Abstract
Friendship choice in university campuses has persisted as a perennial theme in social psychology. Contextualizing the studies with Intergroup Contact Theory can help illuminate the social dynamics of a heterogeneous community. Present work, conducted in Jawaharlal Nehru University, New Delhi, India is an addition to this burgeoning field of literature. This article attempts to distil the role of interaction in the formation of friendship, highlighting the importance of homophily in the process. Furthermore, the chosen setting is a conscious attempt to overcome the significant western bias that exists in this field of enquiry. 495 respondents have been surveyed with the help of a structured questionnaire during the course of the research accounting to a total of 1,426 dyads that formed the unit of analysis. The study found strong evidence of homophily in the group formation within the university campus, even though classroom interaction emerges as the most significant factor. This article outlines that even though the degree and nature of interaction remain identity neutral, friendship choice is highly predicated upon homophilic association. Moreover, it is noted that duration of stay in the university does not, in any significant way, influence the factors of friendship choice.
Friendship must be taken as seriously as sex, aggression and marriage. (Robert Brain, 1976: 264)
Introduction
‘Human groups lie at the heart of social life’ (Hogg, 2012). Influence of group membership, especially to primary groups, in psycho-social development is of paramount importance; and it has remained a significant contributor to the social psychology literature for over a century (Levine and Moreland, 2006). Throughout the last century, the idea of small, close-knit social groups remained a mainstay of analysing social structures and processes. Sociologists have identified family, close friends, romantic partners as types of primary social group (Forsyth, 2010). Our present endeavour, however, focuses exclusively on friendship groups.
Friendship is usually conceptualized as ‘a social relationship where people feel a personal predisposition to one another, exchange affection and establish values, norms and lines of conduct in their relationship tacitly and autonomously’ (Nedelmann, 1991: 387). However, friendship, particularly intimate friendship, is usually relegated to the realm of individual and private. This is the reason why the majority of literature on friendship is dedicated to childhood and early adolescent friendship and its impact on psychosocial development. Small groups as analytic tools are, however, not inherently constricted within the domain of psychoanalysis and personality development. Its trans-disciplinary relevance in understanding group dynamics has been recognized for a long time now (Harrington and Fyne, 2000). Despite this, psychological interests in friendship have failed to proliferate into structural analysis of society. Eve (2002) lamented that friendship has often been a neglected topic in sociology, owing to it being considered ‘personal’ as opposed to having a larger structural consideration.
Who becomes and remains friends with whom is a question that has always intrigued social scientists. Leenders (1996) found that being of same sex was the predominant factor behind best (true) friendship between school children. Analysing a large cross country (USA) sample of ‘High school and beyond’ survey, Hallinan and Williams (1989) noted that same-race friendship is over six times more common than cross-race friendship.
Another essential factor to consider in friendship choice is the preference-opportunity dynamics. In analysing intergroup friendship, one of the hardest tasks is to separate the role of homophilic preference from the opportunities of having an in-group friendship owing to greater in-group interaction (Zeng and Xie, 2008). In their study, Leszczensky and Pink (2015) inferred that friendship within the same ethnic group is more common across classrooms than within the classroom; they argued homophilic tendencies are less prominent due to increased contact. The role of spatial proximity is further evident in the work of Mouw and Entwisle (2006) when they uncovered the role of residential segregation to account for same race friendships.
This is, however, not to say that factors extrinsic to the dyad have been perennially shunned in the considerations of intimate friendship. Adams and Allan (1998) perfectly captured this sentiment when they reflected, ‘Relationships have broader basis than the dyad alone . . . these relationships need to be interpreted from a perspective that recognizes the impact of this wider complex, rather than from one that treats the dyad in relative isolation’ (Adams and Allan, 1998: 3). As McPherson et al., (2001) asserted that the most important preference is homophily, the preference of friends who are similar to oneself on salient attributes (McPherson et al., 2001).
The question remains, however, whether concerns of homophily is strictly intrinsic to friendship dyads alone or can they be used as a mirror to study larger social designs. In his seminal paper, Allport (1954) argued that the possibility of contact which permits individuals to learn about one another as individuals will lead to a reduction of prejudice. It is conceived that with increased active contact, people get to learn about the out group which in turn reduces prejudice (Schiappa et al., 2005). The ideal conditions for intergroup contact, reduction of prejudice, intergroup attitudes and cross-group friendships have been extensively researched. It remains clear that Allport’s work is highly relevant in friendship studies, even 60 years after its publication (Dovidio et al., 2003, 2008). The true significance of intergroup studies, perhaps lie not in its intrinsic values but as a mirror to society; as Pettigrew (1998) puts it, ‘Oddly, for a discipline that focuses on face-to-face interaction, social psychology rarely decomposes situations into their basic components. Allport’s attempt is a prominent exception’ (Pettigrew, 1998: 66).
A more detailed deliberation on intergroup contact theory usually arrives at a discussion on cross group friendship, and the role it can play in reducing prejudice, beyond intergroup interaction. It has been established that cross group friendship is more effective than other forms of contact (Pettigrew, 1997; Pettigrew and Tropp, 2006). Moody (2001) found strong support for the contact hypothesis in his study of American schools when he concluded that without ideal circumstances, simply ‘concentrating minorities within large schools may accentuate friendship segregation’ (Moody, 2001:679). In their meta-analysis, Davies et al. (2011) concentrated on how cross-group friendship aids in eliminating biases, and suggested employing behavioural methods for ‘a greater understanding of why’.
It is easily conceivable that within the framework of intergroup contact, friendship studies mostly engage with adult friendships. It is only at college level that an individual first truly engages with people from an outgroup. University campuses, thus, provide a unique opportunity and freedom to study intergroup relationships in a multifaceted setting. There have been several studies directed towards friendships and general social openness on university campuses. Nix (1993) who focussed on social distance between different groups, found it to be related to contact with other ethnic groups, home-size, education level, subject of study, duration of stay and nationality. Odell et al., (2005) analysed cross racial friendship using social distance and came to conclusion that ‘ . . . programming that encourages a sense of common efforts and goals are likely to be much more effective in reducing racial prejudice . . .’ (Odell et al., 2005: 303). In their analysis, Summers et al., (2002) observed, using openness to diversity and challenge scale, that in a large research university students generally have a socially inclusive mentality across social groups. Using a similar method, Pascarella et al., (1996) inferred that several factors such as living on campus, attending racial or cultural workshops, grades and peer engagement may have an effect on the degree of openness and lack of prejudice in an end of year student. In a study on American college students, it was found that students who had more outgroup friends developed less ingroup biases and intergroup anxiety at the end of college (Levin et al., 2003). In a study on making new friends in the first year of college, Fischer (2008) found that increasing racial diversity in the college increases friendship heterogeneity.
A striking feature of the review is the abundance of literature based on the United States of America. It is expected, given the melting pot culture of USA, and the attention racial diversity has received in both academic and political circles. This is, however, not to say that there does not exist other fertile grounds with sufficient demographic heterogeneity necessitating the study of homophilic association in friendship choice, cross-group friendship and contact theory. Our present work is based in India, a place in many regards significantly more culturally diverse than western societies. It is an attempt to analyse friendship choice and friendship segregation in a society characterized by several fault lines along gender, caste, religion, language and ethnicities. The main objective of this article is to analyse the effect of homophily in friendship choice in a university campus, and subsequently examine the role different demographic similarities play in friendship segregation.
Research design
The study is conducted in Jawaharlal Nehru University campus in New Delhi, India. Out of a total 8,000 (approximate) registered students, more than 7,000 reside within the university campus itself. Added to this is the fact that it is one of the most socially diverse places in the country with several significant socio-cultural fault lines dividing the student community, making it an ideal place to examine homophilic association in friendship choice.
The work is carried out with a structured questionnaire survey. A total of 495 individuals are surveyed. Respondents are asked to provide several background (demographic and socio-economic) details of self and friends, as well as university specific information such as academic or residential details. The survey is entirely self-reported and no background verification is done to authenticate the response. Each respondent is asked to give the details of up to three of their closest friends. This is done for convenience and practicality, also to mitigate the fact that no definition of friend is provided by the researcher and each individual may perceive friendship separately and having a limit within a given number will make the study more uniform. Out of 495 respondents to have mentioned at least one friend, 450 provided details for all three friends and 481 have listed atleast two friends. As such with 495 respondents, the survey finished with a total of 1,426 dyads which will form the basic unit of analysis.
Selecting around 500 people from a total pool of 7,500 students necessitated using a robust sampling mechanism to ensure sufficient diversity is achieved. The survey is conducted exclusively in the residential hostels as they are deemed the least segregated spaces (apart from gender) and the most private to obtain unbiased responses. Furthermore, having only 18 hostels made it feasible to conduct surveys in all the hostels, and draw samples proportional to the hostel population. Within the hostels, rooms are selected randomly. This method of sampling excluded the non-residential population, which is in tune with the research design of focusing entirely on residential population. A total of five non-residential students (or day scholars, as they are popularly called) are surveyed as and when they are found present in another hosteller’s room. Overall the derived sample is considered representative to the target population, both in terms of demographic and socio-economic backgrounds, as well as academic details such as course, discipline and duration of stay.
Table 1 is a concise summary of the surveyed respondents.
Sample Description (Respondent Details).
Results and discussions
As already noted, the greatest challenge of assessing homophily manifests in discerning the identity association with the increased in-group interaction. Recognizing this, subsequent discussion of the results is clubbed into two groups; demographic and socio-economic criteria representing the background characteristics of the dyad members, and the university specific parameters facilitating contact between them.
Age
This study examines a strong lineage towards same-age friends. Among the sample, 50.6% of students are within the 21–25 age group, followed by 37.6% in the 26–30 age group, while the remaining 11.6% of students are evenly split between below 21 and above 30 age groups. For the analysis, people having friends for two years more or less are considered to be in the same cohort; so 78% of dyads are friends within the same age group. 11.6% have friends who are older than them and 10.4% reported having younger friends. As Dowd (1980) pointed out friendship develops amongst age peers as they share more in common and can provide valued exchanges within the relationship (Dowd, 1980). Only the highest age group (above 30) shows disproportionately lower occurrence of same age cohort friends (with only 40.9%) while for the other age groups it is 78.8%, 84.1% and 74.2% respectively. This is to be expected as people in the highest age group are almost exclusively Ph.D. scholars who have no regular classes, and are thus less prone to interact with people from same age cohort.
Sex
Research has established that ‘social interaction in non-romantic relationships rarely crosses gender lines’ (Kalmijn, 2002: 101). In this study, across all the identity categories, sex is the only category having an institutional mechanism of clustering in the form of sex selective residential hostels. As table 2 depicts, 74% of the total respondents reported having same sex friends. The high share of same-sex acquaintances is constant across the variant, classmate or not to duration of stay in the university campus. Not only in gender segregated living spaces are people bound to choose friends from a homogeneous defined set but even in classrooms or outside hostels, they prefer to have a sex-segregated group or hang out with a same-sex friend. One hypothesis states that preferences for same sex peers arise from sex role socialization and the development of gender roles and gender identification (Martin, 1994). On a similar note, women have also reported that ‘opposite sex friendships provide less intimacy, less acceptance, less nurturance and less emotional support’ (Kalmijn, 2002: 102). It has also been observed that interaction in mixed dyads is similar to that of male dyads but female dyads have a broader intimacy and communication (Rose, 1985; Aukett, Ritchie and Mill, 1988; Sapadin, 1988). The study found that men who reported having opposite sex friends are younger-aged women and similarly the majority of women who have male friends are older men. In the survey, over 50% of the respondents having opposite-sex friends abide by this norm. This is in tandem with the findings from Booth and Hess (1974).
Gender Association.
Significant at 99.9%.
Religion
This is such an abstract group that is neither defined biologically like sex nor by appearance like race but is primarily carried forward ancestrally. Humans are compartmentalized on social, cultural, spiritual and moral lines and clubbed into some existing frameworks (Pargament, Magyar and Murray-Swank, 2005; Silberman, 2005). Across a list of socio-cultural identities that humans have, religion is probably the most fundamental and hence, it is analysed as a separate category. In the present analysis, 67.3% respondents are Hindus, followed by 12.5% Muslims, 7.7% Christians and 2.4% Buddhists. 8.3% of the people stated to follow no religion where some have mentioned atheist or even agnostic (refer to table 3). Other than the major religious groups, 1.8% is grouped as others which include Sikh, Sanamahi, Jain and indigenous faiths. Religion, unlike caste or race does not have a vertical hierarchy but tends to create an in-group norm and prejudices about other religious faiths. This leads to people flocking within a similar religious identity as there is a dominant in-group preference (Newheiser and Olson, 2012). In the study, 74% of people reported to have friends with similar religious beliefs—of which, 84.4% of Hindus have said to have only Hindu friends, but this can also be attributed to the fact that over two-thirds of the respondents are Hindus. Surprisingly, people who said they were atheist or simply had ‘no religion’ also have a tendency to flock with non-religious people. Although the study did not canvass any question on the ‘degree of religiosity’ but hanging out together during religious ceremonies is another excuse for having friends from same religious identity; like Christians participating in the Sunday mass which is held in the campus and Muslims going to nearby mosques for Friday prayers can create favourable circumstances within the in-group. The finding is in line with Dunham et al., (2014) that religion serves a unique ‘protective role in bolstering the intergroup attitude of its members’.
Religious Association.
Significant at 99.9%.
Caste
Caste is predetermined from birth, immutable across lifespan and is strictly hereditary. The caste system comprises of an ‘explicit hierarchy of groups, which are inherited by children from their parents, defined historically by traditional occupational roles and associated with distinct rights and obligations’ (Dunham et al., 2014: 312). However, in this study, we did not deal with the Hindu varna system (Brahmins, Kshatriyas, Vaishyas and Sudras) but with the caste categories as mandated by the Indian reservation system (Scheduled Castes, Scheduled Tribes, Other Backward Classes and Unreserved). With the advent of modernization, role of caste in society has relaxed but it remains significantly correlated with economic status (Thorat and Newman, 2007) and is ‘salient in social relationships’ (Das et al., 2010). The sample comprises of 40.8% of unreserved, 37.4% of OBC, 10.7% SC and 9.3% ST as summarized in Table 4. A strong correlation is observed amongst people having friends from the same caste category. 62.8% of UR have reported to have only UR friends, followed by 60.3% OBC-OBC dyad and 64.1% ST-ST dyad. The chi-square value (712.119) is significant at 99.9% for the degree of association between caste categories. This strong association can also be attributed to the several unofficial caste organizations and tribal groups operating in the university campus; like the United OBC forum, the Oraon Group, or even the Chakma and the Naga Association. These informal groups unite the members of a particular caste identity or tribal identity and become a forum for interaction. Regular meetings and gatherings manifest into friendships and some dyads even become close friends.
Association across Caste Categories.
Significant at 99.9%; Df = 9.
Region
The selected university has representations from all socio-cultural zones and encompasses within itself the diverse nature of the country. To avoid statistical dilution, 36 federal units are clubbed into six geographical regions to mimic the broad cultural realms; namely, North-Western, Central, Eastern, North-Eastern and Southern (Table 5). Our survey, although not exclusively focused on languages, confines to the data from https://www.flingjnu.com/ as 42.6% of the respondents reported Hindi to be their mother tongue, distantly followed by Bengali, Urdu and Malayalam. This staggering linguistic diversity that appears to be one of the fascinating aspects of the social climate also renders its use as a statistical tool to be tenuous at best. With 39 distinct mother tongues reported by 495 respondents belonging to four major linguistic families dissuaded us from incorporating language within the fold of analysis and replace it with the state of domicile. This makes sense as India’s state boundaries are drawn predominantly along linguistic lines, and they can near perfectly represent the major language zones of the country. In this study, people coming from different identity groups, instead of making friends from diverse backgrounds, were observed to clutter with people from similar backgrounds. This is presumably on account of the ease in communication, food habits, faith, beliefs and even adhering to similar norms and prejudices.
Regional Association.
Significant at 99.9%; Df = 16.
Income
Social status shapes ‘people’s subjective views and their habitus’ (Bourdieu, 1984), therefore, ‘people of similar objective positions are more likely to feel subjective affinity towards each other’ (Pellandini-Simanyi, 2017: 593). After friendship formation, if there arises a divergence in class positions, then ‘friendship tends to dissolve’ (Burt, 2000). In this study, it was not possible to capture the actual income of the respondents and their friends, so we highlighted the perceived income on a Likert scale (high, higher middle, middle, lower middle, low). The study assumes that since a particular individual will be noting their own, as well as their friends’ income, so it will negate subjective perceptions. As per table 6, over 50% (53.6%) of the respondents reported to have friends of a similar economic background and 31.2% consider their friends to be of a higher income group than their own. The data further reveals that a minimal of 10% dyads have a higher divulgence in the economic differences; the rest reported to have maintained a ‘cross-class friendship’, which is within a singular cell movement (i.e., middle income respondent having a higher middle income friend or a lower middle income friend). Sayer’s (2005a, 2005b, 2011) seminal work highlights that income status is an important aspect to form a homophily not just because of the instrumental benefits of belonging to a particular class but ‘because class distinctions are enmeshed with moral distinctions of the relative worth of people and competing definitions of a good life’ (Pellandini-Simanyi, 2017: 593).
Economic Association.
Significant at 99.9%; Df= 16.
Friendship formation in classrooms
A classroom is an interactive forum where students from different socio-economic background are bound in a defined space for the purpose of learning. In a place where there are people from different socio-cultural settings, there is evidence of the formation of dyads within a relatable milieu. Putting forward a version of Rational-Choice Theory, Windzio and Bicer (2013) commented ‘ethnic homophily is more pronounced in high cost than in low cost situations’. It argues that maintaining friendship within a classroom across ethnic boundaries costs less whereas one has to pay more to maintain cross-friendships outside of the classroom.
This survey reveals that 40% of people have responded to have first met each other in a classroom but then they maintained friendships in the same cultural demographic; be it sex, mother tongue, religion or state of domicile. This is such a space where there is no demarcation of socio-economic-cultural boundaries and yet people tend to cluster within a similar identity. This space is different from the various caste organizations like the United OBC Forum or the cultural associations like the Rajasthani Freshers’ Team, Bengali New Year committee or the Chakma association; where people of a certain identity cluster together.
Furthermore, table 7 contrasts the association categories between friendship between and outside classmates. 56% of the dyads in this study are classmates, yet 60% of them are friends from the same region, 72% are from the same religion, 57% share the same mother tongue as well as caste group and 73% of the classmates are friends within the same sex. As table 7 notes, homophilic association appears to be indifferent whether or not a friend is a classmate with both classmate as well as non-classmate friends having almost similar degrees of homophilic association. Table 7 clearly indicates that academic interaction as part of being classmates cannot transcend the effect of homophily and even within classrooms, people found friends who share common identity with them. It has also been argued that peers of same socio-cultural background are ‘likely to share similar experiences, attitudes and values; which leads to increased mutual understanding and shared interests, and thus to stronger emotional support and social affirmation’ (Leszczensky and Pink, 2015: 19).
Comparing Classmates and Non-classmates.
We asked our respondents where they first met their reported friend in order to ascertain the importance of identity spaces in bringing people together. As can be noted from table 8, cultural activity, which is taken as a proxy for identity spaces, has been reported by only 3.9% of the dyads as the first meeting point. An overarching dominance of classroom (39.6%), academic activities (8.1%) and hostel (9.8%) highlights identity neutral spaces as the first meeting point among potential friends. This finding, coupled with the high degree of homophily thus establishes that although people meet in an identity-neutral forum, they tend to make friends who share similar characteristics of their own.
Location of First Meeting.
Friendship formation in hostels
‘Residential segregation captures the spotlight . . . from discrimination in housing market institutions and attitudes to neighbourhood diversity’ (Ellis et al., 2004: 620). The hostels, thus present themselves as an integral component of non-academic interaction among the students. In the campus, hostels are mostly situated in clusters which can be regionalized on a broader scale. Since there are 17 hostels in the campus, we have grouped them in five regions; Uttarakhand, Dakshinapuram, Paschimabad, new hostels and faraway hostels.
Table 9 reflects that a considerable number of hostel inmates are close friends; mostly if they are roommates or if they share the same washroom, mess or corridor. Sharing the same physical space makes people see each other more often and on a regular basis and it also leads to a certain level of interaction and engagement in dialogue; which in turn develops a level of intimacy amongst the residents. A similar work by Whyte (1956) after surveying the residents of the Park forest, deduced that friendship was predominantly centred around the residential blocks that were situated in close proximity. They think interaction is a function of both physical and functional distance with physical distance being the degree of actual spatial separation and functional distance was the extent to which contact was likely because of environmental features (Paulus and Nagar, 1987). Although only 207 respondents have reported to have met their friends for the first time within residential premises, an interesting pattern emerges from this number. Most of the dyads are members of a similar identity group, be it place of origin, cultural affinity or mother tongue.
Friendship in Residential Places.
Conclusion
India is undoubtedly one of the most socially diverse places on earth (Hardgrave Jr., 1993). Major urban centres of the country often exhibit a staggering variety of class, caste, religion and ethnicity making intergroup contact an intriguing field of enquiry. This endeavour is an attempt to shed some light in the neglected field of homophilic association in friendship group formation in the Indian society. Situated at the heart of the capital city, Jawaharlal Nehru University presents itself as the ideal setting to realize this potential, being a microcosm of India within its campus and providing ample opportunities for unhindered intergroup contact. Present work is an attempt to ascertain the role of homophily in the formation of small friendship groups.
Throughout the conceptualization and realization of the project, one of our primary aims has been to isolate the role of interaction in the formation of groups. The questionnaire is carefully crafted to assess the relative importance of background identity characteristics and interaction within the university in selecting friends. Interaction in the academic sphere has emerged as a key element in the formation of friendship. As noted earlier, 795 out of 1,426 dyads were between classmates. Moreover, academic spaces also occupied the largest share of the pie in the first met category with 39.6%. Classrooms, however, are far from being a uniform space, and all the socio-cultural-economic diversities of the university also manifest themselves within the confines of the classrooms. Thus, although classroom interaction was found to be the most significant in the formation of friends, we tried to look into the identity association among classmates and found strong evidence of homophily in the choice of friends within the classrooms. Furthermore, degree of homophily is found to be the same among both classmates and non-classmates, implying either classroom interaction is not geared towards or not being pervasive enough to penetrate the identity association in choosing friends. Overall, we found a significantly greater proportion of friends sharing similar identity characteristics, with gender, religion and state of domicile (taken as a loose proxy for language) being the most dominant ones. More importantly, it is found that even though the majority choose their friends in identity neutral forums, they seem to form personal bonds with people they share a homophilic association with.
It was also found, as table 10 illustrates, duration of stay in the university campus has little to no effect on the degree of identity association while choosing friends. None of the identity traits appear to have any bearing on the duration of stay in the university, signifying the degree of homophily remains unchanged between the newcomer and students who have been at the institute for a prolonged period.
Friendship Association and Duration of Stay.
Overall, the study presents a strong case for homophily being the dominant factor in selecting friends. Findings from this study can offer significant insights and further avenues of research in other interactive spaces including residential segregation, formation of ethnic enclaves and even ghettoization in larger cities. The same research design used here can be forwarded to other universities and larger colleges, especially to spaces with the existence of a majority demographic community to shed more light in this arena.
