role of a "stronger" contender in a game with two large, equal
factions and one small faction. Four explicit strategies were defined on
the basis of the subject's joint choices in the two types of games and

the divisibility of payoff: Maximization, Competition, Security and Intra-
coalition Compatibility. The following hypotheses were examined:

1. Subjects seek to maximize their share of the payoff with

respect to their coalition partners, regardless of the
probability of winning or the ease of divisibility of the
payoff.

2. Subjects seek to form coalitions in which the division of

the payoff can be negotiated with a minimum of intracoalition
friction. When the payoff is only unequally divisible, the
probability of choosing the unequal contender is higher than
when the payoff is easily divisible. When the payoff is
easily divisible, no intracoalition incompatibility is en-
gendered in a coalition between unequals, so the maximization
decision rule is used.

3. Subjects seek to form.coalitions that will allow maximum

grounds for conflicts, that is, coalitions for which the
payoff structure of the game suggest no obvious division
of the payoff.

4. When the payoff to a coalition can be obtained only with some

probability less than unity, subjects will seek to maximize
their chances of winning by forming the largest coalition
possible.

Twenty-seven triads of male college students were run in each of three
experimental payoff conditions: easily divisible-certain (Here the payoff
was 100 patronage positions at a mid-term party convention); indivisible-
certain (The payoff was nomination for either the governorship or lieutenant

governorship in a one-party state); and indivisible-uncertain (Nomination

for governorship or lieutenant governorship in a two—party state).

Hypothesis 1, Maximization, and 2, Intracoalition Compatibility were con-
firmed, substantiating the initial findings of Phillips and Nitz (1968)
and Nitz and Phillips (submitted for publication) that indivisibility of
payoff tends to elicit strategies that seek to reduce conflict over pay-
off division.

Harvey's (1961) conceptual systems theory defines four rank-ordered
modes of processing information that would be expected to affect decision-
making behavior. Persons with the more cognitively complex information
processing skills would be expected to reject irrelevant social cues and
select more task-related maximization strategies than persons with less
cognitively complex information processing patterns. Tuckman's (1964)
Interpersonal Topical Inventory was used to identify subjects in each
complexity level. A factorial partition of contingency tables was used
to analyze the effect of complexity differences on strategy selection.

The hypothesis relating complexity level to strategy selection was dis-
confirmed. The most complex and the third-most complex of the four groups,
however, performed according to the prediction for the more complex per-
sons. The second—most complex and the least complex groups selected
strategies predicted for the less complex subjects. Moreover of the most
and third-most complex groups of subjects, those who picked Maximization
strategies perceived their opponents as likely to demand less than half

of the payoff. Revisions of conceptual systems theory were suggested.

Christie's (1962) Machiavellian is expected to select strategies that
maximize either payoff (regardless of conditions) or conflict. No strategy
choices could be predicted with the dichotomous classes of subject scoring
above and below the median Mach V score. A post-hoe discriminant function

analysis, however, was able to discriminate Maximization from Competition,

Security and Compatibility; Security from Compatibility; but was not able
to discriminate Compatibility from either Competition or Security. The
extremely strong degree of association of the predicted by observed con-

tingency tables for strategies (X2 = 73.09) as well as the discrimination

9df
among strategies with only four Mach V items in the discriminant function

equations (X2 = 27.86; .005 > p > .001) indicates that the Mach items can

9df
effectively predict strategy selection in an abstract game, but not if

they are taken as an additive scale. The high level of discrimination

obtained among the discrete abstract strategies defined across different
payoff conditions in the political convention game suggested that it may
be inappropriate to assume that any particular attitude scale should pre-
dict political behavior across situations. This study suggested that it
may be fruitful to examine those skills that permit a political actor to

select strategies appropriate to the particular situation; i.e., that per-

mit discrimination among strategic situations.

STRATEGIES UNDER NON-TRANSFERABLE UTILITY:
AN EXPERIMENTAL STUDY OF THE EFFECTS OF DIVISIBILITY 0F PAYOFF,
COGNITIVE COMPLEXITY, AND MACHIAVELLIANISM 0N STRATEGY SELECTION

IN A.MIXED MOTIVE GAME

by
3(VQ ‘ I
Lawrence H. Nitz

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Political Science

1969

A C KNOWIE DC E ME NTS

I wish to thank my dissertation committee members, Professors Rufus
Browning and James Phillips and my Chairman, Frank Pinner for the patient
guidance and thoughtful criticism they have prQVided throughout this
study. I thank Frank particularly for proviuing the initial intellectual
stimulus that led me first to lose myself in experimental and social
psychology for two years, and then to find my way hack to essential issues
in modern political theory. During my tenure as a graduate student he
served far beyond my highest role expectations of a doctoral program
chairmen.

In studying social psychology 1 incurred other debts that I wiSh
someday to be able to repay. Professor Donald Johnson helped me develop
an appreciation for significant problem areas in the study of human cogni-
tion; Professor Eugene Jacobson labored to convey to me an appreciation of
thoroughness in the study of human organization. Professor Frank Marzocco
helped eliminate the less elegant segments of earlier proposals.

To my fellow colleagues at MSU's Human Learning Research Institute,
particularly Steven Cole, Charles Kenoyer, and David Kline, I owe much of
whatever skill in the study of human behavior I have acquired. The in-
formal discussions I have had with them have helped me clarify my ideas.

I wish to acknowledge the support of the Human Learning Research
Institute throughout this project. Professors Frank Marzocco, former
Director, and Ted Ward, current Director of the Institute have consist-
ently supplied not only the facilities and services of the Institute, but

also thoughtful advice and administrative support, without which this

study could not have been done. Computer time has been generously
applied by tlu:lNJlitical Sciewuleyepartment of lmnftnul the Computer
Laboratory of the University of Hawaii. I have already thanked those
graduate students and faculty members who Supplied subjects for the

eXperiments in this study.

I wish to thank thcse who have taken on the tethnical tasks of this
study. Each of them has ptrformed with thoroughly professional compe—
tence. Clinton Tarkoe was responsible for assembling t perimental materials
and coding finished questionnaires. Mrs. Cecelia Highstreet typed the
experimental materials and the Sinai manuscript--much of it from hand-
writing no one should have to rtad. Duane Lawton was kind enough to put

his computer program into working order well ahead of schedule so that I
could complete the contingency table analyses. Steve Cole supervised
production of the final copies of the dissertation.

James Phillips, for intellectual stimulation, insistent attempts to

force me to concentrate m' labors on rohlems that were both significant
3 P s

5' -4
C.
¥~ ,
C-
H
r-Q

and manageable, cogent criticism (even when want it), and open

1

friendship, I hold responsible for much of m} prOgress toward completion
of the Ph.D.

Finally, this work could not have been begun without twenty-odd years
of support and guidance from my parents, Mr. and Mrs. harvey Nitz. It
could not have been completed in the short year it took without the con-
sistent support, encouragement and sexiness of my wife, Kiyoko Kurusu

Nitz.

iii

LIST OF

LIST OF

LIST OF

CHAPTER

TABLE OF COLTENIS

FIGURES.
TABLES

APPENDICES

Introduction

The Identification of Social Contact Strategies in
the Political Convention Paradigm.

III Situation Structure, Cognitive Complexity, and
Machiavellianism as Determinants of Strategy Selection
IV. Objectives and Procedure
V. Results and Discussion
VI. Summary and Implications
BIBLIOGRAPHY

APPENDICES .

iv

Page

vi

viii

12

40
57
67
96
105

109

 

Figure

2.1

3.1

4.2
4.3

4.4

LIST OF FIGLRES

Minimum resource predictions in Gamson's second con-
veI‘lt ion. 0 O O O O O O 0 O O O O O O O O I O O O O 0

Joint choices in Type 2 and 3 triads . . . . . . . .
Theoretical strategies by divisibility condition . .
Experimental payoff conditions . . . . . . . . . . .
Type 2 resource distribution . . . . . . . . . . . .
Type 3 resource distributions. . . . . . . . . . . .

Counterbalance of labels and resource values . . . .

Page

19
41
43
58
64
65

65

Table

2.1

5.4

5.5

5.6

5.7

5.8

5.9

5.10

5.11

5.12

LIST OF TABLES

Probability of choosing the smaller contender as a
function of resource distribution and divisibility
of payoff. . . . . . . . . . . . . . . . . . . . . .
Tests of the difference between p(W)EXperiment I and
p(w?Experiment 11 for the Type 2 Resource Distri-
bution . . . . . . . . . . . . . . . . . . . . . . .

Strategy selected as a function oi group and payoff
Condition. C U C C I O C I C O O C C 0 O C O O O O .

p(W) with 951 confidence intervals by payoff condi-
tion, experiment and type oi triad . . . . . . . . .
p(W) ’ > 3 - p(W)m, . 0 by experiment and payoff
condligbn. . . . 1y?c.h. . . . . . . . . . . . . . .
LP(W)Type 3 ‘ P(W)Type 21 Condition 1 ' 1P(W)Type 3
'P(w)Type 2] Condition 111 by experiment . . . . . .

Strategy choice by perception in the Type 2 conven-
tion by cognitive complexity . . . . . . . . . . . .

Strategy choice by perception in the Type 3 conven-
tion by cognitive complexity . . . . . . . . . . . .

Strategy choice by perception in the Type 3 conven-
tion by cognitive complexity categories I + 111 v.
II + IV. C O O 0 O O O O O O O O O O O O O I O O I I

Frequency of marginal strategy choices for four
cognitive complexity categories. . . . . . . . . . .

Strategy choice by cognitive complexity and Machia-
vellianism Scores. 0 C I 0 I 0 O I O C I I O O l O O

Marginal strategy choices by Machiavellianism score
group. O O O O O O O O O O O O O O O I O O O O O O 0

Means and standard deviations of Mach V item scores
for each strategy type . . . . . . . . . . . . . . .

Linear function discriminating four strategy types on

the basis of 20 Mach V item scores . . . . . . . . .

Page

31

69

7O

72

73

76

78

79

8O

82

84

84

86

88

 

Table Page

5.14 Number of subjects classified into strategy groups

on the basis of Mach V scores on 20 items. . . . . . . 89
5.15 Summary table of steps in discriminant program . . . . 90
5.16 Number of subjects classified into strategy groups on

the basis of Mach V scores on the eight most discrim-
inating items. 0 o o I a O O O 0 0 O O o O o O I O I O 91
5.17 Number of subjects classified into strategy groups on

the basis of Mach V scores on four most discriminating

items. 0 O O C O I I O O O C O O O O O O I O O O O O O 91
5.18 Items for which means of the distribution of Mach V

item scores differ across groupsa:b. . . . . . . . . . 92
5.19 Keyed items from the Mach V scale. . . . . . . . . . . 94

vii

Appendix
A

B

LIST OF AVPENDICES

Induction for Condition 1
Induction for Condition II.
Induction for Condition III
Complete Questionnaire.

Machiavellianism and Interpersonal TOpical
Inventory Scoring Routines.

Instructions for Experiment II.

viii

Page
109
112
115

118

136

143

‘haptcr 1

Introduction

If the process of forming a coalition to effect a political decision
is conceptualized as an n-pcrson game, we would expect the mathematical
theory of n-person games to shed some light on the strategies that would
be in some sense ”rational” in the game. Lute and Raiffa (1957) note
that the nothematical work on n-pcrson games has generally assumed that
"in addition to receiving the payoffs prescribed by the rules of the
game, the players are permitted to nuke additional transfers....” This
provision for side-payments generally subsumes the much stronger condition
"that utility is unrestrictedly transferable". (Luce and Raiffa, 1957).
This assumption about utility (or in actuality, goods to which individuals
ascribe value) requires the additional supposition that:

there exists an infinitely divisible, real, and desirable commodity

(which for all the world behaves like money) such that any

reapportionmcnt of it among players results in increments and

decrements of individual utilities which sum to Zero according

to some specific set of utility scales for the players. This

can happen when money exists, provided that each player's utility

for money and that the ztro and unit of earh utility function is

so chosen that the conservation of money implies the conservation

of utility. When else it can realistically happen is obscure.

(Luce and Raiffa, 1957, p. 168)

Luce and RJiffa (1957) argue that the assumption of transferable
utility is not essential to n-person game theory and that a solution
function can be defined without it. Their discussion is roughly as
follows: Suppose there is a lottery whose prizes are bundles of goods,
services, obligations, etc. The bundles accrue to each possible winning

coalition as a whole. Let C(S) denote the commodity bundle accruing to

the coalition S and let J(S) denote the set of all possible physical

v

distributions, T, of C(S) over the members of S. A coalition S is defined
to be effective for a set of utilities X:(Xl’ x;,......xn), where n is the
number of actors in S, if there exists at least one distribution, T, in
the set J(S) such that the utility of T is less than Xi for all players i
in the coalition S.

The set of utilities Y=(yl, y,,...yn) is said to dominate the SEC
4 __,_____

of utilities X=(x1, x2,...xn) if there exists a non-empty coalition S
such that S is effective for Y and yi).xi for all i in S. Under these
conditions, S is a solution for the game.

Luce and Raiffa (1957) point out the critical mechanical problem of
making side payments in units of indivisible physical commodities. Suppose
the joint payoff to the coalition Sl=ll,?

k

and C are non-homogeneous, indivisible goods like a house, a painting and

is C(Sl)={},8,d}, where A, B,

\IV’”)

a car. If player 3 were to join S1 to form sz{l,2,i} the payoff might
be C(32)={P,Ei; where D and E are also non-homogeneous, indivisible goods
such as a yacht and an antique oriental chest. The monetary equivalents
for the commodities A, B, C, D, and E may differ from person to person.
So long as no external market mechanism is available, there is no apparent
method either S1 or S2 may employ to divide the payoff. Nor is there a
clear way to decide whether it is profitable to add another player. One
means of circumventing this problem is by aggregating payoffs--either by
bringing a number of small issues into the negotiations, or by contributing
some divisible resource, Such as money, to a common pot. Luce and Raiffa
(1957) note that the problem of indivisible payoffs has neither been
attacked directly, nor has it been approached through systematic deve10pment
of a theory of anregdtion.

This study attacks the problem of predicting strategy choices in games

in which payoffs may not be infinitely divisible. The focus of the present

 

study, however, is not the mathematical derivation of ideal solution sets,
but is rather the empirical examination of the effect of nontransferable
utility on the strategies different subjects select. The remainder of

this introductory chapter will note several significant studies of coalition
strategy that have contributed to the general design of the present research.
The chapters to follow will examine several studies in detail, derive
empirical hypotheses, construct a test situation, and present the findings

(3f tlu: stud)n

Caplow (1959) was the first empirically oriented social scientist to
investigate the problem of Strategies in games with non-transferable pay-
offs. He defined three kinds of competitive environments: continuous, in
which the rewards of the coalition process lie in the activities of forming
the coalition, as in competitive social games; episodic--in which the
rewards are distributed periodically to the coalition in control at pre-
determined distribution times; and terminal--in which the distribution of
rewards permanently ends the game, e.g., by destroying one or more actors
or by establishing an equilibrium condition.

In postulating three different strategies, Caplow (1939) made two
implicit assumptions. The first was that all persons perceived the sit-
uation in the same way. The second was that all persons who perceived
the situation in a given way would select the same strategy. Caplow's
(1959) assumptions have been implicit in much contemporary research on
coalition formation. (e.g., Gamson, 19613, Riker, 1967).

One of the first political scientists to develop a fornwl coalition
theory, William Riker (1962) also makes qualitative distinctions among
types of payoff. Riker (1962) identifies one set of rewards as partic-
ularly appropriate to followers, and another set as constituting the

principal reward of leaders. His distinctions, though, do not lead to

predictions of different strategies. Riker (1962) derived one strategy
from a strictly deductive analysis of the n-person zero—sum game. This
strategy assumes an infinitely divisible payoff. All rational subjects
are not expected to perceive the payoff situation in the same way but
they are expected to select the same strategy. Riker (1962) contends
that an actor reaches a strategy decision only by comparing offers ten—
dered with his own preferences and its Subjective estimation of his
opponent's preferences and alternatives. he claims generality for this
approach since the estimation assumption frees him from the necessity
of postulating interpersonal comparisons of utility. In the derivation
of the "size principle", which postulates that rational actors will
choose to form the smallest winning coalition, Riker (l962) examines
a situation in which a larger-than—minimal winning coalition might form.
If it is possible to increase the total payoff more than proportionately
by adding members to an already winning coalition, a larger than winning
coalition might conceivably form. Riher (l962) examines four rules for
payoff division in this situation, and finds that none of them lead to
a stable coalition larger than the minimal winning coalition-—if they lead
to a coalition at all. The entire analysis, though, examines payoff dis-
tribution rules that operate only with infinitely divisible payoffs. Thus
despite his disclaimer to the contrary, Riker (196“) limits his analysis
to situations which have transferable utility.

Schelling (1960) argues that if a social situation is conceptualized
as a mixed-motive game-—that is, as a situation in which there is something

to be gained by cooperation with some, but not all of the participants--then

it is not possible to construct a totally deductive theory of coalition
strategy. Cues in the environment that may be independent of the abstract
characterization of the situation's payoffs may suggest strategies to
actors that would permit them to coordinate their activities. Moreover,
some empirical knowledge of how the actors assessed their opponent's res—
ponses to sucn cues is necessary to a viable theory of strategy selection
(Schelling, 1960). This argument takes the individual actor's perceptions
of the competitive situation as essential elements of a theory of coali—
tion strategy. Schelling's (1960) approach would suggest that intangible
goods such as agreement on ideological stands, or particular sensitivity
of individual actors to certain outcomes or particular prominence of
specific payoff divisions, would be likely to have cue value to the actor
formulating a strate y—-even if neither the actor nor an outside observer
could assign an exchange value to the good or situation element that pro—
vided the cue.

Riker (1967) provides a contrast to Schelling's (l962) position in
this introduction to an eXperimental study of bargaining in a three-person
mixed motive game:

The scientific expectation is that, by studying the quasi—political

action of games-—where the variations among institutional, psychologi—

cal, and ideological components of the behavior are minimized——one
will be able to understand more profoundly the basic political ac—
tivities of bargaining, forming coalitions, and choosing strategies.

This more profound understanding is a consequence of obtaining an—

swers to the following questions: (1) What is the mathematical sol-

ution, that is, what amount of utility can players he expected to
obtain, when it is assumed that players are rational and wish to
maximize utility? (2) What is the strategy (or method of playing)

that will ensure players of achieving the solution? (Riker, 1967)

Riker (1967) found that the mean payoff to players in each playing

position was not significantly different from the mathematical solution

of the game. He also found that those participants who "undersold" them—

selves in bargaining, that is, those who orfered their opponents a larger
share of the payoff than the opponent could expect to obtain in any other
coalition, won significantly more games than did participants who offer-
ed less to the opponent and demanded more of the payoff for themselves.

Participants tended to change their offers as they played different
positions in the game, and players from different social backgrounds did
differ in bargaining strategy or amount won (Riser, 1967). Riker (1967)
concludes his study with the following remarks:

It is often suggested that the outcomes of political events are

determined by the psychological or sociological characteristics

of the participants. Such considerations seem inconsequential

in these eXperiments where quite different kinds of subjects

behaved in substantially identical ways and where the same sub-

jects behaved differently in different positions . . . subjects

did not let their psychological predispositions toward high or

low aspirations or high or low feelings of dominance (or what-

ever else might be said to force them to behave similarly in

different positions) affect their judgement on the choice of a

strategy . . . I conclude hat the crucial determinants of be-

havior are the subjects' (conscious or unconscious) recognition

of the abstract solution and the strategy dictated by the tem-

poral circumstances.

It should be noted that Riker (1967) reports no individual difference
measures on his subjects other than their social group identification

as political science students, randomly selected students, or businessmen
in evening college. Moreover, Riker's (1967) experiment used only one
form of abstract game and a monetary payoff--thus there was only one
abstract solution.

The contrast between Riker's (1967) argument and Schelling's (1960)
approach can be sharpened by examining one of the principal objections
to the study of the effects of personality on political behavior:

Personality is not an important determinant of behavior because

individuals with varying personal characteristics behave sim-

ilarly when placed in common situations. And it is not useful

to study personal variation, if the ways in which people vary
do not affect their behavior. (Creenstein, 1967)

Greenstein's (1967) critical review of this sort of objection rephrases
it in terms amenable to scientific examination:

Under what circumstances do different actors (placed in common

situations) vary their behavior, and under what circumstances

is behavior uniform?

Two sorts of questions may be asked with this proposition in mind. The
first deals with characteristics of the environment that may facilitate
or hinder personal variability. Sherif (1953) notes that ambiguous sit—
uations tend to leave room for personal variability. Budner (1962)
elaborates the concept of ambiguity to include those environments which
are completely new and offer no familiar cues, those which are complex
and provide a great number of cues, and those which provide contradictory
cues. Greenstein (l9b7) suggests the following proposition relating
personality effects to environmental differences:

The opportunities for personal variation are increased to the

degree that political actors lack mental sets which might lead

them to structure their perceptions and resolve ambiguities.

The second sort of question one might ask to ascertain the conditions
under which individual dif‘erences might lead to differential behavior is,
"What kinds of individual predispositions are sensitized by various sit-
uations? What kinds of skills do different situations draw upon?" These
questions can be directed to those elements of the environment budner
(1962) held relevant to individual variation in behavior. What kinds of
skills would be useful in a situation which is novel, complex or contra—
dictory? What kind of predispositions would be sensitized by such a sit-
nation?

This study examines several theoretical contributions of the study

of coalition behavior in the light of the effects of environmental and

predispositional differences. It proceeds by (l) structuring a well

defined mixed—motive environment and postulating a set of strategies
eXpected to be elicited by the environmental structure; (2) developing
a set of systematic alterations in the environment and predicting the
differences in strategy patterns resulting from environmental changes;
and (3) postulating a set of strategy choices that would be expected
of individuals with selected personal characteristics in specific en—

vironmental situations.

The Experimental Environment

 

The "political convention paradigm" has been used extensively as
an eXperimental setting for testing hypotheses based on various theories
of coalition formation behavior. The political convention paradigm was
first used by Gamson (l96lb). In it subjects are asked to take the
roles of faction leaders or candidates in a political party convention.
The purpose of the convention is to allocate an easily divisible pay-
off, such as a number of patronage positions or a relatively indivisible
payoff such as the nomination to an office on the party's ticket. The
subjects, as contenders, must garner a majority of the delegates' votes
to gain effective decision power over the payoff distribution. Their
activities consist of deciding whom to contact to begin negotiations with,
negotiating, and arriving at some coalition agreement with an eXplicit
division of the payoff. The political conVention paradigm thus provides
the opportunity to observe several forms of social and individual behav-
ior. This stLdy focuses on one phase of the individual's strategy: his
selection of a potential partner with whom he will begin negotiations.

Abilities and Predispositions

 

The prospect that situations which are novel, complex, or contradic-

tory may facilitate the use of particular kinds of skills suggests a

particular area of personality theory that may explain aspects of strategy
selection behavior. Harvey's (1963) work with conceptual systems theory
deals directly with Sdllls essential for handling complex situations.
Harvey, Hunt, and Schroder (1961), Harvey (1903), and Schroder, Driver

and Strcufcrt (1967) developed a theoretical scheme that postulates that
the ability to deal with large amounts of complex or contradictory infor-
mation is a basic individual skill. Individuals who can integrate large
quantities of information and make accurate discriminations can generate
more alternative conceptions of possible outcomes in a situation and can
anticipate a larger number of consecuences of different strategies. These
persons are called cognitively complex. They have been found to be suc-
cessful in several forms of strategy games (Schroder, Driver and Streufert,
1967). Persons who lack these abilities are more limited in their strat-
egy perceptions and thus exhibit more rigid and invariate strategy choice.
The contribution of conceptual systems theory to the prospect of determining
the conditions under which different situations will elicit different
strategies is examined in this study.

The possibility that different competitive environments might sensitize
different predispositions or attitudes suggests that one would do well to
examine some set of attitudes that are theoretically related to strategic
behavior. The concept of the Machiavellian refers to the individual who
has skill in interpersonal strategy and has no scruples about using it to

. . , . _ .. . 1
his own advantage. Christie describes the ideal Machiavellian as follows:

1The develOpment of an objective scale to measure Machiavellianism was
initiated by an informal workgroup at the Center for the Advanced Study of
the Behavioral Sciences in 1953-54. The group consisted of Robert Agger,
Richard Christie, Bruce Melnik, and Frank Pinner. (Christie, 1962)

10

(l) he is not basically concerned with morality in the convention-
al sense.

(2) He is basically ”cool” in the interpersonal relationships--
once one becomes emotionally involved with another person it
is difficult to treat him as an object manipulated.

(3) Since those who manipulate are more concerned with means than
with ends, he might be of any ideological persuasion, but is
more concerned with conning others than with what he is conning
them for.

(4) He functions successfully in the contemporary world. He is not
likely to display signs of irrationality viewed as neurotic or
psychotic, but is more likely to be overly rational in dealing
with people. (Ceis and Christie, 1965)

The Machiavellian might be eXpected to be sensitized by situations
which offered a number of strategies for dealing with the environment at
hand. Moreover, he might be especially sensitive to those strategies
which others might avoid because of questions of the ”fairness” or ”right-
ness". This study examines the effects of Machiavellianism in several

competitive situations which provide such strategies as viable choices.

Implications for Political Thecgy

 

The study that follows attempts to integrate the major questions
raised by two distinctly different arguments in modern political theory:
One, the position that the abstract (mathematical structure of the com-
petitive situation determines the participants' strategy choices; the
other that the participants' perception of significant cues may lead
actors to strategies not defined as mathematically optimal. The key to
integrating these conflicting arguments is an examination of behavior in
a situation which is characterized by non-transferable utility, i.e., in-
divisible payoff. Chapter II of this study reviews four competing theories
of coalition formation, Minimum Power Theory, Minimum Resource Theory,

Anti-Competitive Theory, and Utter Confusion Theory (Gamson, 1964). The

review poses a question suggested by Sche‘ling's (1960) criticism of
purely deductive approaches to theory of interaction in mixed—motive
games: how do changes in the competitive environment alter the strat-
egies chosen? This question provides a means of integrating the con—
flicting predictions of the four theories of coalition formation.
Moreover in dealing with situations which may elicit alternative strat-
egies, it becomes possible to ask if the perception and selection of
specific strategies is in part the result of some particular skill or
sensitization to the competitive situation.

hapter II develops the theoretical basis for predicting alternate
strategies in mixed-motive situations and Chapter III derives hypotheses
predicting strategy selection from individual difference measures.
Chapter IV restates the objectives of the study and develOps the experi-
mental design. Chapter V presents results and Chapter VI summarizes
the findings of the study and briefly discusses their theoretical sig-

nificance.

Chapter II

The Identification of Social Contact Strategies in
The Political ConVention Paradigm

The identification of "successful” strategies in mixed—motive games
is generally contingent on the assumption of a particular theory of strat-
egy behavior. This is most apparent in theories of "rational" choice,
such as Riker's (1962) application of a game theoretic model to the study
of political coalitions. Riker (1962) defines a rational choice as a stra—
tegy that would seek to build the smallest winning coalition. Any deci—
sion to form a coalition larger than necessary to win the contest at hand,
unless it is made in ignorance or the necessary margin or size of the min-
imum winning coalition, is an irrational decision. In an empirical exam—
ination of a theory which assumes a single rationality it is meaningful to
ask whether subjects or respondents chose rational strategies or not.

If the theoretical basis for a study of decision behavior postulates
alternative strategies, that is, alternative goods that a participant may
choose to maximize, then it makes little sense to ask wnether subjects
in the study are behaving rationally. It does make sense, however, to
ask, "With respect to what decision rule are their decisions rational?"
This study investigates several distinct patterns of strategy selection,
each of which maximizes a somewhat different expected utility. Each
strategy pattern is in this sense a rational pattern. lhe question we may
then ask is, "Under what conditions do participants in a mixed—motive
game select alternate rational strategies?"

This chapter reviews some of the empirical findings that have led to
the development of the several theory fragments dealing with coalition for—

lnation process, Shelly and Phillips' (1966) distinction between two

i) o

phases of action in forming a coalition provides a useful analytic tool
for the discussion to follow. They distinguished a temporally prior pro-
cess in forming a coalition as the "social contact process" and a later
phase as the "bargaining process". The contact process will receive prin-
cipal attention in this study.

The social contact process is by no means a unitary event. There are
at least two individual activities in the contact process. One is the in—
dividual's estimation of the demands other participants will make in bar-
gaining. The other is the exercise of a decision rule to select one of the
other participants to contact. Several studies have examined the contact
process with major emphasis on decision rules (Chertkoff, l9bb; Phillips
and Nitz, 1968; Cole and Phillips, 1967; Nitz and Phillips, in preparation).
Little previous research has examined the participant's evaluation process—
es. This study will provide initial data on the evaluative processes in
the mixed-motive game.

Theoretical Background

 

Minimum Power Theory. Minimum Power Theory is derived from a game

 

theoretic measure of power developed by Shapley (1953). The power of any
participant in a mixed motive situation is measured by counting the number
of ways he can turn a losing coalition into a winning one by joining it.
The theory assumes that some decision rule that specifies the total amount
of resources an actor or social unit must control in order to influence
the distribution of payoff in the situation. For any mixed motive situa—
tion with a finite number of social actors and a fixed decision rule, it is
possible to enumerate all possible orders of voting (i.e., joining a coali—

tion) and count the number of times each actor holds the pivotal position.

Shapley and Shubik (1954) have shown that any other internally consistent

measure of power in voting bodies must be a transformation of this one.

With Shapley and Shubik's (1954) simple algorithm for computing
power, we can specify the relative power of each actor and the likelihood
of his being the pivotal member of a coalition in a mixed motive situation.
However, in a situation where resources are distributed as follows:

A: 4 votes B= 3 votes C= 2 votes
and a simple majority (5 votes) is required to win any contest, the
Shapley-Shubik algorithm identifies two potential coalitions for each
participant, such that either will become a winning coalition if it is
formed. Thus the power of all participants is equal -- the probability
of any participant forming a winning coalition is 1/3. In this sort of
mixed motive situation where no participant has either dictatorial pow-
er or veto power, Minimum Power Theory predicts that all coalitions will
form with equal likelihood.

So long as the probability of forming a winning coalition is equal
for all actors, Minimum Power Theory neither suggests nor derives from
any social contact strategy based on the payoff structure of the game.
When the actors' chances of forming winning coalitions vary, though,
Minimum Power Theory may suggest strategies based on order of play.

There has been little experimental support for Hinimum Power Theory.
Vinacke and Arkoff (1957), Vinacke (1959), Phillips and Nitz (1968), and
Cole (1969), however, provide a convincing set of counter-examples. The
one study (Kelley and Arrowood, 1900) that does not support Minimum Power
Theory is based on an experiment that allows several confounding variables

to be uncontrolled.

Anti-Competitive Theory

 

Anti-Competitive Theory was named by Gamson (l96é) in his review of

coalition formation literature. This theorv assumes that participants in

l»—
U1

a mixed-motive situation will attempt to minimize conflict or competition
within an alliance by forming coalitions along lines of least resistance.
The coalitions that form with least resistance will be between those par-
ticipants for whom the distribution of resources suggests an obvious and
unambiguous division of the payoff.1 Gamson (leQ) further specifies

that "this will occur among players who are equal in resources, because . .
players with equal resources will share equally".

Two types of findings can be distinguished among studies supporting
Anti-Competitive Theory. The first type consists of instances of anti—
competitive behavior that arise through the play of the game. Hoffman,
et a1. (1934) found that achiey'ng an early lead in a game where cumula—
tive scoring was important was likely to stimulate opposition. Thus
subjects found it advantageous to avoid taking a commanding lead early in
the game.

Uesugi and Vinacke (1963) observed that females repeatedly attempted
to transform a mixed motive game into a pure coordination game by rotating
winners between plays and forming all-inclusive alliances. Chaney and
Vinacke (1960) found that subjects high in achievement motivation were
coalition members significantly less often than were those low on achieve—
ment motivation. Apparently the more highly motivated subjects presented
the image of a fierce competitor andxvere therefore avoided.

The second type of evidence for Anti-Competitive Theory arises from
the resource distribution and the formal structure in the game, rather than

from the bargaining behavior occurring during the game play itself.

 

This is substantially the argument presented by Schclling (1960).

Willis (1962) found that an even distribution of resources between po-
tential coalition partners tended to lead to the formation of more
counter-coalitions in a four person game than did an unequal distribution
of resources. Gamson (196lh) noted a prevalence of coalitions between
participants with equal resources in a five man game under resource
distribution conditions similar to those used by Willis.

Leiserson (1966) observed that his subjects divided a monetary payoff
equally more often than any other way. He interprets this as division
of payoff along the lines of least resistanCe--by virtue of the prominence

0

of the strategy.h Nitz and Phillips (in preparation) further noted that
subjects were less willing to form a coalition with a potential partner
who had equal resources when the payoff was relatively indivisible. Under
that circumstance the equal resource distribution cannot make it easier
to agree on a division of the payoff.

This latter set of data suggests that a careful analysis of a limited
set of structural variables, such as the resource distribution and the
divisibility of the payoff may provide explanations for some of the be-

havior described as anti-competitive.

Minimum Resource Theory

 

Minimum Resource Theory evolved from the work of CAplow (1956), Vinacke
and Arkoff (1957), and Gamson (1960, 19613, b). The theory was explicitly

formulated as a sociological theory of the coalition process by Gamson

 

I did not manage to secure a copy of Leiserson's (1966) Ph.D. disser-
tation until the work reported here was substantially complete. I shall
touch on his research contribution more lightly than I might have had I
examined his work before independently developing a parallel theoretical
framework.

l7

(l961a, 1064). Camson (l9ola) limits Minimum Resource Theory to mixed
motive games where no participant has dictatorial or veto power. In a
three-person game with participants A, B, and C, this would mean that

the distribution of resources among the participants could not be

 

a > (B+C)
or
A = (5+C).

That is, no one contender (here arbitrarily designated ”A”) can have
sufficient resources to control or block any decision. This exclusion
of situations that have participants with dictatorial or veto power
limits Minimum Resource Theory to situations which are essential games:
Each participant has some stake in the outcome and some possibility

of exercising control over that outcome.

Gamson (1961a) assumes that all participants have the same infor-
mation about the initial resource distribution and payoff conditions.
There is some class of payoffs among which they do not differentiate
on the basis of payoff value, but among which they choose according
to a "non-utilitarian” strategy.

These assumptions define an explicit set of situations for which
the theory is apprOpriate. The first empirical hypothesis of Minimum
Resource Theory is a statement of the goals or expectations the partic-
ipant perceives others to have:

1. Any participant will eXpect others to demand from a coalition

a share of the payoff proportional to the amount of resources
which they contribute to the coalition.
This hypothesis has been referred to as the ”parity norm” (Gamson, 1964).
A succinct development of the notion of parity in interpersonal exchange

is found in homans' (1961) discussion of distributive justice:

*4

f
x

A man in an exchange relation with another will expect that
the rewards of each man will be proportional to his costs--
the greater the rewards, the greater the costs—~and that net
rewards or profits of each man will be proportional to his
investments.... (homans, 1961, pp. 75-77)

Gamson's (1961a) second hypothesis specifies a decision rule for
the participant. It may be expressed in both general and specific forms,
depending on the nature of the payoffs in the game and the relevance of

non-utilitarian strategy choices. The more general form, which applies

. .. . .-. _ 3
when the payoff to each posSible coalition may ditier is:

2a. A participant will choose that coalition in which the
product of the total payoff to the coalition and his
expected share of that payoff is highest. He will not
discriminate within payoff classes on the basis of his
members have the highest mean rank in his evaluation
of non—utilitarian preferences.

When the payoff to all coalitions is eoual, hypothesis 2a may be rephrased
as follows:
2b. When a player must choose among alternative coalition strate—
gies where the total payoff to a winning coalition is constant,
he will maximize his payoff by maximizing the ratio of his re-
sources to the resources of the coalition. Thus he will favor
his cheapest winning coalition.
The final prediction of the theory is:
3. A coalition will form if and only if there are reciprocal
strategy choices between two participants. This hypothesis
assumes that the coalition formation process proceeds in a
pairwise manner.
An illustration of the theory's explicit predictions is found in
Camson's (1961b) study of coalition formation in the political convention
paradigm. Camson (1961b) constructed three political convention games and

asked subjects from two local fraternities to play a series of the games.

Two subjects from one fraternity and three from the other played in each

 

3 . . . . .

Hypothe51s 23 and 2b are my wordings rather than Gamson's. I believe
they convey the sense of Gamson's (1961a) propositions more explicitly than
his original phrasing.

V

game. In one caperimcnt the resources or votes were apportioned among
five subjects in the following amounts: l7, l7, 17, 25, 25. The payoff
in this convention was 100 political jobs that would be divided among
the winning coalition. The non-utilitarian strategy preferences were
established by composing each group of members from two different social
fraternities. The effective decision point was taken to be a simple
majority of the votes.

Since the payoff to all coalitions is the same in this convention,
coalition preferences could be prt~o icted on the basis of the initial re-
source distribution. Thc minimal winning coalition for all subjects is
a coalition that includes two of the 17 vote contenders. When the cases
in which both of a subject's pro-hypothesis choice options were members

of the other fraternity were removed from the analysis, the choice hypo-

 

thesis was confirmed. The minimum winning coalition was the alliance
most often preferred by the three contenders with 17 votes each. The
final coalitions formed as predicted one-third of the time (chance expec-
tation was one-tenth).

In the second convention, both resource base and payoff to the
coalition varied. The Minimum Resource Theory choice prediction are
illustrated in Figure 2.1. (Camson, l9bO)

Figure 2.1

Minimum Resource Predictions in Camson's Second Convention

15 votes\ ’5 votes
90 jObS lOOI jobs

6 votes lOI votes
90 jobs

 

\ no jObS
35 votes””’:’/:’ar

no jobs

 

The nwnber of jobs the coalition was able to divide was the highest of
the payoffs to the members, rather than the sum of the jobs available to
the members if they should win.

In this convention the 6 and 10 vote participants made choices con—
sistent with the prediction of the theory. The two 33 vote participants,
however, chose each other more often than they chose any of the other
contenders. None of the final coalition frequencies differed from chance
expectancies. The latter two findings are contradictory to the prediction
of Minimum Resource Theory.

Gamson (1960) suggests that this tendency of the contenders with
greater resources to choose each other may be a strategy of risk-reduction.
Forming any winning coalition on the first negotiating round may have
been preferred to taking the chances involved in a series of negotiating
rounds necessary to build a 3 member cheapest minimal winning coalition.

The results of Camson's (l9blb) first conVention game tend to support
Minimum Resource Theory. The results of the second game suggest that
Gamson's empirical hypotheses capture only a portion of the substance of
a viable coalition theory.

The three theoretical approaches disvussed all appear to be faulted.
Nonetheless, they provide a sufficient basis for formulating an integrated
explanation of strategy selection.

Identification of Strategy Selection Propositions

 

The most fruitful place to begin an examination of unexplained
strategy selection patterns is with the most clearly articulated theory frag-
ment, Minimum Resource Theory. There are stveral ways in which Gamson's
(19613) Minimum Resource Theory hypotheses may not have been adequate for

the task. The empirical hypotheses assumed (l) a particular distributive

maximization decision rule, and

J
b.)
t";
H.
P-f,
*-
0

justice expectation, (Z) a
(3) required reciprocal choice as a prerequisite to coalition formation.
In the eXperiment subjects chose partners intil reciprocal choices were
made, so no test of the latter hypothesis was possible. The subjects'
evaluations or estimates of their opponents' eXpectations were not measur—
ed, but were inferred from their partner preference choices. Since the
partner preferences did not contirm the theoretical predictions in Gamson's
second experiment we must assume that either (1) the subjects did not enter—
tain parity expectations, or (2) the subjects did not maximize according
to the parity principle, or (3) both. Finally, Minimum Resource Theory
predicts coalition outcomes on the assumption that all participants use
the same decision rule and have the same estimate of other's eXpectations.
This assumption is not upheld in Gamson's (1960, l9blb) second game.

The absence of any significant pattern in the coalitions formed in
the second game, combined with consistent partner preferences in the oppo-
site direction from the theoretical predictions challenges the adequacy
of Minimum Resource as a sociological theory. The significant, though not
necessarily pro-hypothesis, patterns of partner preferences in both the
first and second games indicate that Gamson's (lQhU, lfihla) formulation may
provide a model for a viable theory of strategy selection or individual
choice in coalition formation situations.

Gamson's (1961a, 1901b) work suggests several elements of a strategy
selection model. The first is a prediction that an individual will tend
to hold a parity expectation:

A. Any participant will expect others to drmand from a coalition

a share of the payoff preperticnal to the amount of resources
which they contribute to the coalition.

w

r
r\/‘

Since little previous research on expectations and social judgments is
directly related to coalition processes, I will not elaborate upon this
hypothesis at this point. Its purpose is essentially exploratory in
nature.
"e se ox v o n sis su oes co 5 ithsoi , o 's ' ' '
Tn’ c id hyp tle‘ _g“ t ’ *y (an 1's w rk l‘ a modification
of his maximization decision rule.

Bl. Individuals seek to maximize r share of the payoff with

1'
A.
9 .

ii

t e
respect to the coalition partner
The remainder of this cnapter will explore the effects of two factors
that appear to be systematically related to coalition partner selection
strategies; multiple resource dimensions and differential divisibility of

payoff.

Multiple Resource Dimensions

 

Three studies of coalition processes involving multiple resource dim-
ensions may suggest what sorts of mechanisms may contribute to a viable
theory of individual choice in coalition formation. Chertkoff (1966) exam-
ined the effect that differing probabilities of success would have on
coalitions formed in the political convention paradigm. The payoff or pur-
pose of the convention was to award a nomination and to apportion 100 pa“
tronage positions among the coalition members. The distribution of votes
among three contenders was 40—30-20. The distribution corresponds to
Caplow's (1956) Type 5 resource distribution. The essential characteristic
of the Type 5 distribution is that the resources be distributed among the
contenders such that:

A > B > C
A < (B+C)

Here A represents the number of votes pledged to contender A; B, the number

F.)
\a.‘

J I

pledged to Contender h; and c, the number pleugeo to candidate C. The se-
cond resource dimension, the probability of winning the election if nomina-
ted, was also varied. Four conditions were used. In condition one the pro—
bability of winning the election was not introduced. In condition two each
candidate had a probability of .50 of winning the election if he were nomin-
ated. In condition three contender A bad a probability of .70, while con-
tenders B and C had probabilities of .50 each. In condition four candidate
A had a probability of .90 of winning the election if nominated, while con-
tenders B and C had each probabilities of .30.
The bargaining period began only if two subjects made reciprocal choices.

The subjects were allowed to bargain for a fixed period of time. If they
reached no agreement within the time limit, the process of choice and bar-
gaining was repeated until a coalition formed.

Chertkoff (1966) found that Minimum Resource Theory predictions of

coalitions formed held for condition one, the condition with no probability

 

manipulation. When the probability of winning was introduced, the frequency
of BC coalitions decreased as the probability of A's winning the election in—
creased. The BC coalitions were replaced by AB and AC coalitions. Stryker
and Psathas (1960) and Kelley and Arrowood (1060) contend that misperception
of the real power relationships in coalition situations generates the frequent
coalitions observed between weaker contenders. Under Kelley and Arrowood's
(1960) hypothesis the decrease in frequency of BC coalitions Chertkoff (1966)

"correction" of such a misperception of the

observed could be attributed to a
power relationships.

An explanation of Chertkoff's (l96b) data that does not resort the
mechanism of subject error has been proposed by Cole (1969). Cole (1969)

examined coalition formation in a truel (three person duel) under two

conditions of certainty. In the deterministic condition, any attack

a coalition or an individual directed at an opponent was executed with a
trobability of 1.0. In a probabilistic condition, coalitions or indi-
vidual actors were successful in their attacks with a probability of .50.
Since no coalition in the probabilistic condition could be successful
with certainty, no minimum winning coalition could be defined. To gen-
erate predictions comparable to thase of Minimum Resource Theory, Cole
(1969) suggested the following proposition:

Participants in a probabilistic situatimn will prefer to form that
coalition which will maximize their chances of winning.

Cole (1969) then predicted that the stronger contender would be the
coalition choice preferred by all subjects in the probabilistic condition.
The weaker contender would be the preferred coalition choice in the deter-
ministic condition. Cole's (l9b9) first hypothesis was confirmed, while
his second was not. He attributes subjects' choice of the stronger con-
tender in tne probabilistic situation to the operation of a desire for
security. This desire is seen as a by-product of the same structure. It
becomes salient when subjects cannot be certain of executing chosen game
strategies (Cole, 1969).

Cole's (1969) concept of a desire for security is quite different
from Homans' (1961) distributive justice and Gamson's (1900, l96la) parity
principle. The parity principle is an eXpectation the subjects have of
others' desires. Cole's (l969) desire for security is actually a decision
rule the subject uses to select a coalition partner. it would be consis-
tent with this discussion to rephrase Cole's (lQbQ) security norm as a
decision rule alternative to the Gamson's (l9blb) maximization rule:

2c. When the payoff to any coalition cannot be obtained with

certainty, but only with some probability, individuals will

seek to maximize their chances of winning by forming the
coalition that will maximize that probability of winning.

r J
b"

When the probability of winning is perceived as a function
of the total weight or combined resources of the coalition
members, persons will attempt to form the coalition with
the greater resource weight.

Phrased in this way, the security role describes a strategy found by
Leiserson (1966), namely an urgency on the part of some subjects to get
the benefits of a winning coalition no matter how small a share one had
to accept.

Chertkoff (l9b6) noted that the contact prOCess served as a sensi-
tive indicator of a large portion of a liance forming behavior. In every
position the probability of picking the weaker competitor was greater
than the probability of picking the stronger. The difference from the

_ - 2 ,
null hypothes1s P(N) = .3 was highly significant (X = 18.32, p < .001).

df

Not all subjects, however, followed this Minimum Resource Theory strategy.
A minority favored the stronger competitor. SeVeral types of behaviors
may be represented by these non-minimal winning coalition choices. The
Kelley and Arrowood (1960) hypothesis is that these few subjects correct-
ly perceived the real power of all three contenders to be equal and that
the majority misperceived this power distribution. A positive hypothe-
sis, though, can postulate a decision rule to explain this behavior.

Two possible explanations are snéacsted by these data. Each is con-
sistent with a different interpretation of bamson's (1964) Anti-Competi-
tive Theory. First, the occurrence of a plurality of 33-33 coalitions
may simply be an unintended consequence of the pair-wise negotiating pro—
cedure. Forming a three member coalition requires two rounds, forming a
two-man coalition requires only one. A 33-33 coalition may simply be the

"security" strategies.

result of mutual
Gamson's (l96la, b) origi al study of the convention paradigm sheds

some light on one sort of alternative decision rule that some subjects

may be using here. Camson (lfioA) notes that the number of unexpected

large 35-}5 coalitions formed in his second game suggests that some form

of anti-competitive behavior may have occurred. Second, coalitions be-
tween equally matched contenders suggests an obvious division of payoff--
fifty-fifty. Schelling (1960) noted that psychologically salient out-

comes may suggest distinct Strategies. The saliency of some set of possible
outcome allocations may affect subject's choices in this situation. The
remainder of this chapter will review research developing this line of
argument.

Phillips and Nitz (1968) conducted owo studies of the decision rules
used in the contact process that Suggest an explicit set of alternate
decision rules. The method used was a paper-and-pencil measure called
the Political Decision Questionnaire, or PDQ. The PDQ has taken various
forms, the first of which reads as follows:

Assume that you are the manager for a candidate in a political

party convention. There are a total of 300 votes among the dele-
gates and at least a majority (151) of these are required to win the
nomination. Your man, Candidate A, has votes pledged to him.
Candidate B has votes pledged to him, and Candidate C has

votes pledged to him. Which of the other two candidates, B or C,

will you approach first to try to make a deal? (Phillips and Nitz,
1968)

The following scheme was used to distribute the convention votes
among the three candidates. The Subject always played the role of the
Faction A representative. One of the two remaining factions, B or C, was
designated w, the weaker faction, and the other was designated S, the strong-
er faction. The subject always had the same number of convention votes as
either W or S. The following relationships then existed in the resource

distribution:

 

I»)
\J

A + H + S = Total number of votes
5 ((A + la")
and either
A = W (Caplow's type 2 triad)
or
A = S (Caplow's type 3 triad)

The probability that the subject will choose the weaxer candidate can be
designated p(W). The probability of choosing S, the stronger candidate
can be designated 1 - p(W).

Minimum Resource Theory predicts that A will choose the weaker candi-
date, w, with a probability greater than .50 in both the Type 2 and Type 3
triad resource distributions. Phillips and Nitz's (l908) data confirmed
this prediction. The data also showed that the p(w) was greater when
A = S than it was when A = W. Phillips and Nitz (l908) hypothesized that
this shift in preference toward the smaller contender was the result of
the operation of some kind of anticomnetitive norm. That is, some subjects
sought to reduce intra-coalition competition over division of the relative—
ly indivisible payoff-—a nomination—-by choosing the man whose resources
were not equal to his own.

This hypothesis was examined more rigorously in a study by Nitz and
Phillips (in preparation). The difference between [p(K), A=W1 and
[p(W), A=S] observed in the Phillips and Nitz's (1968) study was seen as
a function of the subject's perceptions of the divisibility of the payoff.
The experimental manipulation in Nitz and Phillips (in preparation) varied
the nature of the payoff or outcome of the convention. Three conditions
were used: an easily divisible payoff, a payoff that could be divided

only unequally, and a payoff that could be divided only extremely

 

 

28

unequally. Each PDQ began as follows:

A political party is divided into thrEe factions or
groups. These groups are designated Fa tion A, Faction B,
and Faction C. The party is having a convention. There

 

are 300 delegates to this convention. Each delegate has
one vote. Faction A has delegates (votes). Faction
B has delegates (votes) and Faction C has dele—

gates TUBIés).

In approximately half of the forms Faction A had 83 votes, Faction B
or C had 85 votes, and the third faction had 130 votes. The subject's
group, Faction A, was thus equal to the smaller of these two competitors.
In the other half of the forms, Faction A had 115 votes, either B or C
had 115 votes, and the third faction had 70 votes. The subject thus had
resources greater than those of the weaker opponent. The three forms
continued as follows:

Form 1. Easily divisible condition.

The major business of the convention is to decide how many
of lOO political jobs each faction will receive. Each faction
would like to get as many of these 100 jobs as possible. It is
standard procedure for two factions to get together and agree on
some division of these jobs. If these two factions have a maj-
ority of the votes of the convention (at least 151 votes) be-
tween them, then the jobs are divided acttrding to their agree-
ment. An alliance between Faction A and Faction B would have

votes. An alliance between Faction A and Faction C would
have votes. An alliance between Faction B and Faction C
would have votes. Assume that you are the representative
of Faction A( votes). Which of the other two factions would
you try to contact first to try to make a deal for the political
jobs?

 

Form 2. Unequally divisible condition.
The text of Form 2 was much the same as that for Form 1, with the ex»
ception of the following critical passage:

The major business of the convention is to nominate a candiv
date for governor and a candidate for lieutenant governor. Bach
faction would like to have its man nominated for governor, but
would not be extremely dissatisfied if he received only the lieu—
tenant governor nomination.

 

Form 3. Very unequally divisible condition.
The critical passage in Form 3 was:

The major business of the conVention is to nominate a
candidate for governor and a candidate for lieutenant gover-
nor. The governor's office is a very powerful position. The
post of lieutenant governor, however, has generally been a
political dead end for the candidates elected to it. (Nitz
and Phillips, in preparation).

The three conditions above provided the experimental situation to
test the operation of four different decision rules in the contest phase
of the political convention paradigm. The subjects were assumed to eval-
uate others' expectations on the basis of the parity principle. Four al-
ternate decision rules the subjects might use are described by the follow—
ing general hypotheses:

1. Maximization

Subjects seek to maximiZe their share of the payoff with respect
to their coalition partners, regardless of the divisibility of
the payoff .

2. Dominance

Subjects prefer to he the dominant member of any coalition,
regardless of the divisibility of the payoff.

3. lntra-Coalition Compatibility

Subjects seek to form coalitions in which the division of

the payoff can be negotiated with a minimum of intra-coalition
friction. When the payoff is unequally divisible, the pro-
bability of choosing the unequal contender will be greater
than when the payoff is easily divisible. When the payoff is
easily divisible, no intra-coalition incompatibilitv'will

be engendered in a coalition between equals or in a coali-
tion between unequals, so the maximization principle will

be expected to hold.

4. Equalitarianism

If the payoff is equally divisible, subjects seeking to
minimize intra-coalition Competition will prefer to form
coalitions with equals more than with unequals-—regard-
less of whether he is equal to the stronger or the weaker
competitor.

These hypotheses are not totally independent. it was anticipated that

the majority of the subject population would choose a maximization strat—
egy in all payoff conditions--but that the divisibility of payoff would
determine the size of that majority and the prevalence of the latter three
marginal strategies.

The results of the Nitz and Phillips (in preparation) study are shown
in Table 2.1. These data support the maximization and the compatibility
hypotheses above. The maximization hypothesis assumption predicts that
p(W) would be greater than .50 for any resource distribution. The data
show that this is the case. The dominance hypothesis is not conditional
on divisibility of payoff. it predicts p(W) will be high-—and constant
across conditions of payoff divisibility. The difference between p(W) for
the easily divisible condition and p(w) for the unequally divisible con—
dition at the point A=85 disconfirms the eXpectation that there will be no
difference. Hypothesis three, intra—coalition compatibility, predicts that
p(W) for easily divisible condition will be greater than p(w) for the un-
equally divisible conditions at A=85. This hypothesis is confirmed by the
data presented in Table 2.1. Finally, hypothesis four assumes that an
equal division is the most easily negotiated division and predicts that
when the payoff is easily divisible, p(w) will be higher if A=N than if
A#W. That is, it predicts an effect exactly the Opposite from the effect
observed: p(W) at A=83 should be greater than p(W) at A=llS. The observed
difference is in the opposite direction.

The data in Table 2.1 indicate that the intra—coalition compatibility
hYpothesis is the most reasonable explanation of the deviation from the

Strict maximization observed in the Philliws and Kitz (l9o5) data.

Table 2.1

Prohahility
of resource

lypv

Payoff Divisibility

Conditions [(W)
l. Easily divisible 49
2. Moderately non-
equally divisible 36
3. Extremely non-
equally divisible 48

letween divisibility conditions

c1 c.

/ , ”

of choosing

2 Triad

= W

fiS) pVW)
l0 .830
19 *333
Ql .676

I) .
X‘ comparisons

t, c1
#.Ubl --
.06

Type 3 Triad

{(W)

31

56

Azs

f(S)

9

10

the smaller contender as a function
distribution and divisinility of payoff

P(W) X

.850 .08

.833

.4

-ntegration of Strategy Selection hypotheses

F

 

The findings of the Gamson (l9hlh), Chertkoff (1966), Phillips and
Nitz (1963) and hitz and Phillips (in preparation) studies suggest the
importance of the contact phase in the process of alliance formation.
Phillips and Eitz (1968) and Nitz and Phillips (In preparation) have
demonstrated the viability of tne maximization assumption strategy of
Hinimum Resource Theory for social contacts in the Type 2 and Type 3
resource distributions. They have also shown that perceived indivisibil-
ity of the payoff causes some persons to make contacts that would tend
to minimize eventual intracoalition conflict (Nitz and Phillips, in prepar—
ation). heir conclusion that the maximization strategy of Minimum Re-
source Theory explains most coalition partner choices and that the major
deviations from the maximization choice can be interpreted as strategies
that would reduce intra—coalition competition is subject to three strong
limitations.

In this section I will examine these limitations, suggest means of
dealing with them, and present the elements of a theory of strategy se—
lection whose major purpose is to eXplain those marginal strategies not
predicted by the simple maximization hypothesis of Minimum Resource Theory.

The first limitation of Phillips and Nitz (lane) and Nitz and Phillips
(in preparation) theoretical interpretation is related to Cole's (1969)
security principle. The security principle asserts that when strategies
can succeed with probabilities less than unity, individuals will choose
Coalition partners so as to maximize their chances of winning. The politi—
Cal convention paradigm can be interpreted as either a probabilistic or a
deterministic game. If subjects conceive of a coalition formed in a

Convention to nominate a candidate as a necessary compaign force for

33

an impending general election, then the security principle would be rele-
vant to their coalition choices. If they placed higher value on winning
the general election than on attaining the top spot on the ballot, they
would attempt to build the largest coalition possible. In the Nitz and
Phillips (in preparation) PDQ such a strategy would be indicated if

p(K) = p(K)
Type 2 Type 3 < l/2

in the unequally divisible nomination condition. This security strategy
asserts a preference for the smaller contender le:s than half of the time.

Another finding might be attributable to security strategies in the
indivisible payoff convention situation. If some subjects perceived the
uncertainty of a future general election as more salient when they are
relatively weak than when they are relatively strong, they might choose
security strategies when in a position of weakness and compatibility
strategies when in a position of strength. This combination of strategies
would lead to predictions like this:

P(W) > P(W)
Type 3 Type 2

That is, subjects would tend to choose the different (smaller) man in the
Type 3 situation, where they were equal to the larger opponent. In the
Type 2 situation, they would tend to cheese the larger man (who controlled
3 different amount of resources). Moreover, if both of these strategies
were found in a population that predominantly used Maximization strategies,
the data would take the form:

p(W) > p(fi) > l/Z
Type 3 Type 2

But this is exactly the effect Nitz and Phillips (in preparation) identified
as Intracoalition Compatibility! Nitz and Phillips' (in preparation) pro-

Vide no means of distinguishing the alternative individual strategies that

3:.

might account for this aggregate statistic. This failure to discriminate
between security and anti-competitive strategies is in part due to an
inadequate conceptualization of the nature of these strategies.

Thus the theoretical implications of the intracoalition Compatibil—
ity hypothesis proposed by Nitz and Phillips (in preparation) are not clear.
A replication of Nitz and Phillips' (in preparation) eXperiment with
alternate unequally divisible payoff conditions would provide a test of
the conditions under which the Intra-coalition Compatability hypothesis
identifies a form of Security strategy and the conditions under which it
identifies a strategy that is independent of security motivation.

Three critical convention situations for a test of the distinction
between Security and Compatibility strategies can be constructed. Condi-
tion I would be identical to the easily divisible payoff condition used
by Nitz and Phillips (in preparation). Condition 11 would be identical
to their unequally divisible condition, with nominations for the ballot
positions for a state governorship and lieutenant governorship. This
condition would represent a probabilistic payoff condition. While this
condition does not directly manipulate the probability of winning, it
replicates the ambiguity about the probabilistic dimension found in Nitz
and Phillips' (in preparation) PDQ. Condition Ill would remove this am-
biguity by presenting a convention in a one-party state. Thus the success-
ful nominee need not be concerned about organizing support for a general
election campaign. If he wins the nomination in the convention, he is
virutally assured of the office itself.

All three conditions will provide tests of the Maximization hypothe-

sis:

35

l. Maximization
Subjects seek to maximize their share of the payoff with
respect to their coalition partners. Their choices are
not altered by the divisibility of the payoff or the pro-
bability of winning in a subsequent contest. This strat—
egy is expected to occur with a probability of .50 or more
under all three experimental conditions.
The following strategies are essentially marginal strategies. These strat-
egies may or may not be independent of each other, but are not generally
independent of the maximization strategy. They are defined so as to ex—
plain a portion of the subject population's behavior that is not accounted
for by the Maximization hypothesis. The proportion of non—maximization
strategy choices they explain is expected to vary as a function of the ex-
perimental condition.
2. Intracoalition Compatibility
Subjects seek to form coalitions in which the division of
the payoff can be negotiated with a minimum of intracoali-
tion friction. 'hen the payoff is unequally divisible,
the probability of choosing the unequal contender is higher
than when the payoff is equally divisible. When the pay-
off is easily divisible, no intracoalition incompatibility
will be engendered in a coalition between equals or in a
coalition between unequals, so the maximization decision
rule will be used.
Here Condition III provides a clearly non-probabilistic form of an unequal—
ly divisible payoff. To the extent that Phillips and Nitz' (1968) and Nitz
and Phillips' (in preparation) subjects perceived the payoff as simply a
larger or a smaller nomination and ignored the probabilistic aspect of
the nomination, their results should be replicated in this condition.
Conditions 11 and III test the independent hypotheses that the
security strategy is elicited by a probabilistic payoff situation:
3. Security
When the payoff to a coalition can be attained only with some
probability less than unity, subjects will seek to maximize

their chances of winning by forming the largest coalition
possible.

36

SubjeCts are expected to choose Candidate W less often in Condition 11
‘than in Condition III in both the Type 2 and Type 3 triad, indicating a
tendency to form larger coalitions under conditions of probabilistic pay-
off. An indication of the contribution of security strategies to the
occurrance of lntra-coalition Compatibility behaviors is given by the

difference

p pm) -p('w') \
Type 3 Type 2) Condition 11

/ ,.
.. ’kaw) -p(w)
Type 3 Type 2 Condition III

The test of this difference, moreover, is independent of the two hypothe—
sis tests discussed above. If the difference given by the above equation
is positive and significantly different from zero, it will indicate that
the Intra-coalition Compatibility strategies identified by Nitz and Phillips
(in preparation) can be attributed more correctly to Security choices.
Otherwise, it will suggest that the effect observed by Nitz and Phillips
(in preparation) is entirely the result of subjects' compatibility strategy
choices.

These tests should resolve the ambiguity between the Security and the
Intra-coalition Compatibility strategies identified by Cole (1968) and by
Nitz and Phillips (in preparation). Three additional marginal strategies
may now be presented. All three are marginal to the simple Maximization
strategy, but two are closely related to lntra—coalition Compatibility.

4. Equalitarianism

If the payoff is equally divisible, subjects seeking to min—
imize intra—coalition competition will prefer to form coali-
tions with equals more than with non-equals. (No stipulation

will be made here as to the effect of probabilistic outcome
on this decision rule.)

This hypothesis provid“s a plausible counter-proposition to explain the
strategy choice Phillips and hitz (1968) designated as weak anti-competi-
tive behavior. Choosing Candidate K more often when the payoff is easily
divisible may be a result of situational cues suggesting an equal split.
Nitz and Phillips (in preparation) argue that an Lqualitarian Strategy
would be indicated by

POW?) > PU‘D > l/2
Type 2 Type 3

for the equally divisible condition, and

p(w) > p(w) > 1/2
Type 3 Type 2

for the unequally divisible condi ion. Yet the Equalitarian prediction
for the equally divisible condition differs from the prediction of the
Intra-coalition Compatibility hypothesis only insofar as Compatibility
predicts

p(W) = p(W)
Type 3 Type 2

A more general strategy that incorporates both would be defined:

P(W) 2 D(W) > l/2
Type 2 Type 3

for the equally divisible Condition l, and

p(W) >p(w) >1/2

)

Type 3 Type L
for the unequally divisible condition, Condition Ill.
The second strategy related to lntra—coalition Compatibility is not
a subset of Compatibility behaviors, but rather is the complement. The
Competition strategy can be defined:
5. Competition
Subjects will seek to form coalitions that will allow maximum
grounds for conflict; that is, coalitions for which the struc-

ture of the resource distribution suggests no obvious division
of the payoff.

1

Tue Competitive strategy hypothesis predicts behavior exactly opposite
of Compatibility behavior:

p(l~') > p(l~’) > 1/2
Type 3 Type 2

for Condition 1, and

p(M) > p(N) > 1/2
Type 2 Type 3

for Condition III. The Competition strategy is obviously not independent
of Compatibility. One is necessarily accepted if the other is rejected.
The final strategy pattern to be examined here is the Dominance Strat-
egy:
6. Dominance
Subjects prefer to be toe dominant member of any coalition.
Their choices are not altered by the divisibility or the
payoff or by uncertain prospects of final attainment of the
payoff.
The Dominance hypothesis is dependent on all of the preceeding marginal

hypotheses. If any of the other hypotheses are Confirmed, then the con-

ditions of the Dominance hypothesis are not met. These conditions are

as follows:

PW) 4' pW) .. 1/2
Type 3 Type 2

for Conditions I, II, and Ill.

The second limitation to the generality of Nitz and Phillips' (in prep-
aration) findings is in a sense methodological. Camson (19613), Chertkoff
(1966), and Cole (1969) gathered their contact data in an experimental sit-

uation in which subjects played in each others'

presence. The PDQ studies
gathered data in a classroom situation in which subjects were asked to

imagine their opponents. A high degree of similarity between the induction

39

of the PDQ studies and the eXperimental induction in other studies of
coalition formation in the political convention paradigm is insufficient
grounds to justify the assertion that the PDQs and the political conven-
tion paradigm experiments yielied the same sort of social contact data.
The comparability of the data from these two forms of the political con—
vention paradigm must be demonstrated empirically. It is therefore nec-
essary to introduce two experimental conditions, one in which subjects
are given PDQs, and another in which they make essentially the same types
of choices in an interactive three person political convention game.

The third restriction on the generality of the Phillips and Nitz (1968)
and Nitz and Phillips (in preparation) hypotheses is that the identifica-
tion of marginal strategies is essentially the identification of strategies
selected by only a portion of the subject population. The development of
an experimental paradigm that will permit positive identification of indi-

vidual strategies is the t0pic of the next chapter.

Chapter III
Situation Structure, Cognitive Complexity, and Eachiavellianism
as Determinants of Strategy Selection

The Nitz and Phillips' (in preparation) study was among the first
to examine a set of systematically related coalition strategy hypotheses
under alternative experimental conditions. Their use of two convention
resource distributions (Type 2:8 <(A+W) A=w; Type 3:8 <<A+W), A28) and
three payoff divisibility conditions provided an experimental design that
would necessarily either confirm or reject the Maximization hypothesis,
and could support one and only one of the marginal strategy hypotheses.

The design of the Nitz and Phillips' (in preparation) study, however,

does not permit identification of individual's strategges. It provides

 

information only on the existence of the strategy in the population. The
two critical statistics, p(W) for the Type 2 triad and p(W) for the Type 3
triad within any one payoff condition, are based on the choices of inde-
pendent samples of subjects. Since a major goal of this study is to
identify individual strategy patterns and to predict these strategies from
individual difference measures, a situation must be constructed that will
provide Type 2 and Type 3 choices for each individual.

Identification of Individual Stratccics

 

Nitz and Phillips (in preparation) were atlc to identify four distinct
strategies on the basis of joint consideration of their subject populations'
choices in Type 2 and Type 3 triad P045. Their suhject's choices of either
Faction (W) or Faction (S) yielded only one hit of information. The four

mutually exclusive strategies they defined, however, were necessarily based

on two bits of information. if subjects were asked to make two choices,
one in a Type 2 triad and one in a Type 3 triad, they would supply two
bits of information. This would be sufficient to identify four alternative
strategies among the individuals in the subject population. Joint consid-
eration of the four possible combinations of the two choices a person could
make in the two coalition situations should permit the identification of
four strategy types.

The choice measures can be obtained for a subject if he plays in a
Type 2 triad game and a Type 3 triad game. Since we are concerned here
only with social contact behaviors, they may be obtained equally well if
he takes a Type 2 PDQ and a Type 3 PDQ, or if he plays a Type 2 game and
takes a Type 3 PDQ. Moreover, if the subject is given no feedback he is
subject to less uncontrolled social interaction that would unpredictably
provide cues for his choices in the second PDQ.

The four possible combinations of choices in the two PDQS, are
illustrated as follows:

Figure 3.1

Joint Choices in Types 2 and 3 Triads

Type 2 Type 3

l. E w
2. E E
3. S W
4. S E

E = A choice of the equal resource opponent;

W = A choice of the weaker resource opponent;

S = A choice of the stronger resource Opponent.

L\
to

These four joint choice patterns can he mapped isomorphically into
a set of four simple strategies. These basic strategies generate a more
complex pattern of strategies when two variables discussed in Chapter II
are added to the design.

Strategy Interpretation

Indivisible or uncertain payoffs will change the interpretation of
these four strategy patterns in much the same way as they did in Chapter II.
I will discuss each of the four choice patterns above and elaborate the
changes in strategy they represent as a function of the divisibility and
certainty of payoff.

Choice pattern No. 1, ”E” in the Type 2 situation and ”W” in the
Type 3 situation, suggests only one individual strategy. According to
the Maximization hypothesis, choosing the weaker contender in both the
Type 2 and Type ; triads would indicate a maximization strategy. This
choice pattern indicates the same strategy regardless of the payoff
condition. In the Type 3 triad a choice of the weaker contender promises
the highest payoff, while in the Type 2 situation, choice of the weaker
contender promises either an equal payoff or at least an equal chance for
the higher of the indivisihle payoff.

Two different strategies may be inferred from choice pattern No. 2(EE),
the choice of the equal contender in both Type 2 and 3 triads. In Condition
1, choosing the equal-resource competitor suggests an explicit division of
the easily divisible payoff: an equal split. This sort of division could
be preferred because of a preference for eoaalitarianism, a dislike of
negotiation, or a desire to avoid intra-ccalition conflict. In the un-
equally divisible Conditions 11 and III choice of the equal contender

would not take advantage of the strategy suggested by the unequal division

or payoff. Thus, such a choice implies competitive motive or desire to
negotiate the division of payoff in the absence of any apparent guidelines.
Figure 3.2 shows how this choice pattern leads to different strategies for

the two payoff conditions.

Theoretical Strategies by Divisibility Condition

 

Condition II Condition 111
Condition I Snequally Unequally
Choice Easily Divisible Divisible
Pattern Divisible Probabilistic Certain
(1) EN Haximization Naximization Maximization
(2) EH Intra-coalition Competition Competition
compatibility
(3) SW Competition Intra-coalition Intra-coalition
compatibility compatibility
(4) SE Security Security Security

 

The third choice pattern, the choice of the stronger contender in
the Type 2 convention and the weaker contender in the Type 3 convention
(SW) suggests two different strategies. In the easily divisible condition,
choosing the different rather than the equal opponent does not take advan-
tage of the obvious payoff division suggested by the situation. Choice of
the unequal competitor suggests a willingness to compete over the division
of the payoff. In Conditions II and III choosing the unequal contender
is consistent with the obvious unequal divisibility of the payoff, and
leads to minimal intra-coalition conflict.

The fourth choice pattern, the choice of the stronger man in the
Type 2 situation and the equal can in the Type 3 situation (SE), suggests

a pure security strategy in Condition II. There would seem to be no strategy

 

 

 

advantage in making these choices in Conditions I and III, where any
coalition is certain to win the payoff. If p(SE) in Condition II is
significantly greater than p(SE) in Conditions I and III, a security
interpretation can he made for p(SE) in Condition II. If there is no
difference, the security hypothesis will not serve as an adequate strategy
description for (SE) choices in the political convention paradigm in
Condition 11.

As noted in Chapter II, the Maximization strategy is likely to be
the most frequently chosen strategy. it is also a strategy which is asso-
ciated with a constant response across all divisibility conditions. It
would be relatively simple to explain a lack of effect due to individual
differences if this were the only strategy examined. The three remaining
strategies are essentially marginal strategies. But these are precisely
the strategies that fluctuate with changes in the structure of the com-
petitive environment. This observation suggests that these strategies in
particular may be susceptible to individual variations.

Erediction of Individudl Strategies

The theory of coalition Strategy selection outlines in Chapter 11
consists of two classes of hypotheses. The first type of hypothesis
deals with the evaluations an individual makes about his competitive
environment. The only hypothesis of this class developed in this study
is stated:

A. Any participant will expect others to dewmnd from a

coalition a share of the payoff prOportional to the

amount of resources he contributed to the coalition.
The second class of hypotheses is a set of decision roles that specify
the strategies to be selected under different environmental conditions:

the divisibility of payoff and the certainty of payoff. Each of these

~ .—

L\
U)

hypotheses assumes that the individual perceives others to have a parity
expectation. Above we develOped a paradigm in which an individual's
decision rule can be identified as one of four mutually exclusive strategy
types: Maximization, Competition, Security, or Intra-coalition Compat-
ibility. This chapter will present the thesis that the strategy an
individual chooses is a function of the way he perceives the situation,
particularly of the estimate he makes of his opponents' expectations in the
situation. The way he perceives the situation is, in turn, a function of
either his cue utilization capacities or his orientation toward power,
control or competition.

Two measures of the style in which people perceive social situations,
namely Machiavellianism and cognitive complexity, are particularly
relevant to the prediction of strategy choices. Information about an
individual's level of cognitive complexity, as an indicator of patterns
of cue utilization should permit the prediction of the manner in which
he perceives a social situation. Machiavellianism, as an orientation
toward power, should also differentiate patterns of estimating the com-
petitive environment.

It would be desirable to predict specific patterns of social per-
ception on the basis of cognitive complexity or Machivellianism scores
and to predict strategies from these perceptions. While social per-
ception has been studied extensively, no systematic research has examined

1

perceptions of coalition formation situations. Th1 study will not under-

(.0

take a major analysis of social perception. It will, however, pose
hypotheses linking cognitive complexity and Nithithllianism to the

subject's strategy selection Patterns.

46

CogniLLVe toggle it:. A Substantial amount of research has linked

individuals' strategy selection patterns to measures of their cognitive

. l . . . ,i, p . . . . . .
compleXity. Cognitive compleXity refers here to an indiVidual s ability
to integrate conflicting information. Four nodal systems of integrative
complexity have been described in detail by harvey, Hunt and Schroder (1961).
These may be summarized as follows:

System I. At this lowest level of integrative complexity, the

rules or schemata for categorizing stimuli are highly fixed and

simple. Simple schemata, norms or authorities help the individual
structure his environment in a complete and unyielding way. System I
individuals are characterized by categorical, black—white thinking,
minimization of conflict and avoidance of ambiguity, self definition

in terms of external anchors, preservation of standards and minimization
of alternatives and over-generalization of fixed approaches or stereo-
types.

System 11. Schemata for categorizing stimuli are still relatively
simple, but more alternatives are perceived. The System 11 individ-
ual perceives his world against a background of self vs. others,

and accepts self, while rejectina others. This leads to an absolut-
istic orientation toward others who, when seen in a position of
potential control are "warded off'fi The individual is detached and
negatively independent.

System III. At this level both the self and other people are

highly differentiated. This enables the System Ill individual

to be highly sensitive to others and to attempt to match his
perceptions to those of others. He is highly capable of putting
himself in the role of others and perceiving himself as others
perceive him. He is oriented toward maintaining close interpersonal
relationships; rejection is threatening.

 

System IV. At the highest level of integrative complexity a
diverse world filled with many alternatives is perceived. The
System IV individual is highly autonomous and reacts to people

as a source of information. He generates a large variety of
alternative interpretations of events and can thus react to the
subleties of his environment with appropriate and novel responses.

 

1Realatively complete bibliographies of these studies are found in
Schroder, Driver and Streufert (1907), Harvey, Hunt and Schrdoer (1961)
and Harvey (1966). Vannoy (1907) provides an informative comparison
of a number of different measures of cognitive complexity.

Harvey (1966) presents preliminary results of a number of studies
designed to validate these conceptual systems constructs. Subjects were
categorized as System I, II, III or IV on the hasis of their responses
to a projective measure, the ”This I Believe” test (TIB). In addition
subjects in the various studies took some subsert of a group of standard
attitude scales or measures on tasks designed to indicate change of set.
The four systems may be characterized in terms of some of the more
significant differences on these measures as follows.

System 1 persons tend to be cognitively simple on Kelley's (1956)

Role Rep Test, high on Authoritarianism (Adorno, 1950) and Dogmatism
(Rokeach, 1964), highly rigid (Cough-Sanford, 1954), high in Edward's

EPPS (1934) Deference subscale and low on the Change and Autonomy sub-
scales. System 11 persons are the second most simple group on the Kelley
(1956) test, low on EPPS Deference and Affiliation, high orlAutonomy and
Aggression and high on Machiavellianism. System 111 subjects ranked
second most complex on the Kelley (1954) test, highest on EPPS Affiliation,
low on Autonomy and Change. Finally, System IV persons ranked highest in
complexity on the Kelley test, lowest on Authoritarianism, Dogmatism and
Rigidity, low on EPPS Deference and high on Autonomy and Change.

Four different tasks measured change of set, the Cottschaldt—Embedded
Figure Test, Asch's (1932) impression formation task, and two tasks requiring
subjects first to describe discrete units, then integrate them into a single
description. System 1V persons performed more successfully than any other
system on all four tasks (Harvey, 1956). On the Denny Doodlebug Problem
(Rokeach, 1960), System I subjects took more time, requested more help from

the experimenter, and used fewer cues appropriately than System 1V suhjects.

j\
’3!

harVey (190:) examined the ability of the four system types to shift
set in a more direct social setting. Subjects were asked to construct
and record arguments that contradicted their personal beliefs about a
topic. Their performance was evaluated on 18 content dimensions. System
IV subjects outperformed all other categories on all 18 dimensions.

System 11 subjects ranked second when they believed their arguments were
to be private, and last when their tape recorded irgumenLS‘were to be
played to a faculty committee.

Felknor and Harvey (1964) examined cue utilization in concept formation.
Subjects in the four systems were scored on the redundancy of the information
they requested in trving to solve tne problems. They were also scored on
whether they guessed at answers on the basis of knowledge of the relevancy
of the cues or on the basis of hypothesis disconfirmation. System I sub-
jects were significantly more redundant than Systems 11 and 1V. System IV
subjects used relevancy of cues as a basis for guessing significantly more
often than System 1 persons. System 111 persons, however, performed poorly
in the private condition, but they ranked second in the public condition.

Tuckman (1966) develOped an objective forced-choice measure of
harvey's (1961) four conceptual systems. The test presents subjects
with paired statements drawn from populations representative of the four
system types. Tuckman's instrument, the ”Interpersonal Topical Inventory”
(1T1), yields four scores, each representing the number of items endorsed
in a system category. Since these stores are not independent, the sub-
ject's highest score defines his conceptual system.

Tuckman (1966) administered the 1T1 and Schroder's (1907) Sentence
Completion Test to 146 Naval enlistees. 0f the 136 subjects, 94 were

classifiable into one of the four systems on both measures. The contingency

49

coefficient for these classifications was .54 out of a maximum C of .87

(p< .01). In a study of the relationship of the ITI to other individual
difference measures, Tuckman (1965) found that System 1 persons had the
highest F-scale Scores and Systems IVS the lowest (F=3.68, df=3/207, p <.05).
Also, System 115 were the highest on Mach IV and System Is the lowest
(F=4.78, df=3/307, p<:.Ol). System 11 persons were lower than all others

on Affiliation (F=3.24, df=3/207, p <.05; EPPS, Edwards, 1954).

hese brief characterizations suggest that the conceptual systems
constructs have sufficient fact validity to justify further research. The
findings reviewed by Harvey (1966) do not, however, demonstrate that the
four systems necessarily identify four distinct behavior patterns--in
fact, they suggest that some forms of behavior may be highly relevant to
one or two systems. To date the analysis of the effect of cognitive com-
plexity on strategy selection has dealt only with extreme subjects on the
complexity-simplicity dimension (Schroder, Driver and Streufert, 1967).

A more fruitful line of inquiry would seem to be an investigation of
strategy patterns of all four systems rather than just the two extremes.
This study examines the effect of each of the four systems on individual
strategy selection.

The ability of the cognitively complex individual to make relatively
fine discriminations would be a strategic asset only if it gave him greater
control over his environment. If, for e ample, the complex individual
perceived a large number of availatle stratEgies and selected strategies
on the basis of this perception alone (without regard to their relative
appropriateness to the situation), the complex person might well be at
a strategic disadvantage. Likewise, if he lost sight of his original goals

in the face of a host of alternative strategies suggested by environmental

50

features, complexity would be dysfunctional to successful strategy selection.
To diScriminate between these two behavior patterns, the following general

hypothesis can be advanced:

Cl. A larger proportion of relatively complex persons (Systems
111 & 1V) than of relatively simple persons (Systems I & II)
will select Maximization strategies.
Harvey's (1961) system concepts suggest the following specific strat-
egy selection patterns in addition to the general hypothesis C1. The
rigidity and need for a structured environment that characterizes the

System I person suggests the following hypothesis:

C2. A larger preportion of System 1 subjects will choose Security
strategies than will choose any other marginal strategy.

The aggressiveness and competitiveness of the System 11 person suggests
the following hypothesis:

C}. A larger proportion of System ll persons will choose a com-
petitive strategy than will choose any other marginal strategy.

The sensitivity to interpersonal relationships that characterizes the
System 11 person leads to the following hypothesis:

C4. A larger proportion of System 111 persons will choose Intra-
coalition compatibility strategies than will choose any other
marginal strategy.

ihe exploratory behavior exhibited by the System IV person suggests a

final Specific strategy hypothesis:

C5. System 1V persons will choose marginal strategies with equal
frequency.

Machiavellianism. The concept of Machiavellianism employed by Christie
is most aptly described by eeis (l903) in her validation study of the Mach
Scale:

Machiavellianism is usually understood as the willingness
and ability to use guile, deceit, and other opportanistic

strategies in order to manipulate others. It might also be
described as a love of power and a methodology of interpersonal

51

strategy for the use of power. To the extent that an individual

is Machiavellian he responds to other people impersonally, as

objects, to he used for his own ends, without regard for their

interests or for such ideals as fair-play, justice, or equality.

(Ceis, 19b3)

Geis (1963) examined the effect of Machiavellianism on competitive
behavior in a coalition formation eXperiment. Geis (1963) used a parchesi-
type game (Vinacke and Arkoff, 1957) in which players advanced markers
along a playing board. The number of Spaces advanced on a turn was equal

to the value of a ”

power card” he chose to play. The first player or
players to reach the end of the board won a monetary prize. The players
would bargain openly to form a coalition agreeing on some split of the
payoff. A coalition could be formed at any time, but it could also be
broken on any turn. When a coalition was formed, it started from a
position midway between the position of the markers of the partners. The
pair then advanced a number of Spaces equal to the product of the higher
die value and the sum of the power cards each meaber chose to play.

Three variables were eXperimentally controlled. The first, power
position, was determined by assigning decks of power cards to each of the
three subjects in each game. The second, Machiavellianism was varied by
selecting subjects who scored in the upper, middle, and lower quartile
on one form of the Mach Scale and who also scored in the same or adjacent
quartile on another form of the Mach Scale. The third variable was
amount of information available. This was varied by having subjects place
all power cards face-up on the table (an unantiguous condition) or having
them hold the cards like a hand in a normal card game (ambiguous condition).
Each subject played in all six po551ble games with different opponents in

each game.

Ceis (lBOJ) found that High Machs were more successful than Low Machs
within all power conditions in the ambiguous condition. Collapsing over
power conditions, she also found that High Machs were more successful than
Low Machs in the unambiguous condition, (Geis, 1963). No data on contact
preferences, bargaining tactics, or pattern of forming and breaking coali-

2
tions were reported in that study.

Geis, Christie, and Nelson (1963) conducted an experiment in which
subjects (ostensibly) exercised control over other persons. The eXperimental
paradigm was that of a social psychology experiment. The subject was asked
to take “an important personality test” which in fact was a modified version
of Witkin's (1938) Embedded Figures Test. The test administrator was por-
trayed as a student learning how to administer the test. During the test
the administrator made some comments and falsified some of the feedback the
subject was being given on his performance. The experimenter debriefed the
subject and the confederate, telling them that the experiment was really a
study in frustration tolerance. The subject was then asked to serve as
test administrator for another subject. He was induced to lie about his
subject's performance and attempt to distract and upset the subject. The
”second subject” was in fact another confederate who paced his own per-
formance by watching signal lights that were not visible to the subject
administering the test.

Geis, Christie and Nelson (1905) found that High Mach subjects gener-
ated more manipulative or distracting tactics, used more imaginative,
sophisticated and uncommon tactics and performed with much more ”verve"

than the Low Mach Subjects. The Low Mach subjects, however, used more of

 

2This type of data was discussed in two reports, but these have not
been made available to this writer (Ceis, 1964; Ceis and Christie, 1965).

53
the tactics that had been explicitly suggested in the experimental in-
ducement. (Ceis, Christie, and Nelson, 1963). Moreover, the High Mach's
reported that they enjoyed the experience and the opportunity to Hplay
God“, while the Low Mach's did not feel eSpecially comfortable in their
experimental role (Geis, Christie and Ne18on, 1905).

A study of behavior in reSponse to experimental conditions that in-
troduced distributive injustice into a work-relationship provides some
insight into the tactical usage of Machiavellians (Blumstein and Weinstein,
1967). Subjects were given the task of composing attitude scale items
about college life. The task involved working with another subject who
was actually a confederate. The pairs composed one set of statements;
the stooge did either two-thirds of the work or one-third of the work.
Each was asked to write down the portion of the work he performed so the
experimenter could determine how much credit toward a course grade each
participant would receive. The Stooge claimed either one-third or two-
thirds of the credit. The eXperimenter read the claims and gave the pair
another topic. The subject and confederate were asked to compose a set
of items and to record their contributions once more.

On the second claim for credit, High Nachs were willing to take
earned points from a stooge reluctant to claim them for himself. They
tempered their demands when dealing with a partner who had previously
requested the lion's share (Blumstein and Weinstein, 1967).

These three studies all Suggest that Machiavellianism is related in
some way to manipulative or competitive behavior. From Geis' (1963)
study it is not possihle to deterane just what the High Hachs did to
become more successful. heis, Christie and Xelson (1963), though, suggested
that High Machs may he more creative in aenerating interpersonal communi-

cative (or disruptive) acts.

54

A study by DJwSon and Phillips (in preparation) provides some sugges-
tion of Machiavellian strategy choices in the political convention situation.
A Political Decision Questionnaire was given to 104 subjects. The ques-
tionnaire contained eight independent convention questions. The payoff was
relatively indivisible, a political nomination. The total number of votes
for each convention was 300. Three of the questions placed the subject
above the equalitarian point with 1&4, 132, and 102 votes. Three more
questions placed him below the equalitarian point with 98, 85, and 77
votes. One question put him at the veto point with 75 votes and another
in the dictator region with 73 votes. The PDQ also included Form V of
Christie's (1962) Machiavellianism Scale.

Subjects' Mach Scale scores were significantly (p (.05) correlated
with their choices for the points w = 75, 77, 98, and 102 votes. Sub-
jects with High Mach Scores tended to choose candidate w more often than
those with Low Mach scores. According to the strategy system outlines in
Figure 1.1 the choices the High Machs in Dawson and Phillips (in prepara-
tion) study could represent either maximization or competitive strategies.

Dawson and Phillips (in preparation) indicate that High Machs adOpt
a more competitive contact strategy than Low Machs. It would seem that
all three studies confirm some portion of Geis' (1963) definition of
Machiavellianism. Such confirmation still does not systematically ex-
plain what Machiavellians do and the conditions under which they do it.
Returning to Geis' (1963) definition, we find that a Machiavellian may
(1) be deceitful, (2) love poWer, (3) he skillful in the use of power,

(4) act without reference to moral scruples, (5) act without regard to
any particular social norms. This set of descriptors, though, covers an

innumerable range of social activities or interpersonal tactics--but says

nothing at all abOut the situations in which each of the many different
hypothetical forms of Maehiavellian behavior might appear. There is no
explicit algorithm for deciding which kinds of situations are amenable to
Machiavellian tactics, since the tactics described cover almost all of
human behavior.

In the political convention paradigm we can entertain two general
hypotheses about the strategy selection patterns of High or Low Machia-
vellians. The first is comparable to the Maximization hypothesis for
cognitive complexity systems:

C6. The proportion of High Mach persons who choose Maximization
strategies will be larger than the proportion of Low Mach
persons who choose maximization strategies.

The second hypothesis deals with the Machiavellian's choice of

marginal strategies:

C73. A larger proportion of Low Mach persons will choose security
strategies than will choose any other residual strategy.

C7b. A larger proportion of High Mach persons will choose compet-
itive strategies than will choose any other residual strategy.

Measures

The ”This I Believe” test devised by Harvey and Schroder's (1967)
Sentence Completion Test are projective measures which require a group
of well trained coders to render subject's responses capable of analysis.
Tuckman's ”Interpersonal TOpical Inventory” is a relatively simple forced-
choice objective measure. Since this form of test may be scored by an
appropriate computer program, it was used to measure cognitive complexity
in this study.

Two versions of the Machiavellianism Scale have been used in research
to date. The most popular, the Mach IV, is a Likert-format inventory.

The Mach V is a forced-choice test composed of 20 triplets of statements.

5o

Scores for each item Set are obtained by asking subjects to indicate the
statement they most nearly agree with and the statement they most disagree
with. A Machiavellian response is scored whenever a Machiavelli paraphrase
is ranked higher than an equally undesirable unrelated statement. A
Machiavellian response is also scored when a reversal of a paraphrase of
lachiavelli is ranked below an equally socially desirable statement. An
individual's Mach Scale score is the number of item sets scored as Mach-
iavellian responses. The Mach V is counterbalanced to control for social
desirability effects elicited by the negative tenor of many of the Mach

Scale items (Christie, 1962). The Mach V form will be used in this study.

Chapter IV

Objectives and Procedure

Objectives

 

The introduction to this study set two primary goals. The first
was to develOp a theory of strategy selection in coalition situations
based on the distribution of resources and the divisibility of the payoff,
two major aspects of a mixed-motive environment. The second was to pre-
dict individual's strategy choices from measures of his skill in utilizing
situational cues and of his orientation toward power and control. This
chapter will restate the hypotheses developed in the preceding discussion,
outline the research design and present the experimental methods used.
A Resource Based Theory_of Strategy Selection

The purpose of a theory of strategy selection is to predict under
various conditions, the rules or strategies a person might use to select
a coalition partner. The particular strategy selected is to some degree
a function of the participant's perception of the divisibility and certainty
of the payoff. One of the strategies, a Maximization strategy, is defined
simply as the choice of W, the smaller of two contenders, as a potential
coalition partner. This strategy is likely to occur under all variations
in the payoff structure of the game. The remaining strategies are marginal
choices. These are not independent of the Maximization strategy, but may
occur along with Maximization. The mdrginal strategies explain variations
in subjects' selection of potential partners that result from changes in
the payoff structure. With one exception, each of the marginal strategies

is independent of every other marginal strategy.

a
5,
I

58

Two experiments were designed to elicit marginal strategies. Exper-
iment I consists of bao PDQs. One has a Type 2 (A< (W+S), A=W) resource
distribution. The other has a Type 3 (A< (W+S), A=S) distribution.
Experiment II consists of an interactive game with a Type 2 resource
distribution, followed by a Type 3 PDQ. The payoff conditions under which
the marginal strategies are expected to be discriminable are established
in the following experimental paradigm.

Three experimentally manipulated payoff conditions are established:
easily divisible-certain; invisible-uncertain; and indivisible-certain.
One group of subjects is assigned to each condition, and each of the three
independent groups plays a decision-making role in a Type 2 and a Type 3
resource distribution situation. The payoff structure for the three
groups is as follows:

Figure 4.1
Experimental Payoff Conditions

 

 

Condition I Condition 11 Condition III
Payoff 100 Political Nominations for Nominations for
representation appointments Governor's and Governorship and
Lt. Governor's Lt. Governorship

Position on party in one party

ticket in compet- state
itive state

 

 

 

Divisihility Easily divisible Unequally divi- Unequally divi-
of payoff sible sible
Certain of Certain Uncertain Certain

 

 

 

a ttf—l in ing1 payoff

 

 

The experimental inductions for Conditions I and II followed Nitz and

Phillips (in preparation).1 The induction for Condition III introduced

 

lThese inductions are excerpted on pp. 28 and 29 above and are found
in Appendices A and B.

U)
NC

certainty of achieving payoff by portraying the payoff as nominations for
the governorship and lieutenant governorship in a one-party state.2
The theory's strategy predictions can be tested against the data pro-
vided by Experiment I. All three conditions provide tests of the Maximiz-
ation strategy hypothesis:
Bl. Maximization.

Subjects seek to maximize their share of the payoff with
respect to their coalition partners. Their choices are not
altered by the divisibility of the payoff or the probability
of winning in a subsequent contest. This strategy is expected
to occur with a probability of .50 or more under all three
experimental conditions.

The Maximization hypotheses may be phrased in terms of the subjects choices,

, ,
pr) l/‘Conditions i, 11, 111'

The marginal strategy hypotheses are tested by means of comparisons
between experimental conditions. The Intracoalition Compatibility hypo-
thesis states:

BZ. Intracoalition Compatibility

Subjects seek to form coalitions in which the division

of the payoff can be negotiated with a minimum of intra-
coalition friction. When the payoff is unequally divisible,
the probability of choosing the unequal contender is higher
than when the payoff if (maxilly divisiblE. When the payoff
is easily divisible, no intracoalition incompatibility will
be engendered in a coalition between equals or in a coalition
between unequals, so the Maximization decision rule will be
used.

An unambiguous test of the viability of this hypothesis is provided by a

comparison between Conditions I and III. The hypothesis may be stated in

terms of the experimental data as:

 

2The Condition 111 induction will to found in Appendix c and an
example of the complete experimental questionnaire will be found in
Appendix D.

bO

paW}Type 2 = p<w>Type 3) l/;
for Condition 1, and
p<w>TVp€ 3:> p<w>TVp8 2:. 1/2

for Condition III.
Another marginal strategy is closely related to the Intracoalition
Compatibility strategy hypothesis. This is the Competition hypothesis:
. . '3
B3. Competition
Subjects will seek to form coalitions that will allow
maximum grounds for conflicts, that is, coalitions for which
the payoff structure suggests no obvious division of the
payoff.
The Competitive hypothesis thus predicts behavior that is the complement
of Compatibility behavior:
pO‘DType 3 > p(w)Type 2 > 1/2
for Condition I, and
p<w>Type 2 > p<w>Type 3 > 1/2
for Condition III.
These two hypotheses are not independent, but are completely dependent.
A single test will determine which of these two marginal strategies is a
plausible explanation for non-maximization strategy choices.
The final strategy hypothesis examined here is the Security hypothesis:
B4. Security
When the payoff to a coalition can be attained only with
some probability less thin unity, subjects will seek to maxi-

mize their chances of winning by forming the largest coalition
possible.

 

3 - _

Tests of hypotheses A, 5 and 6 were shown to be dependent on the tests
of hypotheses l, t and 3 in Chapter ll and have been omitted here. The
original numbering system has been retained here to permit cross referencing.

bl

A comparison between probabilistic Condition II and non-probabilistic
Condition 111 provides the data to test this hypothesis. The hypothesis
predicts the following relation:

p(W)Type 2> Murry},e 22 1/2
Condition III Condition II

These tests constitute the examination of the theoretical hypotheses at
the level of population behaviors, the level of behavior that has been
observed in almost all previous studies of coalition formation. The next
two sections of this chapter will deal with the comparability of the PDQ
to interactive group games and the prediction of individual strategy
choices.

The PDQ and the Political Convention Came

 

The comparability of the social contact data obtained with the PDQ
to the contact data obtained from a group of subjects sitting in the same
room choosing a potential partner is examined quite simply. The second
strategy experiment was administered to another group of subjects. Exper-
iment 11 consists of two parts; the first is a political convention game
similar in content to the Type 2 distribution PDQ of Experiment I. The
subjects choose a potential partner to bargain with, bargain and divide
the payoff. The first choice of a potential partner provides the social
contact data for the Type 2 triad. The second part of the experiment was
a Type 3 PDQ. The payoff conditions were identical with those in Figure 4.1.

A comparison of Subjects' choices in the Type 2 triad conventions of
Experiments I and II examines the iSomorphism of the contact data of the
PDQ and the interactive political convention paradigm experiment.
Individual Differences and Individual Strategies

An individual playing the two political convention games of Experiment

I or II can make four possible rcSponses: Kw, RE, SW, and SE, that is, he

62

may choose the equal contender in the Type 2 triad and the weaker in the
Type 3 triad, etc. The t er oi strategy selection presented above maps
four mutually exclusive strategies onto these reSponse patterns. The
mapping (Table 3.2) is a function of the divisibility of the payoff. By
defining the strategies associated with specific choice patterns differ-
ently for Condition I and for Conditions 11 and III, the divisibility
variable is incorporated into the strategy definitions, rendering the
strategies independent of the competitive environment. Each strategy
pattern represents individual behavior consistent with the theory of strat-
egy selection, regardless of the subject's experimental condition. These
individual strategy choices can thus be predicted from individual differ-
ence measures alone.
Cognitive Complexity

Harvey's (l96l) constructs of the four conceptual systems suggest
patterns of strategy selection different from the patterns of social per-
ception. The Maximization strategy selection pattern predicted is:

Cl. System 11 and IV subjects will select more Maximization and
fewer marginal strategies than System I and III Subjects.

EXplicit predictions are also made (or the marginal strategies selected by
each System.

C2. A larger preportion of System I subjects will choose Security
strategies than will choose any other marginal strategy.

C3. A larger proportion of System 11 sukjects will choose Compe-
titive strategies than will choose any other marginal strategy.

C4. A larger prOportion of System III subjects will choose Intra—
coalition Compatibility strategies than will choose any other
marginal strategy.

C5. System IV subjects will choose all three marginal strategies
with equal frequency.

Machiavellianism
An orientation toward the exercise of power or toward competition
leads to a maximization and a marginal strategy prediction:
C6. The proportion of High Mach subjects who choose Maximization
strategies will be larger than the proportion of Low Mach

persons who choose Maximization strategies.

C7a. A larger proportion of Low Mach subjects will choose Security
strategies than will choose any other marginal strategy.

C7b. A larger proportion of High Mach subjects will choose Com-
petitive strategies than will choose any other marginal
strategy.
Method
Subjects. Subjects for Experiment I were 226 male undergraduate social
science and political science students who participated in classroom groups.4
Subjects for Experiment 11 were 243 male volunteers from introductory psycho-
10gy and political science courses. These volunteers participated in 1-4
hours of experiments for extra credit toward their class grade.
Procedure
Experiment I. Each subject took a questionnaire in a classroom situa-
tion. The questionnaire included: (I) a brief statement of the American
Psychological Association Code of Ethics relevant to individual difference
and opinion measures; (2) a Type 2 PDQ; (3) a Mach V Scale revised for

machine scoring;5 (4) Tuckman's ITI Scale, revised for machine scoring,

and (5) a Type 3 PDQ.

 

4Two hundred-thirty females also took the questionnaire for Experiment I.
Since female strategy behaviors have not been discussed in this study, these
data will be analyzed at another time.

5The Mach V Scale is found on pp.121-l26 and the ITI is found on pp. 142
of Appendix E. The answer sheets for these two measures are also found in
Appendix D. FORTRAN scoring subroutines are found in Appendix E.

64

Three payoff divisibility conditions were used in the PDQS. Condition I
had an easily divisible payoff, 100 political jobs or patronage appointments.
It is essentially the same as the easily divisible condition used by Nitz and
Phillips (in preparation). The experimental induction is found in the PDQ of
Appendix A.

Condition 11 has a payoff which was divisible only unequally, the nomi-
nations for the governor's and the lieutenant governor's places on the party
ticket. This condition is also the same as Nitz and Phillips' (in preparation)
indivisible condition. Condition 11 may also be considered a probabilistic
payoff condition, since some subjects will attend to the expectation of an
impending election. An example of this condition is found in Appendix B.

Condition Ill offered the same payoff as Condition 11, but explicitly
discounts the importance of the impending general election by Specifying
that the party holding the convention has been in control of state govern-
ment for several years. Subjects are told that any candidate who secures
the nomination is virtually assured of winning the general election. The
text of the Condition Ill induction is found in Appendix C.

Within each payoff condition, subjects received both a Type 2 and
Type 3 PDQ. The resource distributions for the two PDQs have been selected
so that the difference between the odd man and the other two is equal for
both types. Two counterbalanced forms are used in each PDQ. The distri-
bution for the Type 2 resource distribution is as follows:

F igure 4 . '2
Type 2 Resource Distributions

B C
la 100 150 100
lb 113 124 113
23 113 124 113

2b 113 113 124

65

The distribution for the Type 3 triad is given in Figure 4.3.

Figure 4.3
Type 3 Resource Distributions

A B C
1a 134 134 82
1b 13 82 134
2a 120 120 110
2b 120 120 120

Subjects receive one PDQ of each form number; e.g., la, Type 2 and la,
Type 3.

Each form asks for the subject's first contact choice, the minimum
bargaining share of the payoff he will accept, the reasons for his choice,
what he expects the other faction leader to demand, and his estimate of
his own bargaining ability. Since the length and quality of responses to
these questions have been highly variable in the past, no explicit hypo-
theses will be formulated at this point.

Experiment 11. Subjects were told they would participate in a series
of mock political conventions. They actually participated in two acti—
vities, but only the first was an interactive convention. Each subject
played in only one of the three payoff conditions in the convention. With-
in each payoff condition, the labels assigned to the subjects were counter-
balanced according to the following scheme:

Figure 4.4
Counterbalance of Labels and Resource Values

Subject Designation

X Y 2
Resource (1) 150 100 100
Distribution (2) 100 150 100

Order (3) 100 100 150

66

Within each condition, seven triads were run in each of the three assign-

ment orders. Twenty-seven triads comprised one experimental condition.

The second activity consisted of a Type 3 PDQ and the Mach V Scale and

Tuckman's (1964) ITI.

The PDQ administered to the EXperiment 11 Subjects contains an in-

struction to treat the factions in the PDQ as essentially different from

and not identified with the people in the live experiment.

The order of play in the political convention game was as follows:

Instructions were read to the subjects.7
Subjects were randomly assigned to candidate positions.

Subjects were asked for their written choices of contender with
whom they wish to start negotiations.

If two subjects made reciprocal choices, these two bargained
(in the presence of the experimenter, but out of the presence
of the third subject) for three minutes. The bargaining
session was recorded but is not analyzed in this study.

If an agreement was made, it was to be written and given to
the experimenter. If no agreement was reached, the subjects
were asked again to make choices.

If no reciprocal choices were made, the subjects were also
asked to choose again.

After completing the questionnaire, subjects filled out the
questionnaire described above.

 

6(me subject in each triad, the odd man, was lost because he had no
meaningful choice. The number actually used was smaller than 54 in some
cases because of subject errors.

7

The experimental instructions are found in Appendix F.

Chapter V

Results and Discussion

The hypotheses presented in the previous chapters deal predominantly
with marginal strategy choices. These are expected to be minor strategies,
that under Specifiable circumstances account for a portion of coalition
formation behavior. The hypotheses are relatively easy to test; most call
only for comparisons among proportions. The conditional nature of the
hypotheses, though, would pose a problem of interpretation if these were
the only analyses performed. We would have no indication of the conditions
under which individual difference measures predict strategy selection, the
relationship between Subject's perceptions and strategy selection, or of
any of the interaction effects among subsets of variables if only tests
for differences were used.

The principle dependent variable, strategy choice, is a nominal level
measure. Thus, most interval level statistics are inappropriate for this
dependent variable. Contingency tests, however, are applicable. The
major analysis used here will be a factorial partition of a contingency
table (Sdtcliffe, 1957). Since this method does not permit the use of an
observation which has incomplete data, such as an unanswered question, the
numbers used in the tables will decrease slightly as the number of levels
in the factorial design increases. It will be possible to examine fairly
complex interaction patterns, but most fourth level interactions will have
many cells with expected values less than 5.0. To avoid the X2 estimation

problems encountered in such cases, some variables will be collapsed into

67

68

a smaller number of categories. In most cases, only the collapsed tables
for lower order interactions will be shown, since the entire table is
generally too large to interpret visually.

Popula t ion Behavior

 

Comparison of the PDQ and the Political Convention Game. The most
direct indication of the comparability of the PDQ social contact data with

that of the political convention game is a test of the difference

p(w)Experiment 1 ' p("Ulixperiment II
for the Type 2 triad, the resource distribution for which administration
conditions differ. Table 5.1 presents the results of this test. The
observed differences were not statistically significant under any payoff
condition.

Another indication of the comparability of the PDQ and the political
convention game is provided by a comparison of the distribution of strategies
selected in Experiment 1 with those selected in ExPeriment II. Table 5.2
gives these distributions for the three payoff conditions. The Group X
strategy effect is clearly significant, indicating that subjects in the
two experiments did not select identical distributions of strategies. The
Condition X strategy effect is also significant, but the Group X Condition
X Strategy effect is not. The payoff condition, then, does affect the
strategy distribution, but has no significant differential effect on the
two experimental groups.

The inconsistency between the Croup X Strategy effect in Table 5.2
and the lack of effect in Table 5.1 indicates that the two experiments
elicited different distributions of individual strategies. These differ-

ences in individual strategies are masked in the population statistic p(W).

69

Table 5.1 Tests of the differences between p(W). , and P(W)
hxperiment 1

, for the Tv a 2 Resource Distribution.
hxperiment II ‘ -p“ 5

 

 

 

 

 

 

 

 

 

 

 

Condition p(W) p(W) Difference Pooled df t
Experiment I EXperiment ll Sp
1 .797 .722 .075 .0747 93 1.00
*
II .711 .588 .121 .0628 115 1.92
111 .605 .500 .105 .0621 117 1.69
k
.05:p_<_ l0

70

Table 5.2 Strategy selected as a function of group and payoff condition

Strategy Type

Group Condition 1 2 3 4
Maximization Competition Security Compatibility

Experiment I 1 38 6 6 9
II 52 7 12 12

III 45 4 13 19

EXperiment II I 21 7 3 5
II 19 1 2 11

Ill 17 2 l 18

1 r a a w . . . a
Group X Condition X Strategy Factorial Contingency Ana1y81s

2

Effect X df p

A Group Fixed 0

B Condition Fixed 0

C Strategy Fixed 0

AxB 2.15 2 ns
AxC 11.53 3 5 .01
BxC 13.53 6 5 .05
AxBxC 9.08 6 .25< p: .10
Total 36.34 17 < .005

 

8The method of computation is from Sutcliffe (1957). Lawton (1968)
programmed the analysis routine.

71

B1. Max1mization:

The Maximization hypothesis asserts that p(W) will be greater than
1/2 for all experimental conditions. The observed value of p(W) is pre-
sented, for each experimental condition, in Table 5.2. The .95 confidence
interval has been computed for each point. Table 5.3 indicates that three
experimental conditions have .95 confidence intervals that include the
point p(W) = .50. In Experiment I, Condition III, the lower bound of

is .498; in Experiment ll, Conditions II and III, the lower

bounds of p<w>Type 2 are .416 and .334, reSpectively. In all other con-
ditions, the lower bounds of the confidence region exclude p(W) = .50.
Thus the Maximization hypothesis is supported in all easily divisible
Condition I groups and all unequally divisible Condition II groups except
for the Type 2 triad in Experiment II. Maximization is supported in the
Type 3 triads of both Experiment I and II in Condition III, but is not

supported in either Type 2 or Type 3 triads in Condition III.

B2. Intracoalition Compatibility:

The Compatibility hypothesis predicts that

p(w)l‘ype 3 : p("DType 2 1/2
in Condition 1, and
p<w>Type 3 , P(W)Type 2 1/2

in Condition III.
Table 5.4 presents tests of the difference p(W)Type 3 - p(W)Type 2
for payoff Conditions I and III in Experiments 1 and II. There are no
significant differences in Condition I of either experiment, as predicted
by the Compatibility hypothesis. The differences in Condition III are
significant for both experiments, again in accord with the Compatibility

hypothesis.

72

.uouuo wawvasou can uwumaumum u mamamm Hausa ou one ecumsaumouu>on

.oom. deuce use muesaoea

unwoa menu you ~m>houaH mofievamdoo mm. 058

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

m
noao.a Nam. mmeo. Ham. poao.a omm. cmao. m~a. mam. has. mace. was. mass .HHH .axm
mm am on
com. memm. Name. cow. ooh. mode. «awe. wwm. «mm. cum. memo. «mm. maze .HH .axm
0mm. go“. quo. ooh. New. owe. moqo. an“. wmm. «no. memo. man; came .H .axm
Hm mm mm
was. mama. demo. moo. mom. mac. oaao. Han. oom. ace. ammo. mom. oa>e .H .axm
venom venom venom page - venom venom
noun: neaoa a Home: ~03 a Home: HoBQA a
IIIIIL a
2:335 m z 9: 3535 m 2 96¢ Jaime: m z Azj _
moaopwmcoo eocenamcou oucopfiueoo

 

 

esoouso awmuumo
ofinwmw>ua maamsvmcs
HHH :Oauwvcoo

 

eeoouso :kuumona
manama>aa saamsame:
HH coauaeeoo

manama>ao magnum
H coauaeaoo

 

pmwua mo make
new
uCeEwuwmxm

 

meoauaeeoo «Cosme

 

 

pmwuu mo memo cam usesfiuoaxo .aofluflvcoo wwozma mp mam>hmuaw mocovwwcoo Nmm so“: A3Va m.m essay

73

.emwsuonuo o .3 ma moaono m mama m.x ma H u .xx

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

x mom
.omw3uonuo o .3 ma mowono N mama m.x oomnnsm «H H I «xx paw m Anxxuaxxv w n e mecca . e u u a
a .I _I
So; as.
ammm.¢ mm mmm. Hma. summ.m am cam. awn. Nun. cm can. omo. HH .axm
*Nam.m Hm mmm. me. qu.H mm mum. coo. Nam.ou mm 0mm. fine. a H .axm
unfit du z um> Na3vanmA3va «mac au 2 um> NA3vaumA3va wmwc ea 2 Hm> NA3valma3va .ucmafiumaxm
mEOUUso samuumo osoouao cwmunmoca
waHmH>fia Adawsvoc: oabwma>aa xafimsamc: manflmfl>wo maammm cofiufiecoo

HHH :oHqucoo

 

HH coauaeeoo

 

H coauaeeoo

 

muchmm

 

coauaeeoo uuosam

cowufiecoo muchma cam uaosHuoaxo an N enhemzva u m mamHflzva a.m magma

74

The Competition hypothesis predicts that

PW) p(W) 1/2

Type 3 > Type 2 >

for Condition 1, and

p (W) (W)

Type 2 > p Type 3 > 1/2
for Condition III.

Since this prediction contradicts the prediction of the Compatibility hypo-
thesis, which received strong support, the Competition hypothesis must be
rejected as an explanation of marginal strategy choice.

Conditions I and II provide a test of the Compatibility hypothesis
designed to replicate the conditions of the Nitz and Phillips study (in
preparation). Table 5.4 indicates that the critical difference

Type Type 2
is significant for eXperiment II, the interactive group game, but is not
significant for Experiment I, which most closely replicates Nitz and Phillips
(in preparation). Thus, the Intracoalition Compatibility hypothesis has
received strong support, but not under Nitz and Phillips' (in preparation)
origional conditions.
B3. Security

An alternate interpretation of the finding Nitz and Phillips' (in
preparation) interpreted as a compatibility effect is that the effect they
observed was in part the result of a form of Security strategy played by
subjects in the Type 2 triad who sought to build a larger coalition in
the face of an uncertain outcome.

A test of the plausibility of this interpretation of the operation of
a Security strategy is found in the difference between the effects observed
in Condition II and those observed in Condition III. If the quantity
- p(W)Type 2)Condition II - (p(W)

- p(W) e 2)Condition III

(p (W)Type 3

Type 3 Typ

75

is positive, the difference identifies the pr0portion of the Intracoalition
Compatibility effect observed by Nitz and Phillips (in preparation) that

can be explained as a function of Security strategy choices. If the differ-
ence is zero or negative, it suggests that subjects in Conditions II do not
perceive a greater value in minimizing conflict within the coalition than
the subjects in Condition III. Table 5.5 indicates that there are no signi-
ficant differences between the Compatibility effect observed in Condition II
and that observed in Condition III. The differences observed (t - -l.57

p.i .13 and t = -.53, p >.50, EXperiments I and II, reSpectively) are

negative, and do not support the Security hypothesis.

‘_£ediction of Individual Behaviors

Behaviorgpredicted from cognitive complexity. The subjects perception
of their Opponent's payoff expectations were obtained as written answers to
the question, "what do you expect the other faction leader to demand?"
These answers were coded into one of three classifications:

(a) The opponent would demand less than 1/2 of the payoff.

(b) The opponent would demand 1/2 of the payoff.

(c) The opponent would demand more than 1/2 of the payoff.

Category b also included several answers that opponents would demand some-
thing in a range of outcomes, such as 40-60%.

The cognitive complexity measures were scored according to Tuckman's
(1964) recommended procedure: A subject was classified into the category
in which he scored above the 75 percentile, provided he scored lower in
all other categories. If a subject scored at or above the 75th percentile
in more than one category, or below the 75th percentile in all categories,

he was classified into category 0. Category 0 subjects were not used in

analyses which used complexity as an independent variable.

76

 

 

 

 

 

 

 

 

 

Amzv M
On. A M mm. . mo moa. moo.- Mme. eNm. HH .oxm
M
Amzv n
ma. em.a- Nos eoeo. mNH.- mma. ooo. H .oxm
Anoaaoo NV W so
i HHH eoaoaoeoo Ha eoaoaoooo
a H o up oocmpoMWHQ unmawuoaxm
wOO .
s a n Nisan - Mason Nixon - mason
unmaauoaxo up
HHH noaoaoooo HH ooaoaoeoo
mam max duh oak
N e Axon- m sxzvni MN aAsvou n aflzvo_ m.m oases

77

The distribution of cognitive complexity categories, perceptions, and
strategies chosen is given in Tables 5.6 and 5.7. Table 5.6 presents
subject's perceptions on their first (Type 2 triad) choice and Table 5.7
represents their perceptions on the second (Type 3 triad) choices. Cogni-
tive complexity appears to be unrelated to strategy choice or to perception
of the opponent's expectations in either political convention game. The
only effect that approaches significance is the interaction between com-
plexity, perception in the second game (Type 3 triad) and strategy choice.

Hypothesis Cl. predicts that cognitively complex System III and IV
persons will choose a greater preportion of Maximization strategies than
will cognitively simple System I and 11 persons. Tables 5.6 and 5.7 pre-
sent data relevant to this hypothesis. There is no difference in the
proportion of Maximization strategies selected by the complexity groups
(X23df = 5.65, .25 >p>.10). Moreover, controlling for perception of
opponent's expected demands does not render a significant Complexity X
Strategy effect. Table 5.7 suggests, though, that there may be differences
in strategy as a function of Complexity and Perception or differences in
perception in the Type 3 game as a function of Complexity and Strategy in the
preceding_(Type 2) game. The complexity X Perception in the second (Type 3)
game X Strategy effect approaches significance (Xzédf = 11.39, .10:>p:>.05).

Table 5.8 indicates the source of the effect observed above. The data
here are those of Table 5.7 collapsed over complexity Systems I and IIIv.

II and IV. The perception information in this table comes from the subject's
second game which had a Type 3 resource distribution. There is no signi-
ficant difference here in the perceptions of the two complexity groups

(x22df = .588, ns). Of those who perceive the opponent's payoff demand

to be less than half of the total payoff, four times as many System II

and IV subjects select a Maximization strategy as choose a Marginal strategy.

78

Table 5.6 Strategy choice by perception in the Type 2 convention by

cognitive complexity

 

 

 

 

 

 

 

 

 

 

 

 

 

Cognitive Perception Strategy, Choice
of Opponent's

Complexity, Demands Maximization Marginal
System I < 1/2 17 18

2 1/2 5 9

> 1/2 20 9
System II < l/2 18 5

2 1/2 4 4

> 1/2 14 8
System III < 1/2 3 6

Z l/2 3 3

> 1/2 5 4
System IV < 1/2 13 8

I 1/2 9 3

> l/2 9 4

Factorial Anal sis for Strategy Choice x
Perception x Cognitive Complexity
2

Effect X df p
A Cognitive Complexity Fixed 0
B Perception .1 Fixed 0
C Strategy Fixed 0
AB 3.47 6 ns
AC 5.65 3 .25 >p >.10
BC 2.08 2 ns
ABC 6.39 6 ns
Total 17.61 18 ns

 

79

Table 5.7 Strategy choice by perception in the Type 3 convention by
cognitive complexity

———--“_.

 

 

 

 

 

 

Cognitive Perception Strategy Choice
of Opponent's

Complexity Demands Maximization Marginal

System I < 1/2 18 22
I 1/2 15 8
> 1/2 9 6

System II < 1/2 24 6
I 1/2 10 6
> 1/2 2 5

System III < 1/2 4 7
2 1/2 6 5
> 1/2 1 1

System IV < 1/2 20 6
I 1/2 6 6
> 1/2 5 3

 

 

 

 

 

Factorial Analysis for Strategy Choice x
Perception x Cognitive Complexity

 

Effect _§__ _£§;_ __2__'

A ,Cognitive Complexity Fixed 0

B Perception 2 Fixed 0

C Strategy Choice Fixed 0

AB 4.36 6 ns

AC 5.65 3 .25 >p >.1O
BC .75 2 ns

ABC 11.39 6 .10 {p >.05

 

Total 22.15 18 ns

 

80

Table 5.8 Strategy choice by perception in the Type 3 convention by
cognitive complexity categories I + 111 v. II + IV

 

 

 

Cognitive Perception of ,

. . Strategy Ch01ce
Complex1ty Opponent S Maximization Mar i a1
System Demands ' g n
I + III (4% 22 27

*% 21 13
>% 10 7
II + IV < % 44 11
I l6 13
:>% 7 9

 

 

Factorial Analysis of Strategy Choice x

Perception in Type 3 Triad x Complexity Category I + III v. II + IV

 

 

 

Effect x2 df p

A Complexity Fixed Effect 0

B Perception Fixed Effect 0

C Strategy Choice Fixed Effect 0

AB .558 2 ns
AC 4.736 1 <.05
BC 1.223 2 ns
ABC 11.392 2 <.OOS
Total 17.909 7 <.025

 

81

The System I and III subjects who see the opponent as demanding less than
half of the payoff chose Maximization strategies less often than they chose
marginal strategies.

Hypotheses C2. through C5. predict specific marginal strategies for
each of the conceptual system types. Each hypothesis assumes as its null
hypothesis that each marginal strategy will be chosen with equal probability.
Table 5.9 presents the observed marginal strategy choices for each con-
ceptual system along with the test of the deviation of the observed distri-
bution from that predicted by the null hypothesis.

Hypothesis C2. predicts that System I subjects will choose more Security
than any other marginal strategy. The observed strategies for System I per-

sons differ from the null hypothesis slightly (X2 = 4.75; p‘: .10), but

2df
this difference is in the wrong direction. The Security strategy is not
the most frequently chosen. Hypothesis C2. thus lacks support.

Hypothesis C3. predicts that System II subjects will tend to choose
Competitive strategies. The observed distribution is significantly differ-
ent (X22df = 18.0; p < .001) from the null distribution, but in the opposite
direction from that predicted by C3. Hypothesis C3 is thereby disconfirmed.

Hypothesis C4. predicts that System 111 persons will select more Com-
patibility strategies than any other marginal strategy. Table 5.11 indicates
that this prediction is confirmed (X22df = 6.71, p < .05).

Hypothesis C5. predicts that System IV subjects will select all three
marginal strategies with equal probability. Since this prediction is equiv-
alent to the null hypothesis, a goodness-of-fit test rather than an indepen-

dence test is required. The observed distribution fits the prediction of

the null hypothesis moderately well (.90:> p > .75).

Table 5.9 Frequency of marginal strategy choices for
four cognitive complexity categories

 

 

 

 

 

 

 

 

Complexity ,
System Strategy Ch01ce
Competition Security Compatibility
1 6 10 16
II 1 4 16
III 5 3 12
IV 6 4 ' 4
Contingency Tests Against the
Null Hypothesis p(Sl) = p(Sz) = p(S3)
2 Confirmation
Complexity X df p of
System Hypothesis
I 4.75 2 .10) p) .05 NO
11 18.00 2 <.001 Opposite
Direction
III 6.71 2 <.05 Yes
IV .43 2 .90 >p> .75 Yes
Total 29.89
Strategy
Main Effect 18.83 2 <.001

Overall Com-
plexity x Strategy 11.06 6 .10> p> .05

 

a J'll-CI W

 

83

Only one of the four marginal strategy hypothesis received support
here. The overall Complexity X Strategy effect is given by partitioning
the data of Table 5.9 to remove the main effect for Strategy. The Come
plexity X Strategy effect approaches significance (X26df = 11.06;

.10 > p > .05).

Behavior Predicted from Machiavellianism Scores
Hypothesis C6. predicts that High Mach subjects will choose a larger F‘
proportion of Maximization strategies than will Low Mach subjects. Table

5.10 presents strategy selections by cognitive complexity groups and High- .

 

Low Mach score groups. Machiaevellianism is not significantly related to
selection of Maximization or Marginal strategies (ledf = 1.26, ns). This
finding disconfirms hypothesis C6. Nor is Machiavellianism related to
cognitive complexity (X23df = 4.90; .257’ p>'.lO). The interaction of
Machiavellianism, complexity and strategy choice is in the predicted direc-
tion but is not significant (X23df = 1.01, ns). The total effect in

Table 5.10 is not significant (leodf = 12.83; .25 > p >.10).

Hypotheses C7a. and C7b. predict that High Machs will tend to prefer
Competitive Marginal strategies. Table 5.11 presents the distribution of
marginal strategy choices by Machiavellianism groups. High and Low Mach
subjects do not differ in their preferences among the marginal strategies
(x22df = .512; ns). Hypotheses C78. and C7b. are not supported by these
data.

The data in Tables 5.10 and 5.11 indicate no differential strategy
choices attributable to Machiavellianism. The studies by Geis (1963),
Geis, Christie and Nelson (1963), and Blumstein and Weinstein (1967) though,
lead us to expect High Machs to behave differently from Low Machs in com-

petitive situations. The finding above raises two questions about the

84

Table 5.10 Strategy choice by cognitive complexity and
Machiavellianism scores

Strategy Choice

Complexity Mach Maximization Marginal
System
I Lo 23 22
Hi l9 14
11 Lo 13 8
Hi 23 9
111 L0 5 5 #1
Hi 6 8
IV Lo 13 8
Hi 18 7 .

 

Factorial Analysis of Strategy Choice x
Complexity x Machiavellianism.

2

Effect X df p
A Complexity Fixed Effect 0 8 '
B Machiavellianism Fixed Effect 0
C Strategy Choice Fixed Effect 0
AB 4.90 3 .25>;)>.10
AC 5.65 3 .25 >p >.10
BC 1.26 1 ns
ABC 1.01 3 ns
Total 12.83 10 .25 >p >.lO

Table 5.11 Marginal Strategy Choices by Machiavellianism
score group

Strategy Choice

 

 

Machiavellianism Competition Security Compatibility
Lo 16 17 34
Hi 10 15 29

 

 

 

 

 

X = .512 df

ll
N

I'lS

85

nature of Machiavellianism. First, does the construct measured by the
Mach V scale identify an ability to select appropriate abstract strategies
or does it identify a nonstrategic ability to deal with others only in
close interpersonal situations? Second, does dichotomizing the Mach V
scale scores or does the operation of summing the item scores mask the
dimensions of the scale that predict behavior most effectively?

These questions can be answered with an additional multivariate
analysis. If it is possible to predict strategy choices in these data
from some weighted combination of Mach item scores, then it will be clear
that the Mach scale can predict abstract strategy choices. If a differ-
ent prediction equation is necessary for each of the four strategies,
however, the utility and appropriateness of summing the item scores and
assuming a unidimensional scale must be questioned. The UCLA BMD program
for stepwise multiple discriminant analysis provides a simple means for
making these tests (Dixon, 1967).

Table 5.12 presents the means and standard deviations of the Mach V
items for each strategy type. The analysis program fits a linear regres-
sion equation to each of the strategy categories in such a way that the
function constructed maximizes the discrimination of subjects falling into
each category. The program adds one predictor variable (Mach item) to the
regression equations on each iteration, always selecting that item on
which the mean item score across strategies differs most. During the
addition process the effects of the variables previously added to the
prediction equations are partialled out. The resulting discriminant
function weights for 20 predictor variables are presented in Table 5.13.

Each function is of the form

Strategy(j) : a.qxj2 + ... +

. 4.. r O .
ajO + ajlle . J“ ameJm

“A <r—-.r.--—I
A...

 

 

 

 

Table 5.12 Means and standard deviations of Mach V
item scores for each strategy type
Strategy Type Means
Mach V I 11 III IV Mean over
Item No Maximization Competition Security Compatibility Strategies
Variable n = 190 n = 27 n = 36 n = 76 n = 329
1 0.52105 0.51852 0.36111 0.42105 0.48024
2 0.60000 0.77778 0.61111 0.55263 0.60486
3 0.66316 0.77778 0.47222 0.64474 0.64742
4 0.40000 0.55556 0.36111 0.38158 0.40426
5 0.55789 0.44444 0.55556 0.42105 0.51672
6 0.74211 0.70370 0.61111 0.78947 0.73556
7 0.51579 0.29630 0.52778 0.48684 0.49240
8 0.45263 0.77037 0.38889 0.31579 0.40729
9 0.23684 0.07407 0.22222 0.11842 0.19453
10 0.46316 0.55556 0.52778 0.52632 0.49240
11 0.57895 0.62963 0.75000 0.65789 0.62006
12 0.69474 0.66667 0.72222 0.71053 0.69909
13 0.53158 0.48148 0.55556 0.60526 0.54711
14 0.38421 0.25926 0.38889 0.40789 0.37994
15 0.49474 0.59259 0.47222 0.53947 0.51064
16 0.68421 0.51852 0.72222 0.60526 0.65653
17 0.90000 0.96296 0.94444 0.92105 0.91489
18 0.73684 0.59259 0.66667 0.76316 0.72340
19 0.71579 0.70370 0.63889 0.72368 0.70821
20 0.64737 0.59259 0.47222 0.67105 0.62918
Strategy Type Standard Deviations
1 0.50087 0.50917 0.48714 0.49701
2 0.49119 0.42366 0.49441 0.50052
3 0. 7387 0.42366 0.50631 0.48177
4 0.49119 0.50677 0.48714 0.48900
5 0.49794 0.50637 0.50 95 0.49701
6 0.43863 0.46532 0.49441 0.41039
7 0.50107 0.46512 0.506 1 0.50315
8 0.49906 0.49210 0. 49441 0.46792
9 0.42626 0.26688 0.42164 0.32525
10 0.49995 0.50637 0. 50631 0.50262
11 0.49503 0.49210 0. 43915 0.47757
12 0.46173 0.48038 0.4542 0.45653
13 0.50031 0.50917 0.50395 0.49204
14 0.48769 0.44658 0.49441 0.49471
15 0.50129 0.50071 0. 50631 0.50175
16 0.46605 0.50917 0.45426 0.49204
17 0.30079 0.19245 0.23231 0.27145
18 0.44151 0.50071 0.47809 0.42797
19 0.45222 0.46532 0.48714 0.45015
20 0.47905 0.50071 0.50631 0.47295

02’
\1

where a , is the constant term for the jth strategy and a .. is the
.10 Jl
regression coefficient for the ith item and the jth strategy.

The effectiveness of these prediction equations is indicated by the
classification Table 5.14. The X2 of the classification matrix is 73.09
with 9 df (p’ .001). The discriminant functions are highly successful
compared to the reSults in Tables 5.10 and 5.11.

The presence of as many as 20 items in the prediction equation, though,
raises the question of the relative importance of any particular item or

subset of items. Table 5.15 summarises the steps in the present analysis.

.. "7""? 7 7

No single variable discriminated categories with a conditional probability
level appreciably less than .10, and only 9 variables discriminated with
conditional p..,:3, Nevertheless, Table 5.15 indicates that the accuracy
of overall discrimination increases rapidly. After the fourth item added,
the F approximation of Wilks' Lambda to test the equality of group means
is significant at the .01 level (F = 2.30; df = 12,852). The effects of
limiting the prediction equations to the more significant items are shown
in Tables 5.16 and 5.17. Table 5.16 presents the classification matrix
for discriminant functions based on the first eight variables listed in
Table 5.15. The distribution of predicted strategies across observed
strategies shows a strong degree of association (X29df = 49.88, p <.001).
Table 5.17 shows the classification matrix for prediction equations using
only the first four Mach items. The level of association is relatively
high in this table also (ngdf = 27.86, .005 up >.001).

An indication of the nature of the ”Machiavellianism” underlying each
of the strategies can be seen by examining the items on which the strategies

differ most markedly. Table 5.18 summarizes a series of tests which identify

(r
(I

Table 5.13 Linear function discriminating four strategy types on the

basis of 20 Mach V item scores

Linear Coefficients for Each Strategy Type

I II 111 IV
Mach Item No. Maximization Competition Security Compatibility
1 1.87703 1.95503 0.99802 1.26568
2 1.44775 2.22581 1.59916 1.25567
3 2.12372 2.94689 1.08141 2.10734
4 0.79066 1.86258 0.72249 0.91590
5 0 98275 0.59006 1.15272 0.51356
6 1.54095 1.23982 1.00355 1.75744
7 1.00361 0.01019 1.21310 0.91430
8 1.37535 1.14201 0.91424 0.70363
9 1.92810 0.86398 1.77829 1.18258
10 1.06840 1.49766 1.40395 1.33324
11 1.53996 1.73637 2.42302 2.16943
12 2.87590 2.82723 2.86919 3.05498
13 2.16363 2.12922 2.27833 2.51544
14 1.00424 0.55182 1.28434 1.29258
15 1.29153 1.89076 1.23761 1.48925
16 -0.42412 -1.39225 -0.14017 -0.82861
17 10.45706 11.45313 10.84952 10.51994
18 1.62254 0.85477 1.33652 1.82874
19 3.23176 3.35551 3.09557 3.20752
20 2.24840 1.95887 1.60844 2.45587
Constant

-13.25240 -l4.06542 -12.75385 -l3.52576

89

Table 5.14 Number of Subjects classified into strategy groups
on the basis of Mach V scores on 20 items

Predicted Strategy Group

 

 

 

 

 

 

 

Observed I II III IV
Strategy Maximization Competition Security Compatibility
Group
Maximization 76 41 38 35
Competition 3 18 2 4
Security 6 3 19 8
Compatibility 8 20 19 29
2

x 9df = 73.09 p< .001

90

OH.
ca.
m0.
mo.
25.
as.
”o.
Ho.
#0.
Ho.
Ho.
so.
Mo.
2c.
#3.
so.
Ho.
Ho.
no.
0H.

qmm
0H3
ram
Amy
mm:
mmo
0mm
wma
mm:
mmo
mmo
owa
nmm
0H0
n00
owm
mmw
cwm
was
mmm

.oe
.Nm
.qm
.am
.mq
.ma
.Na
qfim.
.OM
.mm
.on
.am
.am
.sm
.ma
.ma
.Na
.0

.o

.m

Eopeoum

WC

moepmoo

U

a:3«ucEax0pagm~n~

m

to.

cg.

mm.m
Om.m
wM.N
aq.N
nm.m

f\
o
A‘OJNI—‘HHHI—‘i—Jy—‘l—‘HHH-fl

Emuwoua osmcfiefiuomfln an maeum mo eapmu unmeasm

Hmwm.o
cmwm.o
ewwm.o
mmmm.o
moom.o
waow.o
mNow.O
wman.o
mflmw.o
qo«w.o
mmmm.o
noqx.o
chm.o
ofimm.o
acmw.o
omom.o
~®H©.o
mom©.o
ammm.o
Nam©.o

sa.mxaa2

.AwoaHV axooum cam cease can Aaoaav coxao umuuau
.pmuouce xawsofl>oua meabmfipm> ecu Se>fiw .eeuomaem oHanuo> ecu mo :oHusnfluumflp

Hmcowowpcoo ecu mo .masouw xwmumuum use“ “Hm uw>o .moflamsvo wo momma oflomu woonflaexfla ego mum emmzan
.peuooce :moo uo: m>m£ once emozu mo msfim> m umecwwz ecu Lows manmfium> ecu pause cu uem coflueufluom

WC
WC
m5
m5
WC
m5

II.

mm.
mm.
mm.
mm.
OH.
OH.
OH.

Eocovum
mo

mmepmea

n nannana
menmfi-Ommfiméefirnmmmmmmmm

A

h a H A a a H G I a

I

mooo.o
HHM~.0
mmw¢.o
wamm.o
quo.o
odom.o
Acmm.o
qoqa.o
OHNH.H
wmwa.fi
noom.H
mo~¢.~
Rcwo.a
mxwm.H
maoc.m
wmfim.a
mwao.w
weqa.m
wOm¢.N
qawm.m

ueocm ou
oaam> e

mH.m waan

 

tn~7<31\lwcurw
HH—‘r—c—I—a—l

N 00 m \o
N v—4

r-fi

imrmr\.4sa'n~o.q'3
F4

peueucm
manmflno>

ponesz
amum

91

Table 5.16 Number of subjects classified into strategy groups on
the basis of Mach V scores on eight most discriminating
items

Predicted Strategy Group

 

 

Observed I II III IV

Strategy Maximization Competition Security Compatibility
Group

Maximization 68 43 45 34
Competition 3 l6 2 6
Security 7 l 19 7
Compatibility ll 19 21 25

 

 

X29df = 49.88 p< .001

Table 5.17 Number of subjects classified into strategy groups on
the basis of Mach V scores on four most discriminating

 

 

items
Predicted Strategy Group
Observed 1 11 III IV
Strategy Maximization Competition Security Compatibility
Group
Maximization 66 53 '48 23
Competition 6 17 3 1
Security 8 7 16 5
Compatibility 19 30 15 12

 

ngdf = 27.86 .005. p> .001

Table 5.18 Items for which the means of the distrigugions of
Mach V item scores differ across groups ’

Strategy Type

 

 

 

Maximization Competition Security
Competition 9,3,7,16,4,20,2,18,
5,6,15,10,17,13,12
3.7.1.16,4,20,2,
Security 8,11,16,20 18,5,6,15,17,
13,12,19
Compatibility 9,8 None None

 

 

 

 

 

 

8This table represents those sets of k items for which the sum of
Euclidean distances between strategy groups on items included was
significantly different from zero. (p = .05: Actual F and df
vary as the number of items increases.)

Items listed in order of decreasing variance explained.

\I

 

1 Table 5.19 identifies

the strategies each Mach V item differentiates.
the keyed Mach V items represented by the item numbers in the discussion

that follows. Because of the number of items involved that discriminate
with relatively low probability, only the first four in each cell will be
discussed here.

Competitive subjects were less likely than Maximizing subjects to view
flattery as a useful tactic (9), more likely to tell real reasons only if
they are useful (3), less likely to give real reasons for asking a favor (7),
less likely to feel honesty is always the best policy (16), and more likely
to believe that the best way to handle people is to tell them what they
want to hear (4).

Subjects who chose Security strategies differed from Maximizing sub-
jects as follows: Security subjects are more concerned about morally
right action (8), humility and honesty (11), honesty as the best policy
(16), and believe in trusting others (20).

Subjects who made Compatibility choices differ from Maximizing subjects
by feeling flattery an inappropriate tactic (9), and are more concerned

about morally right action (8).

 

1The test is based on the statistic:

F = (n-g-r+1) nm “k r _. _
mk \. it (ami - aki) (xm - Xk)
r(n-g)(nm+nk) i=1

df = r, r, n-g-r+l

where
= number of groups

n = number of subjects

nm= number of subjects in group m
r = number of variables included at time of test

10.

ll.

12.

13.

14.

15.

16.

17.

18.

19.

20.

Table 5.19 Keyed items from the Mach V scale
Most men forget more easily the death of their father than the
loss of their property.

People suffering from incurable diseases should have the choice
of being put painlessly to death.

Never tell anyone the real reason you did something unless it
is useful to do so.

The best way to handle people is to tell them what they want to
hear.

Most people are basically good and kind.

Most people who get ahead in the‘world lead clean, moral lives.
When you ask someone to do something for you, it is best to give
him the real reasons for wanting it rather than giving reasons
which might carry more weight.

One should take action only when sure that it is morally right.

It is wise to flatter important people.

The biggest difference between most criminals and other people is
that criminals are stupid enough to get caught.

All in all, it is better to be humble and honest than to be impor-
tant and dishonest.

There is no excuse for lying to someone else.

Generally Speaking, men won't work hard unless they are forced to.
Most men are brave.

It is hard to get ahead without cutting corners here and there.
Honesty is the best policy in all cases.

It is possible to be good in all respects.

Barnum was probably right when he said that there's at least one
sucker born every minute.

It is safest to assume that all people have a vicious streak and
it will come out when they are given a chance.

Anyone who completely trusts anyone else is asking for trouble.

 

Indicates a Machiavellian item.
Indicates a reversed item.

Finally, the competitive subject can be distinguished from the
Security strategy subjects by their preferences for never telling the
real reasons for actions unless useful (3), for giving reasons that carry
more weight than the real reasons (7), their beliefs that most men value
their prOperty more than the death of their father (1), that honesty is
not always the best policy (16), and that the best way to handle peeple is
to tell them what they want to hear (4).

The preceding analysis of the relationship of the Mach V items to
the strategy patterns defined in Chapters 11 and Ill Suggests that Machi-
avellianism is not a useful concept so long as it is thought of as a
unitary scale. If the items that predict Specific strategies are iso-
lated, however, distinct subsets of items effectively discriminate be-
tween alternative behaviors even when these behaviors are highly abstract

strategy choices.

 

Chapter VI

Summary and Implications

The major objective of this study was to examine the effects of one
of the conditions of nontransferable utility, namely indivisible payoffs,
on coalition formation strategies selected in a mixed motive game. Two
distinct analytical approaches were used. The first probed for evidence
of consistent social contact strategies across the population of subjects
as a function of experimental manipulation of divisibility and certainty
of payoff. The second examined individual social contact strategy patterns
and sought to predict them on the basis of individual difference measures
that could be eXpected to have a bearing on strategic behavior. In addition,
two different experimental measures were designed to identify social con-
tact strategies, the Political Decision Questionnaire and an interactive
political convention game. Differences between these two measures suggest
an important consideration for future strategy research.

Payoff Conditions as Determinants of Strategy Selection

 

The analysis of social contact strategies in the subject population
provided support for the Maximization and the Intracoalition Compatibility
hypotheses: Under all conditions of payoff the most frequently chosen
strategy is a maximization strategy; but under conditions of indivisible
payoff, deviations from the Vaximization strategy can best be attributed
to a strategy that seeks to reduce conflict within the coalition about to
be formed. Intracoalition Compatibility, however, was not supported in
all eXperimental conditions. In a condition that replicated Nitz and
Phillips (in preparation) identification of the Compatibility strategy,

support was lacking.

96

 

97

This condition, the indivisible uncertain Condition II in Experi-
ment I differed from the corresponding condition in the Nitz and Phillips
(in preparation) study in one major reSpect: both Type 2 and Type 3
choices were obtained from each subject in this study. Although the two
PDQ forms were separated by about 20 pages of questionnaire, the value
of Pearson's contingency coefficient for the two PDQ choices in Experi-
ment I is a = .41 (X2=35.18, p <.001). The degree of association between
the Type 2 and Type 3 contact choices in Experiment II is insignificant
(X2=.1O, ns). This lack of independence would account for the reduced
effects in the Experiment I PDQ.

The experimental manipulation of certainty of the payoff did not
induce security strategies in either experimental pOpulation. Moreover,

the fact that p(W)

3-p(W) is larger for the indivisible certain

Type Type 2
Condition III than for the indivisible uncertain Condition II in both
experiments suggests that the Intracoalition Compatibility strategy iden-
tified by Nitz and Phillips (in preparation) is the result of the relative
indivisibility of the payoff rather than the uncertainty of obtaining it.
The fact that the differences observed in Table 6.4 are in the opposite
direction from a Security prediction suggests that the indivisible-uncertain
Condition II subjects may not have attended to the uncertainty cues in the
experimental inductions.

An analysis of the comparability of two means of assessing social
contact strategies identifies a critical shortcoming of coalition strategy
research that has relied on pepulation statistics to test strategy hypo-

theses. A test of the difference[ p(W)EXp I - p(W)EXp II] for all three

experimental payoff conditions and both resource distributions yields no

98

significant differences between p(W)s in the two experiments for corre-
sponding conditions. A factorial analysis of the individual strategy
distribution across experiments and payoff conditions indicates a strong
Experiment X Strategy effect and a strong Payoff Condition X Strategy
effect. There is no significant three-way interaction.

It is clear that by defining strategies based on multiple bits of
information for individual subjects it is possible to detect effects due
to experimental presentation and payoff conditions. These results suggest
that the results of a whole series of studies notably, Riker (1967), Gamson
(1961b), Vinacke and Arkoff (1957), Vinacke (1959), Vinacke, Crowell, Dien,
and Young (1966), and others may be suSpect--since they aggregate behavior
over games and subjects. This study finds that simple aggregation across
subjects (without adding together non-independent successive games) masks
significant behaviors. Moreover, the behaviors that are masked are the
Marginal Security and Compatibility Strategies. Table 5.2 indicates that
Security strategies appeared only in the indivisible conditions of the
PDQ, and were chosen as frequently as Compatibility strategies, but
marginal strategies were chosen only half as often as Maximization strat-
egies. In indivisible Conditions II and III of the interactive game,
however, Security strategies do not appear, but Compatibility, alone accounts
for about half of all strategies chosen.

The above observations suggest that an essential element in the study
of strategic behavior is a highly cautious approach to the possibilities
of confounding effects. Identification of individual strategies as joint
behavior on separate critical tasks, conditional on experimental manipulation
of the payoff permitted positive identification of the Intracoalition Com-

patibility hypothesis as a descriptor of a meaningful strategy elicited by

99

a condition of nontransferable utility, i.e., indivisible payoff. Were
the identity of the experimental tasks neglected, the joint aspect of the
subjects' behavior would be lost and so would be the opportunity to crit-
ically test any hypothesis about the individual's strategy behavior other
than Maximization. In light of this consideration, it seems crucial to
ask whether the study of exclusively divisible payoffs, i.e., transferable
utility situations may do more to hinder the development of political
theory than to facilitate it. Certrainly Maximization cannot be taken on
apriori grounds as the only viable political strategy.
Individual Differences as Determinants of Strategy Selection

Cognitive Complexity. The major complexity hypothesis predicts that
cognitively simple System I and II persons will choose fewer Maximization
strategies than cognitively complex System 111 and IV persons. This hypo-
thesis was disconfirmed. The combinations of categories I + III and
II + IV, however, do select the strategies predicted for cognitively simple
and cognitively complex persons, respectively. An even stronger effect
is found in the interaction among complexity, perception of the opponent's
demands in the second game, and strategy choice. System II and IV persons
select more Maximization strategies than do System I and 111 subjects; but
the System II and IV subjects who do so are those who perceive their oppo-
nents as likely to demand less than half of the payoff (See Table 5.8).
This effect cannot be attributed to the interaction between complexity and
perception, since that effect is minute (x22df = .558gns). The two com-
plexity groups, then do not perceive the situation differently, but act
differently given equivalent perceptions.

An alternative explanation of this effect can be prOposed. Since

‘the perception measures were taken after the subject made his strategy

lOO

choice, they could be a result of his choice rather than a cause. More-
over, the greatest Perception X Complexity interaction effect occurs in
the second game played. This interpretation cannot be dismissed on the
basis of these data. The effect observed here suggests that it may be
fruitful to design a study that controls for the reactive effect of social
perception questions on subsequent strategy choices and of strategy choices
on subsequent perceptions.

The marginal strategy hypotheses for cognitive complexity received
at best only Spotty support. Hypothesis C4 predicted that System III
persons would select more Compatibility strategies than any other marginal
strategy. The test against the null hypothesis indicates that they did so
(XZde = 6.71; p< .05), but the Complexity by strategy interaction over
the entire marginal strategy table (Table 6.10) was not significant
(x26df = 11.06; .lO> p> .5). The distribution of strategies for System
111 persons is thus not significantly different from the distributions
for the other three systems. In light of this observation and in light
of the fact that the reverse ordering of the System 11 prediction is

highly significant (X2 = 18.0; p< .001), it seems reasonable to reject

2df
this entire set of marginal strategy hypotheses.

0f the five conceptual systems theory predictions, only a post hoc
hypothesis predicting Maximization strategies for System II and IV persons
was strongly supported. These two pairs of systems that show the most
nearly equivalent strategy selection behavior, though, are not closely
ordered in level of complexity to Harvey's (1961, 1963) theoretical con-

structs of cognitive complexity. Systems I + III and II + IV do not

differ in the distribution of their perceptions of their opponents but

101

they differ in the way these perceptions are related to strategy selection.
This behavior cannot be attributed to a greater or lesser degree of ab—
stract abilities within the context of conceptual systems theory.

This observation suggests that conceptual systems theory should be
revised, since the predictions we can make in the political convention
paradigm are not related to the presumed level of complexity or abstract-
ness of the four Systems. The conceptual system theorist, though, might
challenge this conclusion by citing an important construct in his theo-
retical framework. The basic conceptual ability that is the referent of
the concept of cognitive complexity is the ability to deal with large
amounts of information. That is, to be cognitively complex is to be able
to work effectively under an information overload or in a complex environ-
ment. The political convention paradigm used here could be described as
an information underloaded or simple environment. The conceptual system
theorist does not expect cognitive complexity to predict differential
behavior in an environment that does not overload even the most concept-
ually simple subject. This sort of argument does not clarify the ambi-
guities we have noted in the concept of complexity. We have observed
differential behavior in a simple environment, and we have predicted this
behavior on the basis of a measure of cognitive complexity. If cognitive
complexity predicts differential behavior only under information overload
conditions, then what we have measured must not be the cognitive com-
plexity originally intended. If we have measured cognitive complexity,
then information overload must not be necessary for the prediction of
individual behaviors from complexity information.

One suggestion as to what Tuckman's (1965) Interpersonal Topical

Inventory may have measured is provided by Harvey's (1966) review of

102

complexity studies. Harvey (1966) notes that System 11 and IV persons
tend to score low on the EPPS Deference subscale and high on the Autonomy
subscales. Moreover, System 115 score high on Aggression. System I
persons were high on EPPS Deference, and low on Change and Autonomy.
System III persons were high on Affiliation and also low on Autonomy and
Change. Not only do System I and III persons choose fewer Maximization
and more Marginal (predominantly Security and Compatibility) strategies
than System II and IV persons, but they achieve similar scores on
measures of affiliation, deference and autonomy. It would be reasonable
to hypothesize that Tuckman's ITI provides a measure comparable to
certain subscales of the EPPS.

These considerations confront the student of political strategy with
a major problem in the study of cognitive behavior. How does one identify
patterns of cognition? How does one select a measure of these patterns
appropriate to the behavior he wishes to predict? In the long run it
would be fruitful to replicate Vannoy's (1965) study of the comparability
of cognitive complexity measures, but to replicate it with a primary focus
on identifying patterns of perceiving competitive situations and patterns
of strategy selection. In the short run, though, the problems raised in
this report suggest a project of narrower scope: a discriminate function
analysis of ITI items as predictors of strategy patterns.

Machiavellianism. Hypotheses C6 and C7 predict that High Machs will
select Maximization strategies more often than Low Machs, and that the
High Machs will tend to select Competition strategies when they do select
uarginal strategies, while Low Machs will tend to select Security strategies.
Both of these hypotheses were disconfirmed. The possibility that the Mach

Scale might contain subscales which predicted specific strategies led to

103

a discriminant function analysis of the Mach V Scale items. This analysis
identified items which discriminated among the four strategies. The two
strategies most clearly distinguished were Security and Competition.
Subjects who had responded with these strategy selections could be pre-
dicted with quite high accuracy on the basis of only four Mach V items--
items that deal with mores and tactics for dealing with people. The dis-
criminant functions for the Machiavellianism Scale did not permit proper
classification of subjects into all strategy categories, however. While
it was possible to distinguish Maximization from Competition, Maximization
from Security, and Security from Competition, only two items lent to dis-
criminating Maximization from Compatibility and no items discriminated
Compatibility from either Competition or Security. It is apparent that
some Mach Scale items predict some behaviors-~what is important is that
the behaviors the Mach items predict are the strategies that could not be

predicted on the basis of divisibility or certainty of payoff.

Problems for Further Analysis

Lawton's (1968) program of Suttcliffe's (1957) factorial partition
of contingency tables made the three-way interaction analyses in this report
possible. The cell expected values decrease rapidly for four dimensional
analyses and lend to unstable approximations to X2. Thus several desirable
analyses could not be performed. The technique of using dummy variables
to represent discrete independent variables in regression equations, how-
ever, offers promise of completing additional analyses. The dummy variable
technique can be extended to discriminant function analysis to obtain the
joint effects of eXperimental conditions and individual difference item

measures on discrete strategy behaviors. This particular extension, though,

104

Irequires prior knowledge of the pairwise interaction effects of individual
(difference items and the dummy variables representing discrete experimental
condition. 1

The discriminant function analyses of Machiavellianism items and
Specific strategy choices suggests that a similar analysis including ITI
items and dummy variables for experimental conditions would not only be
‘highly enlightening, but may be a step toward answering Greenstein's (1967)
questions about the effects of personality variables. Under what condi-
tions will which personality measures predict what kinds of political be-
havior? The present analysis has identified four distinct political
strategy behaviors and the conditions under which Maximization and Compat-
ibility strategies are elicited. It has also demonstrated a relationship
between particular individual difference measures and strategies. Reanal-
ysis of the data presented here may contribute to an understanding of the
relationship between payoff conditions and individual difference measures

and the interaction among environmental conditions, personality variables

and strategy selection.

1These analyses must be temporarily deferred until an efficient inter-

éiction program, such as Morgan and Sondguist's (1965) AID, is operational
éit the University of Hawaii.

 

BIBLIOGRAPHY

Adorno, T. W., Frenkel-Brunswik, E., Levinson, D. J., and Sanford, R. N.
The Authoritarian Personality. New York: Harper, 1950.

Asch, S. B. Social Psychology. New York: Prentice Hall, 1952.

 

Blumstein, P. W., and Weinstein, E. A. The redress of distributive in-
justice. American Sociological Association meeting, San Francisco,
August, 1967.

Bond, J. R., and Vinacke, W. E. Coalitions in mixed-sex triads. Sociometry,
1961, 33, 61-71.

 

Budner, S. Intolerance of ambiguity as a personality variable. Journal
of Personality, 1960, 22, 30.

 

Caplow, T. A theory of coalitions in the triad. American Sociological
Review, 1956, g1, 489-493,

Caplow, T. Further development of a theory of coalitions in the triad.
American Journal of Sociology, 1959, 63, 488-493.

Chaney, M. V., and Vinacke, W. E. Achievement and nurturance in the triad.
Journal of Abnormal Social Psychology, 1960, pg, 175-181.

Chertkoff, J. M. The effects of probability of future success on coalition
formation. Journal of Experimental Social Psychology, 1966, 3,
265-2770

Christie, R. Impersonal Interpersonal Orientations and Behaviors. Mimeo-
graphed research proposal, Columbia University, 1962.

Cole, S. C. An examination of the power inversion effect in three person
mixed-motive games. Journal of Personality and Social Psychology,
1969,_11, 50-58. -

Cole, 3. G., and Phillips, J. L. A note on verbalized strategies in a
three-person political convention situation. East Lansing: Human

Learning Research Institute, Report No. 10, May 1967.

Dawson, J., and Phillips, J. L. Machiavellianism and manipulative strategy
in a mock political convention. In preparation.

Dixon, W. J. (Ed.) BMD Biomedical Computer Programs. Berkeley and
Los Angeles: University of California Press, 1967.

105

106

Driver, M. J. Conceptual structure and group processes in an inter-nation
simulation. Part OnE: The perception of simulated nations. Ed. Test.
Serv., April 1962, Report RB-62-15.

Edwards, A. L. Manual for the Edwards Personal Preference Schedule. New
York: Psychological Corporation, 1954.

Felknor, C., and Harvey, O. J. Some cognitive determinants of concept
formation and concept attainment. Technical Report No. 10, Contract
Nonr 1147(07), University of Colorado, 1964.

Gamson, W. A. An experimental test of a theory of coalition formation.
American Sociological Review, 1961, 26, 565-73.

Gamson, W. A. A theory of coalition formation. American Sociological
Review, 1961,.26, 373-82.

 

Gamson, W. A. A Theory of Coalition Formation. Unpublished Dissertation,
University of Michigan, Ann Arbor, 1960.

Gamson, W. A. Experimental studies of coalition formation. In L. Berkowitz
(Ed.), Advances in Experimental Social Psychology, Vol. 1. New York:

Academic Press, 1964.

Geis, F. L. Machiavellianism in a three-person game. Ph. D. Dissertation,
Columbia University, 1963.

Geis, F. L. Machiavellianism and the manipulation of one's fellow man.
Paper read at American Psychological Association, Los Angeles, 1964.

Geis, F. L., and Christie, R. Machiavellianism and the tactics of manip-
ulation. Paper read at the American Psychological Association,
Chicago, September 1965.

Geis, F. L., Christie, R., and Nelson, C. In search of the Machiavel.
Paper presented at American Psychological Association meeting, Chicago
1963.

Gillmore, C., Kline, D. K., Anderson, J., Lawton, D., and Phillips, J. L.
A 3400/3600 Fortran program for computing chi-squares and exact
probabilities of all effects in a factorial design. Technical Report
No. Human Learning Research Institute, Michigan State University,
in preparation.

Cough, H., and Sanford, R. N. Rigidity as a psychological variable.
Unpublished Mss. University of California, Los Angeles, Institute
of Personality Assessment and Research, 1952.

Greenstein, F. I. The impact of personality on politics: an attempt to
clear away the underbrush. American Political Science Review, 1967,
-6__1_, 629-6410

107
Harvey, O. J. Cognitive determinants of role playing. Technical Report
No. 3, Contract Nonr 1147(07), University of Colorado, 1964.

Harvey, O. J. Experience, Structure and Adaptability. Chapter 6, New York:
Springer, 1966.

Harvey, O. J., Hunt, D. E., and Schroder, H. M. Conceptual Systems and
Personality Organization. New York: Wiley, 1961.

Homans, G. C. Social Behavior: Its elementary forms. New York: Harcourt
Brace and World, 1961.

Kelley, G. A. The Psychology of Personal Constructs. New York: Norton,
1956.

Kelley, H. H., and Arrowood, A. J. Coalitions in the triad: Critique and
experiment. Sociometgy, 1960, 23, 231-244.

 

Nitz, L. H., and Phillips, J. L. The effects of divisibility of payoff on
confederative behavior. In preparation.

Phillips, J. L., and Nitz, L. H. Social contacts in a three-person "polit-
ical convention" situation. Journal of Conflict Resolution, 1968,.12,
206-214.

Riker, W. H. Bargaining in a three-person game. American Political Scientific
Review, 1967,.61, 642-656.

Riker, W. H. The Theory of Political Coalitions. New Haven: Yale Univer-
sity Press, 1962.

Rokeach, M. The Open and Closed Mind. New York: Basic Books, 1960.

Rulon, P. J., and Brooks, W. D. On statistical tests of group differences.
In D. K. Whitla (Ed.), Chapter 2 Handbook of Measurement and Assess-
ment in Behavioral Sciences. Reading, Mass.: Addison-Wesley, 1968.

Schelling, T. C. The Strategy of Conflict. New York: Oxford University
Press, 1963.

Schroder, H. M., Driver, M. J., and Streufert, S. Human Information Pro-
cessing: Individuals and Groups Functioning in Complex Socia1 Situations.
New York: Holt, Rinehart and Winston, 1967.

Schroder, H. M., and Streufert, S. The measurement of four systems of per-
sonality structure varying in level of abstractness. (Sentence Com-
pletion Method.) Princeton University: ONR Technical Report No. 11,
Nonr 1858(12), 1962.

Shapley, L. S. A value for n-person games. Annals of Mathematical Studies,
1953,'g§, 307-317.

108

Shapley, L. S., and Shubik, M. A method for evaluating for distribution
of power in a committee system. American Political Scientific
Review, 1954,.48, 787-792.

Shelly, R., and Phillips, J. L. A social contact model for coalition form-
ation. East Lansing: Human Learning Research Institute, Report No. 6,
September 1966.

Sherif, M. The concept of reference group in human relations. In M. Sherif
and M. W. Wilson (Eds.), Group Relations at the Crossroads. New York:
Harper, 1963.

Suttcliffe, J. P. A general method of analysis of frequency data for multiple
classification designs. Psychological Bulletin, 1957, 21, 134-137.

Tuckman, B. W. Personality structure, group composition, and group function-
ing. Sociometry, 1964, 21, 469-487.

Tuckman, B. W. Integrative complexity and attitudinal orientation. Perceptual
Motor Skills, 1965,_gl, 838.

Tuckman, B. W. Integrative complexity: its measurement and relation to
creativity. Educational Psychological Measurement, 1966,.26, 369-382.

Tuckman, B. W. Personal Communication. 1968.

Vannoy, J. S. Generality of cognitive complexity-simplicity as a personality
construct. Journal of Personality and Social Psychology, 1965, 2,
385-397.

Vinacke, E. A. Sex roles in a three-person game. Sociometry, 1959.

Vinacke, W. E., and Arkoff, A. Experimental study of coalitions in the triad.
American Social Review, 1957, 22, 406-475.

Witkin, H. A. Individual differences in the perception of embedded figures.
'ggurnal of Personaligy, 1950, 12, 1-15.

 

A ppend ix A

Induction for Condition I

Form I-la PDQ 12

POLITICAL DECISION QUESTIONNAIRE

This questionnaire is part of a study of some basic political abilities.
There are several parts to the questionnaire and each part has its own
instructions. Since various forms of this questionnaire will be given in
a number of different classes, we would like the following information.

Name

 

Class in which this
questionnaire is given

 

Year in college

 

Major

 

Sex M F

In keeping with the American Psychological Association's Code of Ethics,
no information given on this questionnaire will be released at any time
except as part of a statistical average which cannot be identified with a
person. The answers you give will not be available to your instructor,
administrative officers of this university or to investigative agencies for
any reason.

PART ONE

A state political party is divided into three strong factions or groups.
These groups are designated Faction X, Faction Y, and Faction Z. The
party is having a convention. Assume that you are the representative,
that is, the floor leader of one of the three factions in the convention.

There are 350 delegates to the convention, and each delegate has
one vote. Since the factions in this party are quite strong, all of the
delegates in each faction have pledged their votes to the faction leadership.
This enables the floor leader of each faction to bargain as the representative
of his entire faction. The faction will then vote as a bloc, in line with
whatever agreement its floor leader may make. Faction X has 100 delegates
(i.e., votes). Faction Y has 150 delegates (votes), and Faction Z has
100 delegates (votes). The major purpose of this convention is to de-
cide how many of 100 political jobs each faction will receive. Each faction
would like to get as many of these jobs as possible.

110

111

Form I—la Cont.

It is standard procedure for two factions to get together and agree
on the division of the jobs. If these two factions control a majority of
the votes of the convention, that is, 176 votes, then the jobs are divided
according to their agreement. An alliance between Faction X and Faction Y
would have 250 votes. An alliance between Faction X and Faction Z would
have 200 votes, and an alliance between Faction Y and Faction Z would have
250 votes.

Assume that you are the floor leader of Faction X (100 votes). Which
of the other two factions Y or Z, will you contact first to try to make a
deal for the division of the jobs?

Faction Y Faction Z
(150 votes) (100 votes)

(Circle one)

What portion of the jobs are you prepared to offer?

What is the smallest portion of the jobs you would be willing to
accept in a coalition with this faction leader? (That is what is your
rock-bottom low?)

Why did you choose to contact the faction your chose?

What do you expect the other faction leader to demand?

What is likely to be the outcome of the bargaining session? What
portion of the jobs do you think you can realistically obtain?

GO ON TO PART TWO.

Append ix B

Induction for Condition II

Form II-la PDQ 12

POLITICAL DECISION QUESTIONNAIRE

This questionnaire is part of a study of some basic political abilities.
There are several parts to the questionnaire and each part has its own
instructions. Since various forms of this questionnaire will be given in
a number of different classes, we would like the following information.

Name

 

Class in which this
questionnaire is given

 

Year in college

 

Major

 

Sex M F

In keeping with the American Psychological Association's Code of Ethics,
no information given on this questionnaire will be released at any time
except as part of a statistical average which cannot be identified with a
person. The answers you give will not be available to your instructor,
administrative officers of this university or to investigative agencies for
any reason.

PART ONE

A state political party is divided into three strong factions or groups.
These groups are designated Faction X, Faction Y, and Faction Z. The
party is having a convention. Assume that you are the representative,
that is, the floor leader of one of the three factions in the convention.

There are 350 delegates to the convention, and each delegate has
one vote. Since the factions in this party are quite strong all of the
delegates in each faction have pledged their votes to the faction leader-
ship. This enables the floor leader of each faction to bargain as the
representative of his entire faction. The faction will then vote as a
bloc, in line with whatever agreement its floor leader may make. Faction X
has 100 delegates (i.e., votes). Faction Y has 150 delegates (votes),
and Faction Z has 100 delegates (votes). The major purpose of this
convention is to nominate a candidate to run for the office of governor
and a candidate for the office of lieutenant governor. Each faction would
like its man to receive the nomination for the governorship, but would
not be extremely dissatisfied if its man received only the lieutenant
governor's place on the ballot.

113

114

Form II-la

It is standard procedure for two factions to get together and agree
on the division of the nominations. If these two factions have a majority
of the votes of the convention, that is, 176 votes, then the nominations
are divided according to their agreement. An alliance between Faction X
and Faction Y would have 250 votes. An alliance between Faction X and
Faction Z would have 200 votes, and an alliance between Faction Y and

Faction Z would have 250 votes.

Assume that you are the floor leader of Faction X (100 votes). Which
of the other two factions, Y or Z, will you contact first to try to make

a deal for the division of the nominations?
Faction Y Faction Z E
(150 votes) (100 votes)

(Circle one)

 

Which nomination are you prepared to offer them?

Which nomination would you accept as a rock-bottom bargain in a
coalition with this faction's floor leader?

Why did you choose to contact the faction you chose?

What do you expect the other faction leader to demand?

What is likely to be the outcome of the bargaining session? What
nomination do you think you can realistically obtain?

GO ON TO PART TWO.

Appendix C

Induction for Condition III

 

Form III-la PDQ 12

POLITICAL DECISION QUESTIONNAIRE

This questionnaire is part of a study of some basic political abilities.
There are several parts to the questionnaire and each part has its own
instructions. Since various forms of this questionnaire will be given in
a number of different classes, we would like the following information.

Name

 

Class in which this
questionnaire is given

 

Year in college

 

Major

 

Sex M F

In keeping with the American Psychological Association's Code of Ethics,
no information given on this questionnaire will be released at any time
except as part of a statistical average which cannot be identified with a
person. The answers you give will not be available to your instructor,
administrative officers of this university or to investigative agencies for
any reason.

PART ONE

A state political party is divided into three strong factions or groups.
These groups are designated Faction X, Faction Y, and Faction Z. The
party is having a convention. Assume that you are the representative,
that is, the floor leader of one of the three factions in the convention.

There are 350 delegates to the convention, and each delegate has
one vote. Since the factions in this party are quite strong, all of the
delegates in each faction have pledged their votes to the faction leader-
ship. This enables the floor leader of each faction to bargain as the
representative of his entire faction. The faction will then vote as a
bloc, in line with whatever agreement its floor leader may make. Faction X
has 100 delegates (i.e., votes). Faction Y has 150 delegates (votes),
and Faction Z has 100 delegates (votes). The major purpose of this
convention is to nominate a candidate to run for the office of governor
and a candidate to run for the office of lieutenant governor. Each faction
would like its man to receive the nomination for the governorship, but
would not be extremely dissatisfied if its man received only the lieutenant
governor's place on the ballot. Since this party has effectively controlled
State government for several years, any man who receives the nomination
is virtually assured of winning the general election against the opposition
party. '

116

117

Form III-1a

It is standard procedure for two factions to get together and agree
on the division of the nominations. If these two factions have a majority
of the votes of the convention, that is, 176 votes, then the nominations are
divided according to their agreement. An alliance between Faction X and
Faction Y would have 250 votes. An alliance between Faction X and Faction Z
would have 200 votes, and an alliance between Faction Y and Faction Z
would have 250 votes.

Assume that you are the floor leader of Faction X (100 votes). Which
of the other two factions, Y or 2, will you contact first to try to make
a deal for the division of the nominations?

Faction Y Faction Z
(150 votes) (100 votes)

(Circle one)

Which nomination are you prepared to offer them?

Which nomination would you accept as a rock-bottom bargain in a
coalition with this faction's floor leader?

Why did you choose to contact the faction you chose?

What do you expect the other faction leader to demand?

What is likely to be the outcome of the bargaining session? What
nomination do you think you can realistically obtain?

GO ON TO PART TWO.

 

 

Appendix D

Complete Questionnaire

Form II-Zb PDQ 12

POLITICAL DECISION QUESTIONNAIRE

This questionnaire is part of a study of some basic political abilities.
There are several parts to the questionnaire and each part has its own
instructions. Since various forms of this questionnaire will be given in
a number of different classes, we would like the following information.

Name

 

Class in which this
questionnaire is given

 

Year in college

 

Major

 

Sex M F

In keeping with the American Psychological Association's Code of Ethics,
no information given on this questionnaire will be released at any time
except as part of a statistical average which cannot be identified with a
person. The answers you give will not be available to your instructor,
administrative officers of this university or to investigative agencies for
any reason.

PART ONE

A state political party is divided into three strong factions or groups.
These groups are designated Faction X, Faction Y, and Faction Z. The
party is having a convention. Assume that you are the representative,
that is, the floor leader of one of the three factions in the convention.

There are 350 delegates to the convention, and each delegate has
one vote. Since the factions in this party are quite strong, all of the
delegates in each faction have pledged their votes to the faction leader-
ship. This enables the floor leader of each faction to bargain as the
representative of his entire faction. The faction will then vote as a
bloc, in line with whatever agreement its floor leader may make. Faction
X has 113 delegates (i.e., votes). Faction Y has 113 delegates (votes),
and Faction Z has 124 delegates (votes). The major purpose of this
convention is to nominate a candidate to run for the office of governor
and a candidate for the office of lieutenant governor. Each faction would
like its man to receive the nomination for the governorship, but would
not be extremely dissatisfied if its man received only the lieutenant
governor's place on the ballot.

119

120

Form II-2b

It is standard procedure for two factions to get together and agree
on the division of the nominations. If these two factions have a majority
of the votes of the convention, that is, 176 votes, then the nominations are
divided according to their agreement. An alliance between Faction X and
Faction Y would have 226 votes. An alliance between Faction X and Faction Z
woul have 237 votes, and an alliance between Faction Y and Faction Z would
have 237 votes.

Assume that you are the floor leader of Faction X (113 votes). Which
of the other two factions, Y or Z, will you contact first to try to make
a deal for the division of the nominations?

Faction Y Faction Z
(113 votes) (124 votes)

(Circle one)

 

Which nomination are you prepared to offer them?

Which nomination would you accept as a rock-bottom bargain in a
coalition with this faction's floor leader?

Why did you choose to contact the faction you chose?

What do you expect the other faction leader to demand?

What is likely to be the outcome of the bargaining session? What
nomination do you think you can realistically obtain?

GO ON TO PART TWO.

121

PART TWO: MACH V. ATTITUDE SCALE
Instructions:

Below are twenty groups of statements. Each group contains three state-
ments labeled A, B, and C. Each statement refers to a way of thinking about
people or things in general. The statements reflect opinions, not matters
of fact, and different people have been found to agree with different items.

Read the three statements in each group. First decide which of the three
statements, A, B, or C, comes the closest to describing your own beliefs. On
the red answer sheet, make a mark in the plus (+) space for that statement.
This is the way you would indicate that statement B was the closest to what
you believe.

+ _
A— __
.B‘ —_
C——

Then decide which of the remaining two statements is farthest from your beliefs.
Mark the minus (-) space next to this letter:

4. ..
A———
.13-:-
C‘: -

Here is a set of example statements. On the questionnaire you might find
these statements:

A. It is easy to persuade people but hard to keep them persuaded.

B. Theories that run counter to common sense are a waste of time.

C. It is only common sense to go along with what other people are
doing and not be too different.

0n the answer sheet you will find the answer space next to the circled
question number.

3>
||+
II'

B

 

 

122

If you agreed with A, you would mark your answer sheet like this:

+ -

A- —_

B —— —
C

Then, out of the remaining two statements, if you disagreed most with B, you
would finish marking the question like this:

+ _
A-':_

g
B '-.

C

 

Be sure that you have marked one plus (+) and one minus (—) space in every
group of three statements. Do not omit any group of statements.

123

Please answer all of the sets of statements on your answer sheet, not on
the questionnaire booklet.

1.

A.

It takes more imagination to be a successful criminal than
a successful business man.

The phrase, "the road to hell is paved with good intentions"
contains a lot of truth.

Most men forget more easily the death of their father than
the loss of their preperty.

Men are more concerned with the car they drive than with the
clothes their wives wear.

It is very important that imagination and creativity in
children be cultivated.

People suffering from incurable diseases should have the
choice of being put painlessly to death.

Never tell anyone the real reason you did something unless
it is useful to do so.

The well-being of the individual is the goal that should be
worked for before anything else.

Once a truly intelligent person makes up his mind about the
answer to a problem he rarely continues to think about it.

People are getting so lazy and self-indulgent that it is bad
for our country.

The best way to handle people is to tell them what they want
to hear.

It would be a good thing if people were kinder to others less
fortunate than themselves.

Most people are basically good and kind.

The best criteria for a wife or husband is compatibility--
other characteristics are nice but not essential.

Only after a man has gotten what he wants from life should
he concern himself with the injustices in the world.

Most people who get ahead in the world lead clean, moral lives.

Any man worth his salt shouldn't be blamed for putting his
career above his family.

People would be better off if they were concerned less with
how to do things and more with what to do.

 

10.

11.

12.

124

A good teacher is one who points out unanswered questions
rather than gives explicit answers.

When you ask someone to do something for you, it is best to
give the real reasons for wanting it rather than giving
reasons which might carry more weight.

A person's job is the best single guide as to the sort of
person he is.

The construction of such monumental works as the Egyptian
pyramids was worth the enslavement of the workers who built
them.

Once a way of handling problems has been worked out it is
best to stick with it.

One should take action only when sure that it is morally right.

The world would be a much better place to live in if people
would let the future take care of itself and concern them-
selves only with enjoying the present.

It is wise to flatter important people.

Once a decision has been made, it is best to keep changing it
as new circumstances arise.

It is a good policy to act as if you are doing the things you
do because you have no other choice.

The biggest difference between most criminals and other peOple
is that criminals are stupid enough to get caught.

Even the most hardened and vicious criminal has a spark of
decency somewhere within him.

All in all, it is better to be humble and honest than to be
important and dishonest.

A man who is able and willing to work hard has a good chance
of succeeding in whatever he wants to do.

If a thing does not help us in our daily lives, it isn't very
important.

A person shouldn't be punished for breaking a law which he
thinks is unreasonable.

Too many criminals are not punished for their crime.

There is no excuse for lying to someone else.

 

 

13.

14.

15.

16.

17.

18.

125
Generally speaking, men won't work hard unless they're
forced to do so.

Every person is entitled to a second chance, even after
he commits a serious mistake.

People who can't make up their minds aren't worth bothering
about.

A man's first responsibility is to his wife, not his mother.
Most men are brave.

It's best to pick friends that are intellectually stimulating
rather than ones it is comfortable to be around.

There are very few people in the world worth concerning
oneself about.

It is hard to get ahead without cutting corners here and there.

A capable person motivated for his own gain is more useful
to society than a well-meaning but ineffective one.

It is best to give others the impression that you can
change your mind easily.

It is a good working policy to keep on good terms
with everyone.

Honesty is the best policy in all cases.

It is possible to be good in all respects.
To help oneself is good; to help others is even better.

War and threats of war are unchangeable facts of human life.

Barnum was probably right when he said that there's at
least one sucker born every minute.

Life is pretty dull unless one deliberately stirs up
some excitement.

Most people would be better off if they controlled their
emotions.

 

19.

20.

A.

126
Sensitivity to the feelings of others is worth more
than poise in social situations.

The ideal society is one where everybody knows his
place and accepts it.

It is safest to assume that all people have a vicious
streak and it will come out when they are given a chance.

People who talk about abstract problems usually don't
know what they are talking about.

Anyone who completely trusts anyone else is asking for
trouble.

It is essential for the functioning of a democracy that
everyone vote.

Now check over your answer sheet to see that you have answered every
set of three statements. You should have a total of 20 marks in the
plus (+) columns, and 20 marks in the minus (-) columns.

Write your name on the line provided on the answer sheet and mark the
appropriate space to indicate your sex.

 

127

PART THREE: Topical Inventory
Form N

INSTRUCTIONS

This inventory gives several tapics or situations and a number of
different ways that people react to them. The reactions are presented
in pairs. Your task here is to choose the one member of each pair that
most closely fits your Opinion or feeling about the general topic. Some
of these choices will be easy to make, while others may be rather difficult.
All of the choices are statements of opinion or feeling, so there is never
any "right" or "good" choice in any pair. If you do not agree with
either of the responses in a pair, choose the one that is least disagree-
able of the two.

 

The items in the inventory will be presented like this:
Pair Number

1. When I am confused - - -
a. I try to find a solution and end the confusion.
b. I completely ignore the fact that I am confused.

ii. When I am confused - - -
a. I break out into a nervous sweat.
b. I remain completely calm at all times

How to respond:
Find space on your purple answer sheet with the same number as the

pair number.
It will look like this:

Then decide which response, g_or b, you agree with most. If you
agree most with response a, mark your answer sheet like this:

Make your mark heavy and dark. When you have finished the first item
go on to the next. Decide which response most fits you. For example,
on pair two if response b, "I remain completely calm at all times" best

 

128

describes your behavior, you would mark your answer sheet like this:

ii a b c d e

Be sure to choose only one of the responses in each pair. Do not
skip any pair, even if it is difficult to make a decision. Once you have
marked your choice for an item, don't go back to it; first impressions
are usually the most reliable in this inventory.

There are six situations or topics in the inventory. Each situation
or topic his six pairs of responses. Be sure to pick one and only one
response from each pair. When you have finished, you should have 36 marks
on your answer sheet.

Before you begin, put your name and indicate your sex in the appro-
priate place on the answer sheet. Work at your own speed, but work
straight through the inventory without stopping. Once you have completed
an item do not return to it.

129

Questions 1-6:

Imagine that someone has criticized you. Choose the response from each
pair that comes closest to your feelings about such criticism. Indicate your
choice by marking either the A or the B space on your answer sheet.

1. When I am criticized - - -

a. I try to take the criticism, think about it, and value it for what
it is worth. Unjustified criticism is as helpful as justified
criticism in discovering what other people's standards are.

b. I try to accept the criticism but often find that it is not justified.
PeOple are too quick to criticize something because it doesn't fit
their standards.

 

2. When I am criticized - - —

a. I try to determine whether I was right or wrong. I examine my be-
havior to see if it was abnormal. Criticism usually indicates that
I have acted badly and tends to make me aware of my own bad points.

b. It could possibly be that there is some misunderstanding about some-
thing I did or said. After we both explain our viewpoints, we can
probably reach some sort of compromise.

 

3. When I am criticized - - -

a. I listen to what the person says and try to accept it. At any rate,
I will compare it to my own way of thinking and try to understand
what it means.

b. I feel that either I'm.not right, or the person who is criticizing
me is not right. I have a talk with that person to see what's
right or wrong.

 

4. When I am criticized - - -

a. I usually do not take it with good humor. Although, at times,
constructive criticism is very good, I don't always think that the
criticizer knows what he is talking about.

b. At first I feel that it is unfair and that I know what I am doing,
but later I realize that the person criticizing me was right and I
am thankful for his advice. I realize that he is just trying to
better my actions.

 

5. When I am criticized - - -

a. I try to ask myself what advantages this viewpoint has over mine.
Sometimes both views have their advantages and it is better to com-
bine them. Criticism usually helps me to learn better ways of
dealing with others.

b. I am very thankful. Often I can't see my own errors because I am
too engrossed in my work at the time. An outsider can judge and
help me correct the errors. Criticism in everyday life usually
hurts my feelings, but I know it is for my own good.

 

130
When I am criticized - - -
a. It often has little or no effect on me. I don't mind constructive

criticism too much, but I dislike destructive criticism. Destruc-
tive criticism should be ignored.

 

b. I try to accept and consider the criticism. Sometimes it has
caused me to change myself; at other times I have felt that the
criticism didn't really make much sense.

Questions 7-12:

Imagine that you are in doubt. Choose the response from each pair that

comes closest to your feelings about such doubt. Indicate your choice by

marking either the A or the B space on your answer sheet.

7.

10.

11.

When I am in doubt - - -

a. I become uncomfortable. Doubt can cause confusion and make one
do a poor job. When one is in doubt he should ask and be sure
of himself.

 

b. I find myself wanting to remove the doubt, but this often takes time.
I may ask for help or advice if I feel that my questions won't
bother the other person.

 

When I am in doubt - - -

a. I don't get too upset about it. I don't like to ask someone else
unless I have to. It's better to discover the correct answer on
,your own.

 

b. I usually go to someone who knows the correct answer to my question.
Sometimes I go to a book which will set me straight by removing the
doubt.

When I am in doubt - - -

a. I first try to reason things out and check over the facts. Often
I approach others to get ideas that will provide a solution.

 

b. I think things over, ask questions, and see what I can come up with.
Often several answers are reasonable and it may be difficult to
settle on one.

When I am in doubt — - -
a. I realize that I'll have to decide on the correct answer on my own.

Others try to be helpful, but often do not give me the right
advice. I like to judge for myself.

 

b. I usually try to find out what others think, especially my friends.
They may not know the answer, but they often give me some good ideas.

When I am in doubt - - -
a. I look over the problem and try to see why there is a doubt. I try

to figure things out. Sometimes I just have to wait a while for an
answer to come to me.

 

b. I try to get some definite information as soon as possible. Doubt
can be bad if it lasts too long. It's better to be sure of yourself.

131
12. When I am in doubt - - -

a. I consider what is best in the given situation. Although one should
not rush himself when in doubt, he should certainly try to discover
the right answer.

b. I act according to the situation. Sometimes doubt can be more
serious than at other times and many of our serious doubts must go
unanswered.

 

Questions 13-18:

Imagine that a friend has acted differently toward you. Choose the
response from each pair that comes closest to your feelings about such an
action. Indicate your choice by marking either the A or the B space on your
answer sheet.

13. When a friend acts differently toward me - - -

a. I am not terribly surprised because people can act in many different
ways. We are different people and I can't expect to understand all
his reasons for acting in different ways.

b. I am usually somewhat surprised but it doesn't bother me very much.
I usually act the way I feel towards others. People worry too much
about others' actions and reactions.

 

 

14. When a friend acts differently toward me - - -

a. I find out why. If I have doen something wrong I will try to straighten
out the situation. If I think he's wrong, I expect him to clear things up.
b. I feel that I may have caused him to act in a different way. Of
course, he may have other reasons for acting differently which would
come out in time.

15. When a friend acts differently toward me - - -

a. I first wonder what the trouble is. I try to look at it from his
viewpoint and see if I might be doing something to make him act
differently toward me.

b. It is probably because he has had a bad day, which would explain
this different behavior; in other cases he may just be a changeable
kind of person.

 

16. When a friend acts differently toward me - - —

a. It is probably just because something is bothering him. I might try
to cheer him up or to help him out. If these things didn't work I
would just wait for him to get over it.

b. I try to understand what his different actions mean. I can learn
more about my friend if I try to figure out why he does things.
Sometimes the reasons may not be very clear.

 

17. When a friend acts differently toward me - - -

a. There has to be a definite reason. I try to find out this reason,
and then act accordingly. If I'm right I'll let him known it. If
he's wrong, he should apologize.

b. I usually let him go his way and I go mine. If a friend wants to
act differently that's his business, but it's my business if I
don't want to be around when he's that way.

 

 

,ilftlrlli Ir
I‘l ‘Ill: all. Ills! I." III... all!

132

18. When a friend acts differently toward me - - -

a. I don't get excited. People change and this may cause differences.
It is important to have friends, but you can't expect them to always
be the same.

b. I like to get things back to normal as soon as possible. It isn't
right for friends to have differences between them. Whoever is at
fault should straighten himself out.

 

Questions 19-24:

Think about the topic of people in general. Choose the response from
each pair that comes closest to your thoughts about people. Indicate your
choice by marking either the A or the B space on your answer sheet.

19. This I believe about people - - -

a. Whatever differences may exist between persons, they can usually get
along if they really want to. Although their ideas may not agree,
they probably still have something in common.

b. People can learn from those who have different ideas. Other people
usually have some information or have had some experience which is
interesting and can add to one's knowledge.

 

 

20. This I believe about people - - -

a. People can act in all sorts of ways. No single way is always best,
although at certain times a particular action might be wiser than
others.

b. Each person should be able to decide the correct thing for himself.
There are always a few choices to be made and the individual himself
is in the best position to pick the right one.

 

21. This I believe about people - - -

a. Some people think they know what's best for others and try to give
advice. These people shouldn't make suggestions unless asked for
help.

b. There are certain definite ways in which people should act. Some
don't know what the standards are and therefore need to be straight-
ened out.

 

22. This I believe about people - - -

a. I can tell if I am going to get along with a person very soon after
meeting him. Most people act either one way or another and usually
it is not difficult to say what they are like.

b. It's hard for me to say what a person is like until I've known him
a long time. PeOple are not easy to understand and often act in
unpredictable ways.

23.

24.

133

This I believe about people - - —

a. People have an outside appearance that usually isn't anything like
what can be found on the inside, if you search long and hard enough.

 

b. Each person is an individual. Although some people have more good
or bad points than others, no one has the right to change them.

This I believe about pe0ple - - -
a. People can be put into categories on the basis of what they're really

like. Knowing the way a person really is helps you to get along with
him better.

 

b. People are unlike one another in many respects. You can get along
with people better and better understand them if you are aware of the
differences.

Questions 25-30:

Think about the general t0pic of leaders. Choose the response from each

pair that comes closest to your thoughts about leaders. Indicate your choice
by marking either the A or the B space on your answer sheet.

25.

26.

27.

28.

Leaders - - -

a. Leaders do not always make the right decisions. In such cases, it
is wise for a man to look out for his own welfare.

 

b. Leaders are necessary in all cases. If a leader cannOt make the
right decisions another should be found who can.

Leaders - - -
a. Leaders cannot provide all the answers. They are like other people

--they have to try to figure out what action is necessary and learn
from their mistakes.

 

b. Leaders make decisions sometimes without being sure of themselves.
We should try to understand this and think of ways to help them out.

Leaders - - -

a. I like a leader who is aware of how the group feels about things.
Such a leader would not lead any two groups in exactly the same way.

 

b. A person should be able to put his confidence in a leader and feel
that the leader can make the right decision in a difficult situation.

Leaders - - -
a. There are times when a leader shouldn't make decisions for those

under him. The leader has the power to decide things, but each
man has certain rights also.

 

b. A leader should give those under him some opportunity to make de-
cisions, when possible. At times, the leader is not the best judge
of a situation and should be willing to accept what others have to
say.

 

 

134

29. Leaders - - -

a. Some leaders are good, others are quite poor. Good leaders are
those who know what is right for the men under them. These leaders
deserve the respect of every man.

b. Leaders cannot be judged easily. Many things go to make up good

 

leadership. Most people fall short in some way or another, but that

is to be expected.

30. Leaders - - -

a. Leaders are needed more at certain times than at others. Even though

people can work out many of their own problems, a leader can some-
times give valuable advice.

b. Some people need leaders to make their decisions. I prefer to be
an individual and decide for myself, when possible. Most leaders
won't let you do this.

 

Questions 31-36:

Imagine that someone has found fault with you. Choose the response
from each pair that comes closest to your feelings about such a situation.
Indicate your choice by marking either the A or the B space on your answer
sheet.

31 When other people find fault with me - - -

a. It means that someone dislikes something I'm doing. People who
find fault with others are not always correct. Each person has
his own ideas about what's right.

b. It means that someone has noticed something and feels he must
speak out. It may be that we don't agree about a certain thing.
Although we both have our own ideas, we can talk about it.

 

32. When other people find fault with me — - -

 

a. I first wonder if they are serious and why they have found fault with

me. I then try to consider what they've said and make changes if
it will help.

33. When other people find fault with me - - -

a. They have noticed something about me of which I am not aware. Al—
though criticism.may be hard to take, it is often helpful.

b. They are telling me something they feel is correct. Often they may
have a good point which can help me in my own thinking. At least
it's worthwhile to consider it.

 

34. When other people find fault with me — - -

a. I may accept what is said or I may not. It depends upon who is

ypointing out the fault. Sometimes it's best to just stay out of sight.
b. I accept what is said if it is worthwhile, but sometimes I don't feel

 

like changing anything. I usually question the person.

35.

36.

135

When other people find fault with me - - -

a.

I like to find out what it means; since people are different from one
another, it could mean almost anything. A few peOple just like to
find fault with others but there's usually something to be learned.

 

b.

There is something to be changed. Either I am doing something wrong
or else they don't like what I'm doing. Whoever is at fault should
be informed so that the situation can be set straight. '

When other people find fault with me ... -

a.

I don't mind if their remarks are meant to be helpful, but there
are too many people who find fault just to give you a hard time.

 

b.

It often means that they're trying to be disagreeable. People get
this way when they've had a bad day. I try to examine their re-
marks in terms of what's behind them.

Please count the marks on your answer sheet to see that you have made
36 choices, one from each pair.

 

 

Illl“: Ill-lil-

 

Appendix E
Machiavellianism and Interpersonal Tapical

Inventory Scoring Routines

 

 

Form II—2b

PART FOUR

This part of the Political Decision Questionnaire is a convention
problem somewhat different from the first one. Please work through it
without referring to other portions of the questionnaire.

There are 350 delegates to the convention and each delegate has one
vote. Since the factions in this party are quite strong, all of the delegates
in each faction have pledged their votes to the faction leadership. This
enables the floor leader of each faction to bargain as the representative
of his entire faction. The faction will then vote as a bloc, in line with
whatever agreement its floor leader may make. Faction X has 120 delegates
(i.e., votes). Faction Y has 110 delegates (votes), and Faction Z has
120 delegates (votes). The major business of this convention is to
nominate a candidate to run for the office of governer and a candidate
to run for the office of lieutenant governor. Each faction would like its
man to receive the nomination for the governorship, but would not be extremely
dissatisfied if its man received only the lieutenant governor's place on
the ballot.

It is standard procedure for two factions to get together and agree on
the division of the nominations. If these two factions have a majority of
the votes of the convention, that is, 178 votes, then the nominations are
divided according to their agreement. An alliance between Faction X and
Z would have 240 votes, and an alliance between Faction Y and Faction Z
would have 230 votes.

Assume that you are the floor leader of Faction X (120 votes). Which
of the other two factions, Y or 2, will you contact first to try to make
a deal for the division of the nominations?

Faction Y Faction Z
(110 votes) (120 votes)

(Circle one)
Which nomination are you prepared to offer them?

Which nomination would you accept as a rock-bottom bargain in a

coalition with this faction's floor leader?

What do you expect the other faction leader to demand?

What is likely to be the outcome of the bargaining session? What
lumnination do you think you can obtain for your faction?

THIS IS THE END OF THE POLYTICAL DECISION QUESTIONNAIRE. Thank you for
Your cooperation. If you wish to find out the results of studies similar to
fills one, you may get a c0py of one of the reports from the Human Learning
ReSearch Institute, 202 Erickson Hall, M.S.U.

137

 

 

all!

138

Three coding procedures are presented here. SUBROUTINE MACH counts
the number of Machiavellian orderings of the items in each of the 20
triplets of Mach V items. AM is an N by 60 array of subjects' responses
to each item, where N is the total number of subjects, and the entries
are 8.0 for "Agree" and 9.0 for "Disagree". (This coding is dictated by
the position of the Mach V response spaces on the machine-score sheet.)
BM is an N by 20 array of individual item Mach scores. Sum is an N element
vector of total Mach responses.

SUBROUTINE TUCK counts the number of items the subject has endorsed
from each of the four complexity system domains. Here AC is a l by 36
array of one subject's item responses, BC is a l by 4 array of the number
of items in each category he endorsed, and M is a 2 by 36 matrix containing
the scoring Key-the number identifying the complexity system each item
represents.

SUBROUTINE CATAG determines whether a subject has any system score
which is in the upper 25% of the population distribution. If he has one
and only one such score, CATC is set equal the number of the complexity

category in which he has the high score.

 

 

1'

 

 

 

(3061(3065

C

L

C

139

SUBROUTINE MACH‘IvA”.EM.SLF'Nbe1
DIMENSICN5V1196J198V1ltZCloSUMiVl
RCAD FACH DATA, SUBJECIS TAKEN If BE ARRAY Bra;
CL 5 KZ=1920
5 BMII.KZ)=C.O
REAC‘5298121(Aplloleleoégl
312 FORVAT1/6X.oCF1.3/)
DC 10 J=lv63
1F lA”(IyJ).NL-O.Cl CC 1C 10
AN119J138.5
1C CUNIIVUE
THE ABOVE CPLRATICN FILLEL 1N ALL NC? RFQPFN't VDACiS WITH THE MIUPUINI
PREFERENCE RANKING IMPLIED BY THE RANK 4 CUT Cf ‘ PHLC‘DURE

th FCLLCHINC CPERAIICNS WILL GtNEKaTL a», Ihi vaca SLPRFS FOR
E'Ch IRIPLET 0F IIEMS. A HIGH MACH RtSPCXSt WILL ME scwxan 1.3. A LOW
P'CH RESPLNSE nR AN INCCRRECTLY MARKED RtSPLUL. WILL tr SCORED o.
21 IF IAMII.II.GI.AMII.S)I sMII.II=I.:
22 1F (AM‘I,71.GT.AM(I'11)1 BN11921=103
24 IF (AMII.I3I.LI.A~II.I?II BMII. 3)=1..
25 IF‘AMl [.19).GI.AP(I.21)) 9M1194131.3
28 [F (AMII.25I.GI.A~II.27II erI.5I=I.¢
38 IF (AMII.3II.CI.AM(I.35II BMII.6I=I..
32 1F (AM(1937).LI.AN(1'3911 “~(117,;lo£
54 IF (AM(I,45).LT.AN(I.47D) BF(I.8)=1.C
36 [FIAM(I.51).LI.A“(I.53)) thI. 9I=I.:
38 1F IAMII.55I.GT.A~II.57II BP(I.101=1.;
4: IF (AM(I.2).GT.AM(I.4)) RNI I.III=I.-
42 IFIAMII.10I.LI.AMII.12)I sNII.12I=I.:
44 IF(AMII.I4I.LI.AMII.I&II 8N(I.l3)=1.c
«e IFIAVII.22I.GI.AMII.24II BH(I,14)=1.£
as IF(AM(I .28).LT.AP(I.3011 s~II.ISI=I.:
5C IF(AMII.34I.LI.A~II.3¢II sHII.IsI=I.t
52 IFlAM(I'38).GI.AM(I.4c)) BM(I.17)=1.
54 IFlAMlI.44).LI.ANiI.46)) s~(1.18I=1.c
56 IF(AM(I.521.GT.AM(I.54)) sMII.19I=I.c
58 IF (AMII.5¢I.cI.AMII.58II BP(1.201=1.C
ac CONTINUE
THIS COMPLETtS IHE RECCCE FCR SUBJECTS I. lbt wyxr ovIRATION
u'LL SUM THE MACH SCORES.
SUMINVI=C.C
DC 70 JJ=I.20
7C SUMINVI=SL~INVI+BMII.JJI
l‘-Is COMPLETiS THE Macs RtCCCt nun SCCRr LUNLLAIILN Eta SUBJ. I.

h"IIE ROUTINE TC STCRE PACH SCORES ON TAPE
HRIIE‘5119(BMlloJleJJ=11291
RtTLRN
END

1.

Li

L
l

i1

 

815
816

99

140

SUBRUUIINE IUCK (IoACoBCthtn)
DIMENSION AC‘lv3blvBC119419M12936)
REAC152’81611AC110K19K31936,
FURNAT‘I/llxv36Fl.OI33X1
BC(I.HT)=C.O

THIS CLEARS THE TEMPORARY COUNTER

101

102

20C

201

DC 200 K=1v36

MT=C

IFlAClIoK 1-1.0) 200'10191C2
HT=NlltKl

BC‘IgHT138C119HT1 *loC

GO TO 200

MT=P129K1

BC‘IvHT138C11vHT1 +1.0

CONTINUE
HRIIEIS3).(BCII.MI).FI=1.4)
HRITE‘6222511NVy13C111M119H131v4’
FCRFAT11594F1000’

RETLRN

END

 

I‘ll-ll 1“ mill I'i’il 1" 1‘1!

 

141

SUBROUTINE CATAGIBCoCATCoNIZvIoIT)
DIMENSION 8C!IIQIICATCINI,ITIN.4).AVB(4)
1 ’TESTI4)

REHIND 53

DC 400MT3194

SUMEC=0.0

SSQBC=0.0

SDBC=0.0

AVBIHT)=O.C

DO 300 NV=I,N

CATCINVI=C.O
REACIS3IQIECII’MSI'M5=194)
SUMBC=SUMBC +8C‘I’MT,

SOC SSQEC=SSCBC+I8C(I,"TI1092.0
VARBC=SSQBC/(N-II-(SUPBCG'ZIII(N-1)'N)
SDBC=SQRTIVAR8CI
REWIND 53

C THE FCLLCWINC STATISTIC PROVIDES A STES tOR THE

C UPPER 25 PERCEAT CF EACT BCIHTI DISTRIBUTION.

C THIS TEST WILL IDENTIFY PERSONS CF EACH SYSTEM TYPE.
C

 

TESTIMT)=Z'SDBC
AVBIMTI=SLV8CIN
400 PRINT 5000, VARBC: SDBC: AVBIHTI. MT
500C FURFATISHOVAR= ,F7.3,IOX,3H SD.F7.3,2X.10HAVERAGE 8= 9
1F7.3'7HSYSTEM 913)
C
C
C THE FCLLCHING PROCESS IS THE TEST ITSELF
REHIND 53
DC 600 NV=1.N
READID319IHCII'MZ)9”Z=194)
DC 500 MT=194
ITIRVyMTI=C
IF!(BCII.FTI-AV8(MT)).GT.TEST(PT)I ITINV,NT)=I
500 CONTINUE
ITS=O
DO 601 HT=194
601 ITS=ITS+ITINV9MTI
IFIITS.E0.11 602:603
602 DU 604 MT=1o4
IFIITTNV,FT).EQ.II 6109604
603 CATCINV)=C.O
GC TO 600
610 CATCINVI=FT
604 CONTINUE
600 CCNTINUE
RETLRN
END

 

 

I. T I I- I‘ll-III I ' 1'1 III l 1 III"

142

ITI SCORING KEYa

some areas
Pair First Item Second Item Pair First Item Second Item
at... A .B. 219.1. A I:
1. 3 2 l9. 3 4
2. l 4 20. 4 2
3. 3 1 21. 2 1
4. 2 l 22. 1 4
5. 4 3 23. 3 2
6. 2 4 24. 1 3
7. 1 3 25. 2 1
8 2 1 26. 4 3
9 3 4 27. 3 l
10. 2 3 28. 2 4
11. 4 1 29. 1 4
12. 2 4 30. 3 2
13. 4 2 31. 2 4
14. 1 3 32. 3 1
15. 3 2 33. 3 4
16. 3 4 34. 1 2
17. 1 2 35. 4 l
18. 4 1 36. 2 3

 

aThese items are read in row order into array M of SUBROUTINE TUCK,
where they provide a basis for classifying responses.

 

 

 

Append ix F

Instructions for Experiment II

 

CONDITION I

INSTRUCTIONS

The experiment you will participate in today consists of a series of
competitive games. The game you will play will be a political convention
game. In this game, a state political party is divided into three strong
factions or groups. These groups are designated Faction X, Faction Y, and
Faction Z. Each of you will be the representative, that is, floor leader,
of one of the three factions in the convention.

There are 350 delegates to the convention, and each delegate has one
vote. Since the factions in this party are quite strong, all of the delegates
in each faction have pledged their votes to the faction leadership. This
enables the floor leader of each faction to bargain as the representative
of his entire faction. The faction will then vote as a bloc, in line
with whatever agreement its floor leader may make. Faction X has __delegates
(i.e., votes). Faction Y has __de1egates (votes), and Faction Z has
__delegates (votes). The major business of this convention is to decide
how many of 100 political jobs each faction will receive. Each faction
would like to get as many of these jobs as possible.

It is standard procedure for two factions to get together and agree
on the division of the jobs. If these two factions control a majority
of the votes of the convention, that is, 176 votes, then the jobs are
divided according to their agreement. An alliance between Faction X and
Faction Y would have __ votes. An alliance between Faction X and
Faction Z would have __ votes, and an alliance between Faction Y and

Faction Z would have __ votes.

144

 

 

CONDITION II
INSTRUCTIONS

The experiment you will participate in today consists of a series of
competitive games. The game you will play will be a political convention
game. In this game a state political party is divided into three strong
factions or groups. These groups are designated Faction X, Faction Y, and
Faction Z. Each of you will be the representative, that is, floor leader,
of one of the three factions in the convention.

There are 350 delegates to the convention, and each delegate has one
vote. Since the factions in this party are quite strong, all of the dele-
gates in each faction have pledged their votes to the faction leadership.
This enables the floor leader of each faction to bargain as the representa-
tive of his entire faction. The faction will then vote as a bloc, in line
with whatever agreement its floor leader may make. Faction X has ____delegates
(i.e., votes). Faction Y has____ delegates (votes), and Faction Z has
___ delegates (votes). The major business of this convention is to nomin-
ate a candidate to run for the office of governor and a candidate to run
for the office of lieutenant governor. Each faction would like its man
to receive the nomination for the governorship, but would not be extremely
dissatisfied if its man received only the lieutenant governor's place on
the ballot.

It is standard procedure for two factions to get together and agree
on the division of the nominations. If these two factions have a majority
of the votes of the convention, that is, 176 votes, then the nominations

are divided according to their agreement. An alliance between Faction X

and Faction Y would have votes. An alliance between Faction X and Faction
Z would have votes, and an alliance between Faction Y and Faction Z
would have votes.

145

 

 

146

These cards list the votes controlled by each faction leader.
The second card indicates the faction assigned to you.

(Pass out vote distribution and identification cards.)

The convention will proceed like this: First, each of you will fill
out a "choice form" to indicate the faction you wish to negotiate with in
the first round of negotiations. The form also asks some questions about
what kind of offer you are willing to make, and what you think they will
accept as a final agreement. This information is for use in the analysis
of the study. It does not constitute an actual opening offer or any
part of the bargaining process. I will not disclose these answers, but
I will tell you if any two of you have chosen to bargain with each other.
If two of you have chosen each other (regardless of what offers you intend
to make) you will have three minutes to verbally negotiate the division of
jobs. The third man will leave the room during the bargaining session.

If the two bargainers reach an agreement about the division of the jobs,
they will fill out an "agreement form." The third man will be called
back, the jobs will be divided according to the agreement, and we will go
on to the next game.

If the two bargainers do not reach an agreement during the negotiation
period, the third man will be called back and all three of you will fill
out another set of choice forms. Another round of negotiation will then
follow.

Do you have any questions before we begin?

If not, here are the choice forms. Please pass the form back through
the slot in the divider when you have filled it out.

(If a reciprocal choice has been made.)
Factions ___and ___have chosen to negotiate with each other on the

first round of negotiations. Faction leader __3 would you please step

 

147

across the room to 201D. When you shut the door we will begin the
three-minute bargaining session. I will come and get you at the
end of three minutes.

(To the remaining subjects)

Your task now is to come to some verbal agreement, if you can,
on the division of the 100 jobs. Please remain seated while bargaining.
You have three minutes.

(Upon reaching anagreement)

Have you reached an agreement? Would you please fill out these
forms to record your agreement while I call the other man back.

(With third man present)

 

As I mentioned before, we will go on to another game when an agree-
ment has been reached. Since you have already played this game, and since
two of you have had some bargaining experience, this is likely to affect
the way you play the next game. To control for this, we have constructed
a questionnaire that presents you with a convention situation somewhat
different from the one you just played. We would like you to fill out
this questionnaire visualizing your opponents as people you have never
met. Don't take them to be the people you played with today. The
questionnaire is in three parts. The first is the convention situation,
and the second two are attitude scales like others you have taken before.
You should finish it with enough time left in the hour to talk a bit
about some of the purposes of this study.

(Assign them to different rooms.)

 

 

CONDITION III
INSTRUCTIONS

The experiment you will participate in today consists of a series of
competitive games. The game you will play will be a political convention
game. In this game a state political party is divided into three strong
factions or groups. These groups are designated Faction X, Faction Y, and
Faction Z. Each of you will be the representative, that is, floor leader,
of one of the three factions in the convention.

There are 350 delegates to the convention, and each delegate has one
vote. Since the factions in this party are quite strong, all of the dele- =a.
gates in each faction have pledged their votes to the faction leadership.

This enables the floor leader of each faction to bargain as the representa- T
tive of his entire faction. The faction will then vote as a bloc, in line 1
with whatever agreement its floor leader may make. Faction X has ____dele-
gates (i.e., votes). Faction Y has ____delegates (votes), and Faction Z has
____delegates (votes). The major business of this convention is to nominate
a candidate to run for the office of governor and a candidate to run for the
office of lieutenant governor. Each faction would like its man to receive
the nomination for the governorship, but would not be extremely dissatisfied
if its man received only the lieutenant governor's place on the ballot.
Since this party has effectively controlled state government for several
years, any man who receives the nomination is virtually assured of winning
the general election against the opposition party.

It is standard procedure for two factions to get together and agree on

 

the division of the nominations. If these two factions‘have a majority of
the votes of the convention, that is, 176 votes, then the nominations are

divided according to their agreement. An alliance between Faction X and

Faction Y would have votes. An alliance between Faction X and Faction Z
would have votes, and an alliance between Faction Y and Faction 2 would
have votes.

148

 

149

These cards list the votes controlled by each faction leader. The
second card indicates the faction assigned to you. (Pass out vote distri-
bution and identification cards.)

The convention will proceed like this: First, each of you will fill
out a "choice form" to indicate the faction you wish to negotiate with in
the first round of negotiations. The form also asks some questions about
what kind of offer you are willing to make, and what you think they will
accept as a final agreement. This information is for use in the analysis
of the study. It does not constitute an actual opening offer or any part
of the bargaining process. I will not disclose these answers, but I will

tell you if any two of you have chosen to bargain with each other. If two

 

of you have chosen each other (regardless of what offers you intend to

make) you will have three minutes to verbally negotiate the division of
nominations. The third man will leave the room during the bargaining session.
If the two bargainers reach an agreement about the division of the assign—
ment will fill out an "agreement form." The third man will be called back,
the nominations will be divided according to the agreement, and we will go

on to the next game.

 

If the two bargainers do not reach an agreement during the negotiation
period, the third man will be called back and all three of you will fill
out another set of choice forms. Another round of negotiation will then

follow.

 

Do you have any questions before we begin?

If not, here are the choice forms. Please pass the form back through
the slot in the divider when you have filled it out. (If a reciprocal choice
has been made)

Factions ___ and ___ have chosen to negotiate with each other on the

first round of negotiations. Faction leader , would you please step

 

150

across the room to 201 D. When you shut the door, we will begin the three-
minute bargaining session. I will come and get you at the end of the
three minutes.

(To the remaining subjects)

Your task now is to come to some verbal agreement, if you can, on the
assignment of the nominations. Please remain seated while bargaining.

You have three minutes.

(Upon reaching an agreement)

Have you reached an agreement? Would you please fill out these forms
to record your agreement while I call the other man back.

(With third man present)

As I mentioned before, we will go on to another game when an agreement
has been reached. Since you have already played this game, and since two
of you have had some bargaining experience, this is likely to affect the
way you play the next game. To control for this, we have constructed a
questionnaire that presents you with a convention situation somewhat differ-
ent from the one you just played. We would like you to fill out this question-
naire visualizing your opponents as people you have never met. Don't take
them to be the people you played with today. The questionnaire is in three
parts. The first is the convention situation, and the second two are
attitude scales like others you have taken before. You should finish it
with enough time left in the hour to talk a bit about some of the purposes
of this study.

(Assign them to different rooms.)

 

 

151

CHOICE FORM
I represent Faction X Y Z.
(circle one)
I wish to negotiate with Faction

X Y Z. I am prepared to
(circle one)

offer them I of the jobs as
my Opening offer. I expect him to
demand I of the jobs on a

final agreement.

AGREEMENT FORM

Factions and agree
to pool their resources and form a
coalition. It is agreed that Faction
will get jobs as its
share of the convention outcome and
Faction will get jobs
as its share.

 

 

 

 

I “““““

 

 

i

I
I

fl
1'
ll
L"
H
T”
III
E“
H

H

Will” I

169 0138