5. 6 CONCLUSIONS FROM THIS STUDY
During the development of this paper we have been drawing some
explicit conclusions which were stated at the end of each chapter. They were
then used for justifying and introducing our effort in the subsequent chapter.
We present now an overview of the whole study in the form of a combined series
of the earlier statements and some concluding remarks.
The reviewed EDP literature does not present definitions of
quality of information, in the sense that no explicit support is found for the
formulation of operational definitions of the concept.
The quality of information, however, is of fundamental
importance for the development and use of data-banks and information systems:
this is the opinion implied in the reviewed EDP literature and it also is
implied by the lack of a scientifically justified cost-benefit analysis of
data-banks and information systems.
We have reviewed empirical results and reported experience
intuitively or explicitly related to quality of information in EDP. Their
quantitative content assumes a concept of quality in terms of communication
theory - theory of signal transmission.
The utilization of such results and experience in the context
of a particular information system, as well as the development of other
necessary measures, require a broader concept of quality.
It is possible to illustrate some of the consequences of the
communication approach to quality by observing that it may easily lead to the
uncritical acceptance of aggregated data in the context of high-level
decision-making. It may also lead to a technical interpretation of the coding
issue disregarding the possibility to consider it as a source of symptoms of
inadequate model building or systems design.
The search for an adequate concept of quality leads to
regarding information systems and data-banks as integrating different theories
or models at different levels of "maturity". This integration requires the
development of an overall concept of quality of information.
It is possible to meet this requirement by redefining accuracy
and precision as two aspects of overall quality of information, with the purpose
of allowing inferences on the reproducibility of the computational
results.
Our study provides a starting point and a set of suggestions
on how to proceed in order to develop a complete and detailed quality-control of
information in the context of a particular information systems.
A fundamentally important overall conclusion from this study
is that the quality-control effort must be concetrated on designing into the
systems those features which allow for THE STRONGEST
DISAGREEMENT.
Eventually, this study raises suggestions concerning the
existence and possible solution of some important quality problems. In a more
informal way, and in different degree of justification the suggestions are
presented in appendix A11 in the form of comments, questions, and
proposals for further action. Some of the suggestions, like regarding the right
to know and disagree about personal attributes, stem directly from the main
arguments of our study and should be regarded as strong recommendations for
immediate action. Other suggestions are more loose speculations about
exceedingly complex and important matters: they are presented in order to
stimulate debate and further research.