Lies, Damned Lies, and Statistics

The older I get the more cynical I become towards statistics. Statistics can be a valuable tool for identifying trends. However, the trends revealed by statistics often have multiple possible explanations. Case in point, a lot of media outlets have been making a big deal about the supposed rise in hate crimes, especially against Muslims. They have been quick to blame the election of Trump. However, another cause of this trend could be methodology:

There were 271 more incidents deemed hate crimes in 2016 than the previous year, according to the latest Uniform Crime Reporting (UCR) data. There were also 257 more law enforcement agencies reporting last year, so that increase could largely or even entirely be a matter of getting more complete statistics. The higher numbers mostly represent small increases in incidents classified as anti-Hispanic, anti-Jewish, anti-Muslim, or anti-white.

[…]

Some will surely blame the beginning of Donald Trump’s political ascendancy, and that can’t be ruled out. But another explanation is as likely, if not more likely: The FBI changed how it classified certain hate-crime incidents in 2015.

Before this period, crimes based on someone’s ethnicity or national origin were simply sorted into Hispanic or non-Hispanic bias incidents, leaving us with a cache of uncountable incidents that could’ve been based on someone’s perceived Middle Eastern or Arabic status. But in 2015, ethnicity was lumped in with the racial-bias category. This means that some of the incidents previously attributed to a general sort of anti–Middle Eastern bias could either be categorized as anti-Arab racial/ethnic bias or anti-Muslim religious bias, possibly spiking the anti-Islamic incident stats.

More law enforcement agencies providing data may be influencing the results. Moreover, the category being mentioned most frequently by the media, hate crimes against Muslims, is a recent addition. Going from zero incidents before 2014 to incidents in 2015 will necessarily show an increase in incidents.

None of this is to say that Trump’s election hasn’t played a contributing factor. But there are also alternative explanations for the increase in hate crimes that cannot be ignored. Perhaps the increase in hate crimes is a combination of Trump’s election and changes to methodology. Statistics can reveal a trend of the methodology is solid. But even if a trend is revealed, statistics can seldom point to a specific cause or provide an effective solution.