Statistical Analysis of the Consequences of Using the Phrase “Chinese Virus”

Trump used it. Reader wrote “The Chinese virus is not racist, it’s placist.” I believe that since there was already a well-known term for the virus, there was no need to resort to a made-up nickname. Khan, Riddell & Piquero (2023) explores what happened after the popularization of the term.

Source: Tam, Riddell, Piquero (2023).

The authors estimate the regression of the number of hate crimes as a function of time constant, a dummy value that takes the value of one after March 16, 2020, and a time trend starting from March 16, 2020.

The time factor (where each observation is weekly) is 0.075 with a standard error of 0.056. The authors conclude that none of the coefficients is statistically significant using two-tailed tests. Note that the t-statistic is 1.32, is is significant at 10% of the mean difficulty level for a one-tailed t-test, which is somewhat more appropriate if one considers that the relevant alternative hypothesis is that crime increased faster after using the terms “kung flu” or “Chinese virus”. Note that these are results based on data from four major cities, as described in the paper.

Aggregate national (but probably underestimated for the AAPI group) FBI data for 2020 is here.

Source: FBI (February 2023).

That’s 77% more hate crimes against people of Asian descent in the US in 2020.

Here is a time series that shows a discrete jump in hate crimes against Asians in 2020.

Picture 1: Hate crimes against Asians (blue) and hate crimes against blacks or African Americans (yellow) in cumulative percentage change from 2016 (logarithmic terms). Source: FBI Crime Data Explorerand the author’s calculations.

Unlike Han et al. research, there is some evidence of persistence of reports of hate crimes against Asians.

For more on the limitations of FBI hate crime data and findings from other datasets, see Voice discussion.