Deceiving Data

A great benefit of the digital age is that large, multivariate data sets can generate insights that would have been impossible to find in earlier days.

Unfortunately, all that Big Data can also be misleading. Popular magazines and TV advertisments are frequently found guilty of coming up with misleading information, in an attempt to lure consumers into their fold.

Simpson's paradox (or the Yule-Simpson effect) is a phenomenon in probability and statistics, whereby a trend appears in different groups of data, which disappears or reverses when these groups are combined.

An example of Simpson's Paradox is when the University of California, Berkeley was sued for bias against women who had applied for admission to graduate schools in 1973. Admission figures showed that men applying were more likely than women to be admitted, and the difference was so substantial that one would conclude that discrimination existed.

However, when examining individual departments, it appeared that no department was significantly biased against women (Bickel, Hammel, and O'Connell 1975). In other words, there was no significant difference between the number of men and women admitted when looking at departmental groups; but this finding reversed (suggesting that more men were admitted than women) when the groups were combined.

Other issues with data analysis or visualization that can lead to misunderstanding and incorrect analytical outcomes are:

• Not showing the entire scale of the data
• Using scales that are irrelevant to the data
• Using a subsection of the data to support a given conclusion
• Showing data out of context
• Using irrelevant or opportunistic subgroups
• Falsely implying relationships between variables