Quality & Quantity 34: 65–86, 2000.
© 2000 Kluwer Academic Publishers. Printed in the Netherlands.
Computerized Decision Support Systems
and Text Analysis: Evaluating CETA
HARRY VAN DEN BERG and KEES VAN DER VEER
Vrije University Amsterdam, Faculty of Social-Cultural Sciences, Department of Social Research
Methodology, De Boelelaan 1105, NL-1081 HV Amsterdam, The Netherlands
Abstract. Computers can play several different roles in text analysis. We will examine one of these,
computerized decision support systems; ﬁrst giving a brief introduction to the general features of
this family of software and then focusing on CETA (Computer Aided Evaluative Text Analysis) –
developed by Van Cuilenburg et al. (1988) – a well known example of this type of computer assisted
text analysis tool. After outlining the principles and recommended procedures for the use of CETA,
we will formulate several hypotheses concerning its strengths and weaknesses. These will be tested
by means of a case study: an in-depth analysis of news coverage by the New York Times of the
air trafﬁc controllers’ strike against the Reagan administration in the early 1980s. We will focus
primarily on one key NYT editorial to illustrate the potential and limitations of CETA.
1. The Role of Computers in Text Analysis
There are, essentially, three different roles computers can play in text analysis, each
of which can be identiﬁed with a speciﬁc type of software. It is possible to order
these three functions by the degree to which, in each, the computer is substituted
for the human coder (see also Van den Berg, 1994, 1997).
(1) Computers can be used in the development of information storage and retrieval
systems for text analysis. In this case there is no question of substitution. The
computer’s main function is to facilitate searches for coded text fragments,
enabling the researcher to locate and compare fragments with equal or dif-
ferent codes in an efﬁcient manner. The role of the computer remains rather
limited. Although valuable in terms of efﬁciency, the reliability and validity of
the coding process are not substantially affected (Van den Berg, 1997).
(2) Computers may also be used in the development of decision support systems
for text analysis. In this case the human coder remains the ultimate decision
maker in the coding process, but the computer plays an important role in, as
far as possible, structuring coding decisions. In other words, the coder is forced
to follow steps speciﬁed by the computer, which then delivers information to
him or her of possible relevance to coding decisions.
(3) The most ambitious application of computers to text analysis is that of
automatic or semi-automatic text analysis-systems, whereby a computer is