Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Child Dev. Author manuscript; available in PMC 2014 November 21.
Published in final edited form as:
Child Dev. 1983 June; 54(3): 549–556.
doi:  10.2307/1130041
PMCID: PMC4240624

Developmental Changes in Young Children’s Conceptions of Friendship


The present study examined the development of friendship conceptions from 4 to 7 years of age. Subjects were administered an open-ended interview, a picture recognition task, and a forced-choice rating task in which they identified the most important characteristics of friendship. Common activities, affection, support, and propinquity were all found to be salient aspects of most children’s conceptions. Friendship expectations concerning affection and support increased m frequency with age, while references to physical characteristics decreased. In general, parallel findings were found on the 3 measures, although the results were not as strong on the open-ended interview. The findings suggest that children first learn the overt characteristics of the occupants of the role of friend, but as they grow older they place increasing emphasis on affectively based characteristics.

During the course of development, children’s conceptions of friendship undergo marked qualitative changes. The pattern of changes is well documented for school-aged children and adolescents (cf Bigelow, 1977, Bigelow & LaGaipa, 1975, 1980, Reisman & Shorr, 1971, Selman, 1980, Berndt, Note 1, Furman & Bierman, Note 2). When asked to describe friendships, young elementary school children emphasize the importance of concrete behaviors such as common activities and helping. As they grow older, children begin to stress the significance of more abstract characteristics of friendship, such as acceptance, understanding, and loyalty.

Much less information is available about the early development of these conceptions during the preschool years. Several investigators have asked young children to describe friends or other persons they know, but their results have been disappointing (Livesley & Bromley, 1973, Watt, 1944). Generally the children’s descriptions were sparse and lacking in coherence, they usually reported little else but the other’s physical appearance. Recently Hayes and his colleagues elicited more extensive responses by specifically asking children why they liked their best friend (Hayes, 1978, Hayes, Gersham, & Bollin, 1980). The reasons commonly given included mutual activities, general play, propinquity, and physical possessions. One limitation of these studies, however, is that the children were only asked the basis for their attraction to one specific child. Consequently, it is not clear whether young children have a general conception of friendship which transcends a particular relationship.

The limited amount of research and the mixed nature of the findings point out the need for further work on young children’s friendship conceptions. Several important issues particularly need attention. To date, no one has examined developmental trends in this age range. Previous investigators have shown that the conceptions of young elementary school children contain a broad range of expectations, including common activities, propinquity, support, and affection (Bigelow & LaGaipa, 1980, Furman & Bierman, Note 2), but it is not yet known when these expectations emerge or what the sequence of development is.

Additionally, although several authors have advocated the use of multiple methodological approaches (Berndt & Heller, in press, Damon, 1977), previous investigators have relied exclusively on open-ended interviews. It is unknown if the same results would be obtained if other methodological approaches were employed. In fact, some of the difficulties encountered in research on preschool children’s conceptions may be due to the use of interviewing techniques. Children’s responses to open-ended questions depend heavily on their expressive skills. Thus, the sparse descriptions by the young children may reflect their limited verbal skills rather than the absence of a coherent concept. If so, more positive results should be yielded from alternative methodological techniques which are influenced less by the children’s expressive ability.

Furman and Bierman (Note 2) used such alternative methodological techniques in a recent study of the development of friendship conceptions during the grade school years. Three different methods were employed —an open-ended interview, a story recognition task in which subjects heard stories about friends and then were asked what made the story characters friends, and a questionnaire in which they rated the importance of each of 10 commonly reported characteristics. In general, the three methods yielded similar results, but each also provided unique information about the children’s conceptions. The use of such alternate methodological procedures may be particularly valuable for studies of preschool children’s conceptions because of their limited verbal skills.

The purpose of the present study was to examine the development of young children’s friendship conceptions with a series of methodological approaches. First, previous research was reviewed to identify friendship expectations which appeared to be present in the preschool years or at least by the early elementary school years (Bigelow & LaGaipa, 1980, Hayes, 1978, Furman & Bierman, Note 2). Five commonly reported expectations were selected for examination: (1) common activities, (2) propinquity, (3) physical characteristics, (4) support, and (5) affection. On the basis of previous psychohnguistic research on early concept development (Clark & Clark, 1977), it was hypothesized that children would first learn to recognize the perceptual attributes of the people who could appropriately be called occupants of the role of friendship. Having once identified the overt characteristics, children may then learn the functional, social, and cultural factors which more clearly specify and define the characteristics requisite for occupying that role. Thus, it was expected that very young children’s conceptions of friends would be based on observable attributes such as propinquity or physical characteristics. The feature of common play activities is also readily observable and thus should appear early. Features of affection and support were expected to emerge subsequently, because they require the recognition of prosocial or affectively based motives and hence are less obvious. Finally, it was anticipated that the acquisition of these latter expectations would refine children’s conceptions such that concrete physical features would decrease in importance. These developmental hypotheses were examined with three different methodological approaches an open-ended interview, a picture recognition task, and a forced-choice rating task.



The subjects were 32 boys and 32 girls recruited from three day-care facilities in a large metropolitan area. The sample was dichotomized on the basis of age. The younger group consisted of 4- and 5-year-olds, the older group of 6- and 7-year-olds. The ethnic and socioeconomic composition of the sample was mixed.


The experimental procedure consisted of three parts an open-ended interview, a picture recognition task, and a forced-choice rating measure. Undergraduate research assistants individually administered each task in a small private room at each school. The younger children were seen in two 15-mm sessions and the older ones were seen in one 30-min session. All children spent the same amount of time on each task, but the divided session was implemented with the younger children to assure their continued attention. In the first session, children were administered the open-ended interview and the first half of the picture recognition task. In the second session (or second half) the children completed the picture recognition task and were administered the forced-choice rating task. Three undergraduate research assistants served as experimenters.

The order of presentation of tasks was held constant for several reasons. The interview task was always presented first so that the children’s spontaneous reports of their conceptions would not be influenced by the experimenter’s presentation of specific features during the recognition and rating tasks. Similarly, the recognition task preceded the forced-choice rating task to ensure that the children would feel free to mention more than one characteristic of friendship rather than feeling limited to mentioning only the one characteristic most important to friendship.

Dependent Measures

Open-ended interview

Previous investigators have relied on single open-ended questions to elicit children’s friendship conceptions. In the present study we used a series of four open-ended questions to provide a more complete assessment of children’s conceptions. These questions were also designed to focus on general characteristics of friendship rather than on particular friendships. The four questions were “What is a friend?”, “A friend is someone who,” “Why do people need to have friends?”, and “Tell me what you do with friends”. To ensure that the children provided complete answers, standardized probes were included after each question (e g, “Tell me more”). The procedures used to code children’s answers are described subsequently.

Picture recognition task

The interview task provided a measure of the children’s verbal production of friendship expectations, while the picture recognition task provided a measure of their comprehension or recognition of characteristics of friendship. The task was based on Furman and Bierman’s (Note 2) story recognition measure. Subjects were shown 10 sets of three pictures that depicted children engaging in various activities (e g, “these kids share with each other,” “these kids sit by each other,” and “these kids make things together”). As each picture was presented, the activity was described, then the children were asked which activities would make the pictured children friends. If they gave only one response, they were prompted (e g, “Sometimes there is more than one right answer. Do you think anything else would make them friends or just that one?”).

The pictures were all line drawings and were designed to depict the stimulus behavior but to be devoid of other unique stimulus properties. Pictures were constructed to represent two examples of each of the five features being studied. The features and specific examples were (1) affection a, love each other, b, care for each other, (2) support a, help each other, b, share with each other, (3) common activities a, play together, b, make things together, (4) propinquity a, are together a lot, b, sit by each other, and (5) physical characteristics a, wear blue hats, b, have black hair. The pictures were divided into sets that contained examples of three different features. A total of 10 sets of three pictures were constructed to represent all possible combinations of the five different features under study. The order of set presentation was counterbalanced across subjects. To ensure that the children understood the task, they were first trained with three sets of pictures depicting typical and atypical behaviors of mothers (e g, a woman taking care of children, cooking dinner for her children, and riding an elephant). The children were shown each set and then asked which pictures showed things a mother would do. Four sets of training pictures were also interspersed among the stimulus sets and were used to cue children if necessary.

Forced-choice rating task

In the picture recognition task the children were encouraged to report all of the features of friendship they had recognized. In contrast, in the final task the children were asked to identify the more important features. Children were shown 45 pairs of pictures. These pictures were identical to those used in the previous task. Each picture was labeled and then the subjects were asked to select the one that was more important for friends to do. All possible pairs of the 10 examples were presented. Three pictures depicting the activities of mothers were presented initially to teach the children the task, and four training pictures were interspersed among the 45 pairs. The order of presentation was counter-balanced across subjects. Scores on the five characteristics consisted of the number of times that feature was selected as more important.


Children’s responses on the interview and picture recognition task were coded into the following six categories: (1) affection statements of liking, loving, caring for, or other references to emotional attachment or admiration, (2) support statements specifying sharing, helping, giving support, or comforting, (3) common activities statements referring to doing things together, playing together, having fun together, or being involved in specific activities together, such as riding bicycles, (4) propinquity references to being near one another, hanging around together, living nearby, going to the same school, or otherwise being physically close, (5) physical characteristics statements describing particular physical characteristics or aspects of appearance, and (6) miscellaneous naming friends (e g, “Susan”) and other statements not included in one of the above categories. Two naive undergraduate research assistants served as coders. Interrater agreement was assessed on 30% of the interview responses and was calculated separately on each of the six categories. To correct for chance agreements, κ coefficients were calculated using the formula (P0Pc)/(1 − P0), where P0 was the percentage agreement and Pc was the percentage of chance agreements. Interrater agreement on the interview task ranged from 0.72 to 1.00 with a mean value of .85.


Saliency of Specific Friendship Expectations

First, the interview data were examined to determine whether the five characteristics of interest were commonly reported by the children. Table 1 presents the proportion of children mentioning each one. In general, these characteristics were found to be significant aspects of the children’s conceptions of friendship. The feature of common activities was mentioned by almost everyone (96%) and both the features of affection and support were mentioned by the majority of children at each age. The features of propinquity and physical characteristics were reported by the majority of the younger group (50% and 69%, respectively) but were less salient for the older children (31% and 38%, respectively).

Saliency of Friendship Expectations.

The five features also incorporated most of the descriptors mentioned by the children. Fully 85% of the children’s comments could be classified into one of these categories. Examination of the miscellaneous comments revealed that references to companionship (e g, “buddy”) and the listing of friends were the most common of these responses (7% and 5%, respectively).

Age Differences

Next the data were examined for age and sex differences. Initially, the scores on the 17 variables were subjected to a multivariate analysis of variance in which age and sex were factors. This analysis revealed a highly significant effect for age (p < .001). Additionally, the interaction between sex and age approached significance (p < .10). The main effect for sex was not significant and hence was not examined subsequently.

To determine the nature of the significant age effect, univariate analyses of variance were conducted on scores in each category. A summary of these analyses is presented in Table 2.

Age Differences in Friendship Expectations.

As predicted, the characteristics of affection and support gradually increased in importance with age. For affection, significant age effects were revealed on both the picture recognition and the forced-choice rating tasks, F(l,60) = 6.80, p < .05, F(l,60) = 20.18, p < .001, respectively. On the open-ended interview, the interaction between age and sex was also found to be significant, F(l,60) = 7.90, p < 0. Follow-up analyses, using Student Newman-Keuls tests, revealed that the boys’ scores increased with age (p < .01) but the girls’ did not.

A similar pattern of results was revealed in the analyses of the expectation of support. Highly significant increases with age were found on both the recognition and forced-choice rating tasks, F(l,60) = 15.38, p < .001, F(l,60) = 20.07, p < .001, respectively. The age effect also approached significance on the open-ended interview, F(l,60) = 3.38, p < .06. This effect was, however, moderated by a significant interaction between age and sex, F(l,60) = 7.60, p < .01. In this case, significant age increases were found for the girls (p < .01) and not the boys.

Few age effects were observed on the common activities or propinquity features. On the forced-choice rating task, however, when children were forced to rate the importance of common activities relative to affection and support, a near significant decrease with age was noted, F(l,60) = 3.74, p < .06. Apparently, the older children recognize common activities as characteristic of friendship but rank them as less important than the other characteristics.

Analyses of the physical characteristics feature revealed highly significant age effects on each of the three tasks, F(l,60) = 8.38, p < .01, on the interview, F(l,60) = 18.92, p < .001, on the picture recognition task, F(l,60) = 25.60, p < .001, on the forced-choice rating task. In all cases significant decreases were observed. It appears that the acquisition of additional features such as support and affection reduced the significance of the physical characteristics feature.

Finally, the frequency of miscellaneous responses was examined. No significant differences were observed on the interview task, F(l,60) = 1.26, p < .10, although scores on the picture recognition task tended to decrease with age, F(l,60) = 3.81, p < .06. (This response category was not possible on the forced-choice rating task).


The current findings provide important information about the early development of friendship conceptions. Previous research indicates that by grade school, children have formed conceptions based on common activities, affection, support, and propinquity (cf Bigelow & LaGaipa, 1980, Furman & Bierman, Note 2). Hayes (1978) showed that preschool children also mention some of these dimensions (common activities and propinquity) as reasons for particular friendships. The present study expands upon this work by providing more information about young children’s general conceptions of friendships. In particular, the present study used multiple methods to elicit more complete responses and examined developmental changes in general friendship conceptions.

During the current study’s interview the majority of children spontaneously reported the features of common activities, affection, and support. Although less common, the feature of propinquity was mentioned by a sizable minority (40%) as well. Consistent with these findings, children selected dimensions of common activities, affection, support, and propinquity on the recognition task. Finally, when asked to select the most important features on the forced-choice rating task, children chose common activities and affection most frequently, suggesting that they view them as the most important characteristics of friendship.

Several developmental changes in the composition of children’s friendships were observed. On each of the three measures the characteristics of affection and prosocial support showed increases with age. On the interview task slightly over half of the young children mentioned the friendship expectation of affection, but almost all of the older children reported it. Comparable trends were found with the prosocial support expectation. Moreover, on the rating task the younger children indicated that affection, prosocial support, and common activities were equally important, in contrast, the older children showed greater differentiation, rating affection and support as more important than common activities. Compared to the younger children, the older ones place greater emphasis on the affective or motivational aspects of friendly behaviors. That is, although helping and sharing refer to specific actions, they imply prosocial intentions and positive affect. The developmental decreases which were observed on the feature of physical characteristics are also consistent with this explanation. Once the children begin to emphasize the significance of affectively based expectations, their conceptions become more refined and the expectations concerning physical characteristics become superfluous.

The observed developmental changes may also reflect the children’s growing recognition of the concept of a stable relationship. Selman (1980) suggests that preschool friendships are unstable because they are based on physical characteristics or transitory play. As children become aware of thoughts, feelings, and motives underlying friendly behaviors, the concept of a stable relationship may emerge. That is, the children may begin to recognize that friendship involves more than just the current interaction but instead is an ongoing relationship which transcends any specific interchange.

In general, the present results are consistent with those found by previous investigators such as Bigelow (1977), Bigelow and LaGaipa (1975), and Hayes (1978). The major difference between the present study’s findings and previous work is in the proportion of young children mentioning the various features. Higher proportions were observed in the present study. For example, in Hayes’s (1978) study propinquity and physical characteristics were mentioned by 28% and 13% of the children, respectively, while in the present study they were reported by 50% and 69% of the young children. This difference may reflect the difference in the interviewing procedure. Hayes (1978) asked a single question (with follow-up probes), while here the children were asked a series of questions with probing after each question. If the proportions on each of the present questions were examined separately, they would be more similar to those found by Hayes (1978).

Additionally, the preschool children in this study reported affection and prosocial behaviors although previous investigators have not found these dimensions to emerge until early grade school (Bigelow, 1977). Again, this difference may stem from differences in the interview procedure. Young children are likely to interpret questions in a literal, concrete, and narrow fashion (Bierman, in press). Both Bigelow (1977) and Hayes (1978) focused their questions on one specific friendship. Children may have interpreted their questions as requests to describe the particular characteristics of or expectations for that friendship. In the present study, children were asked about friendships in general and not about any one relationship. Additionally, a number of questions were used, so that even if the phrasing of one question led to specific or narrow response by the children, the interview overall would provide a more complete assessment of their general conceptions. These differences in the interview technique could account for the observed differences in proportions, but some of the variation may also reflect differences in the coding schemes or samples in the various studies.

The current study underscores the importance of multiple measures which have been suggested (but not employed) by previous investigators (Damon, 1977). Generally, the findings were relatively consistent across the three tasks, providing evidence of the robustness of the results. The two new tasks also supplemented the findings on the interview task. Each of the measures assessed different facets of the development of friendship conceptions. The interview task measured production, the picture recognition task, comprehension, and the forced-choice rating task, the relative importance of different characteristics. When discrepancies did occur, they usually involved the open-ended interview measure. For example, for the characteristic of affection both sexes displayed developmental increases on the picture recognition and forced-choice rating task, but only the boys showed such change on the interview measure. A similar pattern of results was found on the feature of support, except that it was the girls who showed the developmental change on the interview measure. These differences may indicate that the developmental increases occur on a comprehension task (e g, the picture recognition measure) before they appear on a production task (e g, the open-ended interview). Alternatively, the interview task may simply not be as sensitive as the other two. The pictorial tasks do not require as much verbal facility of the children and thus may provide a more precise and richer means or assessing young children’s conceptions of social role.

Any recognition or rating task is prone to the criticism that the features incorporated in the task are unimportant ones in children’s conceptions. The fact that all of the features in these tasks were frequently reported in the interview indicates that this is not the case here. In fact, since it was found that almost all of the children’s descriptors could be incorporated in this list, one may conclude that the picture recognition and forced-choice rating tasks provided relatively complete characterizations of the children’s conceptions even though the selection of features to be measured had to be done m advance. Children did give a few descriptions (5%–7%) which were not captured in the prespecified categories. These primarily consisted of two types of responses (a) a listing of names of friends, and (b) a characteristic only mentioned by one or two children (e g., “a pal,” “a friend misses me when I am away”). The former type of response was given by some of the younger children. Possibly these children named specific friends because they had not yet developed a general concept of friendship that superseded specific individuals. Alternatively, these responses may have reflected an egocentric and concrete misinterpretation of the question, “What is a friend?”. Instances of the latter type of miscellaneous response did appear to reflect meaningful friendship expectations, but none were common enough to warrant coding as a separate category.

The current results suggest several directions for subsequent research. First, research is needed to determine how the different friendship expectations are related to each other. In the present study there were some suggestions that the concept of friendship may be hierarchically organized, but further work is required to delineate the structural characteristics of the concept. To date, no research has examined whether any of these or other friendship expectations are either necessary or sufficient indices for children to infer that a friendship is present. The fact that on the forced-choice task children gave the highest ratings to affection and common activities suggests that these are likely candidates for being necessary conditions for a friendship to exist, but this hypothesis requires empirical verification. Additionally, our results seem to suggest that during the early school years children begin to recognize that friendship is a relationship which transcends specific interchanges, the present evidence is inferential, however, and a direct test of this hypothesis is warranted.

Although the age range incorporated in the present study was relatively narrow, marked developmental changes were observed. This study did not, however, examine the underlying mechanisms responsible for these changes. It would be interesting to investigate the relationship between these developmental changes and changes in social cognitive variables such as perspective-taking ability. It would also be important to study the relationship between children’s conceptions of friendship and their behaviors with friends. The fact that most children’s friendship conceptions incorporated the feature of common activities provides indirect support for the hypothesis that young children’s friendships are focused around mutual play activities (Furman, 1982, Hartup, 1970, 1978, Sullivan, 1953). As children grow older, one may also expect to see indications of support and affection in their patterns of interaction with friends.

Future work should also consider alternative means of measuring friendship conceptions than those used in the present study. All of the present measures required a verbal response. Nonverbal measures are often easier for young children and thus warrant consideration. For example, children could be presented with a series of pictures of others engaged in various activities thought to be characteristic or uncharacteristic of friendship. Then they could be asked simply to point to the pictures that illustrate friendships rather than to verbalize the reasons why? Additionally, research on attributional processes suggests that children may display developmental advances earlier when stimuli are presented on videotape than when they are presented in hypothetical stories (Shultz & Butkowsky, 1977). Certainly the present study has demonstrated that measures other than the open-ended interview have great promise. In particular, the inclusion of the recognition and rating tasks served as a means of replicating the results obtained with a traditional interview, enabled a closer examination of early developmental trends, and helped demonstrate that young children have more sophisticated conceptions of friendship than previously recognized.


Reference Notes

1Berndt, T. J. Children’s conceptions of friendship and the behavior expected of friends. Unpublished manuscript, Yale University, 1979.

2Furman, W., & Bierman, K. L. Children’s conceptions of friendship. A multimethod study of developmental changes. Unpublished manuscript, University of Denver, 1981.


  • Berndt TJ, Heller KA. Predictions of future behavior, trait ratings, and responses to open-ended questions as measures of children’s personality attributions. In: Yussen SR, editor. The development of reflection. New York: Academic Press; in press.
  • Bierman KL. Cognitive development and clinical interviews with children. In: Lahey B, Kazdin A, editors. Advances in clinical child psychology. Vol 6. New York: Plenum; in press.
  • Bigelow BJ. Children’s friendship expectations. A cognitive-developmental study. Child Development. 1977;48:246–253.
  • Bigelow BJ, LaGaipa JJ. Children’s written descriptions of friendship. A multidimensional analysis. Developmental Psychology. 1975;11:857–858.
  • Bigelow BJ, LaGaipa JJ. The development of friendship values and choice. In: Foot HC, Chapman AJ, Smith JR, editors. Friendship and social relations in children. New York: Wiley; 1980.
  • Clark HH, Clark EV. Psychology and language. An introduction to psycholinguistics. New York: Harcourt Brace Jovanovich; 1977.
  • Damon W. The social world of the child. San Francisco: Jossey-Bass; 1979.
  • Furman W. Children’s friendships. In: Field T, Finley G, Huston A, Quay H, Troll L, editors. Review of human development. New York: Wiley; 1982.
  • Hartup WW. Peer interaction and social organization. In: Mussen PH, editor. Carmichael’s manual of child psychology. 3d ed. Vol 2. New York: Wiley; 1970.
  • Hartup WW. Children and their friends. In: McGurk H, editor. Childhood social development. London: Methuen; 1978.
  • Hayes DS. Cognitive bases for liking and disliking among preschool children. Child Development. 1978;49:906–909.
  • Hayes DS, Gershman E, Bolin LJ. Friends and enemies. Cognitive bases for preschool children’s unilateral and reciprocal relationships. Child Development. 1980;51:1276–1279.
  • Livesley WJ, Bromley DB. Person perception in childhood and adolescence. London: Wiley; 1973.
  • Reisman JM, Shorr SI. Friendship claims and expectations among children and adults. Child Development. 1971;42:1685–1699.
  • Selman RL. The growth of interpersonal understanding. Developmental and clinical analyses. New York: Academic Press; 1980.
  • Shultz TR, Butkowsky I. Young children’s use of the scheme for multiple sufficient causes in the attribution of real and hypothetical behavior. Child Development. 1977;48:464–469.
  • Sullivan HS. The interpersonal theory of psychiatry. New York: Norton; 1953.
  • Watt AF. The language and mental development of children. Boston: Heath; 1944.