VISUALIZING SOCIAL GROUPS

Linton C. Freeman, University of California, Irvine

2143 SSPA, School of Social Sciences, Irvine, CA 92697-5100

Key Words: Social Networks, Visualization almost 70 years. And they continue to do so. In this
paper I will review some of the earlier work and present
Jacob Moreno [18, 19] introduced the idea of an up to date description of contemporary approaches to
using visual images to reveal important features of visualization.
social patterning. As he ([20], pp. 95-96) put it: Much of the earlier work was based on the
intuitive view that social connections are binary, or
A process of charting has been devised by the on/off. From this perspective, a pair of individuals is
sociometrists, the sociogram, which is more either linked or not linked. Some of the older and much
that merely a method of presentation. It is first of the newer work, however, has been based on a
of all a method of exploration. It makes quantitative conception of social links. Pairs of
possible the exploration of sociometric facts . . individuals are seen as relatively more or relatively less
. It is at present the only available scheme strongly linked. These two approaches tend to yield
which makes structural analysis of a
different kinds of group images. They will be treated
community possible.
separately in the next two sections.
Although they might not agree with the precise terms of IMAGES OF BINARY STRUCTURES
Moreno’s statement, most contemporary structural
analysts would still subscribe to the view that visual Moreno viewed social linkages as binary. To
images continue to play a key role in both exploring reflect that binary view, he constructed his images out
and presenting the patterns displayed by structural data. of points and lines. Individuals were pictured as points,
Among the many kinds of structural patterns and social ties linking pairs of individuals were shown
that have been explored and presented visually, one as lines connecting pairs of points. One of his early
stands out. Structural analysts have devoted an images—one designed to show group structure—is
enormous amount of attention to the problem of shown in Figure 1.
uncovering social groups. This attention to the group Asking children to choose two others that should
concept makes sense, since groups have been a concern be assigned seats next to them produced the data.
of sociologists from the beginning (see Durkheim [4], Directed lines indicate choices. Moreno drew all the
Tönnies [27], Simmel, [25] and Cooley [3]). boys (triangles) on the left and the girls (circles) on the
All of these early sociologists saw groups in right. This arrangement emphasizes the fact that there
structural terms—as a patterning of interaction. are many links within each gender, but only one link
Freeman and Webster [10] described this view: bridges the gender gap. Boys and girls, in effect,
formed almost completely distinct groups.
. . . whenever human association is examined,
Other investigators adopted Moreno’s general
we see what can be described as thick spots—
relatively unchanging clusters or collections of approach, but many were uncomfortable with his ad
individuals who are linked by frequent hoc method for locating points. Since it did not involve
interaction and often by sentimental ties. any well-defined procedures, Moreno’s method meant
These are surrounded by thin areas—where that no subsequent investigators could reproduce the
interaction does occur, but tends to be less results of earlier investigations. For that reason, Bock
frequent and to involve little if any sentiment. and Husain [1], Proctor [22], Laumann and Guttman
[15] and Levine [16] all introduced computational
In addition, group members are believed to vary in procedures that were designed to locate points in such a
terms of the degree to which they are involved in group way that members of the same group would
activities. Some are core members; some are automatically be placed in close proximity to each
peripheral. other. These procedures were all based on one of two
Visualization and the group concept, then, form a approaches: 1) multidimensional scaling (Kruskal and
natural pair. Visualization is a tool for structural Wish [14]) or 2) singular value decomposition (Weller
analysis, and group—in the sociologists’ sense—is a and Romney [30]).
structural concept. Investigators have been using visual
tools to uncover and to communicate about groups for
 Instead of drawing a point and line image with
points placed in arbitrary locations, we can use one of
our standard computational procedures to construct an
image of the game-playing matrix. Let us begin with
multidimensional scaling (MDS).
There are various versions of MDS, but they are
all essentially search programs. They accept as input a
matrix of proximities or a matrix of distances. Then
they seek to arrange the points in such a way that the
distances between pairs of points in the image
correspond to the distances between individuals in the
data matrix. With binary data, the aim is to place points
close to each other if they are associated with
individuals who are directly connected and to place
points at a greater distance from each other if they are
associated with individuals that are not directly
connected. The resulting picture should display the
Figure 1. An Attraction Network in a Fourth structure contained in the data.
Grade Class (from Moreno [19], p. 38). The investigator begins by specifying how many
spatial dimensions the final image should have. For
We can learn something about how these two actual viewing, one, two or three dimensions are
procedures work by applying them both to a single data desirable. The program then produces a picture with
set. In the 1930s Roethlisberger and Dixon [23] the desired number of dimensions and reports an index
directed an extensive study of the organization of a part of stress that tells how much distortion the picture
of the Western Electric Company. As a part of that produces.
study, they placed an observer for several months in a A two-dimensional MDS image of the game-
“wiring room” where telephone switches were playing data is shown in Figure 2. The reported stress
constructed. The observer collected ethnographic data level is 0.0, so the image turns out to reproduce the
and described interaction patterns among nine wiremen, patterning of who played games with whom quite well.
two solderers and an inspector in detail. The same And it is fairly obvious from looking at the figure that
observer also collected systematic data on various kinds these individuals are divided into two groups. There
of links among the workers. One data set recorded who are a good many ties linking the individuals within each
played games with whom during off-hours. The data group, and only one cross-group link.
are shown in Table 1.

Figure 2. MDS of Game Playing at Western

Electric (Raw Data).

This analysis was based on proximities. An

alternative standard practice creates a distance-based
image by calculating the graph theoretic distances—
based on the shortest paths linking each pair of points.
An MDS based upon these distances is shown in Figure
3.
Table 1. Game Playing at Western Electric
(from Roethlisberger and Dixon [23]).
 game-playing data. The image is shown in Figure 3.
The analysis was generated by a version of SVD called
principal components analysis. But any other version
would have produced much the same image.
This image seems again to divide everyone up
into two clusters. Here, however, some sets of
individuals are collapsed; they are represented by a
single point. This is the result of the fact that the
Figure 3. MDS of Game Playing at Western collapsed points are structurally equivalent; they
Electric (Graph Theoretic Distances). exhibit the same pattern of connections with the same
others.
So, we have two images that seem to find
The image in Figure 3 is very like the one shown groups, but it would be nice to have some baseline to
in Figure 2. Going through the intermediate step of compare them to. At this point all we know is that the
calculating graph theoretic distances seems to have had algorithms both produce structures that seem to divide
little impact here. This is probably the result of the fact everyone up into two clusters, but we no nothing about
that we are using non-metric MDS and it is producing whether these clusters represent groups or not. What
only a monotone, not a linear, mapping of the input we need is some sort of criterion that we can use in
matrix to the visual image. evaluating these displays.
We can also use singular value decomposition Two such criteria are available: 1) groups as
(SVD) to assign locations to the points. Like MDS, specified in ethnographic reports and 2) groups based
there are several versions of SVD. But they are all on computations resulting from formal specification of
alike in that they transform a data matrix into its basic group properties. I will use both of these criteria to
structure. That transformation maps each observed evaluate the images produced by MDS and SVD.
variable (in this case one variable is associated with First, let us examine the ethnographic report
each of the 12 workers) into a new variable, or vector. based on the observer’s experience in the wiring room.
These new vectors are produced in such a way that The observer concluded that the workers were divided
most of the variance in the original data is associated into exactly two groups. One contained four of the
with the first new vector, next most with the second and wiremen (designated W1, W2, W3 and W4), a solderer
so on. (S1) and an inspector (I1). The other also contained
The result is a new matrix of the same size as the four wiremen (W6, W7, W8 and W9) and the other
original, but arranged in such a way that each solderer (S4). He concluded, moreover, that each of the
successive variable accounts for less variance. If we groups had core and peripheral members. In the first
are to produce a good image of structure of the data, the group, W3 was a “leader” and W2 was “marginal.” In
first two or three new vectors account for most of the the second, W6 “was not entirely accepted by the
important patterning of the original data. When that group” and S4 “was socially regarded as inferior.”
occurs, SVD will produce a good image of the structure Finally, the remaining wireman, W5, reportedly was not
of the original data. fully accepted by either group; he was designated an
“outsider.”
In broad terms, this report corresponds to both
figures quite closely. In both cases the clusters shown
match those produced by the ethnographic description,
with the exception of W5 who was assigned a
peripheral position in the first group in both images.
W3 is certainly put in the center of the group in Figure
2, but is not singled out in Figure 3. W2, W6 and S4
also appear to be more marginal in Figure 2. In this
case, then, both MDS and SVD succeed in reflecting
the overall structure reported by the ethnographer, but
MDS captures more of the subtle detail.
Figure 3. SVD of Game Playing at Western Luce and Perry [17] introduced the first formal
Electric. definition of social groups. They defined a clique as a
maximal complete subgraph. That idea had, and
continues to have great intuitive appeal. Yet, when
In this case, the first two vectors of the new cliques are calculated and compared with groups
matrix captured almost 90% of the variance in the defined by ethnographers or by the participants
themselves, they consistently show four weaknesses: 1)
there are too many of them, 2) they are too small, 3) IMAGES BASED ON QUANTITATIVE
they overlap too much and 4) because they are ASSUMPTIONS
complete they can display no internal structure.
There is, however, a procedure based on Galois When network data come in the form of
lattices that has been used to define bounded groups in quantitative indexes of social proximities or distances,
terms of the patterning of overlap among cliques we would like to continue using MDS and SVD to
(Freeman, [6]). The Galois lattice constructs a partial uncover groups. As an example of quantitative network
order among the individuals. That order for the data, consider the observations of 13 male dolphins
Western Electric workers is shown in Figure 4. The shown in Table 2. Records of the frequencies of pairs
five cliques, labeled 1 through 5, are shown at the top of dolphins observed swimming together were made
of the figure. The 12 individuals are shown below. over a period of time in a clear shallow bay.
Clique overlaps are shown in terms of their
memberships; Cliques 4 and 5, for example, share three
members, W7, W8 and W9, and each has a single
defining member (S4 for clique 5 and W6 for clique 4).

Table 2. Male Dolphins Observed Swimming

Together (from Connor, Smolker and Richards, [2]).

When the data of Table 2 are entered into non-

metric MDS, the two dimensional solution has a stress
of .001. The resulting two-dimensional image is shown
in Figure 5.
Figure 4. Galois Lattice of the Western
Electric Cliques.

The cliques form two non-overlapping groups.

Dependencies are bottom up. S1, for example, never
plays games unless W1, W3 and W4 are playing. And
W5 never plays unless those three and S1 are involved.
So, one group contains W7, W8 and W9 as core
members and W6 and S4 as peripheral. The other has a
core of W1, W3 and W4, a semi-periphery of S1 and
W2, and a periphery of W5 and I1.
This lattice-based structure corresponds closely
both to the pattern described by the ethnographers and
that revealed in the pictures produced by MDS and the
SVD. All three methods apparently converge on the
same specification of groups.
Recent research has increasingly embodied the
idea that inter-actor links are not binary, but
quantitative. Pairs of individuals are now viewed as
linked by ties of varying in strength and each pair is
assigned a number indicating the strength of their tie. Figure 5. Two Dimensional MDS
This has led to the development of another approach. Representation of the Association among Dolphins.
This is examined in the next section.
 When the principal components form of SVD spends more time with monkeys in group G
was used with the data of Table 2, the first two axes than with other monkeys that are not.
were associated with 72% of the variance. The
resulting image is shown in Figure 6. This idea specifies clearly the conditions under
which a collection of individuals can or cannot be
considered to be a group. But it does not provide a
procedure for uncovering such subsets. Given N
individuals, it would be natural to examine all possible
subsets and thus to reveal those that met the defining
criterion. The problem with such an approach is that
the number of subsets grows exponentially with N.
That means that whenever N is larger than some very
small number, computation is not possible.
It turns out, however, that a simple genetic
algorithm provides an effective way to search for the
subsets that meet the group criterion (Freeman, [5]). In
the case of the dolphin data, the algorithm finds three
non-overlapping groups, one of which is divided further
into three partially overlapping groups. The non-
overlapping groups are {a,b,c,d,e,f,g,h},{i.j} and
{k.l.m}. The first of these sets divides up into {a,b},
{c,d,e} and {f,g,h}which overlap to some degree.
The images displayed in both Figure 5 and
Figure 6 arrange the points in such a way that they both
reveal this group structure. In both cases, we can draw
Figure 6. Two Dimensional SVD a convex polygon that contains all the members of each
Representation of the Association among Dolphins. group. But the image produced by principal
components analysis displayed in Figure 6 has an
advantage. As Webster [29] showed, SVD images are
Both figures appear to break the dolphins up into oriented to rectangular X and Y coordinates; they place
well-spaced subsets. The question is, do these subsets the points in such a way that the groups can all be
correspond to groups. To answer this question requires bounded by rectangles. See Figure 7.
that we specify the structural properties of groups in
terms of this kind of quantitative measure of social
linkages.
Homans ([11], p. 84) suggested a definition of
groups that was based on a quantitative conception of
interpersonal links:

If we say that individuals A, B, C, D, E, . . .

form a group, this will mean that at least the
following circumstances hold. Within a given
period of time, A interacts more often with B, C,
D, E, . . . than he does with M, N, L, O, P, . . .
whom we choose to consider outsiders or
members of other groups. B also interacts
more with A, C, D, E, . . . than he does with
outsiders, and so on for the other members of
the group.

Sailer and Gaulin [24] made this idea a bit more

precise when, in discussing groups of monkeys, they
said: Figure 7. The Group Structure Contained in
the Dolphin Data.
On the basis of co-occurrence, monkey
m could be said to belong to group G if it
 THREE DIMENSIONAL IMAGES explore its details from a range of different
perspectives.
I used VRML to illustrate these features in an
It should be remembered here that the first two earlier web publication [8]. There I took published data
axes of the SVD of the dolphin data accounted for 72% on the sexual links among the individuals in a gay
of the variance. But the first three axes account for group and constructed a VRML image (shown there as
more than 93% of the variance. It might be reasonable, Figure 3). Most modern web browsers will permit their
then to display this pattern using three rather than two users to explore the details of that structure.
dimensions. Other programs, like Xgobi from A.T.& T. Labs
Three-dimensional figures are difficult to (see Swayne, Cook and Buja, [26]) and a great many
produce for the printed page. Klovdahl [13] used an statistics and data exploration programs can create the
early program, ORTEP, that had originally been same kind of illusion of three dimensions simply by
developed for the display of molecules (Johnson, [12]). slowly rotating the image. In addition, many of these
He adapted it in order to produce three-dimensional programs incorporate various “grand tour” options that
images of social networks. permit viewers to explore the structural properties of
images that contain more than three dimensions.
ORTEP images are drawn in such a way that
differences in the sizes and shapes of points and COLOR AND MOTION
convergences in the lines depicting links are used to
create the impression of perspective. As Figure 8 Color is extremely important, not only in
shows, a third dimension is at least crudely displayed. presenting images to others, but also in using them to
gain new insights about structural properties of social
network data. Unfortunately, in this black and white
medium, I cannot illustrate the point. But I have often
found it illuminating to produce an image (using MDS
or SVD) and to color each point according to a series of
properties of the individuals—age, sex, education and
the like. If a color representing a property is spread out
throughout the image, it is probably unrelated to the
structure. But, to the degree that a property maps to a
particular cluster in the image, it may provide a basis
for, or be a consequence of the structural patterning
displayed in the image.
In addition, animated images can provide
powerful devices for uncovering structural processes.
Like colors, animated images cannot be effectively
displayed on the printed page. But also like colors, they
are easy to display on a computer monitor. A
Figure 8. Apparent Perspective in an ORTEP particularly useful program for social network
Image. animation is called MOVIEMOL. It was produced by
Ojamäe and Hermansson [21].
More recently a large number of programs are In another web publication [7] I have used
available to facilitate producing apparent three- MOVIEMOL to show some of the power of dynamic
dimensional figures on computer monitors. When a images to help to uncover important structural patterns
monitor is used as a display device, the three- in social network data. In that paper, Figures 21 and 22
dimensional illusion can be much more convincing. illustrate how animation can help to reveal patterning
There are, for example, a great number of screen- that is otherwise difficult to find. And in another web
oriented programs that permit the creation of Virtual publication [9], I have use a dynamic display to show
Reality Modeling Language (VRML) images (see Web how a weighted spring embedder can produce a
3D Consortium, [28]). meaningful arrangement of points representing
VRML images create the illusion of a third interconnected individuals. That exercise can be
dimension by providing not only perspective, but also viewed by anyone with a Java-equipped browser.
lighting, shading and color cues as well. In addition,
they enhance the 3D illusion by permitting the viewer
to “walk” around and through the image and thus to CONCLUSIONS
 By way of summary, I would like to suggest that [6] FREEMAN, L.C. (1996), "Cliques, Galois lattices, and
social network analysts have a number of available the structure of human social groups," Social Networks,
procedures for visualization that produce images that 18, 173-187.
correspond quite closely to both ethnographic
descriptions and formal definitions of social structure. [7] FREEMAN, L.C. Visualizing Social Networks
These procedures, particularly MDS and SVD, are (http://carnap.ss.uci.edu/vis.html).
simple and quick and they arrange the points in such a
way that their main structural properties are displayed. [8] FREEMAN, L.C. Using Molecular Modeling Software
In addition, visual displays of this sort often in Social Network Analysis: A Practicum
provide new structural insights. Three-dimensional (http://eclectic.ss.uci.edu/~lin/chem.html).
images and those involving color and animation are
particularly useful in this respect. [9] FREEMAN, L.C. Using Available Java Programs to
And finally, the aim of this essay was to present Visualize Social Networks
a “state of the art” review of visualization in social (http://carnap.ss.uci.edu/java/java.html).
network analysis. When it was presented orally at the
meeting, I used a computer to illustrate and expand [10] FREEMAN, L.C., AND WEBSTER, C.M. (1994),
many of the points I have tried to make here. "Interpersonal proximity in social and cognitive space,"
Unfortunately, in print, I must often describe, rather Social Cognition, 12, 223-247.
than illustrate contemporary visual imagery. So, I
encourage interested readers to use their computers and [11] HOMANS, G.C. (1950), The Human Group, New
to explore more of the images—particularly those York: Harcourt, Brace and Company.
involving color, three dimensions and animation—
described above. The relevant URLs are listed in the [12] JOHNSON, C.K. (1965), ORTEP Report ORNL-3794,
bibliography below. Oak Ridge, TN: Oak Ridge National Laboratory.

[13] KLOVDAHL, A.S. (1981), "A note on images of

ACKNOWLEDGEMENTS networks," Social Networks, 3, 197-214.

This work was supported by MSH-Paris and the [14] KRUSKAL, J.B., AND WISH, M. (1978),
Maison Suger. Their generous support is gratefully Multidimensional Scaling, Beverly Hills, CA: Sage.
acknowledged.
[15] LAUMANN, E.O., AND GUTTMAN, L. (1966), "The
relative associational contiguity of occupations in an
REFERENCES urban setting," American Sociological Review, 31, 169-
178.

[1] BOCK, R.D., AND HUSAIN, S.Z. (1952), "Factors of [16] LEVINE, J. (1979), "Joint-space analysis of "pick-
the tele: a preliminary report," Sociometry, 15, 206-219. any" data: Analysis of choices from an unconstrained set
of alternatives," Psychometrika, 44, 85-92.
[2] CONNOR, R. C., SMOLKER, R. A., AND RICHARDS, A.
F. (1992), "Dolphin alliances and coalitions," Coalitions [17] LUCE, R.D., AND PERRY, A. (1949), "A method of
and Alliances in Humans and Other Animals, eds: A.H. matrix analysis of group structure," Psychometrika, 14,
Harcourt, and F.B.M. deWaal, Oxford: Oxford 95-116.
University Press.
[18] MORENO, J.L. (1932), Application of the Group
[3] COOLEY, C.H. (1909/1962), Social Organization, Method to Classification, New York: National
New York: Shocken Books. Committee on Prisons and Prison Labor.

[4] DURKHEIM, E. (1893/1933), The Division of Labor in [19] MORENO, J.L. (1934), Who Shall Survive?,
Society, trans: G. Simpson, N.Y.: Free Press. Washington, DC: Nervous and Mental Disease
Publishing Company.
[5] FREEMAN, L.C. (1993), "Finding groups with a
simple genetic algorithm," Journal of Mathematical [20] MORENO, J.L. (1953), Who Shall Survive?, Beacon,
Sociology, 17, 227-241. N.Y.: Beacon House Inc.
[21] OJAMÄE, L., AND HERMANSSON, K., MOVIEMOL,
( http://www.fos.su.se/moviemol.html).

[22] PROCTOR, C. (1953), "Informal social systems,"

Turrialba, eds: C.P. Loomis, J.O. Moralis, R.A. Clifford,
and O.E. Leonard, Glencoe, IL: Free Press.

[23] ROETHLISBERGER, F.J., AND DICKSON, W.J. (1939),

Management and the Worker, Cambridge, MA: Harvard
University Press.

[24] SAILER, L.D., AND GAULIN, S.J.C. (1984),

"Proximity, sociality and observation: the definition of
social groups," American Anthropologist, 86, 91-98.

[25] SIMMEL, G. (1902), "The number of members as

determining the sociological form of the group. I,"
American Journal of Sociology, 8, 1-46.

[26] SWAYNE, D.F., COOK, D., AND BUJA, A., Xgobi,

(http://www.research.att.com/areas/stat/xgobi/).

[27] TÖNNIES, F. (1887/1955), Gemeinschaft und

Gesellschaft, trans.:C. P. Loomis, London: Routledge
and Paul .

[28] WEB 3D CONSORTIUM (1999), The VRML

Repository (http://www.web3d.org/vrml/vrml.htm).

[29] WEBSTER, C.M. (1993), "Task-related and context-

based constraints in observed and reported relational
data," Ph.D. Thesis, University of California, Irvine.

[30] WELLER, S.C., AND ROMNEY, A.K. (1990), Metric

Scaling: Correspondence Analysis, Beverly Hills, CA:
Sage.