Page 347 - Invited Paper Session (IPS) - Volume 1

IPS162 Pedro C. et al.
c. Misleading Statistics
Statistics have a crucial role everywhere in the modern society. They appear
in the form of numbers, summaries, tables, and visualizations. Although there
is an old saying about “lies, damned lies and statistics”, the statistics may be
perfectly reliable and accurate but still seriously misleading because of the way
they are presented or because of the way they are interpreted or
(mis)understood. Even simple numbers may sometimes be dramatically
misleading. For an example, see Taub (2018) or Willis (2018).
In general, most people do not feel comfortable with statistics. For
common OSN users but also for journalists, simple numbers or percentages
may often be surprisingly demanding to cope or work with, let alone statistical
summaries like means, standard deviations, and correlations, or some results
of statistical analyses and tests (say, p-values, confidence intervals, and
regression coefficients). Perhaps the most challenging thing here is that
statistics are not just plain numbers (although that is what they typically look
like). Instead, statistics are reflections of complex phenomena behind the
numbers, and hence interpreting statistics requires at least some level of
understanding of those phenomena, which might often be far from trivial.
Understanding the uncertainties and pitfalls related to data gathering and
measurement is also crucial for assessing the validity and reliability of
statistical information.
The classic book How to Lie with Statistics by Darrell Huff (1954) does still
the job of summarizing the typical ways and forms of presenting misleading
statistics. However, the (at least apparently) easy ways of creating visualizations
and extremely easy ways of spreading them rapidly online has multiplied the
number of possible situations where presenting or interpreting statistics
appears misleading in a way or another.
It is also noteworthy that in the current world of OSNs, (dis)information
typically spreads via quick instinctive reactions instead of slow critical thinking.
A recent book Factfulness (Rosling et al., 2018) goes through ten specific
instincts that easily distort our views of the world. Many of those instincts are
indeed closely related to interpreting statistics, especially in visual forms. The
aim of the book is to encourage the reader to replace his/her instinctive
reactions by critical thinking. In addition of old and new books, plenty of useful
materials in various forms are openly available on the web, see for example,
Blauw (2016), Liddell (2016), Rosling (2010), and Steinberg (2017). Excellent
handbooks especially for (data) journalists have been compiled by Gray &
Bounegru (2019) and Gray et al. (2012).

336 | I S I W S C 2 0 1 9