This monograph is for postgraduate students of statistics, statistical analysts,
and other professionals who are interested in the design and analysis of studies
in which responses are elicited from human subjects. Emphasis is placed on
dealing with data that arise in imperfectly conducted studies. The reasons for
imperfection include a sampling plan that cannot be implemented, measurement or elicitation of information by imperfect instruments, poor motivation
of the subjects and their unwillingness to cooperate, and a multitude of other
unavoidable shortcomings in relation to textbook-like settings that would be
easy to analyse.
The subject of statistics is defined as making decisions in the presence of
uncertainty. The context of a population and one or several variables defined
for each member of this population is presented, and complete information is
at first defined as having established the values of these variables for every
member of the population. Making decisions with such complete information
is regarded as a task outside the remit of statistics and is assumed to be a
resolved problem or a problem for another profession. The raison d’ˆetre for
statistics is that the available resources (time, manpower, expertise, funding, respondents’ goodwill, and the like) are not sufficient for collecting the complete information.