Page 52 - The Real Work Of Data Science Turning Data Into Information, Better Decisions, And Stronger Organizations by Ron S. Kenett, Thomas C. Redman (z-lib.org)_Neat

9

Take Accountability for Results

The wide‐angle perspective of data science includes activities as diverse as building trust so
you are asked to contribute to really important problems, clearly stating the problem, con-
ducting the analyses, teaching, supporting decisions in practice, and so forth. This chapter
focuses on one activity that is too often ignored – impact assessment.
Impact assessment is important so data scientists (and CAOs and groups of data scientists)
can show others what they contribute in concrete terms. This in turn can help with funding,
build trust, and more powerfully position the work. Similarly, it helps data scientists learn to
become more effective. It is the last step in the life cycle introduced in Chapter 1.
Importantly, different communities judge impact differently. In science, new ideas must stand
the test of time, and concepts such as statistical significance help guard against results that will not
do so. In business, the criteria are wholly different. They may involve increasing sales, decreasing
costs, improving market share, reducing risk, and so forth. Results, particularly those tested in the
marketplace, need not stand the test of time, but they must stand up to tough competitors, gain
new customers, and keep current ones. In a nonprofit organization, results may be judged on cri-
teria such as improved test scores, reduced homelessness, enhanced national security, and the like.
In our view, none of these criteria are inherently better or worse; easier or more difficult;
more noble or more basic than the others. But they are different, so understanding what the
organization values is essential, as we discussed in Chapter 3.
Practical Statistical Efficiency

Of course, statisticians and others have understood the importance of evaluating the impact
of their work for generations. Researchers of statistical methods have thought in terms of
“statistical efficiency,” comparing, for example, two methods for estimating the mean of a
population. Building on a basic idea, Kenett et al. (2003) proposed the concept of practical
statistical efficiency (PSE) to address the impact of statistical work in a specific problem area.
PSE embraces the following elements:

V{D} = value of the data actually collected
V{M} = value of the statistical method employed (statistical efficiency)

The Real Work of Data Science: Turning Data into Information, Better Decisions, and Stronger Organizations,
First Edition. Ron S. Kenett and Thomas C. Redman.
© 2019 Ron S. Kenett and Thomas C. Redman. Published 2019 by John Wiley & Sons Ltd.
Companion website: www.wiley.com/go/kenett-redman/datascience