Homophily – Classical Origins

            In the sociological literature, the homophily principle is most often traced back to Lazarsfeld and Merton’s (1954) terminological innovation in their study of Hilltown and Craftown.  However, Lazarsfeld and Merton acknowledge that the earlier anthropological notion of “homogamy” (or homophily in the marriage relation) formed a conceptual template for their own work (Burgess and Wallin 1943 and Ellis 1936, quoted in Lazarsfeld and Merton 1954: 23).  They also acknowledge that the tendency of social ties to form between socially similar individuals had already been empirically noted and theoretically implied (Lazarsfeld and Merton: 36; see also Park and Burgess 1921; Simmel 1955; Simmel 1971).

            In a more popular reference, Lazarsfeld and Merton (1954: 37), quote the proverbial expression of homophily, “birds of a feather flock together.”  This proverb is attributed in its modern form to Robert Burton (1927 [1651]: 622).  Like Lazarsfeld and Merton, Burton acknowledges some of his own conceptual predecessors.  Burton’s inspiration is taken from the ancient Western classical tradition.  These include Epicharmus (who held that “the Pig is regarded by the Pig as the most beautiful thing in the world, the Dog by the Dog, the Cow by the Cow, the Ass by the Ass”), Theocritus (who wrote that “the grasshopper loves the grasshopper, likewise the ant”) and Diogenianus (whose own “Jackdaw percheth beside Jackdaw” comes closest to the modern proverb) (quoted in Burton 1927 [1651]: 622).

            Classical notions of homophily are, in fact, even more extensive than Burton notes. Aesop’s fable of the two pots describes social distance in physical terms:

 

A river carried down in its stream two Pots, one made of earthenware and the other of brass.  The Earthen Pot said to the Brass Pot, “Pray keep at a distance and do not come near me, for if you touch me ever so slightly, I shall be broken in pieces, and besides, I by no means wish to come near you.”

 

Equals make the best friends (Aesop 1888).

 

In a more intellectual vein, Aristotle’s Rhetoric and Nichomachean Ethics describe the general law that “we love those who are like themselves” (Aristotle 1934a: 1371; Aristotle 1934b: 1155)[1].  Empedocles is well known for his claim that “like seeks after like” (quoted in Stevenson 1948: 2).  Plato notes in Phaedrus that “similarity begets friendship” (Plato 1968: 837)[2].  Arguing a court case, Cicero notes that defendant Caius Verres and Quintus Apronius became friends due to “the similarity of their manners” (Cicero 1903: 23).  Traveling back even further in time to the Greek oral tradition, Melantheus of Homer’s Odyssey belittles Odysseus: “Lo, now, in very truth the vile leads the vile.  As ever, the god is bringing like and like together” (Homer 1919, Ch. 17).  Elsewhere in the same work, Telemachus supposes that he and Nestor’s son are friends in part because they are of the same age (Homer 1919, Ch. 15).

            These accounts of homophily describe psychological choice of similar others; however, the same works also suggest that friendship occurs as a result of simple contact.  Immediately following his remark that birds of a feather flock together, Burton continues that “for custom, use and familiarity, as if a dog be trained up with a Lion and a Bear, contrary to their natures, they will love each other” (Burton 1927: 622).  In the Nichomachean Ethics, Aristotle quips that “familiarity breeds fellowship” (Aristotle 1934b: 1161).  And Telemachus’ full reckoning of his friendship with Nestor’s son includes contact because of their fathers’ friendship and a journey to be taken together that “shall yet more stablish us in oneness of heart” (Homer 1919, Ch. 15).  In other words, the Classics describe not only psychological homophily but also organizational constraint.  The mention of both ideas in recorded human thoughts thousands of years old belies the impression that the network insight in general and the homophily principle in particular are new or counterintuitive.  Rather, they are part of a long intellectual tradition that has gained new currency with the rise of network analysis and structural theory in the academy.


References

 

Aesop.  1888.  Aesop’s Fables, Literally Translated from the Greek, Townsend translator.  New York: The American News Company.

 

Aristotle.  1934a.  “Rhetoric.”  Vol. 22 in Aristotle in 23 Volumes, Rackham translator.  Cambridge, MA: Harvard University Press.

 

Aristotle.  1934b.  “Nichomachean Ethics.”  Vol. 19 in Aristotle in 23 Volumes, Rackham translator.  Cambridge, MA: Harvard University Press.

 

Aristotle.  1934c.  “Eudemian Ethics.”  Vol. 20 in in Aristotle in 23 Volumes, Rackham translator.  Cambridge, MA: Harvard University Press.

 

Breiger, Ronald.  1974.  “The Duality of Persons and Groups.”  Social Forces 53: 181-190.

 

Burton, Robert.  1927 [1651].  The Anatomy of Melancholy.  New York: Farrar and Rinehart.

 

Cicero, Marcus Tullius.  1903.  “The Third Book of the Second Pleading in the Accusation Against Caius Verres: On the Court Relating to Corn.”  in The Orations of Marcus Tullius Cicero, Yonge translator.  London: George Bell and Sons.

 

Ellis, Havelock.  1936.  Sex in Relation to Society.  New York: Random House.

 

Feld, Scott.  1981.  “The Focused Organization of Social Ties.”  American Journal of Sociology 86: 1015-1035.

 

Homer.  1919.  The Odyssey with an English Translation, Murray translator.  Cambridge, MA: Harvard University Press.

 

Kaduskin, Charles.  1966.  “The Friends and Supporters of Psychotherapy: On Social Circles in Urban Life.”  American Sociological Review 31: 786-802.

 

Lazarsfeld, Paul F. and Robert K. Merton.  1954.  “Friendship as a Social Process: A Substantive and Methodological Analysis.”  pp. 18-66 in Freedom and Control in Modern Society, Berger, Abel and Page eds.  New York: Van Nostrand.

 

Plato.  1968.  “Laws.”  Vol 11 in Plato in Twelve Volumes, Bury translator.  Cambridge, MA: Harvard University Press.

 

Simmel, Georg.  1955 [1908].  Conflict, the Web of Group Affiliations.  Glencoe, IL: Free Press.

 

Simmel, Georg.  1971[1908].  “The Problem of Sociology.”  pp. 22-35 in Georg Simmel, On Individuality and Social Forms, Donald Levine ed.  Chicago: University of Chicago Press.

 

Stevenson, Burton E. 1948.  The Home Book of Proverbs, Maxims and Familiar Phrases New York: MacMillan.

 

Xenophon.  1979.  “Memorabilia.” Vol. 4 in Xenophon in Seven Volumes, Marchant translator.  Cambridge, MA: Harvard University Press.

 


Some Recent Articles About or Relevant To Homophily (Outside the Major Journals)

 

Burt, Ronald S. 2000.  “Decay Functions.”  Social Networks 22: 1-28.

            Burt documents the underappreciation in network studies of tie disappearance, the converse of tie formation.  Studying reported ties between bankers over the course of 4 years (making for a longitudinal study, also rare in networks), Burt attempts to explain tie decay.  Burt finds evidence for an organizational effect on tie decay: both working in the same department and the number of shared ties reduce the risk of decay.  Age, rank and gender similarity also significantly (although intermittently so across models) reduce the risk of tie decay.  Interestingly enough for the choice vs. constraint issue, the inclusion of shared ties (or “embedding”) variables renders the effects of age and gender similarity insignificant.

 

Farmer, Thomas and Elizabeth Farmer.  1996.  “Social Relationships of Students with Exceptionalities in Mainstream Classrooms: Social Networks and Homophily.”  Exceptional Children 62: 431-450.

            Farmer and Farmer describe peer networks of 79 students in three third- and fourth-grade classrooms.  The researchers asked students to identify (if any) by name the sets of students in their classrooms that “hang around together a lot.”  The resulting data is listed in a set of tables, organized by classroom and cluster.  The clusters described by the authors are organized strongly by gender and race, as well as by labels of “prosocial,” “antisocial,” and “shy” offered by peers.  However, Farmer and Farmer argue that academic skill is not an important organizing factor for these network clusters.

 

Gilly, Mary, John Graham, Mary Finley Wolfinbarger and Laura Yale.  1998.  “A Dyadic Study of Interpersonal Information Search.”  Journal of the Academy of Marketing Science 26: 83-100.

            In two studies, the researchers asked students in undergraduate marketing classes or university employees (“Seekers”) to select an individual (“Source”, not necessarily in the class) for advice regarding the purchase of a product.  The researchers hypothesized that the greater the similarity of Seeker and Source of advice, the more influence the Seeker would report that the Source’s advice would have on deciding whether to purchase a product.  Two kinds of similarity, demographic (gender, education, age) and attitudinal (life outlook similarity as reported by Seeker), were measured.  Results: to the extent that demographic similarity was significant at all (not sig. in Study 1, only sig. in Study 2 for certain kinds of purchases), its effect on influence was negative.  On the other hand, attitudinal similarity was both more consistently significant and positive in its effect upon reported influence.  In other words, attitudinal similarity but not demographic similarity leads to perceived influence.

 

Hinds, Pamela, Kathleen Carley, David Krackhardt and Doug Wholey.  2000.  “Choosing Work Group Members: Balancing Similarity, Competence and Familiarity.”  Organizational Behavior and Human Decision Processes 81: 226-251.

Hinds et al. argue that “homophily increases the ease of communication and improves the predictability of behaviors and values.”  It therefore makes [rational?] sense for individuals in a group to choose new members who are like the current members, since that decreases the investment required for interaction.  The researchers gathered information from 33 undergraduate software development groups at Carnegie Mellon University over four years.  Every year, each member filled out questionnaires about who she or he would like (from the larger community of groups) to bring into his or her group.  Noting McP-SmL 1987’s argument about induced homophily and observing the age and education homogeneity of the groups, the researchers limited their test of homophily’s effect on gender, race and “role” (manager, programmer, analyst and technical writer being the four roles)[3].  Relational information was entered in matrix form, and QAP regression was used to analyze the impact of independent-variable matrices upon the dependent-variable matrix (a really nice description of what QAP does here).  Finding: no effect gender homophily or “role” homophily in choosing a work partner; however, individuals are more likely to choose individuals of the same race.

 

Louch, Hugh.  2000.  “Personal Network Integration: Transitivity and Homophily in Strong-Tie Relations.”  Social Networks 22: 45-64.

Studies transitivity (ties between alters) in the ego networks of the 1985 GSS.  With the exception of gender (which is significant but in the direction opposite that predicted), demographic similarity of alters is significantly related to the probability a tie is present between them.  Another set of significant predictors of a tie between alters is the sharing  of some associational “focus” between alters.  The “focus constraints” (both alters being kin, neighbors, coworkers or group members) made the only set of variables that were consistently significant and in the predicted direction across all models. 

I was really impressed by Louch’s study – so, of course, I’ll pick on it J.  Interestingly, anomalous results for gender and age (a curvilinear relationship) are accounted for with “focus” explanations – work-based relationships, the marital association, etc.  In fact, when the marital association is accounted for, the relationship between gender similarity and alter tie formation becomes “weak” (insignificant? we don’t know – the model isn’t shown).  This could be interpreted as a vindication of constraint over choice.  Louch notes the problem with the McP-SmL 1987 approach to contrasting choice and constraint homophily (AKA “focus constraints” and “homophily”): when information on the full set of organizational affiliations isn’t gathered, it’s possible that any apparent choice homophily effects are actually the expression of unobserved organizational effects.  Louch moves on and, since he isn’t able to fully specify the set of individuals’ affiliations, asserts that the effects of demographic similarity affirm a choice perspective and even the strong human agency described by Emirbayer.  So the problem of distinguishing between choice and constraint isn’t really solved here: how might it be better addressed?

 

Rao, Nagesh and Peer Svenkerud.  1998.  “Effective HIV/AIDS Prevention Communication Strategies to Reach Culturally Unique Populations: Lessons Learned in San Francisco, U.S.A. and Bangkok, Thailand.”  International Journal of Intercultural Relations 22: 85-105.

 

Comparing more effective HIV/AIDS intervention programs to less effective ones, the researchers found that programs using counselors who were similar in some fashion (San Francisco: race/ethnicity; Bangkok: occupation) tended to be more effective – measured by an index based on program longevity, collaboration with other organizations, operators’ subjective assessment of success, experience of staff, grant money received, expenditures, length of typical interaction with clients, size of staff, number of clients.  Based on this index, it’s not clear whether it’s effectiveness or bigness or program integration into the larger social service community that’s being measured.  In fact, the results could be an indication that programs that interact with other programs and receive lots of grant money (and therefore are big) tend to employ certain strategies.  In that case, the homophily strategy would be a cultural element diffusing between organizations.

 

Stackman, Richard and Craig Pinder.  1999.  “Context and Sex Effects on Personal Work Networks.”  Journal of Social and Personal Relationships 16: 39-64.

            Stackman and Pinder collected egonetwork data from three companies in Canada to test a series of hypotheses regarding the work-based networks of men and women.  They find that the work-related ties of men (the more numerous category) are 67% to men and 33% to women, while women in the same companies report that just over half of their work-related ties are to men while just under half of their ties are to women.  This is not surprising given that women are a minority in the companies studied.  Stackman and Pinder cite McP-SmL 1987 and declare these results to be supportive of induced homophily.

            Stackman and Pinder also look at “expressive” networks of friendship in the same companies.  They find that these networks display more homophily: men’s alter friends are 71% male and 29% female, while women’s alter friends are 40% male and 60% female.  Stackman and Pinder declare these results to be supportive of choice homophily.

 

 

Volker, Beate and Henk Flap.  1997.  “The Comrades’ Belief: Intended and Unintended Consequences of Communism for Neighbourhood Relations in the Former GDR.” European Sociological Review 13: 241-265.

            Think Everything in Its Path meets To Dwell Among Friends (now wouldn’t that make a great combined book title?).  The authors note that, in the German Democratic Republic (East Germany), communist rulers determinedly relocated individuals in order to create class-heterogeneous neighborhoods.  Residents of those neighborhoods faced a choice: either interact with dissimilar others or interact less with neighbors.  Volker and Henk found that in class-heterogeneous neighborhoods (in Leipzig and Dresden), there was little cross-class interaction, even though respondents were able to supply information about their neighbors to researchers.  Instead, networks became smaller as ties to neighbors withered.  In other words, regardless of spatial proximity and awareness of others, GDR residents did not form heterogeneous ties.



[1] The full quotations follow.  “All things akin and like are for the most part pleasant to each other, as man to man, horse to horse, youth to youth.  This is the origin of the proverbs: The old have charms for the old, the young for the young, like to like, beast knows beast, ever jackdaw to jackdaw, and all similar sayings” (Aristotle 1934a: 1371).  “But there is much difference of opinion as to the nature of friendship.  Some define it as a matter of similarity; they say that we love those who are like ourselves” (Aristotle 1934b: 1155).

Note that Aristotle also identifies the contradictory law that “opposites attract” (Aristotle 1934b: 1155; Aristotle 1934c: 1240).  It would therefore be inappropriate to see Aristotle as strongly endorsing the notion of homophily.  Rather, he was discussing a debate carried out by others during his lifetime (see part four of Xenophon’s dialogue between Socrates and Critobulus in Xenophon 1979).

[2] Like Aristotle, Plato also notes that “friendship occurs between opposites” (Plato 1968: 837).

[3] The sample had a quarter to a fifth women, and a ninth to more than a third nonwhite participants, depending on the year.