A GUlTMAN SCALE FOR THE STRENGTH OF AN INTERPERSONAL TIE *

Similar documents
Social Penetration Theory

reward based power have ability to give you what you want. coercive have power to punish

NEGOTIATION SEVENTH EDITION

HOW STATISTICS IMPACT PHARMACY PRACTICE?


Answers to end of chapter questions

Chapter 14. Social Psychology. How Does the Social Situation Affect our Behavior? Social Psychology

PERCEIVED TRUSTWORTHINESS OF KNOWLEDGE SOURCES: THE MODERATING IMPACT OF RELATIONSHIP LENGTH

BELL WORK. List three words that you think describe the "helping process. Be ready to share

Introduction to Social Psychology p. 1 Introduction p. 2 What Is Social Psychology? p. 3 A Formal Definition p. 3 Core Concerns of Social Psychology

Chapter Seven. Learning Objectives 10/2/2010. Three Good Reasons Why You Should Care About... Interpersonal Behavior

Qualitative and Quantitative Approaches Workshop. Comm 151i San Jose State U Dr. T.M. Coopman Okay for non-commercial use with attribution

An International Study of the Reliability and Validity of Leadership/Impact (L/I)

Identity Control Theory

Behavioral EQ MULTI-RATER PROFILE. Prepared for: By: Session: 22 Jul Madeline Bertrand. Sample Organization

LIMITATIONS OF COMMUNICATION: MECHANISMS OF IMAGE MAINTENANCE IN FORM OF SELECTIVE PERCEPTION, SELECTIVE MEMORY AND SELECTIVE DISTORTION

Is Leisure Theory Needed For Leisure Studies?

Ronald B. Adler, Lawrence B. Rosenfeld, Russell F. Proctor II Interplay : the process of interpersonal communication

Sociology 4 Winter PART ONE -- Based on Baker, Doing Social Research, pp , and lecture. Circle the one best answer for each.


COACH WORKPLACE REPORT. Jane Doe. Sample Report July 18, Copyright 2011 Multi-Health Systems Inc. All rights reserved.

RPAD637 Week 13 Reading Summary: Models for Studying Network Effects & Diffusion Jackson and Nampoothiri

Running head: PRESERVING RELATIONSHIPS WITH AN INTERNET CONNECTION 1

Psychological needs. Motivation & Emotion. Psychological needs & implicit motives. Reading: Reeve (2015) Ch 6

Creating a Positive Professional Image

The Toyota Way Chapters February 13, 2014

Interpersonal Communication in a Changing World: Culture and Social Networking 28

DANIEL KARELL. Soc Stats Reading Group. Princeton University

Introduction to Social Network Analysis for Dissemination and Implementation Research

Members recognize that other members share similar feelings,

PERSON PERCEPTION AND INTERPERSONAL ATTRACTION

Psychological Visibility as a Source of Value in Friendship

Emotional Intelligence and Intimacy in Relationships

Groups in Organizations. Overview of topics

Dr. Mike Christensen, CIH AVP, Risk Mgmt. Services CSU, Sacramento

ISA 540, Auditing Accounting Estimates, Including Fair Value Accounting Estimates, and Related Disclosures Issues and Task Force Recommendations

The four chapters in Part I set the stage. Chapter 1 moves from the implicit common sense theories of everyday life to explicit theories that are

DRUG USE OF FRIENDS: A COMPARISON OF RESERVATION AND NON-RESERVATION INDIAN YOUTH

Groups, norms, and conformity. What s it about?

INTERPERSONAL RELATIONSHIPS (IR)

REPORT ON EMOTIONAL INTELLIGENCE QUESTIONNAIRE: GENERAL

Understanding Interpersonal Trust. Further prevention of occupational injuries beyond current plateaus will occur

SOCI 323 Social Psychology

Abstract. In this paper, I will analyze three articles that review the impact on conflict on

CHAPTER 3 METHOD AND PROCEDURE

A review of statistical methods in the analysis of data arising from observer reliability studies (Part 11) *

C O N T E N T S ... v vi. Job Tasks 38 Job Satisfaction 39. Group Development 6. Leisure Activities 41. Values 44. Instructions 9.

Cambridge Public Schools SEL Benchmarks K-12

The weak side of informal social control Paper prepared for Conference Game Theory and Society. ETH Zürich, July 27-30, 2011

Advanced Code of Influence. Book 10

Bandura s Social Learning & Cognitive Learning Theory

Exploring the Role of Time Alone in Modern Culture

David O Malley, Ph.D., LISW Case Western Reserve University Cleveland, Ohio

Philosophy of Physical Activity

The Effects of Societal Versus Professor Stereotype Threats on Female Math Performance

Funnelling Used to describe a process of narrowing down of focus within a literature review. So, the writer begins with a broad discussion providing b

BACKGROUND + GENERAL COMMENTS

Statistical Analysis of Complete Social Networks

Five Dimensions of Conflict By Gayle Mertz and Carol Miller Lieber

THE TRANSCENDENTAL MEDITATION PROGRAM AND CREATIVITY

EASY A-Z Guide. to Choose the Right MP3 for You. MP3 Series & Title. Empowerment

Reflect on the Types of Organizational Structures. Hierarch of Needs Abraham Maslow (1970) Hierarchy of Needs

GROUP REPORT. Insert Personalized Title SAMPLE. Assessments Completed Between: December 18, 2013 and December 20, 2013

ADMS Sampling Technique and Survey Studies

Promises and Perils of Assessing Character and Social and Emotional Learning. Clark McKown, Ph.D.

Sample Report. Sample Report Report. Fa c i l i tat or s (05/13) 180

Pearson Education Limited Edinburgh Gate Harlow Essex CM20 2JE England and Associated Companies throughout the world

Definitions of Indicators

CHAPTER 6 BASIS MOTIVATION CONCEPTS

Types of Research (Quantitative and Qualitative)

Self-disclosure in intimate relationships: Associations with individual and relationship characteristics over time

Bayesian and Frequentist Approaches

Learning Objectives. Learning Objectives 17/03/2016. Chapter 4 Perspectives on Consumer Behavior

Toward the Measurement of Interpersonal Generosity (IG): An IG Scale Conceptualized, Tested, and Validated Conceptual Definition

Motivation & Emotion. Psychological & social needs

Exploring Network Effects and Tobacco Use with SIENA

Agents with Attitude: Exploring Coombs Unfolding Technique with Agent-Based Models

Human Communication & Verbal Problem Solving

Simultaneous Equation and Instrumental Variable Models for Sexiness and Power/Status

Interpersonal Compatibility and S ocial Loafing: An Introduction to a S tudy

Psychological needs. Motivation & Emotion. Psychological & social needs. Reading: Reeve (2009) Ch 6

Psychological needs. Motivation & Emotion. Psychological & social needs. Reading: Reeve (2009) Ch 6

INTRODUCTION TO SYMBOLIC INTERACTION: SYMBOLIC INTERACTION, PERSPECTIVES AND REFERENCE GROUPS LECTURE OUTLINE

TTI Personal Talent Skills Inventory Coaching Report

Fear of Success in Males and Females Effects of Developmental Level and Sex-Linked Course of Study ~

32.5. percent of U.S. manufacturers experiencing unfair currency manipulation in the trade practices of other countries.

A Formal Theory of Reflected Appraisals in the Evolution of Power. Noah E. Friedkin. University of California, Santa Barbara.

SITUATIONAL STRUCTURE AND INDIVIDUAL SELF-ESTEEM AS DETERMINANTS OF THREAT-ORIENTED REACTIONS TO POWER. Arthur R» Cohen

KCP learning. factsheet 44: mapp validity. Face and content validity. Concurrent and predictive validity. Cabin staff

Evaluation Models STUDIES OF DIAGNOSTIC EFFICIENCY

Clinical Family Psychology. Spring Semester, 2014 Prof. Dr. Dominik Schoebi

(SAT). d) inhibiting automatized responses.

Motivational Affordances: Fundamental Reasons for ICT Design and Use

An Exploratory Study on Consumer Psychological Contracts

Developmental Perspectives on Problem-Solving

Interpersonal coping styles and couple relationship quality: Similarity versus complementarity hypotheses

24/10/13. Surprisingly little evidence that: sex offenders have enduring empathy deficits empathy interventions result in reduced reoffending.

The Impact of Relative Standards on the Propensity to Disclose. Alessandro Acquisti, Leslie K. John, George Loewenstein WEB APPENDIX

Energizing Behavior at Work: Expectancy Theory Revisited

Collaboration VALUE STATEMENT

Transcription:

Social Networks 12 (1990) 239-252 North-Holland 239 A GUlTMAN SCALE FOR THE STRENGTH OF AN INTERPERSONAL TIE * Noah E. FRIEDKIN * * University of California Social psychological theory on interpersonal attachments suggests that the strength of a relationship develops in stages. This paper proposes a measure of the strength of an interpersonal tie that is consistent with this social psychological theory. Three findings are reported from a study of adult (Ego-Alter) dyads in six communities. (1) Ego s claims of frequent discussion with Alter, seeking help from Alter, and friendship with Alter form a Guttman scale: the claim of friendship implies the claims of help seeking and frequent discussion; the claim of help seeking implies the claim of frequent discussion. (2) Ego s level of attachment to Alter is related to Alter s level of attachment to Ego. (3) The likelihood of perceived consensus and acknowledged interpersonal influence in a dyad increases with the strength of the Ego-Alter tie as measured by the total number of claims made by Ego and Alter about their relationship. 1. Introduction Scale construction ought to be guided by extant theory or else explicitly involve the development of new theory (Coombs 1964: 5). This paper proposes a measure of the strength of an interpersonal tie that is consistent with social psychological theory on interpersonal attachments (Levinger and Snoek 1972; Altman and Taylor 1973; Secord and Backman 1974; Backman 1981). Three ideas in particular have guided the development of the proposed measure. First; a relationship is most accurately described as consisting of two directed attachments (Ego -+ Alter and Alter + Ego) and any global assessment of the strength of the relationship must take into account these separate attachments. It is potentially misleading to seek an * This research is based on data collected with Charles E. Bidwell as part of a project, supported by the Spencer Foundation, on decision making in school district policy groups. * * Direct correspondence to Professor Noah E. Friedkin, Graduate School of Education, University of California, Santa Barbara, CA 93106, U.S.A. 037%8733/90/$3.50 0 1990 - Elsevier Science Publishers B.V. (North-Holland)

240 N.E. Friedkin / A Guttmnn scale of tie strength answer to such questions as Are Ego and Alter friends? when two persons separate subjective orientations need not be identical. Second; an Ego + Alter attachment develops in stages. Most proposals about the development of an attachment are consistent with Backman s (1981) schema. A relationship is initiated when Ego becomes aware of Alter; the awareness may be based on direct interaction or, in the absence of such interaction, on the receipt of information about Alter. From the information Ego receives about Alter, Ego estimates the value of pursuing the relationship. Positive valuation of the relationship usually is associated with efforts by Ego to strengthen the relationship through verbal or written contact. However, the relationship may increase in strength without direct contact; for example, Ego s fantasies about Alter may serve to strengthen the tie. To the extent that Ego finds the outcomes of the relationship rewarding, conditions have been established for more intimate processes of selfdisclosure, trust, commitment, and strong positive sentiment. Again, these processes are typically (but not necessarily) based on direct interaction; for example, Ego may seek help from (or disclose intimate personal details to) a mass-media celebrity with whom Ego has become infatuated. This developmental perspective on attachment processes suggests that it may be misleading to view tie strength as a simple linear combination of variables like interaction frequency, emotional intensity, and mutual confiding (Granovetter 1973; Marsden and Campbell 1984). Such a view misses the idea that qualitatively different interpersonal processes occur at different stages of a relationship s development; for example, if interaction is prior to feelings of intimacy, then these two dimensions of a relationship are not substitutable in the production of tie strength, The occurrence of intimacy implies both interaction and a higher level of tie strength than does any frequency of interaction without intimacy. In short, the appropriate scale for an Ego -+ Alter attachment may be Guttman in form. Third; both the attainment of a particular level of tie strength and the maintenance of the tie at that strength depend upon the balance of rewarding and punishing outcomes (anticipated and actual) of the relationship. The main reward is reciprocity and the main punishment is its absence (Gouldner 1960; Kelly and Thibaut 1978; Kelly 1979). Ego and Alter are interdependent insofar as the development of Ego s attachment to Alter is contingent upon the development of Alter s

N.E. Friedkin / A Guttman scale of tie strength 241 attachment to Ego. Regular face-to-face contact is associated with strong ties because of the positive feedback (however minimal or imagined) such contact allows; a strong attachment is difficult to maintain in a strictly unilateral relationship because such a relationship does not permit those minimal cues (e.g. a gesture or word of encouragement) from which an Ego might draw sustenance and reward. The crucial idea is reciprocity, but this reciprocity may be more or less delayed. Where there is no delay, then Ego and Alter s attachments advance in tandem - so that they jointly move, if they move at all, to higher levels of tie strength. Where there is some delay, Ego and Alter s attachments may each develop similarly (through the same set of stages) but at any given point in time their two states of attachment may not be identical. Although the extant theory on interpersonal attachments allows for a loose coupling of interpersonal sentiments and behaviors, the theory suggests that the strength of Ego s attachment to Alter is unlikely to progress to (or be maintained at) a high level in the absence of some movement in the same direction by Alter. Hence, reciprocation and balance are crucial for both the occurrence and durability of strong relationships. These three ideas - the directedness of attachments, their development in stages, and the importance of reciprocity - each have a place in a measure of tie strength that is constructed as follows: Sli = sji = rij + y/i where rij = d,, + hij +hj d,j = 1 if i claims frequent discussion with j (d,, = 0 otherwise); hij = 1 if i claims seeking help from j (hi, = 0 otherwise); and f;, = 1 if i claims j as a close personal friend ( f, j = 0 otherwise). Hence si j is the total number of claims made by Ego and Alter about their relationship. I will report on the properties and construct validity of this measure. We will see that the claims are patterned in a way that is consistent with social psychological theory on interpersonal attachments and that the scale scores ( sij) are associated with perceived consensus and acknowledged interpersonal influence.

242 N. E. Friedkin / A Guttman scale of tie strength 2. Data The paper draws on data collected during case studies of decision-making processes in the policy groups of six school districts. The policy groups of these school districts consisted of persons who had some noteworthy influence on school board decisions during the several years preceding the study. The policy group in each school district was delineated by a snowball procedure which combined positional, reputational, and behavioral selection criteria. The delineation of the group started with a list of influentials that included each school board member, the district superintendent, and all other persons identified in local newspapers or school board minutes as currently or recently active in board meetings, district elections, or other efforts to affect board decisions. These individuals nominated persons on the district staff and in the community who were known or reputed to be currently or recently active in attempts to influence school board decisions. These nominees were asked for further nominations according to the same two criteria. To guard against idiosyncratic nominations, only persons who were mentioned at least twice were added to the list of policy group members. The nominating procedure continued until dual nominations no longer occurred. Each policy group member was then interviewed about the decisionmaking processes in their school district and was asked to fill out a questionnaire. The questionnaire contained, among other items, a list of the policy group members in the school district. Respondents were asked a set of questions concerning their interpersonal relationships with other policy group members to which they responded by checking off names, in separate columns, of those persons with whom they claimed the particular relationship existed. Seven of these network items enter into the present analysis: Frequent discussion. Check off any of the persons with whom you frequently discuss matters having to do with the... Schools. Help seeking. When you need information or advice about problems or issues having to do with the... Schools, to which persons on the list do you turn? Close friendship. Check off any persons who are your close personal friends.

N.E. Friedkin / A Guttman scale of tie strength 243 Perceived consensus. When there are differences of opinions about school district matters, which of the persons on the entire list are usually on the same side of these issues as yourself? Which persons are usually on a different side of these issues from yourself. From these two items, a measure of perceived consensus ( pij = p,,) was constructed as follows: pij = 1 if either Ego or Alter reported agreement and neither reported disagreement; pij = - 1 if either Ego or Alter reported disagreement and neither reported agreement; and pij = 0 otherwise. Acknowledged injluence. Check off the names of any person on the list who you would say probably had some influence on your own opinions on school related matters during the last year or so? From the data derived from this item, a variable measuring the occurrence of acknowledged influence ( aij = aji) was constructed: aij = 0 if neither Ego or Alter acknowledged influence; aij = 1 if Ego or Alter but not both acknowledged influence; and aij = 2 if both Ego and Alter acknowledged influence. The effective response rates are high, ranging from 88% to 98% among the groups. The analysis is restricted to respondent dyads because the measure of tie strength is based on the claims of both members of a dyad about their relationship. Among the n,, respondents of a group, there are n,(n, - 1)/2 dyads available for analysis. Table 1 shows the number of policy group members in each school district, the number of respondents to the network items in the questionnaire, and the corresponding number of respondent dyads. Table 1 Sites of the dyadic relationships Sites Group size Number of respondents A 42 38 703 B 42 37 666 C 60 59 1711 D 61 54 1431 E 59 57 1596 F 67 66 2145 Totals 331 311 8252 Respondent dyads

244 N.E. Friedkin / A Guttman scale of tie strength 3. Findings 3.1. Ego s attachment to Alter Ego may claim having frequent discussions with Alter, Ego may claim seeking help from Alter, and Ego may claim Alter as a close personal friend. These claims form a Guttman scale. Table 2(a) describes the distribution of claim patterns and Table 2(b) reports the coefficients of reproducibility, minimum marginal reproducibility, and scalability for the pattern: r f h d 3 1 1 1 2 0 1 1 1 0 0 1 where f represents the claim of close friendship, h the claim of help seeking, d the claim of frequent discussion, and r = f + h + d. The high Table 2 Claims of friendship (f), help seeking (h), and frequent discussion (d) form a Guttman scale (a) Distribution of claim patterns Patterns f hd Sites 111 20 110 1 101 15 100 10 011 64 010 21 001 104 000 1171 A B C D E F 28 70 99 73 126 0 6 1 2 11 4 47 38 16 73 1 32 25 2 42 127 205 216 331 300 19 100 44 48 116 152 293 246 286 353 1001 2669 2193 2434 3269 (b) Goodness of fit Coefficients Coefficient of reproducibility Minimum marginal reproducibility Coefficient of scalability Sites A 0.98 0.92 0.73 B C D E F 0.99 0.96 0.97 0.99 0.96 0.88 0.89 0.85 0.87 0.87 0.91 0.68 0.83 0.89 0.71

N. E. Friedkin / A Guttman scale of tie strength 245 values of these coefficients indicate that r = 1 tends to correspond with a claim of frequent discussion and that r = 2 tends to correspond with claims of frequent discussion and help seeking; naturally, r = 3 implies all three claims. This claim pattern fits the present data better than any of the eight other possible combinations of claim patterns that form a scale of zero to three. While it minimizes the number of prediction errors in each of the school district sites, it also succeeds in describing approximately 75% or more of the Ego + Alter attachments. 3.2. Ego and Alter s joint attachment Table 3(a) presents the ways in which Ego and Alter s attachments may combine under the constraint that their attachments conform perfectly to the Guttman pattern. In addition, Table 3(a) contains simplified descriptions of the combined claims that Ego and Alter may make. For example, d, indicates unreciprocated claims of discussion, d, indicates reciprocated claims of discussion, and d,(hf). indicates reciprocated claims of discussion plus unreciprocated claims of help seeking and friendship. Note that the Guttman properties of the subscales constrain the interpretation of these descriptors; for example, it should be understood that d,(hf). implies that one member of a dyad is claiming friendship, help seeking, and discussion while the other is claiming only discussion. Now suppose that if a relationship increases in strength it does so in elementary steps, that is, with one person changing their level of attachment at a time. There are just five alternative models of the development of such a relationship that preserve the Guttman characteristics of the two attachments; these five process models are described in Table 3(b). Inspection of the frequency with which different patterns of claims occur indicates that the Guttman errors are not evenly distributed. The relatively large contribution to these errors, in each group, of the (010) pattern (i.e. help seeking without friendship or frequent discussion) stems most likely from the instrumental character of the groups activities. In such groups, members will occasionally draw on the specialized knowledge of others without, at the same time, engaging in frequent general discussions of group affairs. At the same time, the strong association of help seeking with frequent discussion suggests that even in these instrumentally oriented groups help seeking is contingent upon the prior generation trust and credibility.

246 N.E. Friedkin / A Guttman scale of tie strength Table 3 Formal analysis of interpersonal claims under the Guttman constraint (a) Ways of attaining Alter s score 11 0 1 2 3 the total claim score s,, = r,, + r,, Ego s score 0 s,, = 0 S,] = 1 d, s,, = 2 old), s,, = 3 (fhd) u 1 2 3 S,) =1 s,, = 2 s,, = 3 du W, WV u S,, = 2 S,, = 3 s,, = 4 d, ha, (fh)a S,, = 3 S,] = 4 SC,5 ha (hd), f,(hd), S,, = 4 s,, = 5 s,, = 5 @)A, f,old), (fhd), (b) Alternative process models Models Total claim score s,, 1 2 3 4 5 6 Note that reciprocation is the most immediate in Model 1 and the most delayed in Model 5. Also note that there are at most two conditions of a relationship for any given total number of claims (s, j). Total claim scores of one correspond with d,, scores of two correspond with either d, or (hd),, scores of three correspond with either (fhd), or h,d,, scores of four correspond with (fh)d,., or (hd)., scores of five correspond with f,(hd),, and scores of six correspond with (fhd).. It is, of course, the Guttman constraints in combination with the one-stepat-time process assumption that have radically reduced the number of possible interpretations of the total claim score (si,). The data in Table 4 show that, among dyads with identical total claim scores, the proportion of dyads that conform to the predictions of Models l-5 is high in each of the six school district sites. Leaving aside those dyads with total claim scores of zero and six (for which no errors of prediction are possible), the joint claim pattern of approximately 69% or more of the dyads is correctly predicted at each of the sites.

N. E. Friedkin / A Guttman scale of tie strength 241 Table 4 Distribution of dyads among claim patterns Total Dyads Sites claim claim A B C D E F score pattern 0 1 du * 2 &, u * 3 VW u Up, * 4 (fhm, WV, * 5 f,w), * 6 UW, Goodness of Fit ** *** 529 429 1163 918 1084 1385 53 12 143 164 118 180 19 13 92 39 37 119 14 23 38 12 42 29 34 56 92 118 107 137 11 2 19 14 7 55 3 2 10 21 2 2-l 14 26 46 38 69 66 4 6 26 18 9 25 2 5 11 11 8 21 5 17 18 20 59 29 5 1 22 15 9 16 3 6 9 10 23 23 3 1 9 14 5 15 4 I 13 13 17 18 0.94 0.96 0.90 0.93 0.96 0.89 0.75 0.90 0.69 0.80 0.86 0.69 * Response patterns that do not conform to any of the models (l-5) described in Table 3(b). * * Proportion of dyads consistent with the models described in Table 3(b). *** Proportion of dyads consistent with models in Table 3(b) among dyads with total claim scores in the range l-5. Moderate positive associations of the separate claim scores rij and 5, (in light of the Guttman character of each) indicate a tendency toward reciprocation. The values of Goodman and Kruskal s gamma and Pearson s product moment correlation are, respectively, 0.765 and 0.474 in Site A, 0.703 and 0.483 in Site B, 0.727 and 0.488 in Site C, 0.595 and 0.380 in Site D, 0.817 and 0.605 in Site E, and 0.647 and 0.425 in Site F. Along the same lines, from the data in Table 4, it appears that total claim scores of two more often arise from (hd), than d,, that scale scores of three more often arise from hdp, than (fhd)., and that scale scores of four more often arise from (hd), than (fh)~i~; indeed, the odds favoring the one alternative over the other generally are better than 2 : 1. Consequently, these data appear to support Model 2 (in which reciprocation is moderately delayed) as the best single

248 N. E. Friedkin / A Gutiman scale of tie strength description of the joint claim patterns of Ego and Alter. * We should allow for alternative ways of achieving given levels of attachment in a relationship. However, there appear to be grounds for rejecting Model 5 (where reciprocation is the most delayed) given the heavily favored odds of hgp, over (fhd).. Table 5 Tie strength and acknowledged interpersonal influence (a) Zero-order relationships Site A Site B Site C Site D Site E Site F * 0.885 0.781 0.819 ** 0.676 0.622 0.626 0.850 0.825 0.786 0.691 0.652 0.602 (b) Probability of no acknowledged influence s.. Site A Site B Site C Site D Site E Site F 0 0.856 0.839 0.866 0.930 0.808 0.826 1 0.347 0.518 0.494 0.719 0.355 0.388 2 0.085 0.333 0.255 0.285 0.231 0.231 3 0.238 0.235 0.110 0.277 0.125 0.186 4 O.OfMl 0.000 0.020 0.087 0.039 0.091 5 O.ooO 0.000 0.111 0.083 0.036 0.053 6 0.000 0.000 0.077 0.000 O.OCO 0.000 (c) Probability of reciprocal acknowledgment of influence s. Site A Site B Site C Site D Site E Site F.I 0 0.011 0.023 0.015 1 0.083 0.094 0.094 2 0.424 0.185 0.195 3 0.571 0.265 0.402 4 0.917 0.522 0.647 5 0.500 0.571 0.444 6 0.750 0.714 0.461 * Goodman and Bruskaf s g-a. * * Pearson s product moment correlation. 0.065 0.021 0.024 0.034 0.161 0.124 0.111 0.301 0.258 0.325 0.450 0.458 0.565 0.632 0.439 0.500 0.821 0.632 0.846 0.706 0.778 The lack of reciprocity in claims of frequent discussion may simply illustrate the unreliability of individuals reports: after all, persons either engage in frequent discussions or they do not; and if they do, then both parties to such discussions should acknowledge the occurrence. Such a view is too simple, because it ignores the possibility of differences in the subjective significance of relationships for the parties. Frequent discussions may be differentially salient to the parties involved in them; when such differential salience is marked a lack of reciprocity may be expected and should not on theoretical grounds be classified as measurement error.

N. E. Friedkin / A Guttmnn scale of tie strength 249 3.3. Construct validity We should find a positive association between tie strength and interpersonal agreement. Newcomb (1953) has suggested that this correspondence should be particularly strong for perceived agreement (also see Scheff 1967). Along the same lines, we should find an association for acknowledged influence. Our evaluation of the construct validity of the scale focuses on these two subjective measures of social solidarity. Acknowledged interpersonal influence Table 5 reports the values of Pearson s product moment correlation and Goodman and Kruskal s gamma for the association between tie strength ( sij) and acknowledged interpersonal influence ( aij). The magnitudes of the association are Table 6 Tie strength and perceived consensus (a) Zero-order relationships * ** Site A 0.557 0.317 Site B 0.630 0.402 (b) Probability of perceived agreement Site C 0.599 0.394 Site D 0.621 0.414 Site E 0.742 0.445 Site F 0.656 0.428 si, Site A Site B Site C Site D Site E Site F 0 0.287 0.205 0.194 0.218 0.229 0.189 1 0.542 0.588 0.506 0.473 0.671 0.512 2 0.746 0.605 0.664 0.722 0.769 0.715 3 0.809 0.794 0.878 0.795 0.850 0.780 4 1 0.913 0.922 0.913 0.934 0.894 5 1 1 0.889 0.917 1 0.947 6 1 1 0.846 1 1 0.944 (c) Probability of perceived disagreement S j Site A Site B Site C Site D Site E Site F 0 0.113 0.161 0.142 0.178 0.208 0.154 1 0.153 0.082 0.170 0.177 0.129 0.127 2 0.102 0.123 0.107 0.049 0.058 0.077 3 0.048 0.029 0.049 0.060 0.062 0.042 4 0 0.043 0 0 0.026 0 5 0 0 0 0 0 0 6 0 0 0 0 0 0 * Goodman and Kruskal s gamma. * * Pearson s product moment correlation.

250 N. E. Friedkin / A Guttman scale of tie strength moderately high and remarkably uniform across the six school district sites. It is useful to examine the proportions of dyads that report no influence ( aij = 0) at various levels of tie strength. These data also are shown in Table 5. The probability of no acknowledged influence declines rapidly from 0.80 or-more to zero; that is, it covers virtually the entire range of possible probability values as a function of tie strength. Perceived consensus Table 6 reports the values of Pearson s product moment correlation and Goodman and Kruskal s gamma for the association between tie strength ( sij) and perceived consensus ( pii). A consistently moderate level of association appears in these data. Table 6 also shows the proportion of dyads that report disagreement ( pij = - 1) and agreement ( pij = 1) at various levels of tie strength. It appears that the perception of disagreement is systematically related to tie strength and is virtually non-existent among dyads at tie strength scores of four and higher. With respect to the probability of agreement, the relationship is similarly pronounced. 4. Summary The three main findings of this study are these. (1) Ego s claims of frequent discussion with Alter, seeking help from Alter, and friendship with Alter form a Guttman scale: the claim of friendship implies the claims of help seeking and frequent discussion; the claim of help seeking implies the claim of frequent discussion. (2) Ego s level of attachment to Alter is related to Alter s level of attachment to Ego. (3) The likelihood of perceived consensus and acknowledged interpersonal influence in a dyad increases with the strength of the Ego-Alter tie as measured by the total number of claims made by Ego and Alter about their relationship. 5. Discussion A definition of interpersonal relationships in terms of two directed attachments (Ego --j Alter and Alter + Ego) is consistent with a viewpoint on dyads as elementary social systems in which co-orientations

N.E. Friedkin / A Guttman scale of tie strength 251 emerge from an interplay of members actions and reactions. In these elementary social systems the rule of reciprocity helps explain why the movement of one actor s attachment in a certain direction (strengthening or weakening) is often followed by a movement of the other actor in the same direction. To be sure, the rule of reciprocity does not explain Ego s initial movement and, thus, does not provide a complete account of why relationships become strong or dissolve. Nonetheless, reciprocity is exceedingly important because it allows us to see why the separate actions of individuals eventuate in similar co-orientations. Inquiries that seek to build on the present work might probe the scope of the rule of reciprocity. For example, it is plausible that the rule of reciprocity is weaker in the superior-subordinate relationships of formal organizations than in peer relationships. Organizational superiors usually cannot reciprocate the affections and disclosures of their subordinates without some risk to their hierarchical status; the question is how subordinates respond to such imbalance. If the rule of reciprocity extends to these relationships, then superior-subordinate relationships should be unstable with respect to a long-run capacity for regular and effective unidirectional influence; such relationships are likely either to be reduced to minimal strength or to transform into strong solidary ties. Future inquiries might also pursue the idea that social conditions affect the form of the developmental process in dyadic relationships. It is not difficult to imagine a population in which interpersonal relationships tend to proceed from help seeking to frequent discussion, and finally to friendship. For example, members of highly technical organizations may interact mainly in order to solve problems; in turn, these problem-solving relationships may develop into relationships of broader scope and intimacy. If we take the possibility of alternative forms of development seriously, then a line of inquiry is opened that would seek to understand why different developmental forms occur in different contexts and the implications of such variation. Structuralists need to be concerned with tie strength to the extent that the effects of a social network importantly depend on the values of the lines that comprise the network. A cautious viewpoint treats a social network as a structure of opportunities for flows of information and influence. However, if we are to advance our understanding of social network effects, we must begin to replace conclusions about the opportunities of network flows with conclusions about the probabilities

252 N.E. Friedkin / A Guttman scale of tie strength of such flows. Good measures of tie strength are likely to be an important foundation for such theoretical refinement. References Altman, I. and D.A. Taylor 1973 Social Penetration: Development of Interpersonal Relationships. New York: Holt, Rinehart and Winston. Backman, C.W. 1981 Attraction in interpersonal relationships. pp. 235-268 in M. Rosenberg and R.H. Turner (eds), Social Psychology: Sociological Perspectives. New York: Basic Books. Coombs, C.H. 1964 A Theory of Data. New York: Wiley. Gouldner, A.W. 1960 The norm of reciprocity: A preliminary statement. American Sociological Review 25: 161-178. Granovetter, MS. 1973 The strength of weak ties. American Journal of Sociology 78: 1360-1380. Kelly, H.H. 1979 Personal Relationships: Their Structure and Process. Hillsdale, NJ: Erlbaum. Kelly, H.H. and J.W. Thibaut 1978 Interpersonal Relations: A Theory Interdependence. New York: Wiley Interscience. Levinger, G. and J.D. Snoek 1972 Attraction Relationship: A New Look Interpersonal Attraction, Morristown, NJ: General Learning Press. Marsden, P.V. and K.E. Campbell 1984 Measuring tie strength. Social Forces 63: 482-501. Newcomb, T.M. 1953 An approach to the study of communicative acts. Psychological Review 60: 393-404. Secord, P.F. and C.W. Backman 1974 Social Psychology, 2nd edn. New York: McGraw-Hill. Scheff, T.J. 1967 Toward a sociological model of consensus. American Sociological Review 32: 32-46.