COMPUTER APPLICATIONS IN SOCIOLOGYquite common in various forms: e-mail texts, e-
mail attachments, entry forms on the Web, and as
programs in external storage devices like diskettes
and CD-ROMs. Sampling problems and low com-
pletion rates pose the greatest challenges. Ongo-
ing methodological investigations will be neces-
sary to determine the implications of this new
mode of research.
Statistics. Hundreds of computer programs
and articles have been written to address the needs
of statistical computing in social research. Prior to
the 1980s, all statistical work was performed on
large or medium-size, mainframe computers. But
advances in both hardware and software for micro-
computers now make it possible to conduct the
statistical data analysis of most small or moderate-
size research studies on microcomputers. A large
share of ongoing social data analysis, like analysis
of massive census files, would never get done
without computer technology. For example, one
use of LISREL, a computer procedure which ana-
lyzes linear structural relationships by the method
of maximum likelihood, would consume weeks or
months without a computer.
Not only does statistical computing save time
but it offers unique views of the patterns in one’s
data. Without the ability to quickly reorganize data
and display it in a variety of forms, social research-
ers neglect important patterns and subtle relation-
ships within complex data. Some patterns can-
not be observed without special software tools.
For example, Heise’s (1988) computer program
called Ethno gives the researcher a framework for
conceptualizing, examining, and analyzing data
containing event sequences. In addition, several
general-purpose statistical packages offer power-
ful exploratory data analysis capabilities with
bidirectionality through dynamic data links (Steiger
and Fouladi 1990). One type of bidirectionality
puts a graph in one window and frequency distri-
butions in another, and when the user adjusts the
data in one window, it automatically changes in
the other.
Finding a statistical program tailored to a
particular problem or technique is often challeng-
ing as the potential ‘‘user community’’ may be
quite small. The best sources for such software are
the notices and reviews in journals such as the
Social Science Computer Review, Educational and Psy-
chological Measurement, Journal of Marketing Research,
and The American Statistician. Another important
source is the annual Sociological Methodology and
it’s software list on the Web (http://weber.u.wash-
ington.edu/~socmeth2/software.html). Some of
the software noted in these sources can be ob-
tained at no cost or very low cost. No matter what
the cost, one cannot assume that any complex
program is free of errors. Thus is it important to
run test data and to apply data to multiple pro-
grams in order to check for inaccuracies.
Accessing, Retrieving and Managing Data.
While years ago students and researchers had to
use a library or similar institution to gain access to
bibliographic data files, now such services are
available from one’s desktop using the Web or
external storage units such as CD-ROM or DVD-
ROM. Large bibliographic databases including So-
ciological Abstracts and Psychological Abstracts
are available in these forms, as is a vast amount of
data in the form of statistical tables and maps. Now
that devices for ‘‘writing’’ onto CD-ROMs have
become inexpensive, it is expected that data from
even small research projects will be disseminated
in this medium.
One major development is interactive access
to data by means of the Web. A variety of models
are used for interactive access to both preformatted
text files and precoded data files. Among the
systems are GSSDIRS from ICPSR at the Universi-
ty of Michigan (http://www.ICPSR.umich.edu/
gss), IPUMS at the University of Minnesota (http:/
/www.hist.umn.edu/~ipums), QSERVE from
Queens College-CUNY (http://www.soc.qc.edu/
qserve), and SDA Archive from the University of
California at Berkeley (http://csa.berkeley.edu:7502/
archive.htm).
The software technology for archiving and
analyzing social data is less than a half-century old.
But it is very plausible to expect many advance-
ments in the next fifty years. Interactive data analy-
sis sites of the Web hint about what these advance-
ments might be. For instance, using the SDA
Archive Web site (see http address above), one can
get a large crosstabulation table for any three
variables in the full General Social Survey of over
35,000 respondents in less time than it takes to
type in the variable names.
The refinement of such systems faces issues
such as how to balance functionality with ease of
use, the plausibility of standardizing interfaces for