Figure
1. Types of Records (Tables) Related in
HSIS Database.
By
Alvin
W. Wolfe and Guy Hagen
University
of South Florida
(Draws in part on a paper co-authored by Alvin Wolfe, Mary G. Rust and Patricia M. Sorrells, presented at the Fortieth Annual Meeting of the Society for Applied Anthropology, Denver, Colorado, March 22, 1980)
The classic traditional ethnographers such as Malinowski, Radcliffe-Brown, and Firth described cultures that were expressed in the communications of a few thousand persons. Malinowski’s Trobrianders numbered about eight thousand, although through the kula ring they shared information with several thousands more “argonauts” of the Western Pacific (Uberoi 1962). Firth’s Tikopians were a scant 1,300 in 1929; even if each of them communicated with all others, the total number of relationships would be less than 845,000 (Firth 1959).
Now consider what the ethnographer of a modern society faces. There are vast numbers of people, over 280,000,000 in the United States alone, according to the 2000 Census. In addition, communications media make it possible for an incredibly large population to share information instantaneously. This was first brought sharply to our attention twenty years ago, in January, 1980, when we realized that 150,000,000 people saw, simultaneously experienced, a sports event, the Superbowl. About that same time, and more meaningful in its impact on the American sociocultural system, we learned that half the households in America, almost 100,000,000 persons, saw the television program, Roots! In this kind of technological milieu, old-fashioned ethnography cannot do the job. Mere statistics are not enough. Something more is called for.
Technology has enabled unprecedented new levels of communication and interaction between human populations. Even new communities, cultural norms, and language elements have developed through the growth of the Internet. Complex interest groups -- “chat” and “news” groups – form around shared interests, dense webs of electronic “email” communication enable long-distance communication in ways unprecedented, not even imagined by most folks just decades ago. Web pages are electronic representations of companies, communities, and individuals. There are an estimated 800 million web pages today, with approximately one million pages being added every day (Chakrabarti, et al. 1999). Broadcast and broadband media enable the near-simultaneous interaction of millions of people. These are all expressions of human communication that ethnographers must now take into account. Social scientists are just beginning to adapt methodologies to explore this “electronic frontier” (Garton, Haythornthwaite and Wellman 1997, and Kautz, Selman and Shah 1997), and ethnographic analysis must keep pace.
Since the beginning of ethnographic fieldwork, the preparation, collation, indexing and storing of field notes has been among the most important activities the ethnographer engages in. Only if such nits and grits are done well, can the relevant information be retrieved when needed.
Surely it was such tasks of organizing the information derived from observations of hitherto unknown cultures that led anthropologists to conceptualize culture as a complex whole, differentiable into various domains and levels, yet integrated through relations among those domains and levels.
As anthropologists took into account more of the internal variability within traditional cultures, previously thought to be simple, their informal methods of handling data were strained. The matter was still worse with respect to the differentiation and specialization of modern societies. For a long period in the middle of the twentieth century, the major statements about complex societies were derived from the macro-level data sets of economists, political scientists, and sociologists while anthropologists tended to provide only local color. We believe anthropologists can and should do more.
It may
yet be within our capabilities to do holistic ethnographies of complex systems,
analogous to what used to be done for traditional societies. We refer here to the kind of description of
a culture which includes not only (in Goodenough’s words) “the set of standards
whose behavioral manifestations local authorities will accept as appropriate”
(1970:102), but also some account of the variances from those standards, more
or less along the lines of Theodore Schwartz’s distributive model of culture
(1978). It will not surprise anyone to
learn that the amount of data necessary to record not only the norms but the
variations as well is massive. Still,
we should not be dismayed or discouraged.
Technological advances have enabled new data
processing tools that can be used to surmount these challenges, including the
introduction of new levels of computing capacity.
To give a rough idea of the processing power currently available, one of the desk top computers (a Macintosh G4) used to edit this text has a performance speed of over one “gigaflop” – one billion floating-point operations per second (Apple Corporation Web Site, December 2000). Today’s microprocessors are approximately 100,000 times faster than the earliest processors. Industry experts predict that Moore’s Law – that computing power doubles every 18 months – will continue for the foreseeable future. Within 20 years, individual computers may be more powerful than all the combined computing power currently available in Silicon Valley (Patterson 1999:86-88). Also, data storage becomes more efficient because of increased capacity of each unit as well as distribution among units.
It is not just that computers can rapidly and
efficiently handle masses of data, but that computer programs can now model the
more complex relations among data domains and levels. Database
Management Systems incorporate methods for handling large sets of data
logically organized to satisfy a variety of users’ requirements. Instead of
relying on sequentially organized records whose relations derive basically from
the fact that they draw certain attribute values from common domains, modern
databases can be hierarchical or relational in organization. If they are hierarchical, they originate in
a root and the records branch out in one-to-many relations, toward terminal
records, as segmentary lineage systems or bureaucracies do in social life. Or, record structures may have a network
form, wherein each record may be related directly to several other records, a
very complicated architecture that, alas, models the complexity of most
sociocultural phenomena.
At this point, it will be meaningful to show an example of how the data of a reasonably complex system can be organized. The schema for the Human Services Information System that Wolfe developed in the early 1980s can serve as that illustration (Figure 1).
Figure
1. Types of Records (Tables) Related in
HSIS Database.
Each box represents a table (or record type), for example “firm” record has information about an agency. The lines represent direct hierarchical linkages creating a set of tables, one table relating to others in “one-to-many” relations. “Many-to-many” relations, as for example, when one firm (agency) operates programs at several sites and one site houses operations of several facilities. The database permits easy retrieval of information in either direction in the network.
From data so structured it is possible to retrieve the specific information desired and to have that particular report printed in the format desired. Instead of some kind of generalizing statistical statement that so many thousand services were performed by some number of agencies, one can make a more descriptive statement closer to the reality of the situation, preserving the variation among agencies or firms, programs, clients, and so forth, as well as preserving information about their relationships.
An
excerpt from such a report is shown as Table 1.FiFi
Table 1.
Report produced by the Human Services Information System
Department
of Anthropology, University of South Florida, Tampa, Florida 33620
|
Program: Agency for Community Treatment Services HSIS Database
No: 12100 4211
E Busch Blvd Tampa,
FL 33617 - 5916 813
9886096 Type
and Funding: Non-profit,
Incorporated (Voluntary) FEES Program: ACTS, Adult Residential Detox 6806
N Nebraska Ave Tampa , FL 33604 - 813
2389505 Program Services: alcohol
counseling services detox
serv for drug abuse residential
trtmnt alcoholism Program: ACTS, Agcy for Comm Trtmt Svcs 4211 E Busch Blvd Tampa,
FL 33617 - 5916 813 9883533 Program Services: alcohol abuse educa/info serv alcohol counseling services drug counseling services Program: Adolescent Group Home, ACTS 3806 W M.L.King Blvd. Tampa,
FL 33614 - 813 8700578 Program Services: residential trtmnt alcoholism residential trtmnt drug dpndn- Program: Adolescent Receiving, ACTS 8620 N Dixon Ave Tampa,
FL 33604 - 813 9314669 Program Services: alcohol counseling services detox serv for alcoholics detox serv for drug abuse drug counseling services Program: Agency for Community Treatment Services, ACTS 4612 N 56th St Tampa,
FL 33610 - 813 2464899 Program Services: juvenile counselling psychological assessment psychological therapy Program: Alcohol Community Treatment 4403 W M.L.King Blvd Tampa,
FL 33614 - 7604 813 8791649 Program Services: alcohol counseling services internships
residential trtmnt alcoholism transitional serv-alcoholism Program: Dependency Group Home, ACTS 3812 W M.L. King Blvd. Tampa,
FL 33614 - 813 8702236 Program Services: foster care – dependency group home -pre-dlnquent youth |
Program: Halfway House ACTS 4403 W M.L.King Blvd Tampa,
FL 33614 - 813 8759645 Program Services: internships Program: HARP, Residential Svcs, ACTS 8702 Stark Road Seffner, FL 33584 - 813 6216051 Program Services: resid.care,long term,mental r- Program: Homeless Day Center, ACTS 6220 N Nebraska Ave Tampa,
FL 33604 - 813 2376630 Program Services: shltr-newcmrs,travlrs,homeless Program: Outpatient Counseling ACTS 1815
W Sligh Ave Tampa,
FL 33604 - 5811 813 2376630 Program Services: alcohol abuse educa/info serv alcohol counseling services drug counseling services Program: Outpatient Counseling, ACTS 4211 E Busch Blvd Tampa,
FL 33617 - 5916 813 9883533 Program Services: alcohol counseling services drug counseling services Program: Pinellas Domiciliary 3575 Old Keystone Road Tarpon Springs, FL 34689 - 813 4612881 Program Services: alcohol counseling services internships residential trtmnt alcoholism transitional serv-alcoholism Program: Youth Outpatient, ACTS 8620 N Dixon Ave Tampa,
FL 33604 - 813 9314669 Program Services: alcohol counseling services drug counseling services Program: Youth Overlay Services, ACTS 3806 W M.L.King Blvd. Tampa,
FL 33614 - 813 8700578 Program Services: alcohol counseling services drug abuse ed/info serv drug counseling services |
Table 1
(Continued). Report produced by the
Human Services Information System
It is important to note that such a database is not conceived as
fixed. In addition to storage and
retrieval functions, users may continuously insert, delete, update, and amend
as the system that is being modeled changes.
We are dealing with a system open to exchange with the environment and
therefore offering management capabilities beyond its own narrow limits. Feedback from component to component
provides for maintaining the levels of flow and for making adjustments within
the system. There is also the potential
for feed-forward, in that predicted future conditions might influence the state
of the system, an anticipatory quality obviously useful for planning and
forecasting.
Such schemas can model cultural structure, and the computer programs provide the “windows” through which an ethnographer of a complex society may view different domains. The richness of ethnographic detail transcends the aggregation of raw data, our traditional way of viewing information storage. Rather, such a system offers the capability of looking back through aggregate information diachronically while viewing complex relationships on multi-level plans synchronically. We can have our data and digest it too! Such capabilities offer flexibility for decision-makers. Patterns emerge and overviews are feasible; planned change may be designed, based on the small picture as well as the large. Thus, database management systems, designed originally to meet the needs of business and governmental enterprises, can be used to help achieve the holistic perspective characteristic of anthropological studies, theoretical and applied.
The development of the Human Services Information System, a computerized system that would serve the needs of the human services agencies of a metropolitan area, can provide an illustration of anthropologists initiating such an effort. Ethnographic observations by practicing anthropologists in the Tampa Bay area revealed a common need by a variety of public and private agencies for useable, readily accessible information to support the functions of planning, monitoring, evaluation, management, and budgeting. Typically, individual agencies rush about seeking information, as it is needed, from a variety of potential sources. In such circumstances, the results often prove disappointing, and clearly, agency resources are being inefficiently used in such independent endeavors. Both the ethnographic observers and the native actors, in this case agency staff, have difficulty learning who is doing what for whom with what resources and with what effect. To address this common problem, an anthropologist/planner with the City of Tampa (Mary Rust), approached a fellow anthropologist at the University of South Florida (Alvin Wolfe), with a proposal for a Human Services Information System. The HSIS, as it has come to be labeled, is the result of their collaboration along with a number of students, and with representation from more than 25 key agency representatives, of whom 16 participated in a planning workshop. The workshop participants arrived at considerable consensus on a number of points: that the need for information is imperative; that they intended to meet that need; that the system should ultimately serve the widest feasible range of functions (those mentioned above: planning, evaluation, management, budgeting, preparation of proposals, etc.); that its organization should have the form of a consortium of agencies.
It is difficult for an entire human services system such as the system of services for children and families in a metropolitan area to organize for collective action. The HSIS contains information on more than a thousand “firms,” several thousand “programs” providing hundreds of types of “services” to hundreds of thousands of “persons.” Since each political unit wants its own thing, political forces inhibit the expenditure of public funds for truly cooperative endeavors. And since there is very little profit in such services, the system is not structured by the market principle. That narrowness aside, an entire human services system is almost unmanageable anyway, because of the complexity of the system itself. Its overall structure resembles a network rather than a hierarchical bureaucratic organization, even though it is composed of hierarchically structured subsets. Even as a network, it is not beautifully symmetrical like the web of a healthy spider, but asymmetric, with dense clusters and gaping holes – the infamous “holes in the safety net”, or the equally infamous “cracks through which clients fall.” Nonetheless, there is some kind of whole there, and that whole can be grasped by using the electronic and network analytic tools made available to us by technology and mathematics.
Human beings lie behind all the entities in these complex systems we are trying to describe. The people involved in them connect organizations; at the same time, the people are connected by their involvement in the organizations. Analysis of these dual mode networks that have heretofore bewildered observers and participants alike is possible by using the high technology and sophisticated mathematics now available. That is what we think of as electronic ethnography. Just as earlier ethnographers made sense out of what seemed initially to be bewildering chaos, electronic ethnographers will make sense out of the bewildering complexities of our developing social systems. Just as earlier ethnographers learned to use mapping instruments, still cameras, audio recorders, movie cameras and video equipment, so modern electronic ethnographers will learn to use technology and mathematics. We feel very much akin to the traditional ethnographers of the past.
Social network analysis meets
the description of “electronic ethnography” in two ways. It takes advantage of computers to augment
traditional ethnographic techniques, allowing analysis of larger communities
and more complex structures than most manual techniques. Also, it enables exploration of “electronic communities.” As the Internet is used more and more as a
medium for human communication, it is becoming rich territory for electronic
ethnography.
PROJECT: ELECTRONIC ETHNOGRAPHY
This project involves the construction of a simple
network database. Students will explore
an “ego network” of connections within a small group of individuals and will
use simple network analysis techniques to identify social and qualitative
characteristics of that group and its members.
The requirements for this project are limited to
tools that should be available to most students: a computer with a spreadsheet
program (Microsoft Excel, Lotus, Appleworks, etc.), and an email (electronic
mail) account. For more ambitious
students, we recommend use of UCINET5 network analysis software that is
available at http://www.analytictech.com.
Depending on your personal interest, we present two options for experimentation. If your interest lies in augmenting traditional ethnographic techniques, you may explore your own personal network of friends and family. If you are interested in exploring the electronic medium, you may collect data on email communication among your friends.
Step 1: Identifying Your Personal Network.
Prepare a new spreadsheet document. You will record your network in a “matrix”
format; basically, you will create a table in which network members are listed
simultaneously as row labels and column labels. Matrices are the most common format for storing and manipulating
network data.
Make a list of the people you communicate with most
frequently (between ten and twenty), and enter them into the database as listed
in the following table. If you are
exploring an email network, work from your email program’s “contact list” (if
available).
Any cell that contains a “1” indicates that a
network connection exists between individuals whose rows and columns
intersect; “0” indicates a lack of
connection. In our project, a network connection signifies that frequent
communication (email or otherwise) exists between network members. Since we aren’t concerned if anybody sends
email to himself or herself; put a 0 in cells where the row and column labels
are the same.
At this point, you have a list of individuals with
whom you personally communicate.
However, you haven’t determined which of those individuals communicate
with each other. We must now fill out our network.
Step 2: Surveying Network Members
Next, you will need to contact each individual in
your network. Provide each a copy of
your list of members, and ask each of them to indicate every individual that
they have regular communication with (or, which individual is in their email
“contact list”).
Once you
have collected your information, enter connections between individuals as shown
in Table 3:
Table 3: Complete “ego
network” matrix for ego and ten contacts
Step 3: Learning About the Network
The number of individuals in your network is its
size (represented by n); people will
have differently sized networks depending on how socially active, influential,
or popular they are.
Density. Aside from
size, an important consideration is how densely connected the network is. Obviously, it is possible for everybody in
your network to be connected to each other; alternatively, they may not be so
well connected. The proportion of
active links in a network to the maximum possible links is the density of the network.
The first part of our calculation will use the
spreadsheet’s sum function (which may
vary depending on which software you use).
For each column, add up the total number of links (Microsoft Excel
example follows for column b of your matrix of ten individuals)
Number of “in-links” = sum(b2:b11)
Add up the total of these
values. This will provide the number of links in your network. Divide it by the maximum possible links (n x
(n-1)). Results will range from 0 to 1.
If you have 20 members in your network, the Microsoft Excel formula
would look like “=sum(B22:U22) / (20 * (20-1)) )”.
Calculate your network’s density and size. Does your network have a high density? What might a low or high density mean? Does this tell you anything about your group
of friends and family?
Prestige. You’ve
probably noticed that not all connections in the matrix are ‘reciprocal.’ Sometimes one person lists another as a
contact but that other person does not list him or her, does not
reciprocate. Consider the implications
for an individual who lists many persons as contacts but is not often listed by
them. In network terms, we can measure
this, sometimes called prestige – individuals with high prestige are cited more
than others in the network. They are
more likely to have access to information.
Some Internet search engines such as AltaVista.com and Google.com use a
form of prestige to measure citation
relevancy; some business leaders advocate a form of prestige analysis to determine business effectiveness.
Fortunately, you have already calculated each
individual’s prestige score with your
column totals. Who has the highest
score? Can you guess why? Is it a measure of popularity, or can you
find some other explanation?
Sociogram. The final
part of our exercise will be to make a visual graph of the network – a
“sociogram.” Sociograms can be useful
tools for exploring network characteristics.
There are dozens of methods for generating sociograms ranging from
simple to sophisticated.
Start by placing labels for individuals in the
network around a circle. Draw a line
between each pair of labels that are connected (represented by a “1” in a cell
at either the row:column intersection
or the column:row intersection, which may be different). Here is a simple sociogram based on our
example matrix:
Figure 2: Sample Network Sociogram
Draw a sociogram for your network. Does the graph suggest anything
interesting? Are the density and prestige calculations revealed?
Chakrabarti, S., B. Dom, S. Ravi Kumar, P. Raghavan, S. Rajagopalan, and A. Tompkins. 1999. “Hypersearching the Web”, Scientific American, June 1999, pp 54-60.
Firth, Raymond. 1959. Social
change in Tikopia. London: George Allen and Unwin, Ltd.
Garton, Laura, Caroline
Haythornthwaite, and Barry Wellman. 1997. Studying Online Social Networks. Journal of Computer Mediated Communication 3 (1).
Goodenough, Ward. 1970.
Description and comparison in cultural anthropology. Chicago: Aldine Press.
Kautz, Henry, Bart
Selman, and Mehul Shah. 1997. Referral Web: Combining Social Networks and
Collaborative Filtering. Communications
of the ACM (40) 3, 63-65.
Patterson, David A. 1999.“Microprocessors in 2020,” Scientific American Revolutions in Science 86-88 (1999).
Schwartz, Theodore. 1978.
The Size and Shape of a Culture. In Fredrick Barth, ed., Scale and
Social Organization. New York: Columbia University Press, p. 215-252.
Uberoi, J.P. Singh. 1971.
Politics of the Kula ring: An Analysis of the findings of Bronislaw Malinowski.
Manchester, England: University of Manchester Press.
Wolfe, Alvin W., Mary G Rust and Patricia M. Sorrells. 1980. Electronic Ethnography: the Human Services Information System. Paper presented at the Fortieth Annual Meeting of the Society for Applied Anthropology, Denver, Colorado, March 22, 1980.