UNDERSTANDING OF STATISTICAL THINKING 33
this ability to use statistical thinking without data as follows: “the uniqueness of
statistical thinking is that it consists of thought processes rather than numerical
techniques. These thought processes affect how people take in, process, and react to
information.”
Tversky and Kahneman’s insights about how regression to the mean affects
people’s beliefs about the effects of reward and punishment are widely promulgated
in quality management as part of “understanding the theory of variation.” The
setting used to illustrate this is typically the reactions of sales managers to the highs
and lows in sales figures of their staff. According to Joiner and Gaudard (1990),
many managers fail to recognize, interpret, and react appropriately to variation over
time in employee performance data. These statisticians are attempting to get
managers to understand that looking at single time-interval changes and meting out
praise and censure is not conducive to improving performance. The way to improve
performance is to make some system change that will increase the average level of
performance. Managers need to recognize that there will always be variation, and
that unless there is a system change there will be regression to the mean. This
suggests that managers are being asked to take on a world view that allows for
indeterminism.
Statistical thinking in quality management is now seen not only as necessary for
gleaning information from data but also as a way of perceiving the world reality.
From quality management we learn that statistical thinking is, first and foremost,
about thought processes that consider variation, about seeking explanations to
explain the variation, about recognizing the need for data to guide actions, and about
reasoning with data by thinking about the system or process as a whole. Implicit in
their concepts about variation is that system (not people or individual) causal
thinking is paramount. Once the type of variation has been categorized as specialcause
or common-cause, then there are appropriate strategies for identifying the
causes of that variation. The quality management thinking approach is not to leave
variation to chance, but to reduce it in an attempt to improve processes and
performance.
Some Contributions from Statistics Education Researchers
The quality management approach to statistical thinking arose from the
confluence of a focus on empirical data and the need to improve processes. In
contrast, the statistics education field tended to have its origins in mathematics
education and in a deductive rather than inductive culture.
Statistics education research emerged in the late 1970s and focused mainly on
probability (e.g., Fischbein, 1975; Tversky & Kahneman, 1982). It has really only
been in the last decade that statistical thinking has begun to be addressed. We will
now discuss some of these developments.