VITA 

Name: Donald Albert Cross 

Place of Birth: West Tamworth, New South Wales, Australia 

. Date of Birth: June 19, 1930 

Schools Attended and Academic Awards: 

Armidale Teachers' College, Armidale, N. S. W., 1948-1949; 
Teachers' Certificate. 

Sydney Teachers' College, Sydney, N. S. W., 1953-19~5; 
Diploma of Physical Education. 

University of Oregon, Eugene, Oregon, 1958-1959; Bachelor 
of Science. 

University of Oregon, Eugene, Oregon, 1959-1960; Master 
of Science. 

University of British Columbia, Vancouver, B. C., Canada, 
1961-1962. 

University of Oregon, Eugene, Orego,n, 1963-1966; Doctor 
of Education. 

Areas of Special Interest: 

Primary Education; Educational Psychology; Education of 
Exceptional Children; Physical Education 

Relevant Professional Experiences: 

New South Wales Department of Education: 
Teacher-in-Charge, Belbora Public School, 1950-1953. 
Parramatta West Public School, 1954-1955. 
Broken Bay National Fitness Camp, 1956-195,7. 

University of Oregon: 
Teaching Assistant, Physical Education, 19 60. 



VITA- -contin ued 

Vancouver School Board, Va _couve r , B . C. : 
El e mentary Sch ool T e ache r of class for s low learne rs, 

1960-1962. 
University of Ore gon : 

Instru ctor in Edu cation, 19 62 -1 963. 
Graduate As s i stant - s u p e r v i s o r of stu d ent teaching, 

1964-1965. 

ii 



AN EXPERIMENTAL STUDY OF THE EFFECTS OF 

NEGATIVE SOCIOMETRIC CHOICES ON INTER­

PERSONAL RELATIONSHIPS IN 

GRADE FIVE STUDENTS 

by 

DONALD ALBERT CROSS 

A THESIS 

Presented to the School of Education 
and the Graduate School of the University of Oregon 

in partial fulfillment 
of the requirements for the degree o f 

Doctor of Education 

March 1966 



APPROVED: 



ACKNOWLEDGEMENTS 

The writer wishes to express his indeb tedness to the thesis 

advisor, Dr. John deJung , for his encou ragement and helpful 

criticism both before and during the dev elopme nt of this thesis, and 

to the other members of the committee : Drs. W. A. Bricker, 

J. M. Jackson, L. L. Lovell, R. H. Mattson and P. 0. Sigerseth 

for their helpful assistance. 

Also acknowledged is the assistance and cooperation of the 

administrators , principals, teachers and students of Springfield 

School Distr i ct, Or egon, who made this study possible. 

Appreciation is also exte nded to m y w ife, Gai, fo r h e r unti ring 

assistanc e and encourageme nt. 

D. A. C. 



' / 

TABLE OF CONTENTS 

ACKNOWLEDGEMENTS 

LIST OF TABLES ••• 

Chapter 
I 

II 

III 

IV 

V 

INTRODUCTION TO THE PROBLEM 

Statement of the Problem 
Method of Investigation: Overview 

REVIEW OF THE LITERATURE •• 

Sociometry: General Definitions 
Measurement of Sociometric Status 
Rejections and Negative Choices 

PROCEDURES 

Major Investigation 
Sample 
Sociometric Treatment Procedures 
Choic e Expectancy Procedure 
Follow-up Questionnaire 
Interview 

RESULTS OF THE STUD Y 

SUMMARY, DISCUSSION AND CONCLUSI ONS •• 

Limitations 

Page 

V 

viii 

l 

12 

52 

71 

115 



TABLE OF CONTENTS--conti.nued 

Chapter 

BIBLIOGRAPHY 

APPENDIX 
A PAIRED-COMPARISON QUESTIONNAIRE 

B SOCIOMETRIC PROCEDURES. • • • 

C CHOICE EXPECTANCY PROCEDURE • 

D EIGHT-ITEM QUESTIONNAIRE • 

E 

F 

INTERVIEWS • • • • • • • • . 

SCORES RECEIVED ON THE PAIRED­
COMPARISON, CHOICE EXPECTANCY AND 
SOCIOMETRIC PROCEDURES ••••••• 

vii 

Page 

126 

141 

152 

160 

167 

17 3 

221 



LIST OF TABLES 

Table Page 

1. Distribution of Subjects Providing Paired-Comparison 
Data: by Treatment Group, Class and Sex • . 72 

2. Distribution of Least-Preferred Subjects Providing 
Paired-Comparison Data: by Treatment Group, 
ClassandSex................ 73 

3. Differences in Mean Proportions of Peer Preferences 
Received by Least-Preferred Boys and Girls, Com­
bined and Separately, on Pre- and Post-Test 
Administrations of the Paired-Comparison Question-
naire • • • . . . . . . . . . . . . ' . . . . 

4. Summary of Analysis of Variance of Differences in Mean 
Proportions of Peer Preferences Received by Least- . 
Preferred Boys and Girls on the Pre- and Po st-Te st 
Administrations of the Paired-Comparison Question­
naire • • . 

5. Summary of Analysis of Variance of Differences in Mean 
Proportions of Peer Preferences Received by Least­
Preferred Boys on the Pre- and Post-Test Adminis-

77 

78 

trations of the Paired- Comparison Questionnaire • • 7 8 

6. Summary of Analysis of Variance of Differences in Mean 
Proportions of Peer Preferences Received by Least­
Preferred Girls on the Pre- and Post-Test Adminis-
trations of the Paired- Comparison Questionnaire • • 7 8 

7. Differences in Changes in Variance of Proportions of 
Peer Preferences Received by Boys and Girls on the 
Pre- and Post-Test Administrations of the Paired­
Comparison Questionnaire 81 



LIST OF TABLES--continued 

Table 

8. Summary of Analysis of Variance of Changes in Variance 
of Proportions of Peer Preferences Received by Boys 
and Girls on the Pre- and Post-Test Administratiohs 
of the Paired-Comparison Questionnaire • 

9. Summary of Analysis of Variance of Changes in Variance 
of Proportions of Peer Preferences Received by Boys 
on the Pre- and Post-Test Administrations of the 
Paired-Comparison Questionnaire • • • • • 

10. Summary of Analysis of Variance of Changes in Variance 
of Proportions of Peer Preferences Received by Girls 
on the Pre- and Po st-Te st Administrations of the 
Paired-Comparison Questionnaire •• 

11. Distribution of Least-Preferred Subjects Providing 
Choice Expectancy Data: by Treatment Group, 

Page 

82 

82 

82 

Clas s and Sex • • • • • • • . • • • • • 8 5 

12. Mean Proportion Choice Expectancies of Least-Preferred 
Boys and Girls, Combined and Separately, by Group 
and Cl as s • • • • • . • • • • • • • • • • • 8 6 

13. Summary of Analysis of Variance of Mean Proportion 
Choice Expectancies of Least-Pr eferred Boys and 
Gir 1 s • • . • . . • • • . . . . . . . . . . , . 87 

14. Summary of Analysis of Variance of Mean Proportion 
Choice Expectancies of Least-Preferred Boys 87 

15. Summary of Analysis of Variance of Mean Proportion 
Choice Expectancies of Least-Preferred Girls • • • 87 

16. Distribution of Subjects Providing Choice Expectancy 
Data by Treatment Group, Class and Sex • • • 89 

17. Mean Proportion Choice Expectancies of Boys and Girls 
Combined and Separately, by Group and Class 90 

ix 



LIST OF TABLES- -continued 

Table 

18. Summary of Analysis of Variance of Inverse Sine 
Transformations of Mean Proportion Choice 

Page 

Expectancies of Boys and Girls • • • . • • • • • 91 

19. Summary of Analysis of Variance of Inverse Sine 
Transformations of Mean Proportion Choice 
Expectancies of Boys • • • • • • • . • • • . • 91 

20. Summary of Analysis of Variance of Inverse Sine 
Transformations of Mean Proportion Choice 
ExpectanciesofGirls.......... 91 

21. Distribution of Subjects Providing Data Relating to 
Questionnaire Items 1, 2 and 3, by Treatment 
Group and Sex . • . . . . . . . . . . . . 

22. Frequency Distribution of Questionnaire Respondents 
Asking Classmates About Their Sociometric Choices 
or Paired-Comparison Preferences, by Treatment 

93 

Group and Sex • • • • • • • • • • • • • • • • 9 5 

23. Frequency Distribution of Questionnaire Respondents 
Asked by Classmates About Their Sociometric 
Choices or Paired.-Comparison Preference s, by 
Treatment Group and Sex • • • • . • 97 

24. Frequency Distribution of Questionnaire Respondents 
Stating That They Heard Classmates Discussing 
Sociometric Choices or Paired-Comparison Pref-
erences, by Treatment Group and Sex. • • • • • 98 

25. Summary of Chi-Square Results of the Frequency 
Distribution of Responses to Questionnaire Items 
1, 2 and 3, by Boys and Girls, Combined and 
Separately • • • • • • • • • • • • • • • • 100 

X 



LIST OF TABLES--continued 

Table 

26. Distribution of Subjects Providing Data Relating to 
Questionnaire Items, 4, 5, 6 and 7, by Treatment 

Page 

Group and Sex • • • . • • . • • • . • • • • • 102 

27. Frequency Dis tribution of Responses to Questionnaire 
Item 4, by Treatment Group and Sex . . . . . . 103 

28. Frequency Distribution of Responses to Questionnaire 
Item 5, by Treatment Group and Sex . . . . . . 103 

29. Frequency Distribution of Responses to Questionnaire 
Ite;rn 6, by Treatment Group and Sex . • . . . . 104 

30. Frequency Distribution of Responses to Questionnaire 
Item 7, by Treatment Group and Sex . . . . . . 104 

31. Distribution of the Percentages of Ratings of Same-
Sex Classmates, by Boys and Girls • • • • • • 106 

32. Distribution of Subjects Providing Interview Data by 
Group and Sex • • • • • • • • • • • • • • • • 107 

Xl 



CHAPTER I 

INTRODUCTION TO THE PROBLEM 

Educators have _attempted to keep pace with the societal trend 

toward more democratic interpersonal relationships during the past 

generation, and modern educational literature is replete with 

admonitions that teachers acknowledge and accept the responsibility 

of maintaining a classroom climate which facilitates positive social 

and emotional growth. In Laycock' s terms, the school "has no 

choice but to teach pupils how to live, work, and play together i n 

school so that both the needs of the indiv idual and of the group may 

be met" ( 1960, p. 5). How well the miniature society fulfills this 

obligation will be reflected in the cooperative and creative manner 

in which today's children, as adults, work together to achieve 

common purposes in a democratic society. 

Central to the theme of wholesome personal adjustment is a 

description of developmental tasks by Hav ighur st. He proposed 

that the process of learning to get along with age-mates is really 

the process of learning a "social personality" or acquiring social 

stimulus value ( 1953, p. 30)~ In recognizing man to be primarily a 

social being, Dinkmeyer and Dreikur s further developed this concept, 



2 

indicating their belief that the cha racteristics which make man 

distinctly human are "a result of his social interaction with his fellow 

man, 11 as it is "only within the group that he can function and fulfill 

himself" ( 1963, p. 8). 

In the preschool period adults set the standards with which 

youngsters try to cope. During this period the latter develop feelings 

about themselves and about their relationships with _other per sons. 

Security is provided through the child's attachment to significant 

adults who interpret the actions of age-mates to him, as he knows 

and understands little of their expectations. 

By the time he is of school age the child has often developed 

socially to the extent that he can now join in some minor cooperative 

play which calls for loose organization and in which ente rprises 

and membership vary. He is still ve r y much the individualist, as 

is typically indicated by the games he selects, but he finds certain 

requirements being established as prerequisites for participation. 

It is apparent to the youngster, even at this stage, that mere physical 

presence is insufficient criterion for inclusion in activities. 

What of the youngster who has not had an opportunity to enjoy 

the company of other children in the preschool period? Due to lack 

of experience in initiating social contacts with age-mates he may 

encounter early loneliness in the school situation unless the teacher 



3 

assumes her vital role of helping him establish rapport with class­

mates. These first school relationships are of paramount importance 

in building a sound foundation for later experiences. 

Generally, the trauma of exclusion does not begin to manifest 

itself until the group commences to set its own standards. There 

is, however, a concomitant possibility that self-appraisal, subsequent 

to exclusion, may lead to doubts of personal worth and possible with­

drawal from any situation in which the individual or his work may be 

appraised. Also, a reputation of undesirability may follow such an 

individual, further limiting his opportunities to develop profitable 

relationships. That this situation is unacceptable and may fore­

shadow future problems is apparent to Thompson who wrote: "In 

the American culture .•. Social maturity in children is sought at 

the expense of almost every other aspect of psychological growth. 

Social acceptability is the payoff in our society .... 11 
( 19 62, p. 460). 

From early in the intermediate grades the social acceptability of the 

individual to his peer group assumes increasing importance. 

If children demonstrated the same alacrity in setting about 

their academic tasks as they do in seeking ways to attain and maintain 

peer group status, teachers would have few motivational problems. 

According to Havighurst (1953), the chief concern of the child, 

whether the teacher pays any attention to it or not, is with the task 



4 

of learning how to get along with age-mates. Prescott ag reed, in 

stating that "for certain individuals, the task of winning belonging m 

this peer group or achieving certain desired roles and status may be 

the most compelling interest of the school year" (1957, p. 277). 

Later, he proposed that "peer group status and roles are of immediate 

concern to the child, in contrast to the mo re remote interest that 

much of the traditional subject matter has for him" (Ibid., p. 374). 

Such concentration is frequently to the detriment of achievement in 

subject matter areas, for energy expended on d evelopment of satis­

factory relationships precludes the availability of this energy for 

meeting academic requirements in the school situation. 

Despite their efforts for a cceptanc e and positive recognition, 

some children , for a variety of reasons collectively d e scribed as 

"social immaturity," find themselves forced into rol es not 

synonymous with social acceptance. These rejects, isolate s, or 

neglectees experience little opportunity to satisfy either physical 

or social needs in an acceptable manner . They do not develop a 

feeling of security in communi cating with others in the group. This 

is reflected in less than optimal group comn;iunication and in subse­

quent diminution of the effi c iency with which the classroom group, 

as a unit, can function. 



Awar e n e ss of the i n t erpersonal r el ation ships and the ir mani­

fes t ati ons , a nd of the uni4.u e str uc LU r e o f e ach group wi th whi ch h e 

i n teracts may assist the t e acher in e mpa thi c guidance of students 

e xperie ncing poor i nte rpe rson al rel a t i o n s hips. Succe ssful guidanc e 

will not only a i d the i r adjustme nt but may well influence the i r 

achievement level and l e arning potenti al. This point of v i ew is 

supporte d by many write rs who t e stify to the fact that the learning 

and adjustment of individual pupils and the ir position in the group 

structure are inextricably interwoven. Having obse rved this cyclic 

relationship in action, Jennings ( 1948) reflected that 11when the 

emotional shocks due to inadequate or discordant group life are 

r e moved and advantage is taken of the exi sting psychological 

affinities, there usu ally r esul ts a h eig h teni n g a nd rele a s e of 

children's intell e ctual abil i t ie s a long wi th a red irecti o n of the ir 

thinking processe s. The s e outcomes a re related not only to what 

happens to individual p e rsonalities but also to the plac e o f grou p o r 

social moti vati on o n p e rformanc e " (p. 550 • 

Acknowledgemen t of r e sponsib ili ty a nd expre ssi o n of goals 

for education for social competence and soci al acc e ptanc e - -to b e 

developed by learning social skills a nd through experience i n social 

i nte raction - - bring the obligati o n of d evi sing me an s to prov i de 

experiences which fac ilitate such interaction. Many factors promote 

5 



or obstruct acceptance by others, so the educator has the 

responsibility first, of understanding and subscribing to the 

philosophy of educating for social growth; second, of being aware 

of the interpersonal relations and structure of the group; third, of 

attempting to facilitate positive deve lopment in social interaction, 

with due consideration to the various developmental levels existing 

6 

in the group; and fourth, of evaluating progress toward the expressed 

goals and specific objectives. 

Many teachers express confidence 1n their ability to accurately 

describe the social relationships which exist in the groups that 

confront them daily. However, this confidence may have a weak 

foundation, since comparisons of teacher judgment and pupil choice 

patterns indicate a "general inaccuracy of some teachers and 

specifi c inaccuracie s by most teachers in judging individual pupils" 

(Gronlund, 1959, p. 11). Gronlund's review of related studies 

revealed an average accuracy score of approximately • 60 among 

teachers estimating their pupils' social status. Among elementary 

school teachers Moreno ( 1953 ) found individuals teaching first 

grade students to provide the most accurate estimations of socio­

metric status. The degree of accuracy progressively declined as 

estimations by teachers of higher grades were observed. Moreno 

attributed this declin.e to the social cl eavage developing between 



7 

youngsters and adults and to the increasing complexity of groups. 

These findings highlight the desi rability of obtaining an objective 

evaluation of the classroom social structure. A more clearly defined 

awareness of the social structure in the classroom should result 

from an understanding and application of sociometric techniques. 

In general, these techniques are designed to furnish objective 

information on the actual or desired relationships between group 

members and provide a basis for a graphic description of the group 

structure. 

The development of sociometry and its measurement devices- -

sociometric te_chniques - -ai:eintimately linked with Jacob L. Moreno 

and the 19 34 publication of his basic work, Who Shall Survive? 

This volume presented the first report of the use of sociometric 

techniques in the classroom. In employing these techniques the 

teacher will realize that, while both positive and negative preferences 

may be obtained, the consensus in the literature pertaining to the 

school context is that requiring negative nominations is a potentially 

harmful practice from the point of view of subsequ ent pupil behaviors 

and awareness. Specifically, it is contended that the introduction of 

negative nomination procedures in the form of forced identification 

of least-preferred or non-preferred classmates may result in the 

creation of resentment and comment among the group, in more 



severe social maladjustment among the less favored members of 

the grot.ip, and in accentuation of any negative feelings which are 

a l ready part of the group social climate. These proposals, how­

ever, appear to be based more upon assumption than upon data. 

This reviewer of the sociometric literature finds inadequate 

evidence supporting the 11 harmful effects 11 thesis. The present 

study was undertaken to further examine these ''effects. " 

Statement of the Problem 

This study was d e signed to examine the differential effe cts of 

requiring, permitting and not permitting negative sociometric 

nominations with respect to accentuation of negative feelings among 

class members. 

Five sets of hypothe ses , regarding differences between the 

four treatment conditions, were tested for boys and girls, combined 

and separately. These hypotheses were r elated to: 

1. Changes in proportion s of same- s ex peer preferences 

received by least-preferred classmates on the pre- and 

post-test administrations of the paired-comparison 

questionnaire. 

2. Chang es in variance of proportions of same- sex peer 

preferences received by classmate s on the pre- and 

8 



post-test administrations of the paired-comparison 

questionnaire. 

3. Proportions of same-sex peer preferences expected on 

the post-test administration of the paired-comparison 

que stionnaire by l east-pr efe rr e d students and by all 

students. 

9 

4. Frequency of affirmative responses to questionnaire items 

concerning student discussion of sociometric nominations 

following administration of the different sociometric 

procedures. 

5. Frequency of affirmative responses to questionnaire i terns 

concerning perceived changes in interpersonal relation­

ships following administration of the d ifferent sociome tric 

procedures . 

Method of Investigation : Overview 

Twelve classes of fifth grade students were assigned to one of 

four experimental treatment conditions, A, B, C , or D. All classes 

received two administrations, two weeks apart, of a paired-

comparison questionnaire for selection of same- sex classmate s as 

continuing friends. In the intervening period Group A received a 

sociometric questionnaire requiring student selection of three 



10 

same- sex classmates most preferred as continuing friend s ; Group B, 

a sociometric questionnaire requiring student selection of three 

most-preferred and three least-preferred same- sex classmates on 

the same criterion; and Group C, a sociometric rating scale requiring 

students to rate all same-sex classmates on a five-point scale of 

preferred to non-preferred as continuing friends. Group D, desig­

nated as control, received no intervening sociometric. Within each 

same-sex class group students were ranked in terms of the number 

of paired comparisons favoring them on the first administration of 

the paired-comparison questionnaire. Those students in the lowest 

third of their group ranking were identified as least-preferred 

students. 

Following the second paired-comparison administration all 

subjects estimated the preferences made by classmates with respect 

to those pairs containing their names. In addition, they completed 

an eight-item questionnaire focusing on the student's awareness of 

possible change in his classmates' evaluations of him, or in his 

evaluations of them, subsequent to the sociometric procedures. 

Selecte d students were interviewed with respect to discussion 

of sociometric choices, stated perceptions of status change, reasons 

for changing initial sociometric choices or ratings, high and low 



choice expectancies, changes in peer pre ferences given and/or 

received and reactions to the experimental procedures employed in 

the investigation. 

11 



CHAPTER II 

REVIEW OF THE LITERATURE 

The research selections presented are grouped into three 

principal categories: ( 1) general definitions of sociometry, (2) 

measurement of sociometric status, and (3) rejections and negative 

choices. 

Sociometry: General Definitions 

When Jacob L. Moreno presented hi~ initial public exposition 

of sociometry in this country and followed it the following year with 

the publication, Who Shall Survive? ( 19 34) he laid what he described 

as the foundation stone of the sociometric movement. In definitional 

terms, Moreno indicated that 11 sociometry deals with the mathematical 

study of psychological properties of populations, the experimental 

technique of and the results obtained by the application of quantitative 

methods. This is undertaken through m e thods which inquire into the 

evaluation and organization of groups and the position of individuals 

within them 11 
( 1934, p. 10). 

That different d efinitions and interpretations of the term 

11 sociometry 11 have been accepted and used is apparent from the 



13 

literature. Having isolated thirteen different definitions of the term, 

Berkstedt polled "experts" for the most acceptable of them. All 

definitions received at least three first choices and twenty-eight 

experts chose the definition ; 11Quantitative treatment of "preferential" 

interhuman relation s (i.e., relations possible to describe in terms 

of attraction-rejection-neutrality with respect to a choice situation) 11 

(1956, p. 20). In operational terms this definition relates primarily 

to situatio ns requiring the individual to signify his willingness to 

interact with certain members of the group and his unwillingness to 

interact with certain other members of the group. Those not 

specified are regarded as "neutral" with respect to their desirability 

as associates for that choice situation. 

The Sociometric Test 

Moreno ( 1934) described the initial measurement device in these 

terms: 

An instrument to measure the amount of organization 
shown by social groups is called a sociometric test. The 
sociometric test requires an individual to choos e his 
associates for any group of which he is or might become 
a member, ••• (Ibid. , p. 11). 

In the first reported use of the sociometric test in the public 

school it was introduced in the following manner: 



You are now seated according to directions your teacher 
has given you. The neighbor who sits beside you is not chosen 
by you. You a.Lt now given the opportunity to choos e the boy 
or girl you would like to have sit on ei the r side of you. Write 
down whom you would like first best; then, whom you would 
like second best. Look around and make up your mind. 
Remember that next term your friends you choose now may 
sit be side you. (Ibid. , p. 13) 

This instrument, requiring nominations of only part of the 

14 

group, has been called the "partial-rank-order" method of measuring 

sociometric status. (Wityrol and Thompson, 1953). It is also 

frequently known as the "Moreno technique" (Northway, 1952). 

The requirements for designation as a sociometric test as 

initially outlined by Moreno were reinforced by Jennings, who 

emphasized that "a sociometric test is not sociometric unless the 

criterion involved exists for the subjects and the 'tested' individuals 

believe results not only can but will be utilized for the criterion on 

which they have expressed themselves" (1950 , p. 28). To the 

criterion and utilization requirements Lindz ey and Borgatta ( 19 54) 

added four essential components of the test. These were : 

delimitation of the group, an unlimited number of choices or 

rejections, no public identification of individual responses, and, 

questions used should be at the developmental level of the group. 

The desire to retain the term "sociometric test" in the 

specific context of a meaningful criterion and a re structuring of 



15 

the group has met with mixed success. Gronlund appears to have 

retained the two essential elements in stating that the test "requires 

individuals to choose a given number of associates for some group 

situation or activity" ( 1959, p. 3). Berkstedt allowed more fre e dom 

of interpretation in proposing that it "requires individuals to choose 

associates for a special type of interaction" ( 1956, p. 35). 

Many adaptations of the original instrument qualify under 

Northway' s description of the test as "a means for determining the 

degree to which individuals are accepted in a group, for discovering 

the relationships which exist among these individuals and for dis-

closing the structure of the group its elf" ( 19 52, p. l ). One such 

adaptation is the data-gathering procedure employed by Polansky, 

Lippitt and Redl ( 1950 ) to observe the social relationships of youngsters 

in a summer camp. In their report they described it as a "near-

sociometric" proc e d ure and the editor ial board added the following 

note to the report: 

When choice instructions are given which concern notions 
in contrast with specific situations, (as eating at the same table), 
this is in the literature call ed a "near-sociometric test". All 
hypothetical choosing provides, in this sense, "near-so c i ometric" 
data •••• 

• • • • Techniques using a specific criterion but not involv­
ing group rearrangement have been designated as ''near­
sociometric". However, when no criterion is involved in the 
choosing, as in such questions as 11 Who is your be st friend?" 
or 11 Who do you like best?", the data may best be described as 
neither sociometric or near- sociometric, but rather as 



projected 11liking 11 reactions in an undetermined setting. 
(Ibid. , p. 49) 

In general, investigators have defe r red to the use of the term 

16 

11 sociometric test'' in its original meaning. This has resulted in the 

adoption of a variety of descriptive terms which have retained the 

word 11 sociometric 11 but which have substituated another word for 

11 test11 when describing the procedure employed to observe social 

relationships. These substitutes include 11 question 11 
( Croft, 1951), 

11 technique 11 (Spero ff and Kerr, 19 52), 11 instrument11 ( Clark and 

McGuire, 1952), 11 measures 11 (Levi, Torrance and Pletts, 1955), 

11 device" ( Campbell and Fiske, 1959), "questionnaire" (Porterfield 

and Schlichting, 1961), and "choice technique 11 (Hoffman, 1962). 

Each of the procedures cited requir e d positive and n egative evalu-

ations of group members. 

Levels of Sociometric Status 

The specialized terminology associate d with the sociometric 

test and used to describe individuals in terms of the number of 

choices they receive relative to the number r e ceived by other 

members of the group was developed primarily by Moreno and his 

associates. Definitions have been stated in general terms, permitting 

arbitrary limits or classifications to be established by the individual 

investigator. 



17 

In the classroom situation there is a tendency for a small group 

of students to recc:1v~ a disproportiona te number of choices, leading 

to their classification as 11 stars 11 (Bonney, 19 4 6). Conver sely, there 

are some individuals who receive relatively few choices, and others 

who receive none. The former are described as ''neglectees" and 

the latter as "isolates. 11 A fourth term, "rejectee" is used to 

describe "an individual who receives negative choices on a socio-

metric test" (Gronlund, 1959, p. 5). 

Several investigato:i,-s have developed more specific and 

objective means of classification. Bronfenbrenner ( 1944) used 

deviations from chance expectancy as his basis for differentiation 

and Jennings ( 1943) identified high and low status delinquent girls 

as those whose sociometric scores placed them at a position greater 

than one standard deviation from the mean. 

Differentiation between levels of acceptanc e in any group 

depends upon the interpretation of the investigator and the socio­

metric technique employed (Evans , 1962). Any statement made con-

cerning an individual's sociometric status is specific to the group 

and to the criteria or choice situations on which the evaluation was 

made . It should be noted that "group status", "social status" 

(Gronlund, 1959), "group acceptance" and "social acceptance" 

(Northway, 19 52), are terms used interchangeably with II sociometric 

status. 11 



18 

In general, the more spontaneity of response permitted in the 

technique, the more valid the description of the individual I s status 

(Jennings, 1950). A technique requiring a small number of positive 

choices may lead to classification of group members in accordance 

with the definitions stated previously. However, requiring choices 

may preclude the selection of some individuals or result in choices 

being received which would not have been available from an unlimited 

choice sociometric (Borgatta, 19 51 ). Further, as Thompson and 

Powell (1951) have noted, unless provision is made for rejections, 

it will not be possible to positively differentiate between the ''isolate" 

and the "rejectee." 

Various procedures have been employed to establish the social 

status of group members. The ensuing discussion will elaborate on 

these procedures. 

Measurement of Sociometric Status 

Sociometric status may be established by several methods. 

Those discussed in this pr e sentation are partial-rank-order or 

partial-rank, paired- comparison and rating scale. The present 

investigation is primarily concerned with the method of partial­

rank-order and the rating scale as they have been employed to 

obtain negative or "rejection" evaluations of group members. 



Studie s employing rating scale procedures to eli cit negative 

evaluations of grou..l::' members are presented as a unit in the 

literature revi ew section : Rej e ctions and Negative Choices. 

Method of Partial-Rank- Orde r 

This method originally (Mo r e no, 1934) required a specific 

number of choices of group m embers , listed in prefer ential order, 

as associates in a specified activity or situation. Seatmate(s) and 

work companion( s) were two choice situations frequently used in 

this method. Modifications of the method emanated from the desire 

19 

to allow freedom of choice and from the desire to employ criteria 

which would not necessarily result in reassignment of group members. 

Numb er of Cho i c es 

In 19 56 B erkstedt summarized a s eries of one hundred socio -

m etri c investigations in which subjects made choices on a social 

preference situation. The series was drawn from the j ournal 

"Sociometry'' for the period 1945-1954. In his summary B erkste dt 

indicate d a m ethodological trend toward use of unlimited choices. 

Twice as many unl i mited as limited choice procedures were used 

i n the latter half of the period compared with equal popularity of the 

two types of choice procedure in the former half. Thi s d evelopment 



was attributed to a growing awareness of research potentialities 

in unlimited choice l::l. 

There has also been a growing awareness of the limitations 

20 

of using restricted choice procedures. Advocates of freedom of 

choice claim it yields a "truer" sociometric picture of the group 

"since people will only choose those who they really feel positive 

towards and will reject only those they actually dislike or reject, 11 

(Lemann and Solomon, 1952, p. 15), permitting the potential number 

of choices or rejections made by one person to range from n-1 to 

zero. Observations by Gronlund (1959) and Northway (1952) suggest 

that the statement by Lemann and Solomon should perhaps be couched 

in less positive terms. The former observed that unlimited choices 

are essential when attempting to determine an individual's drive for 

social interaction. One cannot, however, make the assumption that 

each nomination made by the socially expansive individual repre­

sents only those to whom that person "really feels positive." 

Northway quest·oned whether individuals may expect to receive as 

they give: do group members expe ct others to choose them more 

because they, in turn, have chosen freely? 

A further limitation of restricted choices is the possible lack 

of reliability, derive d on the one hand from restriction of choices 

in instances where individuals have "clear-cut'' feelings over and 



21 

above the numb e r stipul ated ; and on the other from the forced desire 

for interaction, i nh r ent in the proce dur e (Borgatta, 19 51). Forced 

choices, then, may lack reliability, and they may also lack 

sponaneity, which was emphasized by Jennings ( 1950) as a pre-

requisite to sociometric nominations. 

Gronlund ( 1959) has pointed out that when provision is made 

for unlimited positive and negative nominations most students will 

not choose everybody, for there are group members with whom they 

do not interact and about whom they do not have well defined feelings. 

By a process of elimination the observer is enabled to differentiate 

those to whom the ranker is indifferent from those to whom he 

displays a positive or a negative reaction. 

Choice limitation has its cri tics. It also has its supporters. 

Utilization of a specific type of m e asurem e nt is contingent upon the 

aims of the inve stigator. It is possibl e a limited choice procedure 

will best m ee t his n e eds. Reasons w hich support a decision to 

employ limite d choice s include: ease of administration (Bronfen­

brenner, 1944), the fact that scores b ased on chance expectancy may 

provide confusion (Northway, 19 52), and Berkstedt' s observation 

that "empirical status scores worked out from the two primary 

methods ( limited and unlimited choice) usually show very small 

differences." ( 1956, p. 57) This last reason, however, gets little 



22 

support from a study by Eng and French ( 1948) who demonstrated 

a wide variation in status scores of a group of college women when 

two choice, five choice and unlimited choice rankings were obtained 

in terms of others' desirability as roommates. Their criterion was 

a ranking derived from administration of a paired-comparison 

procedure on the same question, and on this basis unlimited choice 

showed a relationship of. 90, five choice a relationship of. 73 and 

two choice a relationship of . 54. 

Reviews of investigations using limited choices have noted a 

predilection for three choice, followed by five choice and two choice 

procedures (Berkstedt, 19 56; Gronlund, 1959). Gronlund suggested, 

in a more specific manner, that if interest lies in reliability of 

choice in the upper elementary grades, more stable results may be 

obtained from a five-choice question (Gronlund, 1955; Gronlund and 

Barnes, 1956). Evans supported this suggestion in stating that "if 

an even moderately accurate assessment is needed, less than five 

choices would appear to be unsatisfactory" ( 1962, p. 32) . 

Same Sex Choices 

A characteristic of the social development of boys and girls 

is the cleavage which occurs between the two sexes when making 

choices of companions or associates for various activities. The 



23 

extent of the cleavage is contingent upon the type of criterion activity, 

the age of the group and the prevailing attitude in the class, school 

and community toward h e terosexual relationships. Homogeneity of 

choice is noted at the pre-school level (Lippitt, 1941) and is 

markedly accentuated when youngsters reach the 11 gang age 11 

(Campbell, 19 39), which typically occurs during the intermediate 

grades (Stone and Church, 1957) of the elementary school. 

Moreno ( 1934) reported intersexual choices to be virtually 

non-existent among fifth grade pupils. A similar observation in 

choices made by sixth grade pupils was made by Bronfenbrenner 

( 1944). Kuhlen and Lee ( 1943) obtained choices from sixth, ninth, 

and twelfth grade students and found a progressively higher per­

centage of cross-sex choices among older students. This finding 

was corroborated to some extent by Gronlund ( 1959) who showed 

boys 1 choices of girls to vary from e leven to eighteen percent in 

grades thr ee through ten, followed by a relatively progressive 

increase to almost one third of the choices made in a college 

population. Cho i c es by girls demonstrated a greater overall range 

and much mor e fluctuation after grade eight. These observations 

were made on criteria of s eating companion in the elementary school, 

work companion in the secondary school and teaching companion among 

the college students. 



Buswell ( 1950) noted that instances of best-liked students 

rating others negative on the Ohio Social Acceptan ce Scale were 

infrequent. Howeve r, she also noted that some boys automatically 

classified all girls as disliked. By using same - sex populations in 

ranking and rating sociometrics McCandless et al., (1956) and 

Re eEe (1961) did not encounter this ''problem" . 

Number of Criteria 

A further decision made by the investigator employing the 

partial-rank method concerns the number of criteria or choice 

situations on which responses are elicited. Again, the purpose of 

the inve stigation will determine the number of crite ria to be 

employed. 

24 

Kerstetter and Sargent ( 19 40) used a single criterion prepara­

tory to making cl assroom reassignments in accordance with a plan 

for therapy. A singl e question was also asked by Byrd (1951) in a 

study of validity and constancy of sociometric choice. Responses 

have been requested of school students on as many as nine criteria, 

as in a study of same- sex social acceptability among adolescents 

by Kuhl en and L ee (1943) . 



25 

Method of Paired Comparisons 

A second method employed to observe the social preferences 

of group members is the method of paired comparisons ( Thur stone, 

1927 ). In this method 11 the name of every individual in a social group 

is paired with each of the others in the group in all possible combi­

nations, the potential number of pairings being defined by the formula 

n (n-1) where n equals the number of individuals. Each member of 
2 -

the group then expresses preferences on all the pairings 11 (Wityrol 

and Thompson, 1953, p. 243). 

The number of relevant sociometric studies that has employed 

this approach is relatively small. Both Koch ( 1933) and Lippitt ( 1941) 

used paired comparisons to observe popularity among young children. 

\. 

Koch furthe r employed this procedure in a study of social distance 

betwee n the sexes in the elementary school ( 1944) and in a survey of 

the social position of minority groups of children in public schools 

( 1946). In 1953 Wityrol and Thompson administered a paired-

comparison instrument to sixth grade children, requesting that they 

select the indivi dual they liked better in each pai r . Data reported 

by these writers in describing the stability of the paire d-comparison 

scores is discuss e d in the presentation of procedures in Chapter III. 



2 6 

Rejections and Negative Choices 

The terms "negative choices'' and "rejections" tend to be used 

interchangeably in current sociometric literature. Negative choices, 

or rejections, have been obtained using different data-gathering 

techniques, including identification by "least like" and "last choice" 

instructions, and by permitting spontaneous positive or negative 

choices. 

Different Connotations of "Rejection" 

Gronlund defined a rejected person as "an individual who 

receives negative choices on a sociometric test. 11 He continued by 

defining negative choices as "those resulting from a sociometric 

question requesting individuals to indicate those whom they least 

prefer for a group activity" ( 1959, p. 5). A similar viewpoint was 

expressed by Northway and Weld who termed the question 11 Who 

would you least like to associate with? 11 
( 19 57, p. 13) a rejection 

criterion. Further support for this viewpoint was gained from 

McLelland and Ratliff ( 1947), Norman ( 195 3) and Shaw ( 1962), all 

of whom equated "least like" and "rejection, 11 and from Thompson 

and Powell ( 1951) who suggested that when an individual completes 

the negative portion of a partial-rank-order scale he is expressing 

social rejection. 



27 

Three studies reporting use of last choices are illustrative of 

investigator prerogative in classification of individuals identified by 

this procedure. Last choices in Northway' s ( 1940) summer-camp 

investigation were classified by the investigator as rejections. They 

were obtained by asking campers their last choices for cabin-mates 

and canoe-trip mates from the camp population. A somewhat 

similar request was made by Gronlund ( 1955), whose group of 

teacher trainees indicated the five classmates they would consider 

first and the five they would consider last as future teaching 

companions. Through this procedure he attempted to "obtain both 

positive and negative sociometric choices without implying rejection 

of any group members" ( 1955 c., p. 123). A rating scale devised 

by Thompson and Powell to overcome criticism directed at the use 

of the negative portion of the partial-rank-order procedure included 

the category "would be the very last one I would choose" (1951, 

p. 444). In this instance the last choice was not considered to be a 

rejection. 

Rejections have been stated in more specific terms. Bedoian 

required the "rejectee" to be "actively disliked by his peers" ( 1954, 

p. 516). Trent ( 1957) and Clarke and McGuire ( 1952) defined a 

rejection as "an expressed desire to avoid interaction" ( 1952, p. 130). 

Trent's use of "desire" is difficult to interpret for he obtained 



r e jections from institutionalized d elinquent lads through the 

instruction, "l want yo u to give me the names of three boys in your 

cottage who you don't like." (Ib id. , p. 381 ) 

28 

More fre e dom of response was apparent when Emer son ( 1953) 

provided youngsters with an opportunity to indicate others with whom 

they d id not like to play, and when Smucker ( 1949 ) asked college girls 

to list others i n the dormitory with whom they preferred not t o 

associate . 

Inferences drawn from positive -only choices have resulted in 

some individuals, accorded very low or zero scores on sociometric 

tests, being identified as ''rejected" (Flo rence B. Mor eno, 1942) 

and "repulsed" (Kuhlen and Lee, 1943). It i s probabl e that some 

were, in fact , r e jected, but inferences d rawn from positive-only 

nominations must, a t b e st, be d eemed spe c ulative (Bronfenbre nner, 

1944). 

On logical grounds it would be difficult to d efend the inferenc e 

that every person identifie d by the "least liked" and " l ast choi c e" 

nomination procedur e s is psychologically rejected by the subject 

making the nomination. It is conceivable that an adaptation of a 

distinction made by Snoek may have application in sociometric 

terminology. He stated: 



••. we will speak of invidious rejection whenever exclusion 
from a group can be taken to mean that the individual is not 
worthy of membership in it. We shall call the rejection 
non-invidious when it is based on other reasons that do not 
reflect on the individual's self-e.steem; when an individual is 
excluded because all membership positions in the group are 
filled we have an exampl e of non-invidious rejection. ( 1962, 
p. 17 5) 

Support for Inclusion of Negative Choices 

A progressive growth in the inclusion of rejection reports in 

29 

investigations was observed by Berkstedt ( 1956) in his examination 

of one hundred articles from "Sociometry" for the period 1945-1954. 

These articles were divided into thirds relative to their date of 

publication. In the early period thirty-three percent of the studies 

employed negative reports. This percentage grew to fifty-two in 

the middle period, and further increased to sixty-five in the mo st 

recent period. This d evelopment, however, does not appear to have 

been reflected in attitudes of researchers toward employing negative 

evaluations in investigations in school s ettings ( cf. Thompson and 

Powell, 1951; Gronlund, 1959). 

Although the use of negative choices in sociometric procedures 

has not met with general approval in educational circles, advocates 

for their inclusion may be found at each level of educational endeavor. 

At the preschool level Dunnington' s observations led her to conclude 

that "sociometric status is not accurately measured by a system 



30 

which does not include rejection and 'forced' responses" (1957, 

p. 100). A similar conclusion wai:; rt:ached by Phillips and De Vault 

( 19 55) from their study of the relationship between choices received 

and pupil adjustment. These writers reported their understanding 

of seventh grade pupils to be "distinctively enhanced" through the 

use of both positive and negative evaluations. At the college level 

Smucker ( 1947) investigated the possibility that spontaneous positive 

and negative choices received and made by students with adjustment 

problems may provide important bases for guidance. His results 

furnished basic information helpful in planning programs of guidance 

therapy. 

The non-use of negative sociometrics in spite of research 

evidence of more comprehensive student identification is evidenced 

in Cornwell' s ( 19 62) report of ambivalent feelings regarding the 

inclusion of rejections among teacher trainees in England. Although 

Cornwell' s prospective teachers "recognized that the inclusion of 

rejections as well as choices gave a truer picture of the group and 

was fairer to the great majority who, though not chosen, were also 

not rejected" (p. 14), and although they realized that identification 

of rejected individuals made it possible for them to be helped, 

sixty-two percent of them voted against the inclusion of rejections 

in a sociometric procedure. 



31 

Practical utilization of information received from negative 

nominations is to be noted in a number of studies. Utilizing 

information derived from negative choices Gronlund ( 1959) evaluated 

efforts to integrate minority group members and to locate subtle but 

revealing interpersonal conflicts in the classroom. Brickell ( 1950), 

seeking more efficient working groups, used negative sociometric 

data to locate secondary school students in groups in which members 

were not antagonistic toward one another. Particularly interesting 

in this vein is Roff' s ( 1961) study of servicemen cited for bad 

conduct. From a review of their child guidance records Roff found 
I 

the best predictor of this type of behavior to be reaction of peers to 

these persons when in grade school. Unacceptable social behaviors 

had led to thei r rejection at that time. It should b e noted that the 

behaviors identified by Roff were parall eled in reports of behav iors 

unacceptable to fellow-delinquents in a girls' training school 

(Jennings, 1947; 1950). 

Subsequent to their observation that a substantial number of 

rejects in the grade school population are emotionally disturbed, 

Sells and Roff ( 1963) suggested that initial observations should be 

conducted early in the school life of the child in order to identify 

children rejected by their peers. They further suggested that, 



following identification, these rejected children should take part in 

a therapy program designed to assist them to develop socially 

acceptable behaviors. 

32 

Two of the pioneers of sociometric study in this country have 

recommended that negative choices have an integral place in socio­

metric evaluation. Jennings has proposed "that the negative aspect 

of choice merits equal consideration with the positive'' ( 1950, p. 21) 

as they form not a dichotomy, but "one choice process and bear 

particular relationship to each other" ( 1943, p. 58) and in 1954 

Moreno reported that the prevailing custom in sociometric laboratories 

was to ask for both choices and rejections. 

Support for the inclusion of negative choices, then, is based 

upon desire to measure sociometric status accurately, to obtain 

more productive evaluations of adjustment, to differentiate between 

those n,ot chosen and those actively disliked , to identify undercurrents 

in the social structure of the group, to structure groups in which 

members are not antagonistic, and to identify potentially disturbed 

or socially maladjusted children. Evans summarized the desire to 

incorporate negative evaluations in sociometric administrations by 

stating that "if a complete study of a group is to be made, rejections 

and choices should both be obtained, whether the aim is social 

engineering, therapy, or research" ( 1963, p. 55). 



Opposition to Inclusion of Negative 
Sociometric Choices 

The use of negative choices in sociometric studies has not 

met with general approval in educational circles . The ensuing dis-

cus sion elaborates on an investigation which has greatly influenced 

attitudes toward inclusion of negative choices. In addition, it 

outlines the reservations individuals have expressed concerning 

this aspect of the study of social relationships. 

A survey of bibliographies of sociometric investigations 

indicates the source from which many generalizations pertaining to 

33 

use of negative choices have emanated to be Northway' s ( 1940) study 

of social development in a girls I summer camp. Generation of 

resentment and comment among campers was reported to be a con-

comitant of the request for "last'' choices, convincing Northway 

that "negative" choices were inappropriate in sociometric procedures. 

It should be noted that the investigator asked for "absolute last choice" 

and "next to last choice" ( 1940, Appendix, p. 61). In turn, she has 

influenced sociometric procedures by providing a central reference 

for non-incorporation of negative choices in investigations of group 

relationships. In her discussion of negative choices at that time, 

Northway counseled that they could be eliminated or given in a less 



objectionable form if adapting the procedure for use in diffe rent 

contexts. In a later publication she proposed that last choice 

investigations may be artificial "in so far as most p eople ar e not 

actively interested in those with whom they do not associate" ( 1952, 

p. 5). Artificiality was noted by Northway when reporting camper 

reaction to the sociometric instrument, for one girl had never 

considered last choice: 

There was considerable resentment toward filling in "last 
choices" - in fact, one camper put on her form, "I never 
thought of having a last choice, if I had, I could never write 
it down." When the meeting for the second test was arranged, 
there were a few campers who announced they were not 
coming. • • • Except for the temporary stir created on each 
occasion when the test was given, it seemed to create very 
little impression, and was very seldom discussed ( 1940, p. 17). 

Observation of the distribution of last choices , of which two 

apparent "scapegoats" initial! y received thirty-nine p er cent and 

34 

subsequently received forty-six percent, suggested a reference point 

for Bronfenbrenne r ( 1944) to state his position on incorporation of 

negative choices. He avoided them on the grounds that they may 

focus attention on the less favored, thereby promoting d i scrimination 

and resulting in accentuation of social maladjustment of these group 

members. These sentiments were echoed by Wertheimer ( 1957) in 

an investigation of choice consistency in adolescents. 



35 

That negative choices may be discussed outside the classroom 

(Thompson and Powell, 1951), may crystallize existing antipathies 

(Baller and Charles, 1961), and may undermine the emphasis on 

the development of a positive approach to interpersonal relationships 

(Phillips and DeVault, 1955) are reasons cited for reluctance to 

request negative evaluations in a school context. Individual class­

room teachers have endorsed the contention of Northway and Weld 

that ''children's groups are free enough that an individual is not 

forced to be continually with a person he dislikes" ( 1957, p. 13) . 

This may be the reason why Young ( 1947 ) found elementary school 

children disinclined to "speak ill" of classmates on "reputation­

type" tests even when identificationswere expressed in an anonymous 

manner. 

Lundberg, Hertzler and Dickson ( 1952) e ncounte r e d a 

corresponding response among college women. They administered 

a questionnaire to these students, requesting respondents to list 

the three persons now living in the house they least preferred as 

ro o mmates for the following year. Students wer e also aske d to list 

three on-campus women as least-preferred on the same criterion. 

The question "Why?" followed each choice and subjects were 

informed choices would be taken into consideration when roo m 

assignments were made the following year. The questionnaire, 



36 

administered in this manner, resulted in only a twenty-six p er cen t 

response. It was appar e nt that some respondents were antagonized 

by it for their written responses included: "questions like this 

encourage intoleranc e, 11 ''uneasy feeling about this whole question­

naire ..• entirely contrary to w hat I have been taught •.• I don't 

like to admit even to myself ••• much l es s set down their names 

for other people to see, 11 and 111 think i t very unfair to deliberately 

dig out of my mind three people that I don't like especially" ( 1949, 

p. 164). A follow-up "logical II explanati on, designed to counter 

objections to negative choices, resulted in a twenty percent increase 

1n responses. 

A similar reaction was noted among resi dent students at an 

English teacher training institution (Cornwell, 1962 ). They found 

the making of reje ctions "repellent and repugnant, causi ng bad 

feelings in the mind and leaving a sense of inward guilt" (Ibid. , p. 14). 

Cornwell observed that students found diffi culty in avoiding prejudice, 

an observation leading to the inference that the students were 

instructed to m ake a specified number of "re j e ctions" in a forced­

choice procedure. 

The Lundberg et al. investigation requir e d negative choices. 

A study by Smucker ( 1947 ) with a similar population p ermitted 

negative choices . Each study typifies the reacti on, at this educational 



37 

level, to the stated difference in data-gathering technique. Smucker 

received a ninety perce.1t response when he asked campus girls to 

name their best friends and those with whom they did not wish to 

associate, eliciting the negative data by saying "it is an obvious fact 

we do not like everybody equally well . List here the names of 

campus girls you don't like so well, wouldn't like to run around with, 

or feel that your p e rsonalities clash. List one, two, or more as you 

wish'' (p. 376). No adverse reaction to this procedure was reported 

but the investigator cautioned that subjects must have confidence 

that complete anonymity will be maintained, in order to forestall 

repercussions which may eventuate should rejection information 

become common knowledge. 

Many sociometric studies conducted in the armed services 

have employed negative evaluation s but few report reaction to 

requests for these evaluations. Any reaction to the procedures 

employed was forestalled in a study involving naval personnel (Kogan 

and Tagiuri, 1958), for least like identifications w e re asked only of 

a group that "had bee n exposed to a gr e at many research pro c edures 

and the data [pertaining to a 72-hour libe rty] could be obtained 

without unduly disturbing them" (p. 113). In a different branch of 

the services some air c rews refused to name one of their number 

as the l east d esirable survival companion (Levi, Torrance and Pl ett s, 

1955 ). 



38 

Reluctance to employ negative evaluations, then, has centered 

on the reaction of campers to making "absolute last choices, 11 on the 

possibility that negative choices might be discussed and thereby 

result in deepening r eje ction, on the concern that a positive approach 

to social relationships is desirable, and on antagonism of respondents 

to completing negative evaluations. It was noted that adverse 

reactions of subjects emanated from forc e d r eje ction choices and 

not from sociometric procedures in which the rejection or negative 

choices were optional. These differing procedures have not resulted 

in discrete reactions, however, since no display of repugnant feelings 

by subjects has been reported in the majority of studies which have 

employed forced negative choices. 

Rejec tions on Unlimited Choice 
Sociometric Instruments 

Patterns of peer relationships show considerable variability. 

The individual and his group are influenced by many factors that tend 

to channel responses in particular directions. Some of these factors 

ar e inherent in the current nature of the situation; others are founded 

in the social and emotional climate which prevails in the group. The 

spontaneity with which individuals accord negative choices to fellow 

group members when this type of choice is administratively optional 



in a sociometric procedure may well be manifestations of these 

factors in action. Several studies reporting quantitative data on 

spontaneous rejections at different levels of educational endeavor 

are noted. 

Dunnington ( 1957) asked a group of youngsters questions 

concerning the desirability of others as nursery school playmates. 

The fifteen children in the group volunteered twenty-nine names to 

39 

a positive question and one fewer to the negative question, (Whom 

in nursery school don't you like to play with?). It might be noted 

that greater class agreement- -a few classmates receiving mo st of 

the nominations--was obtained in the case of the negative question. 

Following the spontaneous nominations, children responded to the 

remaining group members in a positive or negative manner. The 

point scoring system developed to incorporate both positive and 

negative evaluations demonstrate d a retest increase of twenty-nine 

percent in negative sco~es and an increase of twenty-four percent 

in " social expansiveness." A comparison of points awarded subse­

quent to two interview s revealed several marked instances of group 

ambivalence. The lad who twice rec eived most rejection points was 

ranked fourth in choice points, and the lass second in rejection 

points on both occasions was ranked fifth and second on choice points. 



Ambivalence has also been reported by Pope (1953) from 

prestige choices among a contrasting socio-economic group of 

children and by Trent ( 1957) who found that aggressive boys in a 

truant and juvenile delinquent population were both chosen and 

rejected frequently. These observations suggest caution when 

interpreting a net-acceptance composite score derived from plus 

and minus tallies (cf., Norman, 1953). 

Jennings ( 1943, 1950) has shown that with an increase in 

40 

socialization there is almost certain to be some increase in negative 

feelings between particular individuals. This finding was sub­

stantiated by Bonney ( 1953) who inquired into the choice patterns of 

roommates in a college dormitory male population. The initial 

inquiry resulted in rej ection of twenty-four percent of roommates. 

In a sub.sequent inquiry this percentage rose to thirty-five. In the 

intervening period the positive choices rose twenty-nine percent. 

A similar rise in positive choices, but not in negative choices, was 

reported by Drawhorn ( 19 56) among a group of twenty-four women 

teacher trainees in their final year of college. These women had 

made 273 choices and only fourteen rejections at the beginning of a 

four-month period. At the end of the period, following close 

association for three days a week the number of choices rose to 348 

while the number of rejections fell to eleven. 



College women al so participated in two investigations of 

friendship evaluations conducted by Smucker ( 1947, 1949). All-

campus choices and rejections of nearly 700 women in the earlier 

study resulted in a choice average of 2. 8 and a rejection average 

41 

of 1. 03. Evaluations were restricted to the dormitory in the second 

study, and the average choice-rejection ratio was 5. 3 to 2. 1. Aware­

ness of rejections was also studied in these inve stigations. In both 

studies Smucker found mutual rejections almost non-existent, con­

firming his original observation that very few people ar e aware of 

others having hostile feelings toward them. A similar observation 

was made by Ames ( 1945), relative to sixth graders' lack of aware­

ness of how well they were liked by classmates. However, obser­

vations to the contrary were reported by Dunnington ( 1957). She 

interviewed preschool children and the recorded ve rbalizations of 

some youngsters clearly indicated their cognizance of the attitudes 

of others toward them. At this developmental age they were un­

inhibited in their reciprocity of rejections. 

Further studies evidencing choice-r e jection ratios are those 

of Goldstone et al. ( 1963) and Zudick ( 1953). In the Goldstone et al. 

study sophomores in medical school responded to two questions on 

prestige and friendship, and a resultant choice-rejection ratio of 



42 

1. 2 to 1. 0 was derived. In comparing identifications at the extremes 

of the acceptance-rejection continuum the investigator noted rejections 

to be comparatively more concentrated, with the three most rejected 

students re c eiving 223 rejections, and the three most chosen receiving 

133 choices. Zudick ( 195 3) inquired into the choice patterns of 

children in the second, fifth and eighth grades and he reported same-

sex and other- sex nominations. The average same-sex acceptance­

rejection ratio at the fifth grade for seatmates and par ty companions 

was 2. 5 to 1. 0 in favor of classmates of the same sex. Examination 

of other - sex preferences revealed that secon d grade children were 

more acc epting , making twice as many rejection s as preferences, 

than were students in the upper grades who rejected other- sex group 

members on a 3: 1 ratio. 

The quantitative results presented in this discussion under ­

score the fact that, given the opportunity, individuals are p repared to 

make negative evaluations of group members. Frymier ( 1959) has 

sugges t ed that these evaluations are meaningful because students 

only provide such information when they feel this way about others. 

It is further evident that there are wide variations in the degree of 

acceptance-rejection manifest in groups and in the willingness of 

group members to signify the lack of desirability of other group 

member s. 



Procedures and Statements Employed 
to Obtain Negative Evaluations 

43 

Throughout this review, questions to which subjects have been 

asked to respond, directions given to obtain the data, and manner of 

scoring this data have been included if they clarify the point at issue . 

Further procedures and statements are outlined to demonstrate the 

variety of approach employed to secure information of a negative 

nature about relationships within existing groups. All studies cited 

have been classified by their inve s tigator as 11 sociometric , 11 and 

have utilized the partial-rank method, or a modification of it. 

Prior to the time when children can recognize or write names, 

identifications for the purpose of sociometric choice may be examined 

through the use of photographs and individual interviews. During 

individual interviews children have responded to photographs of 

individuals (Medinnus, 1962) and of entire social groups (Emerson, 

19 53; Moore and Updegraff, 19 64). The individual interview 

procedure was also used by Zudick ( 1953) when obtaining responses 

from second grade children preparatory to re structuring classroom 

seating in accordance with stated identifications. In addition to 

requesting positive and negative identifications, Zudick asked 

children to name classmates who they felt would rather not sit with 

them, and he grouped perceived and stated rejections when tabulating 

the number and proportion of rejections. 



44 

A fewer numb e r of rejections than choice s has been required 

in certain studies (cf. Croft, 1951; Phillips and DeVault, 1955). 

Porterfield and Schlichting ( 1961) followed this procedure, and ranked 

sixth grade students on a composite score derived from four positive 

and one negativ e questions. McLelland and Ratliff ( 1947) also 

requested fewer negative than positive choices. Further, they 

obtained them at different administrations. Initially their home­

room class of ninth grade students stated first preferences on four 

criteria. Twelve of the thirty-five subjects were unchosen. In 

order to differentiate between the isolates and rejectees negative 

preferences were elicited on two of the previously employed criteria. 

It was reported that greater hesitancy on the part of the subjects 

was apparent when they were making the n e gative identifications in 

the second testing session. 

Perhaps the use of fewer negative choices is a compromise 

between the investigator's desire to obtain negative e v aluations 

because of their utility in de scribing social relationships in the group, 

and his hesitancy to use an experimental proc edur e deemed undesir­

able by others in the field. It is also possible that procedures such 

as those used by Cassel and Saugstad ( 1952) in requesting that 

optional negative choices be made on the back of the questionnaire 



45 

form, and by Young ( 1947 ) whose subjects made the i r identifications 

anonymously, reflect a oimilar concern. 

No studies have focused upon the subject and the psychological 

meaning to him of instructions given during the administration of 

the negative portion of a sociometric instrument. It is clear, how­

ever, that certain instructions have been stated in more direct terms 

than others. The writer assumed "not suitable" and "definitely not 

choose" (Goldstone et al., 1963 ) and "definitely don't want in your 

group" (Jennings, 1943) to be statements with little inherent 

ambiguity. Similarly, directions of "if there aren't any just leave 

thi s space blank," on a negative question (Zudick, 1953), and "if 

you do not feel really close to any girl in the dormitory, write no 

name" (Smucker, 1949), clearly indicate free dom to nominate or 

freedom to abstain from nominating. 

Less definite statements were made by Clarke and McGuire, 

who asked students to name "the ones with whom you would prefer 

almost never to run around," and "if you were going to have a party 

who are the boys and girls you might not prefe r to have along? 

They could go elsewhere, " and finally "who are the ones you probably 

would not choo se for your very best fri e nds ? 11 
( 1952, p. 135). 

Further diverse ways in which instructions to elicit negative 

evaluations have been expressed include oppo r tunity to identify 



persons "liked least" (Fie dl er et al., 1952); naval aviation cadets 

"not wanted'' as part of a unit (Holl ander and Webb , 1955 ); the 

"least likable three, 11 to test an hypoth e sis on similarity of person­

ality profiles prior to acquaintanc e (Izard, 19 60) and 11 least of all 11 

in a class of secondary school students (Croft and Grygier, 1956). 

Trent asked institutionalized delinquent youths for 11 the names of 

three boys in your cottage who you do n 't like. First name the boy 

you dislike mo st. • • • Think carefully b e for e you answer" ( 19 57, 

46 

p. 381). This parallels a request made by Davids and Parenti (1958) 

of emotionally disturbed youngsters in a camp situation. They asked 

for the names of three campers "disliked the mo st. 11 Drawhorn' s 

( 1956) teacher trainees respond e d to the request for others whom 

they "would not choose" as associates for several diffe rent projects. 

In his investigation of group cohesiveness conducted in the armed 

services, Goodacre ( 19 5 l ) varied his e liciting instruction s for 

differ e nt criteria. On d esi r ability of group m e mbers as associates 

for chow or a party, identifications were made on "want" and "not 

want". However, when the s e l e cti on of a ten tmate was requested, 

identifications were made followi n g the dir e ction of 11 choose 11 and 

"not choos e ". No r e ason was give n for the cha nge in terminology. 



Rating Scale Procedures for Identification 
of Negatively-Preferred Group Members 

47 

The controversy over using negative choices in the observation 

of interpersonal relationships is centered on the use of the partial-

rank instrument in educational contexts. Rarely is opposition noted 

when the data-gathering device is what is typically known as a 11 rating 

scale 11
• Generally these scales provide opportunity for raters to 

evaluate other members of the social group on scales ranging from 

three-point to nine-point. The present review is included to provide 

some examples and contrasts of the negative classifications employed 

in these scales in an educational setting. 

Perhaps the mo st popular classification procedure is one 

similar to that used by Ausubel et al. ( 1952), providing for ratings 

of ( 1) 11 do not want as a friend at all 11 and (2) 11would not like to have 

this person as a friend 11 at the negative end of the scale. In this 

study the fifth grade students classified six percent of their class-

mates in the lowest category and seven percent in the next lowest 

category. The same scale classifications were used by Schiff ( 19 54) 

and Reese ( 1961) but no comparative results were reported. 

The original Ohio Acceptance Scale has been used quite 

extensively (cf. Cunningham, 1951; Buswell, 1953; Bond and Brown, 

1955). Evaluations range from 11 very, very best friends 11 to 11 don 1 t 



48 

care for them'' and "dislike them". Buswell observed that in g e neral 

the best liked students did not use these last two categories, although 

some boys automatically declar e d they disliked girls. The Revised 

Ohio Scale has since modified the categories assumed to be indicative 

of negative peer relationships to; "know them b ut they are not friends" 

and "not okay to you" (Fo r lano, 1964). A similar approach by 

McCandless, Castaneda, and Palermo ( 1956) permitted classification 

of same-sex classmates on a five-point scale in which the most 

negative category, 5, was, "is not my friend". 

Two negative classifications were noted on the Classroom 

Social Distance Scale which was employed by Singer ( 1951) and 

Goslin ( 1962) to evaluate the social status of adolescents. On this 

scale a rating of "4" is ; "don't mind him being in our room but I 

don't want anything to do with him" and a "5" means the rater wishes 

the ratee "weren't in our room". In using the Classroom Social 

Distance Scale, Cunningham ( 1951) reported eleven of thirty-two 

children did not classify anyone in the study in the negative categories 

of four and five. A somewhat different negative classification was 

that presented to ninth grade students by Scandrette ( 1958). Three 

of the six evaluation i terns were: "would like to be with him once 

in a while, but not very often"; "don't, or wouldn't mind him being 

in our room but I don't want to have anything to do with him" and, 



49 

"wish he weren't, or glad he isn't, in our room" (p. 368). Bonney' s 

( 19 54) rating device, 11 How I Feel Toward Others" afforded a further 

variation. Each of his categories was defined in terms of rater­

ratee interaction, concluding with the evaluation statements, "know, 

but they are not my friends", and "children I do not want to have as 

friends - as long as they are like they are now" . 

Though somewhat removed from a strict rating scale procedure, 

Tuddenham's (1952) modification of the Guess Who Technique developed 

by Hartshorne and May ( 1929), is of relevance to this review because 

of observer and participant objections to negative identifications 

required by this test. Tuddenham presented children with a 

Reputation Test which included such items as 11 Who are the ones 

everyone likes? 11 and 11 Who are the ones nobody likes very much (the 

ones nobody seems to care about) ? 11 Objections were voiced by some 

participants and by several administrators to certain items on this 

test. Consequently, interviewers were advised to watch for possible 

signs of discomfort in their informants. Tuddenham reported "the 

alacrity with which nominations were volunteered soon made it clear 

that the Reputation Test was merely tapping attitudes already 

crystallized in each classroom group" ( 19 52, p. 7 ). 

Does the rating scale approach overcome some of the alleged 

shortcomings of the partial-ran k procedure? Thompson and 



50 

Powell ( 1951) stated that opposition to use of the n e gative portion of 

the partial-rank instrume nt is based primarily on the premise that 

it causes children to "crystallize the ir opinio ns about r e j e cte d 

children, discuss the m with thei r frie d . and possibly l e ad to an 

eve n gr e ater r e j e ction of c e rtain childr e n " ( 1951, p. 449). These 

investigators have suggested the rating scale does overcome this 

problem and they hav e based this sugg e stion on ob s ervations of a 

comparison between the two approaches with sixth grade students. 

Their sixth grade students nominated thre e companions for four choice 

situations and rated classmates on four different situations. This 

rating involved the following evaluations: 

1. Would be the very first one I would choose. 
2. Would be one of the fir st thre e I would choose. 
3. Would be one of the fir st six I would choose. 
4. Would be one of those I might or might not choose. 

(makes little difference). 
5. Would be one of the last six I would choose. 
6. Would be one of the last three I would choo se. 
7. Would be the very last one I would choose. 

From their observation of the data they conclude d that the "use of 

the rejection or negative portion of the rating scale does not cause 

certain children to be stigmatized as 'rejects' by their associates" 

(p. 452). Further, an increase in the means of low-rated individuals 

over succe ssive admini s trations l e d Thompson and Powell to suggest 

this was the resul t of an incre ased tol e rance on the part of the raters 

for group m e mbers previously rated negatively. 



51 

Reactions by subjects of a quantitative and qualitative nature 

to the request for negative nominations of group members have been 

discussed in this review. Generalizations emanating from results 

of pertinent investigations, focu sing on the "harmful effe cts" of 

inclusion of negative nomin ative procedures, have been extende d from 

summer camp and college populations to i ncl ude school populations. 

The succeeding chapte rs relate to the present investigator's attempts 

to examine the validity of these "har mful effects" generalizations in• 

a sample of fifth grade students. 



CHAPTER III 

PROCEDURES 

Pilot Study 

This investigation began in spring term, 19 64, with a pilot 

study involving two fifth grade classes from Mt. Vernon and Page 

Elementary Schools, Springfield, Oregon. This preliminary study 

focused primarily upon the stability of a paired-comparison 

questionnaire requiring preferences for all classmates in terms of 

their desirability as friends for a long time, with an intervening 

sociometric questionnaire on which students listed in preferential 

order the names of at least three classmates they would most like 

to have as friends for a lon g time. Additional concerns included 

refinement of admini strative procedures, amount of class time 

involved, appropriate analysis of data, practicability of the paired-

comparison procedure with large class e s , u se of other-sex choices, 

and reaction of subjects and teache r s to the p ro c e dures. 

Observations germane to the study prope r were based 

principally on interaction with one class (N=24), data from the 

other being subjected to mini mal analysis following student absences 



53 

at different stages of the study, a smaller N, and the dispropo rtionate 

number of boys in the class. Stability of the paired-compari son sco:i:es 

was examined in terms of a test-retest product-moment co rrelation. 

The co rrelation between the preference scores obtained from two 

d . . . f h . d . . 98 l a m 1n1strat1ons o t e pa1re - c omparison instrument was • • 

Similarly, correlations for same-sex and other-sex p referenc es 

were compute d. Same-sex p referenc es by b o ys c orrelated. 92 , and 

the corresponding corr elation for gi rls was . 88. Correlations fo r 

other-sex choices were. 96 and. 95 for b oys and girls respectively. 

The pilot study called for no negative or rejection identifications. 

Howeve r , on the crite r io n of "good friend for a long time" the 

investigator found a general sex-cleavage, w ith most-preferred 

involving same - sex preferences and least-preferred involving other -

sex preferences. This was observed both in the paired -comparison 

d ata and in the positive sociometri c nomination s . The least-

preferred third on boys' preferences located three boys and five 

girls. A similar classification of girls' preferenc es was c o mpos e d 

entirely of boys, while the intervening sociometric questionnaire 

1
This corre l atio n is similar to those reported by Koch ( 1944 ) 

with fourth and s ixth grad e pupils, and by W itryol a nd Thompson 
(1953 ) with a population of sixth g rade pupils . 



did not furnish a single heterosexual choice. Similar observations 

have been reported in the review of the literature. 

The implications were clear. As the emphasis of the study 

proper related to the least-prefer red third of each sub-group, and 

as there were strong indications that this classification w ould be 

appli e d predominantly to the opposite sex, it was apparent there 

would be less effect, if, indeed, any effect at all would prevail, on 

54 

subsequent expressions of preferenc e behavior if negative nominations 

were required in cross-sex rather than same-sex social groups. 

For this reason the decision was made to measure the changes in 

preference behavior in the mor e socially significant population of 

1 
same-sex classroom groups. R e quiring negative nominations in 

same-sex groups would constitute a more stringent test of the m ajo r 

hypotheses. 

M ajor Investigation 

Principals of selected schools in Di strict 19, Springfield, 

Oregon, were provided with a brief overview
2 

of the planned study 

1
Whil e this proce dur e was a deviation from general sociometri c 

practice, it did not create a precedent, other investigators having 
employed same-sex groupings i n the regular classroom (cf. McCandless 
et al., 1956; Re ese, 1961). 

2
Study outline presented to principals of participating schools is 

presented in Appendix A , p. 142. 



55 

outlining the purpose of the study a n d the pro cedures to be employed. 

Estimates of proj e cted individ ual and clas s time involvement were 

included in this overview. The investigator discussed the study with 

each principal, who, in turn, discus sed it with teachers of fifth 

grade classes. All teachers agreed to par tic ipate in the study. 

Sample 

The major investigation was conducted in five elementary 

schools and involved twelve classes of fifth grade students. The 

schools, and the respective number of participating classes were 

Brattain (2), Maple (3), Mt. Vernon (2) , Page (3) and Thurston (2). 

Three hundred and nine subjects were enrolled in these classes. 

The study extended over a period of approximately seven weeks in 

the months of November and Dece mbe r , 1964. 

Each of the twelve cl asses was as signed to one of four 

experimental treatment conditions, A , B, C, or D. Admini stration 

of sociometric procedures differentiated the treatment groups. 

Group A received a sociometric ques t ionnai r e r e qui ring student 

selection of three mo st-preferred same - sex classmates as friends 

for a long time ; Group B, a sociometric questionnaire requiring 

student selection of three most-preferred and three least-preferre d 

same-sex classmates on the same crite rion; and Group C , a 



sociometric rating scale requiring students to rate all same- sex 

classmates on a five-point scale of preferred to non-pre i crred as 

friends for a long time. Group D, designated as control, received 

no intervening sociometric. 

Assignment to the four treatment groups was random, with 

limitations of partial consideration of socio-economic background 

and boy-girl ratio. In order to obviate possible discrepancies in 

56 

socio-economic background, and for other obvious reasons, classes 

within each school were available only for different treatment groups. 

Further, several classes evidencing disproportionate boy-girl ratios 

of 17 : 8 and 16 : 7 were not considered for the same treatment 

group. Groups were relatively homogeneous in level of a cademic 

h
. 1 

ac 1evement. 

Method of Paired Comparisons 

The method of paired comparisons was selected as a stable, 

relatively unbiased procedure for measuring the social acceptance 

of same-sex classmates. The major consideration in selection of 

1 
Two principals reported a relatively comparable academic 

achievement level in their fifth grades. Three principals reported 
a system of ability grouping. At least one of the 11 comparable 11 

classes was observed in each treatment group. No group was 
composed of more than one high- or low-achieving class. 



57 

a testing procedure was that the procedure would be innocuous in the 

sense of non-emphasis or non-attention to the rejection aspect of 

choosing , permitting least-preferred classmates to be identified 

without the chooser making other than positive choices. The further 

fact that each student would be required to make numerous choices, 

tending to obscure any specific singling out of least-preferred class-

mates, led to selection of the method of paired comparisons as the 

testing procedure. In terms of the present investigation the method 

of paired comparisons has several advantages over the more 

frequently employed method of partial-rank order, including greater 

stability c;,f measurement, provision of complete rather than partial 

data, less likelihood of recall of initial preferences and simplicity 

1 
of the task. 

1
General limitations of applicability to classroom use include 

the "labor involved in arranging the p airs and in ordering them to 
minimize psychological biases of position effect •.• scoring is 
time consuming, and scaling is laborious, difficult and a highly 
specialized technique . • . sociograms cannot be constructed. . 11 

(Witryol and Thompson, 1953(a) , p. 243). Further limitations are 
the geometrically increasing number of choices required, and the 
concomitant class time involved as the group increases (91 choices 
for N=l5, 171 for N=20, and 276 for N=25), the possible lack of 
motivation of subjects on a long task, the necessity that subjects 
can read all names, the possibility that younger children may 
operate on a position-error basis and the fewer number of criteria 
on which data may be obtained in a comparable testing p eriod. 



58 

Koch ( 1944) reporte d test-retest correlations, computed from 

paired-comparison scaled scores, of. 937 from forty fourth grade 

children and. 965 from thirty-five sixth grade children. The period 

between initial and final administrations was one month and the 

criterion wa s selecti on of the more-preferred child from each pair. 

Witryol and Thompson ( 1953), studying friendship in four sixth 

grade classes, obtained product-moment test-retest paired­

comparison correlations of • 903 to • 987 with the median correlation 

approximately. 96. Correlations were reported over periods of 

one, four and five weeks. The present investigator similarly 

obtained a correlation of. 978, calculated from raw scores of paired 

comparisons administered two weeks apart. In general, investigators 

employing similar groups of subjects and similar criteri a have 

reported a tendency for retest correlations for sociometric data 

elicited by the method of partial-rank order to range from. 60 to 

. 90 (Gronlund, 1959; Mouton, et al., 1960 ). 

A second advantage lies in the results providing a complete 

rather than partial picture of intra-group preferences. This occurs 

since the paired comparisons require a reaction to each group 

member in terms of his acceptance relative to that of each other 

group m ember . In contrast, the partial-rank procedure prov.ides 

for nomination of a limited number (usually less than six) of members 



of the respondent's group. A further advantage of particular con­

cern in this study, is that the large number of preferences to be 

made reduces the likelihood that recall of initial preferences 

could influence those made on the subsequent administration. 

Finally, the task is procedurally simple, requiring minimal recall 

and consideration of only two classmates at one time. 

59 

In order to prepare the paired-comparison forms for presenta-

tion, current class lists were procured from school secretaries. 

Teachers indicated the names by which students were known in the 

classroom and the sex of the student when the name was insufficient 

identification. It was considered that expo sure of names and 

assistance in their recognition prior to administration of the paired-

comparison questionnaire would eliminate difficulties in reading these 

names in the 11 test11 situation. Several days prior to the investigator's 

initial visit to each classroom students received a dittoed copy of 

l 
the first and last names of same- sex classmates, randomly orde red 

with the name by which they were known in the classroom underlined 

and preceded by an assigned student number. Names which would 

appear on the paired-comparison form were repeated at the bottom 

1 
An example of a pre-test class handout is presented in 

Appendix A, p. 147. 



of this paper. If first names were identical, or similar, the first 

letter of the last name was added as the distinguishing feature, in 

the manner of Scott H , Scott O; Ricky l' and Rick C. Classroom 

teachers determined the amount of rev iew of these lists necessary 

for complete familiarity with the names. l, 
2 

From these recognition lists supplied to students the combi-

60 

nation of names and numbers were retained for use in the preparation 

of the paired-comparison form. Names of members of same-sex 

b . d 3 ' bl b ' · 4 ' S P · · su groups were pa1re in every poss1 e com 1natlon. airings 

were mechanically arranged to minimize position effect, with both 

time and space factors controlled following the procedure outlined 

by Ross (1934), with minor variations . 

1
0ne class, reportedly comprised of high achievers in an 

ability-grouped grade, did not use the revi ew sheet. 

2
Students were encouraged to use the lists when completing the 

paired-comparison questionnaire and several poorer readers did so. 

3 
In general, the total number of different pairings presented on 

a paired-comparison worksheet is deriv ed from the formula n(n-1) 
when.E_ equals the number of persons or things being compared.2 

However, in this study subjects were not asked to state preferences 
in pairs containing their own names. The formula for the number of 
different pairings available for study, then, reduces to (n-1) (n-2). 

4 
The number of paired comparisons ranged from fifteen (from 

a same- sex subgroup of seven) to one hundred and thirty- six (from a 
same- sex subgroup of eighteen). 

5 
An example of a paired-comparison questionnaire completed 

by subjects is presented in Appendix A, p. 151. 



I n introducing the paired-comparison procedur e
1 

the investi-

gator indi cate d the study was de signed to pro vide informai:" on on 

the d e cisions m a de by fifth grade students when r e q ui red to make a 

choice between two items, both of which may be like d , o r between 

61 

two p e ople, both of whom may be liked. A practi c e proce dure, 

designed to familiarize students with the proce ss for making pre fer-

ences on the paired-comparison form, compared i c e cre am flavors. 

Preferences were stated orally by students. During the progressive 

discussion the positive approach was emphasized. 

Following the practice and distribution of the paired-comparison 

2 
forms, a procedural explanation, specific to the form, w as given. 

Stu dents were instructed; "from each pai r of names make a ring 

around the name of the p erson you would most like to have as a friend 

for a long time." Subje cts were encouraged to work as quickly as 

3 
they c ould, and they we r e advi s ed that erasing was p ermitted. 

1 
Admi nistration of the paired - comparison que stionnaire is 

presented in Appe ndix A , p . 149. 

2 
Stu d e nts we re informed that questions w o uld be answe red 

individually w hen classmate s had commenc e d to compl ete the form. 
Thi s was do n e to pre clude the class being influenced by ques tions 
with a n e gative connotati on, such as: "What do you do if you don't 
like either of them?", asked during the pilot study. 

3
The writer conside red that spontaneous r eaction to each 

pair of names w ould bes t represent the e v aluator 's feeli ngs. 



62 

During the testing session the investigator moved about the room, 

inspecting worksheets and encouraging slower wL,rker s. Papers 

were checked for omissions and irregularities by the investigator 

during and immediately after the testing period. On the infrequent 

occasions when refe rrals back to respondent s were necessary work­

sheets were returned and completed on the d ay of administration. 

Two weeks after the first administration of the paired-

comparison questionnaire it was readministered in all twelve classes. 

On the second occasion only minimal restatement of instructions 

was necessary. Approximately thirty minutes of class time was 

required for completion of the fir st administration in the slowest 

class. Requiring less preparatory explanation, the second 

administration was completed in less than twenty minutes. 

Sociometric Treatment Proc edures 

One week after the first administration of the paired- comparison 

questionnaire subjects in Groups A, B and C were each administered 

a different sociometric procedure. Group D served as a control 

group, rec eiving no intervening sociometri c procedure. 



63 

Group A compl eted a Moreno-ty pe positive partial-rank-order 

sociometric questionnaire , 
1 

listing in preferential order
2 

the three 

same-sex clas s mates they would most like to have as friends for a 

long time. Additional choice s were permitted. Two deviations from 

usual sociometric admin istrative and d esign practices were incorpo-

r ated. The first modifica tion was the use of a combinatio n of limited 

and unlimited choice, requiring three nominations from all students 

and permitting unlimited additional nominations. This procedure 

was adopted to allow the investigator to document the willingness of 

students (in Group B), to make addi tional negative nominations, and 

it was incorporated into each partial-rank sociometric to maintain 

a consistent experimental practice. The second modification, 

included for the same pur pose, involved provision of listings of 

classmates' names to each nominator, in this instan c e to confo r m 

to the administrative procedure in the Group C rating scale. 

The Group B sociometric questionnaire 
3 

invol ved a negative 

as well as the positive M oreno -typ e partial-rank-order sociometric 

1
Administration of the posi tive sociometri c ques tionnair e is 

presented in Appendix B, p. 153. 

2 
An example of the positive sociometric questionnaire form 

compl eted by subj e cts is presented in Appendix B, p. 154. 

3 
Administration of the positi ve -negative sociometric 

questionnaire is pre sented in Appendix B , p. 155. 



64 

questionnaire administered in Group A. 
l 

The te st form for Group B 

provide d a listing of same- s ex classmate s. Under this listing were 

two adjacent columns h eaded "Most'' and "Least" in which students 

were instructed to write, in p refe r ential order, the names of three 

classmates they would most like to have as friends for a long ti.m e , 

and the name s of three classmates they w ould l east like to have as 

friends for a long time. In both instan ces, additional choice s were 

permitted. Instructions for completion of both the positive and 

negative parts of the questionnaire were given prior to students 

making nominations. 

The Group C sociometric procedure differed from the partial­

rank questionnaire completed by Group A and Group B i n that it 

2 
required evaluation of group members on a five-point rating scale. 

On this s cale , compris ed of two positive, two negative and one 

neutral categories, subjects were asked to rate all same-sex 

classmates in terms of their a cce ptability as friends for a long time. 

3 
Subje cts were presented with a rating form and a supplementary 

1 
An exampl e of the positive -negative sociometric questionnaire 

form complete d by sub j e cts is presented in Appendix B , p. 156. 

2 
Admini stration of the rating sociome t ric que stionnaire is 

presente d in Appendix B, p. 157. 

3 ' 
The form for compl etio n of the rating sociometric questionnaire 

is presented in Appendix B, p. 159. 



list of the n am e s of individuals to be rated . Definitions of the five 

cate gori es were r e ad and d i scussed, employing a representation 

65 

of the rati ng form on the blackboard to facilitate explanati on. Each 

scale category on the form was headed by a pictorial illustration in 

the form of a "face''. The faces serve d as substitutes for wri tten 

c ategories. This was explained to the subjects who were informed 

that by writing a name under a face this would indicate how the rate r 

£el t about having that per son as a friend for a long time. For 

example, by writing a name under the face in the first category 

the rater indicated he would very much like to have that per son as 

a friend for a l ong time. Spac e was provided to permit rating all 

classmates within a single category and the a cc eptability of this 

practice was indicated. 

The fi ve cate gori es of the scal e wer e d e fin e d a s ( 1) "very 

much like as a friend for a long time , " (2 ) "like to have as a frie n d 

for a lon g ti m e , but not as m u ch as per sons pl aced in catego ry l, " 

( 3) " don' t know whether I would or would not like as a friend £or a 

long time," (4) "not like this person as a friend for a long ti me , b ut 

don't feel as strongly as I do about persons placed in cate gory 5," 

and ( 5) "very sure I would not like to have as a friend for a long 

time . " 



Choice Expectancy Procedure 

The choice expectancy procedure l was selected to examine 

the students' perceived social status. More specifically, this 

procedure was introduced to the study to permit comparisons of the 

influence of several sociometric procedures on preference expecta-

66 

tions of students. It '?'as proposed that choice expectancy responses 

would be sensitive to change in group social structure following 

administration of different sociometric procedures and that 

accentuation of negative feelings would be subsequently reflected in 

lower choice expectancy scores of least-preferred subjects. 

To provide data to examine this proposal, immediately follow-

ing the s e cond admini stration of the paired-comparison que stionnaire 

all subjects received a choice expectancy form 
2

' 
3 

on which each 

individual was instructe d to e stimate preferenc e s made by all other s 

in his subgroup on the pairs which involved his name on the post-

sociometric administration of the pair e d-comparison questionnaire. 

1 
Administration of the choice expectancy proce dur e is 

presented in Appendix C, p. 161. 

2 
An example of the choice expectancy form completed by 

subjects is presented in Appendix C, p. 165. 

3
The ordering of presentation of names on the choice expectancy 

is prese