Correction to A Disparate Impact

Correction to A Disparate Impact

Following the release of A Disparate Impact, Drs. Wortley and Laniyonu identified a coding error in which civilian race was incorrectly coded in the multivariate analysis. The Ontario Human Rights Commission (OHRC) retained Dr. Maria Jung[1] to independently review the error.

The multivariate analysis of use of force in A Disparate Impact,[2] among other things, described whether racial disparities in “lower-level” use of force and use of force that resulted in death or serious injury persisted after controlling for patrol zone characteristics (i.e. violent crime rate, median household income, percentage of single mother households).

A Disparate Impact[3] and Use of Force by the Toronto Police Service[4] incorrectly stated that when controlling for patrol zone characteristics:

• Black people were 55 times more likely to experience serious and lower-level use of force than White people
• Other racialized people were 13 times more likely to experience serious and lower-level use of force than White people.

The racial disparities in the multivariate analysis decrease significantly when correcting for the error. Further, other racialized people, when grouped together in the data, are now less likely than White people to experience use of force when controlling for patrol zone characteristics.[5] However, there is still a gross racial disparity in the risk that Black people will experience force compared to White people, which remains “consistent with racial bias arguments:”[6]

As a result of this independent review, the analysis shows that Black people are markedly more likely to experience all types of police use of force compared to their White counterparts. These gross racial disparities remain after statistically controlling for patrol zone characteristics, including violent crime rate, median household income, and proportion of single-mother households. This is consistent with the conclusions drawn in the original report. However, the extent of racial disparity is smaller in the corrected analysis than the original report. Instead of 30–58 times the risk of experiencing use of force experienced by Black civilians compared to White civilians, as noted in the original report, the results of this corrected analysis show that Black civilians are 4–5 times more likely to experience use of force relative to their White counterparts. 

However, for civilians of other racial minority groups, the conclusions from the corrected analysis are substantially different from those drawn in the original report. In the original report, civilians of other racial minority groups were 5–14 times more likely to experience use of force compared to their White counterparts, controlling for patrol zone characteristics, including violent crime rate, median household income, and proportion of single-mother households. In the corrected analysis, civilians of other racial minority groups are about 40% less likely than their White counterparts to experience use of force, controlling for patrol zone characteristics.

The independent review confirmed that “White civilians were incorrectly coded as Black civilians; Black civilians were incorrectly coded as belonging to some other racialized minority groups; civilians of other racialized minority groups were incorrectly coded as individuals where race could not be identified; and persons whose race could not be identified were coded as White.”[7] The error occurred when data was transferred from one statistical analysis program (SPSS) and put it into another statistical analysis program (R).[8]

After discovering the error, Drs. Wortley and Laniyonu reviewed and corrected the expert use of force report. The tables presented in the independent review are “essentially and substantially the same as the corrected series of tables”[9] in the expert use of force report. 

For more information, see:

1. Dr. Jung, Independent Expert Review of the Data, Analysis, and Conclusions of “Part E: Multivariate Analysis of Use of Force Cases” of the Use of Force by the Toronto Police Service report (December 2022)
2. A Disparate Impact, which has been corrected (see “Alternative explanations”)
3. Use of Force by the Toronto Police Service, which has been corrected (see p. 116–126)