Analysis of Special Investigations Unit Race-Based Data

Introduction

The opinions expressed in this report are those of the authors alone and have been derived at arm's length from the SIU.

The SIU conducted data collection, in accordance with the privacy standards stipulated in The Ontario Anti-Racism Act [2017] (hencefoth, the Act), and provided de-identified, aggregated data to the research team. The SIU also commented on previous drafts clarifying the data collection methodology and ensuring accuracy of descriptions. The data provided are used to conduct the required analyses under Order in Council 897/2018, and findings, opinions and recommendations have been derived without input from the SIU.

Understanding Race-Based Data at the Special Investigations Unit (SIU)

There are several complications to understanding Race-based data (RBD) as it pertains to the SIU.

The SIU is often erroneously associated with or compared to police services by members of the public. The SIU is a demand-service; unlike police services, the SIU has limited discretionary  authority to investigate incidents invoked under its mandate.

The SIU's mandate is typically invoked through notification by police services or members of the public, although there are some other means through which an investigation may be instigated. This limits, to some degree, inferences that can be drawn about the relationship between Race and investigations by the SIU. The report will expand on this throughout.

Executive Summary

In compliance with the Act, the SIU gathered RBD on Affected Persons and Subject Officials. These data were collected for cases between October 1st, 2020 and September 30th, 2021, and were provided to the authors on June 9th, 2022.

This report summarizes analyses that meet and exceed the requirements of the Act.

The authors address shortcomings of these data and make recommendations for how to improve data collection.

Response Rate

  • 398 surveys were distributed to Complainants/Affected Persons and 460 to Subject Officials for investigations commenced between October 1st, 2020 and September 30th, 2021.
  • Of those, 98 were returned by Affected Persons (response rate 25%) and nine by Subject Officials (response rate 2%). In death cases, the next-of-kin was provided with a survey to complete on behalf of the Affected Person.


The associated response rate affects both the quantity and quality of data collected. This report recommends augmenting the self-report survey with investigator­ produced reports of perceived Race to achieve a more robust dataset.

Separate RBD analyses were conducted for both the Affected Persons and Subject Official categories. 

Executive Summary - Affected Persons

An Affected Person is someone who died or was seriously injured during an incident involving an official, at whom a firearm was discharged by an official, or who alleged sexual assault by an official. For the purpose of data collection, in death cases, the next-of-kin was provided with a survey on behalf of the Affected Person. The data demonstrate (as compared to the proportional representation in the Ontario population):
  • Black and Indigenous people are represented more frequently.
  • People who identify as Black or a racial identity encompassing Black and another racial category are represented nearly 3.5 times more frequently.
  • People who identify as Indigenous are nearly 6.25 times more frequently represented.
  • People who identify as Latino, Middle Eastern, and Other are represented slightly more frequently.
  • People who identify as East or Southeast Asian, South Asian and White are either less frequently or significantly less frequently represented.


The SIU collected self-report information on religious affiliation. The data showed (as compared to the proportional representation in the Ontario population):
  • People who self-identified as Indigenous Spirituality were nearly 77 times more frequently represented.
  • People who identified as Jewish were twice as frequently represented.
  • People who identified as Other religious affiliation are four times as frequently represented.
  • People who identified as non-religious were effectively evenly represented.
  • People who identified as Muslim and Sikh were less frequently represented.

The SIU collected self-report information on gender. The data demonstrate (as compared to the proportional representation in the Ontario population):
  • People who identified as Men were 1.65 times more frequently represented.
  • People who identified as Women were nearly three times less frequently represented.
  • No respondent identified as Other in these data.


Executive Summary - Subject Officials

A Subject Official is, in respect of an incident referred to in Section 15 (1) of the SIU Act, an official whose conduct appears, in the opinion of the SIU Director, to have been a cause of the incident under investigation. In addition to municipal and provincial police officers in Ontario, the SIU mandate includes the investigation of the conduct of special constables with the Niagara Parks Commission and peace officers of the Legislative Protective Services. Collectively, these persons are known as "officials" under the SIU Act.This lack of knowledge significantly hinders the ability to perform Race-based analyses, as there is no known reference population from which to derive the inequality index.

Using Statistics Canada data, this report assumes that Ontario police services are more diverse (~ 15% visible minority representation) than the national average (8% visible minority representation).
  • Too few Subject Officials participated in the survey.
  • Of the nine respondents, eight self-identified as White and one self­ identified as Black.
  • No reasonable inference about Racial bias toward subject officials exhibited by the SIU could be drawn from either the data themselves or the relationship to the reference group.
The report recommends that SIU investigators be instructed to collect RBD on all Affected Persons and Subject Officials and argues that these would be more robust data to draw conclusions about Racial biases (or lack thereof) than self-report surveys.

Recommendations

Introduction


Caution should be exercised in reviewing the statistical details in this report. The SIU has extremely limited discretion on what cases it investigates. Inclusion in the dataset should lead to conclusions about policing in general, rather than conclusions about the SIU.

Policing is virtually defined by individual officers' ability to exercise discretion when dealing with the public.

Police have considerable latitude to decide who they will investigate, how they investigate, and what kind of force is at their disposal. None of this is the case for the SIU.The willingness of Racialized Affected Persons to participate in the data collection might be viewed as an overall positive sign of the SIU's relationship with Racialized people. Racial disproportionality at the SIU is, therefore, attributable to disproportionality in police services, and may not indicate the presence of discrimination.

This also complicates the findings from these data, and the report articulates that the data cannot be used to draw strong inference about the prevalence of systemic racism at the SIU. The data are profoundly voluntary, and disproportionality is as likely to demonstrate a willingness to participate in the research as it is to demonstrate any systemic bias or lack thereof at the SIU.

This limits the conclusions that can be drawn about how Race factors into SIU investigations when the method for gathering information about Race is self­ identification. Data collection shows how individuals identify, not how they are perceived by investigators.

If, for example, an individual identifies as Indigenous but investigators do not perceive Indigeneity, the dynamic between Race and outcome of investigation is obscured. Any data collection technique will present complications, although given the purpose of the Act, the data analysis team proposes that it is preferable to collect data based on investigator perceptions of Race rather than individual's self-report. This is not a solution without its own complications - investigators would need to understand the
purpose of this exercise and recognize the necessity for accurate reporting - but it will almost certainly result in more relevant data for identifying systemic discrimination (see recommendation 2). 

Limitations

If Racial discrimination were occurring at the SIU, it would most likely be evident at the decision stage rather than the investigative stage. For example, if it were the case that the SIU was settling a disproportionate number of cases involving some group by Memo (i.e. prior to full investigation) it may indicate bias. It would then be incumbent upon the SIU and researchers to better understand the relationship between Race and outcome.

However, the data in this report are insufficient to see any such trends. A longitudinal analysis would be required to draw out such phenomena.

This brings us to the serious limitations of this report, and suggestions for strategies to overcome those limitations. Some of these limitations should be very easy to overcome, others likely will require intervention by the Ministry of the Attorney General and the Solicitor General, as well as Ontario police services.

Overall Recommendations

  1. Quantity of Data - insufficient Subject Official data to draw any conclusions, insufficient Affected Persons data to draw statistical inference on outcomes.
  2. Quality of Data - self-report surveys are not an ideal instrument for assessing Racial bias in public service organizations of this type.
  3. Lack of Reference Populations - there is simply too little known about the demographic composition of Ontario police services to form an adequate disparity index.
  4. Not enough Information to Draw Statistical Inference - longitudinal studies of outcomes will be more likely to provide insights.
  5. Lack of Nuanced Understanding - need to better understand how Race impacts SIU investigations, rather than how frequently Racialized people are subjects to investigations.

1. Improve the quantity of data by augmenting data collection methodology.

Survey response rates, particularly among Subject Officials, are simply too low to be able to draw any significant observations about systemic racism at the SIU. There is little reason to suspect that response rates will improve organically. Augmenting the current data collection methodology with alternate data collection will solve this problem.

Looking to the recently released report by the Toronto Police Service in fulfilment of their Act mandate is instructive. Researchers working with TPS used provincially mandated use-of-force and strip-search data that included information on the Race of subjects to these police procedures. This resulted in a large dataset from which reasonable statistical inference about systemic racism at TPS could be derived.

This report recommends that the SIU undertakes similar data collection practices, mandating investigators to standardly report perceived Race of Affected Persons and Subject Officials as a routine reporting mechanism .
This should be done independently, and in advance of, the self-report data collection methodology, and should be used to augment the self-report methodology. By taking this step, the SIU will be in a position to provide more data for analysis, and data that are more relevant to assessing systemic bias (see recommendation 2).

There are benefits and drawbacks to this approach that will be discussed in the next recommendation, but the sparsity of data from Subject Officials does indicate a need for an alternative data collection methodology.

2. Improve the quality of data at the same time as improving the quantity by collecting investigator perception of race.


The objective of the Act is to eliminate systemic racism and advance Racial equality. A more thorough understanding of if and how Race factors into SIU interactions with Affected Persons and Subject Officials is required in order to meet these ends. One major complication with current data collection practices is self-report surveys do not necessarily reflect Race as perceived by SIU investigators. How someone identifies may be distinct from how they are perceived by others.

We again look to the example set by TPS and other Ontario police services to collect RBD in a manner that reflects the actions of the organization rather than the expressions of individuals who interact with the organization. By mandating SIU investigators collect RBD in a systematic manner, the gap in data and knowledge can be filled and systemic racism's presence and effect at the SIU can be determined.

Recommendation one states that using investigator reports of Race will improve the quantity of data. We also argue here that recording investigator perception of race improves the quality of data. Understanding perceived Race gives better understanding of how members of the SIU react to the Race of Affected Persons and Subject Officials.

Accuracy and fairness of reporting are at stake if SIU investigators report Race independently of Affected Persons or Subject Officials self-identifications. However, this would be preferable to the current situation, where too little data exists to draw statistical inference about systemic racism, particularly among Racialized police officers in the province of Ontario.

3. Address lack of reference population by standardizing the collection and reporting of police service demographics in Ontario.

The Racial disparity index relies on known proportions of the population who identify as the Racial categories proscribed in the Act. Knowing if a Racialized group is experiencing disproportionate outcomes relative to representation in the population is a key factor in understanding the prevalence and impact of systemic racism.

Police services in Ontario do not collect and report service demographic information in a standardized manner. This is a huge roadblock to understanding disproportionate outcomes for Racialized police officers who are subjects of SIU investigations.

This report recommends that the SIU consult with the Solicitor General and/or Ministry of the Attorney General to produce accurate, standardized reporting of police demographics. This would also comply with, if not be required by, the Act, since discrimination in hiring is a contributing source to Racial disparity outcomes in policing.



4. Address the information gap with year-over-year reporting of Race-Based Statistics at the Special Investigations Unit.

The SIU's mandate is invoked approximately 300 to 400-times in any year. While this is not an insignificant number of incidents, no single year of reporting is likely to demonstrate the prevalence of Racially disparate outcomes.

This report suggests that statistical methodologies be developed to track the inequality index both in-year and year-over-year to see how trends on Racially disparate outcomes evolve through the data. The data analysis team is committed to this RBD analysis through to 2024, at which point more funding will be required to continue these activities. The data analysis team will work to systematize the standardized statistical reporting procedures required in the Act so the function can be performed by the SIU or another party should future funding for the project not be secured.

However, the capacity to draw theoretically-informed inference from these statistical reporting mechanisms is almost certainly limited to academic practitioners in law, criminology, sociology, and cognate disciplines.

The data analysis team's ambition is to secure long-term funding to continue analyses for the SIU, as well as developing comparative analyses with other civilian police oversight services.



5. Addressing the Understanding Gap through expanded interview­ based research to analyze how Race is perceived to be a factor in SIU investigations.

As we have noted, the SIU is an on-demand service. SIU investigators have very limited discretion to select cases. As a result, nearly all of the statistical observations articulated above are indicative of factors outside of the SIU's control, such as when their mandate is invoked in response to police activity.

This heightens the requirement of the SIU to understand how Race factors into the investigative process, rather than how frequently Racialized people are parties to investigation.

The research team will provide recommendations to the SIU for how to undertake these activities including a recommendation to conduct qualitative interviews with both Affected Persons and Subject Officials who have been parties to SIU investigations. It is important to note the complexities for qualitative data collection and analyses in the context of the SIU. Interview participants would be asked to reflect on challenging and, in some cases, traumatic experiences. They would be asked to share sensitive information, including their perception of how Race impacted the outcomes of investigations to which they were parties. Conducting interviews on Race-based data requires considerable experience, sensitivity, and expertise.

Interviews would assist in better understanding what the consequences of disproportionality means at the SIU.

Methodology

The SIU collected RBD by distributing surveys to Affected Persons and Subject Officials for investigations commenced between October 1st, 2020 and September 30th, 2021. In death cases, the next-of-kin was provided with a survey to complete on behalf of the Affected Person. Recipients were asked to return completed surveys to the SIU, either directly to an investigator or Affected Persons Program staff, or by mail or telephone. If help was requested by the survey taker, the investigator or Affected Persons Programs staff could assist by recording the responses directly into the survey. The SIU entered disaggregated data from these surveys into a spreadsheet and shared these data with the report's authors.




The report aggregates these data as appropriate to conduct the Racial disparity index while retaining disaggregated data to conduct intersectional analyses. See Ontario Order in Council 897/2018, Anti-Racism Data Standards, S. 29, Racial Disproportionality and Disparity Index. Reference populations for Affected Persons were drawn from StatsCan Census Profile 2016.

Some data manipulation was required to gain alignment between StatsCan categories of visible minorities with those mandated in the Act. The Census Profile uses a variety of Ethnic/National categories, such as "Chinese", "Japanese", "Filipino", etc. and these were coded according to the formalized categories provided for in the Act. This produces a relatively robust reference population from which to produce the Racial disparity index.

Subject Official reference populations were derived from StatsCan Police Resources in Canada, 2019 report. The report states that, across Canada, 4% of police officers identify as Indigenous, 8% of officers identify as visible minorities. However, Toronto Police Service and York Regional Police are both cited in this report as having higher visible minority representation (22%, 19% respectively).

The report authors consulted sources including StatsCan, to acquire better reference data for Ontario's police population. No such data were available. Police annual reporting in Ontario occasionally reports officer demographics, but data collection is inconsistent between services, and some services do not publicly report demographic information. Due to the lack of available demographic information about police services in Ontario, this report used an estimated figure of 15% of Ontario police officers as representatives of Racialized and/or visible minority groups. The report acknowledges the shortcomings of this methodology, but justifies it on the grounds that statistical inference could not be drawn from the data returned to the SIU by Subject Officials.

In addition to the Racial disparity index, the report presents intersectional analyses based on SIU investigative outcome for Affected Persons. The intersectional analyses focused on people who identified as Black and Indigenous, as the most overrepresented groups in the data, and people who identified as White as the most frequently occurring group in the dataset. The report analyzes intersections of Race, gender, age, and investigative outcome.

Race categories defined in the Act included Indigenous, Black, East/Southeast Asian, Latino, Middle Eastern, South Asian, White and Other Race. Gender categories collected were Male, Female and Other. The SIU asked respondents for their age. In the dataset Affected Persons have a mean age of 39.3125, a median age of 37.5 and a mode age of
37. Intersectional analyses separate Affected Persons into two categories - older and younger - based on the median age of 37.5, to form two equal groups. The median age of Subject Officials is 39.1111. However, because statistical inference cannot be drawn from the Subject Official data set, no further intersectional analyses were conducted.



The Act stipulates that outcome measures be considered, although it should be noted that approximately 97% of SIU investigations are resolved without charges, thus limiting the statistical inferences that can be drawn by outcome. Of 98 cases reported in the Affected Persons data set, two resulted in charges. Therefore, these data are roughly representative of the statistical average for cases resulting in a charge decision by the SIU.
This would comply with Section 32. Setting Thresholds to Identify Notable Difference, Further Assessments to Understand Potential Racial Inequalities. When complete, the analysis team will deliver this proposal to the SIU.

These interviews would add depth and sophistication to the statistical analyses, providing information beyond the mere quantity of individuals who are parties to investigation.

Results

1.0 Aggregated Demographics of Affected Persons Respondents

The following aggregated data were drawn from the spreadsheet provided by the SIU:

2.0 Disproportionality

The Racial disparity index was prepared for the Affected Persons population. The table articulates relative Racial disparity by group:

 
  Black E/SE Asian Indigenous Latino Middle Eastern South Asian White Other
Black 3.497881 8.410472 0.561356 2.497311 3.102087 29.32325 3.974872 2.497184
E/SE Asian 0.118899 0.415896 0.066745 0.296929 0.368836 3.486516 0.47261 0.296914
Indigenous 1.7814 14.98241 6.231124 4.448709 5.526057 52.23642 7.080835 4.448484
Latino 0.400431 3.367811 0.224784 1.400659 1.242171 11.74193 1.591661 0.999949
Middle Eastern 0.322364 2.71123 0.180961 0.805042 1.12759 9.452747 1.281354 0.805001
South Asian 0.034103 0.286819 0.019144 0.085165 0.105789 0.119287 0.135554 0.085161
White 0.25158 2.11591 0.141226 0.628275 0.780425 7.377156 0.879998 0.628243
Other 0.400451 3.367982 0.224796 1.000051 1.242234 11.74252 1.591741 1.40073

The bolded figures represent disproportionality relative to Ontario population. All other figures represent disproportionality relative to cross-referenced group.
  • People identifying as Black are nearly 3.5 times more frequently represented in this dataset to proportion of people identifying as Black in the Ontario population.
  • People identifying as White are underrepresented in this data 1.14 times less frequently in proportion of people identifying as White in the Ontario population.
  • People who identify as Black are nearly four times more frequently represented in this data than people who identify as White.
The data indicate (relative to proportion of the Ontario population):
  • People who identify as Black or Indigenous are markedly more frequently represented in data collected about Affected Persons relative to the Ontario population of people who identify as Black or Indigenous.
  • People who identify as Latino, Middle Eastern, or Other Race are moderately overrepresented.
  • People who identify as White are moderately underrepresented.
  • People who identify as East or Southeast Asian and South Asian are markedly underrepresented in these data.
Data were also collected on religious affiliation. StatsCan's 2016 census data was used to derive reference populations and a disproportionality index was constructed. The SIU survey had a write-in section for religious affiliation. Respondents who wrote-in "Catholic", "Roman Catholic" or "Protestant" were recoded as "Christian" for analysis. Respondents who wrote in other, non­ formalized religious categories were recoded as "Religious Other":

  Buddhist Christian Hindu Jewish Muslim Sikh Indigenous Spirituality None Other
Buddhist 0 0 0 0 0 0 0 0 0
Christian 0 0.807285 0 0.395697 0.975738 1.107596 0.010549 0.770628 0.197785
Hindu 0 0 0 0 0 0 0 0 0
Jewish 0 2.527187 0 2.04016 2.465872 2.799102 0.026658 1.947522 0.499839
Muslim 0 1.024865 0 0.405536 0.827358 1.135137 0.010811 0.78979 0.202703
Sikh 0 0.902856 0 0.357257 0.880951 0.728862 0.009524 0.695767 0.178571
Indigenous Spirituality 0 94.79999 0 37.51206 92.49996 105.0001 76.53061 73.05555 18.75
None 0 1.297643 0 0.513473 1.266159 1.437264 0.013688 1.047567 0.256654
Other 0 5.056 0 2.000644 4.933332 5.600005 0.053333 3.896297 4.081633

These data indicate (relative to proportion of the Ontario population):
  • Individuals who identify as Jewish, Indigenous Spirituality, and Other religious affiliation are more frequently or extremely more frequently represented.
  • All other religious affiliations, including people who identify as no religious affiliation, are represented as nearly proportionate or underrepresented.
Data on gender were also collected, using the categories Man, Woman and Other:

These data indicate (relative to proportion of the Ontario population):
  • People who identify as Man are moderately overrepresented.
  • People who identify as Woman are significantly underrepresented.
  • No individuals identified as other than Man or Woman in the dataset. 
  Man Woman Other
Man 1.652765 4.366896 0
Woman 0.228996 0.378476 0
Other 0 0 0

3.0 Intersectionality

The report conducts intersectional analyses on Black and Indigenous identifying individuals, as well as White individuals as the majority Race identified in the survey data due to the disproportionality in representation relative to the Ontario population.

Intersectional analyses centered on Race, Gender, Age, Investigation Type and Outcome. These data indicate (relative to proportion of the Ontario population):
  • Indigenous men under 37.5 years of age comprised the most overrepresented respondents.
  • Younger Indigenous men are approx.18-times more frequently represented in this dataset.
  • Older Indigenous men and younger Black men also responded to the survey 15- times and 13-times more frequently.
 
  Black Man
< 37 .5
Black Man
>37 .5
Indigenous Man
<37.5
Indigenous Man
>37.5
White Man
<37 .5
White Man
>37.5
Black Man < 37.5 13.0769 5.613367102 0.737179451 0.897436073 4.921308144 9.755240582
Black Man >37.5 0.178146197 2.3296 0.131325716 0.159874823 0.876712329 1.737859008
Indigenous Man <37.5 1.356521806 7.614654876 17.7391 1.217391603 6.675861809 13.23319657
Indigenous Man >37.5 1.114285496 .254893544 0.82142837 14.5714 5.483742285 10.87012309
White Man <37.5 0.203198006 1.140625 0.149793394 0.18235722 2.6572 1.982245431
White Man >37 .5 0.102509004 0.575420673 0.075567532 0.091995278 0.504478398 1.3405

These data indicate (relative to proportion of the Ontario population):
  • Black women both under and over the age of 37.5 and younger Indigenous women respond to the survey nearly or over twice as frequently.
  • Indigenous women older than 37.5 were not represented in these data at all. Only one younger Indigenous woman responded to the survey.
  • White women both over and under the age of 37.5 were represented in these data at a fraction of the proportion of the population.
  Black Woman
< 37 .5
Black Woman
>37 .5
Indigenous Woman
<37.5
Indigenous Woman
>37.5
White Woman
<37 .5
White Woman
>37.5
Black Woman < 37.5 2.372 1.220855422 1.162745098 0 5.456636761 5.964294694
Black Woman >37.5 0.819097808 1.9429 0.952401961 0 4.469519209 4.885340709
Indigenous Woman <37.5 0.860033727 1.049976839 2.04 0 4.692891649 5.129494594
Indigenous Woman >37.5 0 0 0 0 0 0
White Woman <37.5 0.183263069 0.22373771 0.213088235 0 0.4347 1.093034951
White Woman >37 .5 0.167664418 .204694014 0.19495098 0 .914883828 0.3977
These participation rates can reflect two different phenomena:
  • The disproportionality of participation by Black and Indigenous people may show the higher risk of these individuals being subjects to interactions with officials, or
  • These individuals are more willing to participate in the data collection exercise. 
More needs to be understood about what motivates individuals to participate (or not) in the data collection to draw conclusions about overrepresentation in these data.

4.0 Investigation Types

The SIU's mandate is invoked in any incident where officials are involved resulting in serious injury or death, where an allegation of sexual assault has been made against an official, or where a firearm is discharged by an official at a person. The SIU gathered data on investigation type to which survey respondents were parties.


Affected Persons
Custody Death 18 (18%) Custody Injury 44 (45%)
Firearm Death 5 (5%) Firearm Injury 3 (3%)
Firearm Discharged at a Person 5 (5%) Other Death 3 (3%)
Sexual Assault Alleg. 9 (9%) Other Injury 0(0%)
Vehicle Injury 10 (10%) Vehicle Death 1 (1%)

Subject Officials
Custody Death 2 (22%) Custody Injury 3 (33%)
Firearm Death 0 (0%) Firearm Injury 1 (11%)
Firearm Discharged at a Person 1 (11%) Other Death 0 (0%)
Sexual Assault Alleg. 0 (0%) Other Injury 0(0%)
Vehicle Injury 2 (22%) Vehicle Death 0 (0%)

In both the Affected Persons and Subject Officials datasets:
  • Custody Death and Injury, Firearm Death, Injury and Discharge at a Person, and Other Death comprise the majority of survey responses.
  • 80% of Affected Persons and 77% of Subject Official surveys fall into these categories.
Comparing these results with the SIU Annual Report 2020-21 gives some indication of the proportionality of responses by case type.

In fiscal year 2020-2021, the SIU investigated a total of 390 cases as follows:

Custody Death 34 (9%) Custody Injury 201 (52%)
Firearm Death 12 (3%) Firearm Injury 12 (3%)
Firearm Discharged at a Person 7 (2%) Other Death 3 (<1%)
Sexual Assault Alleg. 63 (16%) Other Injury 1(<1%)
Vehicle Injury 49 (13%) Vehicle Death 8 (2%)


Comparative analysis of survey responses by investigation type against SIU total recorded investigation types can provide insight on the types of cases that are more likely to invoke a survey response.


Survey Response Affected Person Case Types vs. Total SIU Investigation Types
Type Survey SIU Total %Dif  
Custody Death 18 (18%) 34 (9%) +100% CD
Custody Injury 44 (45%) 201 (52%) -13.5% CI
Firearm Death 5(5%) 12 (3%) +66.7% FD
Firearm Injury 3(3%) 12 (3%) 0% FI
Firearm Discharge at a Person 5(5%) 7 (2%) +150% FP
Other Death 3(3%) 3 (<1%) +300% OD
Other Injury 0 1(<1%) 0 OI
Sexual Assault 9 (9%) 63 (16%) -43.8% SA
Vehicle Death 1 (1%) 8 (2%) -50% VD
Vehicle Injury 10 (10%) 49 (13%) -23.1% VI

This analysis shows us that:
  • Affected Persons for official-involved deaths (i.e. deceased's survivors) are proportionately more likely participate in the survey relative to the number of official-involved death investigations undertaken by the SIU.
  • Affected Persons where the outcome is death (CD, FDOD) participated more frequently in the survey than all other groups.
  • All other investigation types participated less frequently in the survey relative to the number of investigations by type conducted by the SIU.
In these data (as compared to proportion of the population.):
  • Younger and older Indigenous men are respectively 13- and 16-times more represented.
  • Younger Black men are 15-times more represented in these data .
  • Younger White men are nearly twice as frequently represented.
  • Older White men appear 1.6-times as frequently.

  • Younger Black women were nearly three-times as frequently represented in these data for Custody Injuries.
  • Older Black women were just over twice as frequently represented.
  • White women, both older and younger, were underrepresented in these data four times less frequently.
Readers should note, these data are for rough comparative purposes only. The collection periods for the RBD was October 2020 to September 2021, whereas the SIU's annual report covers cases investigated between April 2020 and March 2021. As such, the two data sets do not perfectly intersect.

However, the comparative value stands, as the proportion of case types investigated by the SIU year-over-year remains fairly consistent, meaning these data are good enough for devising what case types most frequently result in participation in RBD collection.

From there, the analysis team may be able to make some inferences toward what motivates respondents to participate in RBD collection, an important step to comprehending what disproportionality in Racial representation means at the SIU.

5.0 Resolution by Memo

An area in which the report took an interest was Outcome Memo.

In the SIU Annual Report 2020-2021, 26.2% of cases are resolved by memo. In some closed by memo cases, the SIU determines that there is patently nothing to investigate (i.e. upon reviewing medical documents, discovering there is no threshold injury to accompany the complaint). When this is the case, the SIU does not attempt RBD collection as the complainant does not fall within the statutory definition of "affected persons" as defined by the SIU Act. However, if RBD is collected before it is known that the case will result in memo, these data are retained.

Memo is one occasion where the SIU director can exercise some discretion whether to investigate incidents that fall within the SIU mandate. It can also be the case that, following an initial investigation, it is discovered that the SIU does not have mandate and/or jurisdiction to investigate any further.

The Annual Report 2020-2021 presents two exemplar cases resolved by memo. In these cases, the SIU Director invoked his discretion to close the investigation prior to completion based on preliminary evidence reviewed or found upon initial investigations that there was patently nothing further to investigate. In the first case, the custody suite video evidenced that injuries were caused by the Affected Person, not Subject Officials, and therefore there was patently nothing further to investigate. In the second case, following officer and civilian witness interviews, it was determined Subject Officials did not contribute to the death of an individual to whom those officers administered first-aid.

There are some difficulties with drawing conclusions about the prevalence of systemic racism in memo decisions. In some cases an investigation will have been closed prior to collecting RBD, meaning there will be a report of a commenced investigation closed by memo, but no opportunity for the Subject Official or Affected Person to provide RBD. In other cases closed by memo, RBD will have been collected prior to closure. The result is a data set that does not necessarily accurately reflect the racial composition of cases closed by memo due to data collection limitations. That said, given the Act's mandate to analyze outcome measures, we proceeded with our analysis, acknowledging and accepting the shortcoming.

The report analyzed whether Memo was used more frequently to close cases of Black, Indigenous, or White people in the dataset. Intersectional analyses were run on the above-mentioned groups against the outcome "Memo".

16 cases in this dataset were resolved by Memo. One of those cases did not include age data and was removed from intersectional analysis - identity features selected in that case were "White" and "Woman". A second case self-identified as Black, White, Indigenous, and Male. In accordance with the instructions for RBD Analysis in Ontario Order in Council 897/2018, this individual was taken out of the code "White" but
included twice in intersectional analyses, once for "Black" and once for "Indigenous". 

Therefore, data are included for 17-cases of resolution by Memo( n=l6), one case is removed for missing age data, and one case is counted twice for the two separate self-identifications:

Men
Black
< 37 .5
Black
>37 .5
Indigenous
<37.5
Indigenous
>37.5
White
<37 .5
White
>37.5
2 (13%) 0(0%) 2 (13%) 2 (13%) 3 (19%) 4 (25%)

Women
Black
< 37 .5
Black
>37 .5
Indigenous
<37.5
Indigenous
>37.5
White
<37 .5
White
>37.5
0 (0%) 0 (0%) 0 (0%) 0 (0%) 2 (13%) 1 (6%)

To determine disproportionality of outcomes based on intersectional analyses, the report made as-close-to-accurate-as-possible estimations of demographic characteristics of Ontario. This report was able to derive approximate populations of Black and White Men and Women in Ontario from StatsCan database of visible minority status. The report uses 2016 Census StatsCan Focus on Geography series to derive Indigenous population by age and gender.

Men
  BMU BMO IMU IMO WMU WMO
BMU 16.0256 0 0.589742 0.717948 8.205212 12.19883
BMO 0 0 0 0 0 0
IMU 1.695656 0 27.1739 1.217392 13.91321 20.68501
IMO 1.392859 0 0.821428 22.3214 11.4287 16.99125
WMU 0.121874 0 0.071874 0.087499 1.9531 1.486717
WMO 0.081975 0 0.048344 0.058854 0.672623 1.3137

Women
WRAO BWU BWO IWU IWO WWU WWO
BWU 0 0 0 0 0 0
BWO 0 0 0 0 0 0
IWU 0 0 0 0 0 0
IWO 0 0 0 0 0 0
WWU 0 0 0 0 1.3312 0.228816
WWU 0 0 0 0 4.370322 0.3046

These data indicate (relative to proportion of the Ontario population):
  • Black and Indigenous men are markedly overrepresented in this dataset.
  • White men are marginally overrepresented in this dataset.
  • No Black or Indigenous women had cases resolved by Memo in this dataset.

6.0 Data Collection Limitations 

To draw stronger conclusions about what these data indicate, analyses would need to be conducted of the response rates by Racial group. In other words, it would be necessary to know, of all Black and Indigenous Men who were given the opportunity to participate in the survey, what proportion did so? Without this information, it is impossible to infer the degree to which Black and Indigenous Men are experiencing cases resolved by Memo, and the cause of that disproportionality. These data indicate the need to conduct qualitative interviewing with Affected Persons to better understand the perception of Racial bias and disproportionality by the SIU.

Of the 98 Affected Persons cases in these data, 76 cases were outcome No Charge (78%), 16 outcome Memo (16%), 4 outcome Outstanding (4%), 2 outcome Charge (2%).

7.0 Qualitative Interviewing


The analysis team identified a significant complication with the data represented above; that the SIU's limited discretion to choose what cases it investigates means limited inference can be drawn from the overrepresentation of Racialized persons in the data. The data above likely say more about police interactions with Racialized peoples than SIU interactions, given how the SIU's mandate is invoked. Because police services are more likely to use force against Racialized people the SIU's mandate is more likely to be invoked in those interactions. Inclusion in this data set indicates two things: the respondent had been seriously injured or killed while interacting with an official, is next-of-kin of a deceased Affected Person, had a firearm discharged at them by an official, or alleged sexual assault against an official; and the respondent has volunteered their information to the SIU via the survey.

To better understand how Race factors into SIU investigations, it is important to analyze the relationship between Subject Officials or Affected Persons and the SIU. This requires speaking with these individuals about their experience and learning how the individuals themselves see their Race as a contributing factor to SIU investigations and outcomes.

Instructions from the Anti-Racism Data Standards

In Data Standard 32 (Setting Thresholds to Identify Notable Differences), the Anti­ Racism Data Standards - Order in Council 897/2018 gives the following instruction:

Racial disproportionality or disparities on their own may not be conclusive of systemic racial inequalities.

Methods of further analysis could focus on determining the extent to which a racial disproportionality may be attributed, in whole or in part, to systemic racism. Multivariate analysis is one method used to identify other factors, such as socioeconomic conditions, that may help explain differences in group outcomes.

Draw on other sources of information to help in the interpretation and understanding of findings. PSOs should use multiple methods, such as qualitative information obtained through focus groups, individual interviews with clients, employees, and experts, police and program evaluations, research literature reviews, etc.

It is our opinion that this is a crucial endeavour to achieve the Act's objectives. Consulting with (i.e. interviewing) both Affected Persons and Subject Officials will help improve interpretation of RBD gathered for this report. Disproportionality, in this data, is expected. The question that remains is how that disproportionality is experienced by individuals who are party to SIU investigations.

Given the impact incidents the SIU investigates have on Affected Persons and Subject Officials, the preferred qualitative methodology is individual interviews (as opposed to focus groups).

Conclusion

The SIU's collection of RBD in compliance with the Act is a marked step forward in understanding how Race factors into investigations conducted by the SIU. It is a landmark opportunity to understand more about the SIU's relationship with racialized people, although the current data collection methodology is as likely to be a commentary on policing in Ontario as commentary on the SIU itself. Adding further methods of data collection and analysis will be crucial for better understanding how the SIU is performing in interactions with Racialized people.

The SIU's commitment to Racially sensitive investigative practice is to be commended. This report is an important first step in understanding how the SIU is performing on behalf of Racialized people who are disproportionately subjects to police use-of-force incidents, and Women who are disproportionately complainants of alleged sexual assaults by police officers.

However, there is much more that remains to be done. This report recognizes the accomplishment of gathering and beginning to analyze RBD, and looks forward to developing that understanding in the coming years.

Acknowledgements

We thank our fellow researchers, Dr. Carmen Nave, Dr. Jason Turowetz and Dr. Waverly Duck, and our research assistant, Hannah Feldbloom, for their assistance in preparation of aspects of this report. We are also grateful to Dr. Scott Blandford for his assistance. Report design by S. Reibling.

This report draws on research supported by the Social Sciences and Humanities Research Council.