In May 2025, questions measuring the Labour Market Indicators were added to the Labour Force Survey as a supplement.

Questionnaire flow within the collection application is controlled dynamically based on responses provided throughout the survey. Therefore, some respondents will not receive all questions, and there is a small chance that some households will not receive any questions at all. This is based on their answers to certain LFS questions.

Labour Market Indicators

ENTRY_Q01 / EQ 1 - From the following list, please select the household member that will be completing this questionnaire on behalf of the entire household.

WFH_Q01 / EQ 2 - At the present time, in which of the following locations do you usually work as part of your main job or business?

1. At a fixed location outside the home
2. Outside a home with no fixed location
3. At home

WFH_Q03 / EQ 3 - Among those locations, where do you usually work the most hours?

1. At a fixed location outside the home
2. Outside a home with no fixed location
3. At home

CCOMM_Q01 / EQ 4 - What modes of commuting do you usually use to get to work [when working outside the home]?

1. Car, truck or van — as a driver
2. Car, truck or van — as a passenger
3. Bus
4. Subway or elevated rail
5. Light rail, streetcar or commuter train
6. Passenger ferry
7. Walked to work
8. Bicycle
9. Motorcycle, scooter or moped
10. Other method

CCOMM_Q02 / EQ 5 - What main mode of commuting do you usually use to get to work [when working outside the home]?

1. Car, truck or van — as a driver
2. Car, truck or van — as a passenger
3. Bus
4. Subway or elevated rail
5. Light rail, streetcar or commuter train
6. Passenger ferry
7. Walked to work
8. Bicycle
9. Motorcycle, scooter or moped
10. Other method

CCOMM_Q03 / EQ 6 - How many workers, including yourself, usually ride in this car, truck or van to work?

Would you say:

• 1 worker
• 2 workers
• 3 or more workers

CCOMM_Q04 / EQ 7 - How many minutes does your trip to work usually last?

Number of minutes _____

WRK_Q01 / EQ8 - On which of the following days do you usually go to your worksite in your main job or business?

1. Monday
2. Tuesday
3. Wednesday
4. Thursday
5. Friday
6. Saturday
7. Sunday

OR

  8. It varies from week to week

WFH_Q02 / EQ 9 - Last week, what proportion of your work hours did you work at home as part of your main job or business?

Would you say:

1. All hours at home
2. More than half, but not all at home
3. One quarter to half at home
4. Less than a quarter at home
5. No hours at home

Why are we conducting this survey?

This survey is conducted by Statistics Canada in order to collect the necessary information to support the Integrated Business Statistics Program (IBSP). This program combines various survey and administrative data to develop comprehensive measures of the Canadian economy.

The statistical information from the IBSP serves many purposes, including:

• Obtaining information on the supply of and demand for energy in Canada.
• Enabling governmental agencies to fulfill their regulatory responsibilities in regards to public utilities.
• Enabling all levels of government to establish informed policies in the energy area.
• Assisting the business community in the corporate decision-making process.
• Your information may also be used by Statistics Canada for other statistical and research purposes:
• Supporting the government in making informed decisions about fiscal, monetary and foreign exchange policies.
• Enabling academics and economists to analyze the economic performance of Canadian industries and to better understand rapidly evolving business environments.

Your information may also be used by Statistics Canada for other statistical and research purposes.

Your participation in this survey is required under the authority of the Statistics Act.

Other important information

Authorization to collect this information

Data are collected under the authority of the Statistics Act, Revised Statutes of Canada, 1985, Chapter S-19.

Confidentiality

By law, Statistics Canada is prohibited from releasing any information it collects that could identify any person, business or organization, unless consent has been given by the respondent, or as permitted by the Statistics Act. Statistics Canada will use the information from this survey for statistical purposes only.

Record linkages

To enhance the data from this survey and to reduce the response burden, Statistics Canada may combine the acquired data with information from other surveys or from administrative sources.

Data sharing agreements

To reduce the response burden, Statistics Canada has entered into data sharing agreements with provincial and territorial statistical agencies and other government organizations, which have agreed to keep the data confidential and use them only for statistical purposes. Statistics Canada will only share data from this survey with those organizations that have demonstrated a requirement to use the data. Section 11 of the Statistics Act provides for the sharing of information with provincial and territorial statistical agencies that meet certain conditions. These agencies must have the legislative authority to collect the same information, on a mandatory basis, and the legislation must provide substantially the same provisions for confidentiality and penalties for disclosure of confidential information as the Statistics Act. Because these agencies have the legal authority to compel businesses to provide the same information, consent is not requested and businesses may not object to the sharing of the data.

For this survey, there are Section 11 agreements with the provincial and territorial statistical agencies of Newfoundland and Labrador, Nova Scotia, New Brunswick, Quebec, Ontario, Manitoba, Saskatchewan, Alberta, British Columbia and the Yukon.

The shared data will be limited to information pertaining to business establishments located within the jurisdiction of the respective province or territory. Section 12 of the Statistics Act provides for the sharing of information with federal, provincial or territorial government organizations. Under Section 12, you may refuse to share your information with any of these organizations by writing a letter of objection to the Chief Statistician, specifying the organizations with which you do not want Statistics Canada to share your data and mailing it to the following address:

Chief Statistician of Canada Statistics Canada Attention of Director, Enterprise Statistics Division 150 Tunney’s Pasture Driveway Ottawa, Ontario K1A 0T6

You may also contact us by email at Statistics Canada Help Desk or by fax at 613-951-6583.

For this survey, there are Section 12 agreements with the statistical agencies of Prince Edward Island, Northwest Territories and Nunavut, as well as with the provincial government ministries responsible for the energy sector, Natural Resources Canada and Environment and Climate Change Canada.

For a complete list of the provincial government ministries responsible for the energy sector, you can visit the following link: Information for survey participants.

Note that there is no right of refusal with respect to sharing the data with the Saskatchewan Ministry of the Energy and Resources for businesses also required to report under The Oil and Gas Conservation Act and Regulations (Saskatchewan) and The Mineral Resources Act (Saskatchewan).

The Saskatchewan Ministry of the Energy and Resources will use the information obtained from these businesses in accordance with the provisions of its Acts and Regulations.

For agreements with provincial and territorial government organizations, the shared data will be limited to information pertaining to business establishments located within the jurisdiction of the respective province or territory.

Business or organization and contact information

1. Verify or provide the business or organization’s legal and operating name, and correct information if needed.

Note: Legal name should only be modified to correct a spelling error or typo.

Note: Press the help button (?) for additional information.

• Legal name
• Operating name (if applicable)

2. Verify or provide the contact information for the designated contact person for the business or organization, and correct information if needed.

Note: The designated contact person is the person who should receive this questionnaire. The designated contact person may not always be the one who actually completes the questionnaire.

• First name
• Last name
• Title
• Preferred language of communication
• Mailing address (number and street)
• City
• Province, territory or state
• Postal code or ZIP code Example: A9A 9A9 or 12345-1234
• Country
• Email address Example: user@example.gov.ca
• Telephone number Example: 123-123-1234
• Extension number (if applicable)
• Fax number (including area code)

3. Verify or provide the current operational status of the business or organization identified by the legal and operating name above.

• Operational
• Not currently operational e.g., temporarily or permanently closed, change of ownership

4. Verify or provide the current main activity of the business or organization identified by the legal and operating name above.

Note: The described activity was assigned using the North American Industry Classification System (NAICS).

Note: Press the help button (?) for additional information, including a detailed description of the activity with example activities and any applicable exclusions.

Description and examples

• This is the current main activity
• This is not the current main activity

Main activity

5. You indicated that is not the current main activity.

Was this business or organization’s main activity ever classified as?

• Yes
• No

Main activity

6. Search and select the industry activity classification that best corresponds to this business or organization’s main activity.

How to search:

• If desired, you can filter the search results by first selecting the business or organization’s activity sector.
• Enter keywords or a brief description that best describe the business or organization’s main activity.
• Press the Search button to search the database for an industry activity classification that best matches the keywords or description you provided.
• Select an industry activity classification from the list.

Select this business or organization’s activity sector

Enter keywords or a brief description, then press the Search button

Reporting period information

1. What are the start and end dates of this operation’s most recently completed fiscal year?

• Fiscal year start date Example: YYYY-MM-DD
• Fiscal year end date Example: YYYY-MM-DD

Operating revenue and expense accounts

1. What were this business’ operating revenues and expenses for the fiscal year?

Operating revenue — gross sales

1. Coal of own production
2. Purchased coal
3. All other products

Total gross sales (Sum of a. to c.)

Operating revenue — marketing expenses

1. Outward transportation — road
2. Outward transportation — rail
3. Outward transportation — water

Total outward transportation (Sum of a. to c.)

4. Port handling charges
5. All other marketing expenses

Total marketing expenses (Sum of Total outward transportation + Port handling charges + All other marketing expenses)

Total net sales (Sum of Total gross sales - Total marketing expenses)

Other operating revenue

1. Contract mining
2. Subsidies (operating only)
3. All other operating revenue

Total operating revenue (Sum of Total net sales + Contract mining + Subsidies (operating only) + All other operating revenue)

2. What were this business’ operation, maintenance and administration costs for the fiscal year?

Direct mining costs

1. Salaries and wages
2. Supplementary labour benefits e.g., employer contributions
3. Materials and supplies
4. Contracted services

Repair and maintenance costs

1. Salaries and wages
2. Supplementary labour benefits e.g., employer contributions
3. Materials and supplies
4. Contracted services

Other costs

1. Labour
2. Supplementary labour benefits e.g., employer contributions
3. Purchased fuel and electricity
4. Coal purchased for resale
5. Taxes — non income e.g., royalties, acreage, business, municipal school
6. All other costs

Total operation, maintenance and administration (Sum of Direct mining costs + Repair and maintenance costs + Other costs)

Other expenses

1. Depreciation
2. Income tax
3. All other deductions

Total operating expenses (Sum of Total operation, maintenance and administration + Depreciation + Income tax + All other deductions)

Net income

(Sum of Total operating revenue - Total operating expenses)

Net income (Sum of Total operating revenue - Total operating expenses)

3. What was the quantity of coal produced and purchased in the fiscal year?

1. Coal of own production
   Metric tonnes
2. Foreign purchased coal
   Metric tonnes
3. Domestic purchased coal
   Metric tonnes

Operations payroll

4. What were the salary and wages and total number of employees by category in the fiscal year?

Salaries and wages for the year
Total number of full time employees during the year

1. Executive, administrative, office and sales

   Salaries and wages for the year
   Total number of full time employees during the year

2. Mine and related

   Salaries and wages for the year
   Total number of full time employees during the year

3. Preparation plant and related

   Salaries and wages for the year
   Total number of full time employees during the year

All other payroll

1. All other payroll

   Salaries and wages for the year
   Total number of full time employees during the year

   Total employees at this location (Sum of Operations payroll + All other payroll)
   Salaries and wages for the year

2. Employees at other locations within province

   Salaries and wages for the year
   Total number of full time employees during the year

   Total employees (Sum of Total employees at this location + Employees at other locations within province)
   Salaries and wages for the year

   Total number of full time employees during the year

Changes or events

5. Indicate any changes or events that affected the reported values for this business or organization compared with the last reporting period.

Select all that apply.

• Strike or lock-out
• Exchange rate impact
• Price changes in goods or services sold
• Contracting out
• Organizational change
• Price changes in labour or raw materials
• Natural disaster
• Recession
• Change in product line
• Sold business or business units
• Expansion
• New or lost contract
• Acquisition of business or business units
• Vacation or maintenance periods
• Equipment failure
• Seasonal operations
• Increased or decreased market demand
• Other
• Specify the other change or event
• OR
• No changes or events

Contact person

6. Statistics Canada may need to contact the person who completed this questionnaire for further information. Is the best person to contact?

• Yes
• No

Feedback

7. How long did it take to complete this questionnaire?

Include the time spent gathering the necessary information.

• Hours
• Minutes

8. Do you have any comments about this questionnaire?

Survey purpose

Statistics Canada is conducting this survey to prepare for the 2026 Census.

The purpose of the survey is to confirm the facility's address and contact information, and to collect information on the type of facility in order to improve the efficiency of collecting census data.

The information you provide now may make completing the census questionnaire easier for you next year.

Authority and confidentiality

Your answers are collected under the authority of the Statistics Act and will be kept strictly confidential. Statistics Canada can share your information with your consent or, in limited cases, where permitted by the Statistics Act. The information you provide may also be used by Statistics Canada for other statistical and research purposes.

Record linkages

Statistics Canada may combine your responses from this survey with information from other surveys or from administrative sources.

Voluntary participation

Participation in this survey is voluntary, but your participation is important so that the information collected is as accurate and complete as possible.

Facility and contact information

1. Verify or provide the facility name and correct where needed.

Facility name

2. Is this the civic address of this facility?

Note: If the address below is missing or incomplete, please answer "No" and provide the complete address.

• Yes
• No
  • Please enter the civic address of this facility.
    Note: For a non-civic address, please provide a rural route or land description in the "Street name" field.
    Example: 63532 Range Rd 444 or NW-34-42-4-W3
    • Civic number
    • Suffix
    • Unit number
    • Street name
    • Street type
    • Direction
    • City, municipality, town or village
    • Province or territory
    • Postal code
      Example: A9A 9A9

3. Verify or provide the following information of the designated contact person for this facility and correct where needed.

Note: The designated contact person is the person who should receive this questionnaire but may not always be the one who actually completes the questionnaire.

• First name
• Last name
• Title
• Preferred language of communication
• Email address
  Example: user@example.gov.ca
• Telephone number (including area code)
  Example: 123-123-1234
• Extension number (if applicable)

Dwelling type definitions

In preparation for the 2026 Census, the Collective Dwelling Status and Classification Survey asks about the services provided at this facility, as well as any additional collective dwellings or private dwellings that may be located at this address.

A collective dwelling is a dwelling of a commercial, institutional or communal nature in which a person or group of persons resides or could reside. The collective dwelling must provide care, services or common facilities shared by the occupants, such as a kitchen, dining room or bathroom. Examples include long-term care homes, residences for older adults, lodging or rooming houses, correctional facilities, group homes, hotels, motels, tourist establishments, hospitals, staff residences, military bases, and work camps.

A private dwelling is a separate set of living quarters that shares the same civic address as the collective dwelling but has a different apartment or unit number. The private dwelling must have a private entrance either from outside the building or from a common hall, lobby, vestibule or stairway inside the building and cannot be accessed through another person's living quarters. Residents of the private dwelling do not receive any care or services provided by the facility.

Collective dwelling types

1. Which of the following best describes this facility?

• Hospital
  • If selected, go to Question 4.
    • Is this facility licensed as a hospital?
      • Yes
      • No
• Long-term care home or residence for older adults
  • Select the most applicable:
    • Long-term care home

      A facility that provides 24-hour nursing care or personal care. Residents receive help for most or all daily activities.

    • Residence for older adults

      A facility that offers personal support and assisted living care. Services are provided as part of the rent or available for an additional fee paid to the facility, e.g., retirement homes or assisted living homes. These facilities do not provide 24-hour nursing or personal care.

    • Both long-term care home and residence for older adults
    • No care or services are provided to residents
      • If selected, go to Question 6.
• Residential care facility related to disabilities, mental health, addiction, etc.
  • This facility is for:
    Select all that apply.
    • Primarily children or minors
    • Persons with psychological disabilities
    • Persons with an addiction
    • Persons with physical challenges or disabilities
    • Persons with developmental disabilities
    • Persons with other disabilities
      • Specify
• Shelter
  • This facility is primarily for:
    • Persons lacking a fixed address, such as homeless persons
    • Persons released from custody or on conditional release
    • Victims of domestic violence or abuse
    • Refugees and asylum seekers
    • Other
      • Specify
• Correctional or custodial facility, including municipal detachments
  • What type of facility is this?
    • Young offenders' facility
    • Temporary lock-up (e.g., police holding cell)
    • Provincial or territorial detention centre or custodial facility
    • Federal correctional facility
• Religious establishment
• Establishment with temporary accommodation services
  • What type of establishment is this?
    • Hotel, motel or tourist establishment
    • Campground or park
    • Other establishment with temporary accommodation services, such as a YMCA-YWCA, Ronald McDonald House, or hostel
• Lodging or rooming house
  • If selected, exit survey.
• Hutterite colony
  • If selected, exit survey.
• Other establishment
  • What type of establishment is this?
    • Residence for school or training centre
    • Military base
    • Commercial vessel
    • Work camp
    • Government vessel
    • Other type of establishment
• None of the above
  • If selected, go to Question 2.

If no answer is selected, go to Question 2.

Go to Question 5, unless otherwise specified.

The following question will help determine whether this establishment should be included in this questionnaire or not.

2. Does this establishment allow for a person or group of persons to stay overnight?

• Yes
  • Does this establishment provide care, service or shared amenities?
    Include:
    • any medical service, health care or personal care that is provided by the facility
    • shared amenities, which include access to a common kitchen, dining room or bathroom.
    Exclude:
    • short-term vacation rentals, such as bed and breakfasts (e.g., Airbnb, Vrbo, etc.)
    • services not provided by the establishment (e.g., Uber, SkipTheDishes, HelloFresh, etc.).
      • Yes
      • No
        • If no, go to Question 6.
• No
  • If no, go to Question 6.

Go to Question 3, unless otherwise specified.

3. Based on the answers provided, this facility should be included. Which of the following options best describes the primary purpose of this facility?

• Hospital
  • If selected, go to Question 4.
    • Is this facility licensed as a hospital?
      • Yes
      • No
• Long-term care home or residence for older adults
  • Select the most applicable:
    • Long-term care home

      A facility that provides 24-hour nursing care or personal care. Residents receive help for most or all daily activities.

    • Residence for older adults

      A facility that offers personal support and assisted living care. Services are provided as part of the rent or available for an additional fee paid to the facility, e.g., retirement homes or assisted living homes. These facilities do not provide 24-hour nursing or personal care.

    • Both long-term care home and residence for older adults
• Residential care facility related to disabilities, mental health, addiction, etc.
  • This facility is for:
    Select all that apply.
    • Primarily children or minors
    • Persons with psychological disabilities
    • Persons with an addiction
    • Persons with physical challenges or disabilities
    • Persons with developmental disabilities
    • Persons with other disabilities
      • Specify
• Shelter
  • This facility is primarily for:
    • Persons lacking a fixed address, such as homeless persons
    • Persons released from custody or on conditional release
    • Victims of domestic violence or abuse
    • Refugees and asylum seekers
    • Other
      • Specify
• Correctional or custodial facility, including municipal detachments
  • What type of facility is this?
    • Young offenders' facility
    • Temporary lock-up (e.g., police holding cell)
    • Provincial or territorial detention centre or custodial facility
    • Federal correctional facility
• Religious establishment
• Establishment with temporary accommodation services
  e.g., hotel, campground
  • What type of establishment is this?
    • Hotel, motel or tourist establishment
    • Campground or park
    • Other establishment with temporary accommodation services, such as a YMCA-YWCA, Ronald McDonald House, or hostel
• Lodging or rooming house
  • If selected, exit survey.
• Hutterite colony
  • If selected, exit survey.
• Other establishment
  e.g., residence for school, work camp
  • What type of establishment is this?
    • Residence for school or training centre
    • Military base
    • Commercial vessel
    • Work camp
    • Government vessel
    • Other type of establishment
      • Specify

Go to Question 5, unless otherwise specified.

4. What services are provided at this hospital?

• Short-term care
• Long-term acute care
  • Is there also a care home for older adults at this facility?
    • Yes
    • No
• Both short-term care and long-term acute care
  • Is there also a care home for older adults at this facility?
    • Yes
    • No

Go to Question 5.

Maximum capacity

5a. What is the maximum number of persons who could stay overnight?

5b. What is the maximum number of persons who could stay overnight, including the long-term care home?

Note 1: If the number of persons is unknown, enter your best estimate.

Note 2: Do not report for the Canadian Forces Housing Agency (CFHS) or Personnel Support Program (PSP) military housing. If the number of persons is unknown, enter your best estimate.

Maximum number
If the number is zero, go to Question 6.

Go to Question 7, unless otherwise specified.

Operation status

6. Based on the answers provided, this establishment either does not allow for persons to stay overnight or does not provide care, services or shared amenities and as a result does not meet the requirements for this questionnaire.

Select the option that best describes the operational status.

• Seasonal operations
  • When did this establishment close for the season?
    Example: YYYY-MM-DD
  • When does this establishment expect to resume operations?
    Example: YYYY-MM-DD
• Temporarily inactive
  • When did this establishment become temporarily inactive?
    Example: YYYY-MM-DD
  • When does this establishment expect to resume operations?
    Example: YYYY-MM-DD
  • Why is this establishment temporarily inactive?
• Ceased operations
  • When did this establishment cease operations?
    Example: YYYY-MM-DD
  • Why did this establishment cease operations?
    • Bankruptcy
    • Liquidation
    • Dissolution
    • Other
      • Specify the other reasons why operations ceased
• Private dwelling
  • When did this establishment become a private dwelling?
    Example: YYYY-MM-DD
  • Other
    • Specify

Go to Question 18.

Additional collective dwellings

7. Are there any other facilities that share the same address with this facility?

• Yes
  • Number of facilities
    • If the number is zero, go to Question 12.
    • If the number is greater than 0 and less than 100, go to Question 8.
• No

Go to Question 12, unless otherwise specified.

8. What is the name of this facility?

Facility name

Go to Question 9.

9. Which of the following best describes this facility?

• Hospital
  • Is this facility licensed as a hospital?
    • Yes
    • No
• Long-term care home or residence for older adults
  • Select the most applicable:
    • Long-term care home

      A facility that provides 24-hour nursing care or personal care. Residents receive help for most or all daily activities.

    • Residence for older adults

      A facility that offers personal support and assisted living care. Services are provided as part of the rent or available for an additional fee paid to the facility, e.g., retirement homes or assisted living homes. These facilities do not provide 24‑hour nursing or personal care.

    • Both long-term care home and residence for older adults
    • No care or services are provided to residents
• Residential care facility related to disabilities, mental health, addiction, etc.
  • This facility is for:
    Select all that apply.
    • Primarily children or minors
    • Persons with psychological disabilities
    • Persons with an addiction
    • Persons with physical challenges or disabilities
    • Persons with developmental disabilities
    • Persons with other disabilities
      • Specify
• Shelter
  • This facility is primarily for:
    • Persons lacking a fixed address, such as homeless persons
    • Persons released from custody or on conditional release
    • Victims of domestic violence or abuse
    • Refugees and asylum seekers
    • Other
      • Specify
• Correctional or custodial facility, including municipal detachments
  • What type of facility is this?
    • Young offenders' facility
    • Temporary lock-up (e.g., police holding cell)
    • Provincial or territorial detention centre or custodial facility
    • Federal correctional facility
• Religious establishment
• Establishment with temporary accommodation services
  e.g., hotel, campground
  • What type of establishment is this?
    • Hotel, motel or tourist establishment
    • Campground or park
    • Other establishment with temporary accommodation services, such as a YMCA-YWCA, Ronald McDonald House, or hostel
• Lodging or rooming house
• Hutterite colony
• Other establishment
  e.g., residence for school, work camp
  • What type of establishment is this?
    • Residence for school or training centre
    • Military base
    • Commercial vessel
    • Work camp
    • Government vessel
    • Other type of establishment
      • Specify

Go to Question 10.

10. What is the suite or unit number of this facility?

Suite or unit number

Go to Question 11.

11. Are you the contact person for this facility?

• Yes, I am the contact person
• No, someone else is the contact person
  • Please provide the contact information for this person.
    • First name
    • Last name
    • Title
    • Preferred language of communication
    • Email address
      Example: user@example.gov.ca
    • Telephone number (including area code)
      Example: 123-123-1234
    • Extension number (if applicable)

Go to Question 12.

Private dwellings at this facility

12. Are there any private dwellings that share this address with this facility?

• Yes
  • Number of dwellings
    • If the number is zero, go to Question 14.
    • If the number is greater than 0 and less than 100, go to Question 13.
    • If the number is greater than 99, go to Question 14.
• No

Go to Question 14, unless otherwise specified.

13. Provide the information for each private dwelling.

• Unit or apartment number
• Is this dwelling occupied or unoccupied?
  • Occupied
  • Unoccupied

Go to Question 14.

Additional collective dwellings at a different address

14. Are you the contact person for any other facilities that have not been mentioned?

• Yes
  • Number of facilities
    • If the number is zero, go to Question 18.
    • If the number is greater than 0 and less than 100, go to Question 15.
• No

Go to Question 18, unless otherwise specified.

15. What is the name of this facility?

Facility name

Go to Question 16.

16. Which of the following best describes this facility?

• Hospital
  • Is this facility licensed as a hospital?
    • Yes
    • No
• Long-term care home or residence for older adults
  • Select the most applicable:
    • Long-term care home

      A facility that provides 24-hour nursing care or personal care. Residents receive help for most or all daily activities.

    • Residence for older adults

      A facility that offers personal support and assisted living care. Services are provided as part of the rent or available for an additional fee paid to the facility, e.g., retirement homes or assisted living homes. These facilities do not provide 24-hour nursing or personal care.

    • Both long-term care home and residence for older adults
    • No care or services are provided to residents
• Residential care facility related to disabilities, mental health, addiction, etc.
  • This facility is for:
    Select all that apply.
    • Primarily children or minors
    • Persons with psychological disabilities
    • Persons with an addiction
    • Persons with physical challenges or disabilities
    • Persons with developmental disabilities
    • Persons with other disabilities
      • Specify
• Shelter
  • This facility is primarily for:
    • Persons lacking a fixed address, such as homeless persons
    • Persons released from custody or on conditional release
    • Victims of domestic violence or abuse
    • Refugees and asylum seekers
    • Other
      • Specify
• Correctional or custodial facility, including municipal detachments
  • What type of facility is this?
    • Young offenders' facility
    • Temporary lock-up (e.g., police holding cell)
    • Provincial or territorial detention centre or custodial facility
    • Federal correctional facility
• Religious establishment
• Establishment with temporary accommodation services
  e.g., hotel, campground
  • What type of establishment is this?
    • Hotel, motel or tourist establishment
    • Campground or park
    • Other establishment with temporary accommodation services, such as a YMCA-YWCA, Ronald McDonald House, or hostel
• Lodging or rooming house
• Hutterite colony
• Other establishment
  e.g., residence for school, work camp
  • What type of establishment is this?
    • Residence for school or training centre
    • Military base
    • Commercial vessel
    • Work camp
    • Government vessel
    • Other type of establishment
      • Specify

Go to Question 17.

17. What is the civic address of this facility?

• Civic number
• Suffix
• Unit number
• Street Name
• Street type
• Direction
• City, municipality, town or village
• Province or territory
• Postal code
  Example: A9A 9A9

Go to Question 18.

Comments

18. Please use this section if you have concerns, suggestions or comments.

For example, you may have concerns, suggestions or comments about:

• the steps to follow or the content of this questionnaire (a question that was difficult to understand or to answer, etc.)
• the characteristics of the online questionnaire (the navigation, the online help, the design, the format, the size of the text, etc.)
• any technical issues encountered.

Canadian Centre for Justice and Community Safety Statistics
June 30, 2025

Table of contents

• Executive summary
• Acknowledgments
• Background
• Police-reported Indigenous and racialized identity data engagement activities
• Engagement to operationalize data collection by police: What we heard
  • Engagement strategy
• Findings from engagement with police services and police representative bodies
  • Data collection and data standards
  • Legislative and regulatory considerations
  • Data analysis and dissemination
  • Education and awareness
  • Community engagement
• Findings from engagement with community organizations
  • Data collection and data standards
  • Data analysis and dissemination
  • Education and awareness
  • Community engagement
• Findings from engagement with academics
  • Data collection and data standards
  • Data analysis and dissemination
  • Education and awareness
  • Community engagement
• Final recommendations
• Translating PIRID Recommendations into Action
  • System updates at Statistics Canada
  • Development of Key Documents: Operational Guidelines and Analytical Framework
• References
• Appendix: Discussion Guides
  • A: Discussion guide for community organizations
  • B: Discussion guide for police organizations
  • C: Discussion guide for police associations

Executive summary

In recent years, a growing demand has emerged for more granular, or disaggregated, data to shed light on the diverse experiences of individuals across multiple social domains, including the justice system. Disaggregated data are crucial for understanding the experiences of marginalized populations, and for identifying and addressing systemic inequities, discrimination and racism within society. In the context of the criminal justice system, those with concerns about the different treatment and overrepresentation of Indigenous and racialized individuals have highlighted significant gaps in the availability of disaggregated data, particularly related to the identities of those who interact with law enforcement for various reasons, including criminal incidents. These gaps have restricted research on racial inequality and discrimination in policing and hindered evidence-based policy making to address the aforementioned issues within the criminal justice system.

In response to these increasing demands, in July 2020, Statistics Canada and the Canadian Association of Chiefs of Police (CACP) proposed a joint initiative to support the collection of data on the Indigenous and racialized identity of victims and accused persons through the Uniform Crime Reporting (UCR) Survey. The UCR Survey is an instrument through which police-reported data on criminal incidents are collected by Statistics Canada. These data are used for research and statistical purposes to monitor the nature and extent of police-reported crime in Canada.

As a part of the joint Statistics Canada–CACP initiative, Statistics Canada initiated a public engagement process to gather feedback from diverse stakeholders, including Indigenous and racialized organizations, police services, and academics. The goal was to assess the valued added in collecting data on the Indigenous and racialized identity of individuals involved in criminal incidents; explore appropriate methods for data collection, reporting and use; and identify related concerns. Based on feedback received through the engagement process, six draft recommendations were developed and publicly released in September 2022.

Beginning in the summer of 2022, a second phase of public engagement to support the implementation of data collection and reporting by police was initiated. The current article synthesizes and presents the findings of this second phase of engagement, highlighting key takeaways from discussions with Indigenous and racialized community organizations, police services, academics, and other parties. Status updates on the initiative and a discussion of next steps are also presented in this article.

The current report concludes with 12 final recommendations, based on the public engagement conducted under this initiative. These recommendations are for guiding the development and implementation of a national strategy for the police collection of Indigenous and racialized identity data through the UCR Survey.

Acknowledgments

Statistics Canada would like to thank all those who participated in the engagements it has held since 2020. This includes all participants who gave feedback throughout the initiative from Indigenous and racialized community organizations; academia; police services and representative organizations; government agencies at the federal, provincial, territorial and municipal levels; and the general public.

Background

In recent decades, there have been numerous calls for action to address the overrepresentation of and discrimination against Indigenous and racialized individuals in Canada's criminal justice system (Owusu-Bempah et al., 2023; Saghbini, Bressan & Paquin-Marseille, 2021; Clark, 2019; Department of Justice, 2017; Owusu-Bempah & Wortley, 2013; Cotter, 2022; Truth and Reconciliation Commission of Canada, 2015a;  Royal Commission on Aboriginal Peoples, 1996 ). These calls emphasize the need for disaggregated data from the justice system to uncover and address issues of systemic racism (Canadian Race Relations Foundation, 2020; Black Canadian National Survey, 2023; Clark, 2019; National Action Plan, 2021; Willick, 2021).

This information has also been called for by communities, organizations and academics as an essential component to measuring and understanding the extent to which individuals from different populations are represented in Canada's criminal justice system, beginning with their interactions with the police (National Inquiry into Missing and Murdered Indigenous Women and Girls, 2019 ; David & Mitchell, 2021; Millar & Owusu-Bempah, 2011; Owusu-Bempah & Jones, 2023; Owusu-Bempah & Jones, 2024; Samuels-Wortley, 2021; Statistics Canada, 2020; Truth and Reconciliation Commission of Canada, 2015).  

In response, in July 2020, Statistics Canada and the Canadian Association of Chiefs of Police (CACP) announced a joint commitment to collect data on the Indigenous and racialized identities of victims and accused persons ^{Footnote 1} as they pertain to criminal incidents through the Uniform Crime Reporting (UCR) Survey ^{Footnote 2}. This police-reported Indigenous and racialized identity data (PIRID) initiative aims to provide insights on and further understanding of experiences faced by Indigenous and racialized individuals to address systemic issues of racism, discrimination and inequity within the Canadian criminal justice system.

Although this initiative focuses solely on police-reported criminal incidents collected through the UCR Survey, it will nonetheless help shed light on the experiences of Indigenous and racialized communities in relation to policing and the criminal justice system. Specifically, PIRID can be used for:

• identifying differences and inequities in criminal justice system pathways and outcomes
• providing quantitative indications to support evidence-based policies and programming
• developing targets and benchmarks to monitor progress and assess the effectiveness of policies and programs and their impacts on specific populations
• in combination with linked data from other sources, identifying the role of systemic issues in the inequitable experiences of Indigenous and racialized persons in the criminal justice system
• providing communities with analytical findings to support programs and initiatives.

The Uniform Crime Reporting Survey

Statistics Canada's Uniform Crime Reporting (UCR) Survey is the official national statistical tool for collecting police-reported crime data and the main source used by government agencies, policy makers and researchers for information on crime trends in Canada. The UCR Survey data reflect crime that has come to the attention of Canadian police services. Since the UCR Survey was implemented in 1962, Statistics Canada has collected data on all criminal incidents that have been reported to federal, provincial, territorial and municipal police services in Canada involving offences under the Criminal Codeand other federal statutes.

The Incident-based UCR Survey covers 99.9% of the Canadian population and collects information on characteristics of criminal incidents and victims and accused persons involved. The UCR Survey is periodically revised to improve the quality of the information collected, to respond to changes in the definitions of different types of crime, and to better reflect emerging types of crime and information needs.

Police-reported Indigenous and racialized identity data engagement activities

Following the July 2020 announcement: Collection of data on Indigenous and ethnocultural groups in Canada's official police-reported crime statistics, Statistics Canada began working toward enabling the collection of Indigenous and racialized identity data by police through the UCR Survey. In July 2021, Statistics Canada embarked on an engagement process to seek feedback and diverse perspectives on this initiative, including from Indigenous and racialized community organizations; police services; academics; and other parties at the national, provincial, territorial and municipal levels. The purpose of this engagement was to assess the value-added in collecting this sensitive information, what information should be reported by police, how police should collect and report the information, and how the data should ultimately be used and accessed.

The engagement provided key insights on how to proceed with the initiative, leading to the development of six draft recommendations to guide the initiative forward. Statistics Canada published these draft recommendations and the results of the engagement in September 2022 in the Report and Draft Recommendations: Police-Reported Indigenous and Racialized Identity Statistics via the Uniform Crime Reporting Survey(hereafter referred to as "the September report"). This report outlined feedback and insights from participants, which led to the creation of the recommendations, as well as some reassurance and guiding principles for advancing this initiative effectively.

Following the development of the September report and draft recommendations, the need for further e ngagement became apparent, to inform specifics related to operationalizing data collection. As a result, in the summer of 2022, Statistics Canada embarked on a second phase of engagement, with the goal of seeking further feedback on the September report and draft recommendations, gauging concerns related to operationalizing data collection, and seeking further guidance to support the development of concrete data collection and analysis methods and plans.

Engagement to operationalize data collection by police: What we heard

Engagement strategy

From July 2022 to August 2023, Statistics Canada embarked on a second phase of engagement with Indigenous and racialized community organizations, police services, academics, and other parties of interest. This phase, the "operationalization phase," specifically aimed to

• gauge operational needs and concerns
• seek guidance on how best to collect these data, including when it would be most appropriate to collect them
• take stock of legislative and regulatory considerations, including acts and directives, that may affect data collection
• strategize on how data could be used after collection.

Generally, the engagement approach began with the development of discussion guides, allowing for an option to provide written feedback or participate in virtual engagements. Discussion guides included background information, a questionnaire to complete and space for respondents to provide any further information.

• Three discussion guides were created to ask specific questions of
  • community organizations (Appendix A)
  • police services (Appendix B)
  • police associations (unions) (Appendix C).
• Two facilitation approaches, or a combination, were used to guide the meetings:
  • asking every question in the exact format detailed in the discussion guide
  • describing the initiative, gauging feedback on it, asking key questions from the discussion guide and then encouraging participants to expand on their feedback from the meeting by submitting a completed discussion guide.

Invitations to participate in Phase 2 of the engagement included outreach to more than 780 contacts. In this effort, all organizations and partners originally identified for Phase 1 of the engagement were included. Additionally, efforts were made to increase outreach by connecting with organizations with which Statistics Canada already had an established relationship, conducting web searches to identify organizations serving or representing Indigenous and racialized communities, and asking key partners to provide contact information for individuals and organizations they believed should be included in the engagement. Moreover, extending the length of the engagement period in Phase 2 led to improved outcomes, compared with the first phase of engagement. O utreach included multiple invitations, reminders and phone calls to encourage participation.

Overall, 132 respondents (individuals or organizations) participated in the Phase 2 engagement, an increase of 39 respondents from the previous phase. Community organizations (including Indigenous organizations and representative bodies) made up 48% of the respondents, followed by police services (42%) (Chart 1).

Engaging police services and representative bodies

The engagement with police services and other police representative bodies occurred from July 2022 to September 2023. During this time, more than 125 police services were invited to participate. These comprised all police services reporting to the Incident-based UCR Survey, including 15 First Nations police services. Engagement with police services in Quebec was facilitated through that province's ministère de la Sécurité publique.

Engaging police bodies also involved reaching out to other police representative bodies, including police associations at the national, provincial and municipal levels, and selected committees under the CACP. Selected police services were consulted to review the Phase 2 engagement guide for police associations. In all, Statistics Canada's outreach yielded feedback from

• 55 police services from across the regions (Table 1)
• eight police representative bodies and committees (including police associations and committees)^{Footnote 3}
• three Indigenous police services or representative bodies.

Table 1: Regional breakdown of respondent police services, including provincial police services

Region	Number of respondents
National	1
Atlantic	8
Quebec	7
Ontario	21
Alberta	4
Manitoba	3
Saskatchewan	3
British Columbia	8
Total	55

Engaging community organizations

Engagement with community organizations began in May 2023, with invitations to 435 non-Indigenous organizations that represent or serve racialized or marginalized communities. This phase saw a significant increase in participation, with 47 organizations providing feedback, up from 18 in Phase 1, and with representation spreading across regions.

In addition to the engagement of non-Indigenous community organizations mentioned above, engagement with Indigenous organizations also began in May 2023, with outreach to more than 130 contacts from First Nations, Métis and Inuit communities, including national, regional, rural and urban organizations. In Phase 2, participation increased to 17 organizations, compared with Phase 1, when seven out of 83 contacted Indigenous organizations provided feedback.

In all, 63 Indigenous and racialized community organizations participated in the second phase of engagement.

Table 2: Regional breakdown of community organizations in Phase 2

Region	Number of respondents
National	16
Atlantic	4
Quebec	2
Ontario	18
Alberta	12
Manitoba	2
Saskatchewan	4
British Columbia	2
Territories	3
Total	63

Engaging academics

Academics consulted in the second phase of engagement were identified during Phase 1. Of the 10 academics who were contacted, five participated in the second phase (other academics were also consulted through the engagement with police services, police representative bodies and committees, and feedback from these academics is included within the police services and representative bodies feedback).

Academics were also consulted in Phase 2 as part of an expert advisory panel for reviewing community engagement guides.

Findings from engagement with police services and police representative bodies

Discussions with police services and police representative bodies aimed to gauge concerns and perspectives related to the collection of Indigenous and racialized identity data through self-identification and officer perception, exploring current police agencies' policies, data usage concerns and system requirements. They considered how to develop guidelines for data collection, collection timing, training needs and community engagement recommendations. The discussions also touched on the role of Statistics Canada, other operational considerations, and the ideal data collection process to ensure success and sensitivity. The full discussion guides used for these engagements can be found in Appendix B and Appendix C.

Data collection and data standards

Respondents were asked to identify any concerns regarding the collection of Indigenous and racialized identity data, particularly relating to data standards. The following points were the most widespread concerns raised by police:

• Data collection may have negative impacts on the relationships between police services and communities. Specifically, there were concerns that, because of the sensitivity of the data being collected, police officers may be accused of racial profiling.
• Individuals may hesitate to provide accurate identity information for various reasons, such as fears of discriminatory treatment.
• Too many identity categories within records management systems may cause misidentification and data quality issues.
• There may be difficulties collecting the data in incidents that are sensitive or when emotions are heightened.
• Reassurance that collected data will not be used in criminal investigations is necessary.
• Proper training on police sensitivity when asking these questions is needed.

Police respondents were also asked to identify the circumstances in which police officers should refrain from collecting these data:

• in volatile or active situations—according to the responding officer's assessment
• in situations with a potential for escalation because of self-identification data being requested
• in situations when an individual is experiencing mental health issues, they are intoxicated or there is a language barrier between the individual and the officer, and they are therefore incapable of providing and/or gathering the information
• in situations when the person involved experienced a traumatic event or when asking them to self-identify could lead to additional trauma.

Some respondents raised concerns over the appropriateness of collecting this information when incidents involve youth.

Furthermore, based on engagement feedback from Phase 1, Statistics Canada has recommended that Indigenous and racialized identity information be collected using both the officer perception method and the self-identification method (Recommendation 1). Respondents were asked for their thoughts on this recommendation, and below are the main areas of concern raised by police.

Officer perception data collection

• There were concerns related to the potential inaccuracy of officer perception data, compared with how accused persons and victims self-identify, and this may affect police relationships with communities.
• A possible challenge is that no data may be reported if the police is unable to determine identity based on very specific population groups.
• Concerns were raised over the specific racialized categories police will list during the collection of Indigenous and racialized identity data, with some police services explaining that some communities may not want to be generalized into certain racialized categories.

Self-identification data collection

• It would be best to give police officers the discretion to decide when self-identification information should be requested.
• The ideal time to collect is during booking (for accused individuals) and during the victim's statement (for victims).
• Collection should take place after the initial interaction, such as during the booking process or when emotions are lowered.
• Data should not be collected during arrest; some advised that the information should be collected after the initial investigation process has concluded.

Legislative and regulatory considerations

Police respondents were asked whether there are any current policies within their service, municipality, province or territory that would impact or prevent the collection of Indigenous or racialized identity data. Further, respondents were asked whether their police services currently collect any data (self-identification or officer perception) on Indigenous and racialized identity and, if so, what type of data they currently collect.

• The majority of respondents indicated that they did not have policies in place that explicitly prevent the collection of Indigenous or racialized identity data.
• Most respondents also highlighted that they do collect data on Indigenous and racialized identity, but that there are some limitations to these data collection practices:
  • These data are not being collected in a systematic or consistent manner.
  • There is no differentiation between officer perception and self-identification.
  • The data are not consistently collected in all types of incidents (e.g., typically during traffic stops). However, there is usually no way to distinguish how or when the data were collected.

Police service respondents also expressed several other regulatory concerns related to the data collection. These included concerns related to

• geographical comparisons, as some provinces have legislation that governs data collection, while many do not
• the impact of former "carding" legislation on relationships between police services and communities ^{Footnote 4}
• the use of the data within police services as performance measures
• how data collection could be enforced within jurisdictions, and the specific mandates that would govern this data collection
• information privacy and the need for establishing rules around data access, sharing, release and oversight.

Data analysis and dissemination

Police services were asked whether they had any input regarding the use of the data after they have been collected (e.g., analysis and dissemination). Broad support for this initiative was accompanied by cautions and recommendations regarding data analysis and the dissemination (or public release) of analytical results. These are highlighted below.

Data analysis

There was a general caution that analysis has the potential to reinforce stereotypes about Indigenous and racialized persons, further leading to mistrust of police-collected data. Several police respondents raised awareness about the potential misuse and misinterpretation of collected data. Respondents stated that these data alone should not and cannot be used to measure discrimination and racism without appropriate context related to Indigenous and racialized communities' experiences, as well as about the work of the police in those communities.

Further, respondents expressed some concerns related to discrepancies between officer perception and self-identification data. Officers raised concerns that by misidentifying individuals through officer perception, officers could potentially be blamed for poorer data quality overall. Concerns were also shared related to the officer perception data collection method and the potential for bias. Respondents indicated that this is especially true in situations where the identity information has already been predetermined prior to the police response at the scene (for example, an individual may call dispatch and describe someone's racialized identity over the phone).

Data usage and dissemination

When asked about concerns related to the usage and dissemination of the data collected through this initiative, police respondents expressed a need for contextualizing the data and putting parameters in place to pre-empt data misuse. To help prevent such misuse, respondents emphasized that the collected data should be presented as transparently and completely as possible.

Police respondents also raised questions about the challenges in determining which data type takes precedence (officer perception or self-identification) and how the data will be published and accessed. Additionally, there were cautions about skewed statistics leading to incorrect conclusions, if only specific interactions are recorded, and challenges in presenting and analyzing the data. Without context, the representation of policing within certain communities may be inflated (for example, when there are repeat encounters with the same individual).

Overall, with the collection of such sensitive information, there were strong suggestions by police respondents for best practices and guidelines to support informed and responsible usage of the collected data before dissemination.

Education and awareness

Education and awareness were central topics in the engagement with police services. Police respondents were asked whether they had any concerns related to training officers on how to collect the data and suggestions on how the training should be conducted. The following summarizes the feedback and recommendations received from respondents, categorized into several subtopics.

Transparency and communication

An overwhelming majority of the police respondents indicated that communicating the "why" of the data collection is a significant factor for gaining both community and police personnel support for the collection of data.

• Communities and police officers need to know the rationale behind the data collection, how the data will be used, and the purpose and value of collecting these data. This will more likely lead to better interactions with community members and provide reassurance to communities and individuals fearing that the data will be used to reinforce stereotypes.
  • If the "why" of the data collection is not clear, it may potentially create difficult interactions between officers and community groups.
  • Several respondents suggested that local police associations could reach out through their annual meetings to achieve buy-in from front-line officers.
  • It was suggested by many respondents that a comprehensive public information campaign is needed, with clear messaging and marketing efforts to convey the goals, benefits and importance of data collection to different communities.
  • Several respondents emphasized the importance of engaging with racialized and Indigenous advisory committees, using social media platforms, and providing information in multiple languages to rebuild trust.
• Police services also suggested that community groups should be informed of individuals' right to refuse to provide information about their identity during self-identification data collection.

Officer training

Police respondents were asked for any suggestions on types of training and why the training is important. The most common feedback from police is highlighted here:

• Training should provide information on when during an interaction data should be collected, what data should be collected and how they should be collected. Training should also ensure that officers understand the category definitions being used and how to apply them (officer perception).
• There was an emphasis on the importance of clarifying the purpose and benefit of the data collection to police officers to remove potential hesitancy to collect the information.
• Training should be standardized across the country, and communities should be informed about what officers are being trained on.
• Training should include a focus on diversity and understanding cultural differences, providing officers with the necessary skills to navigate diverse interactions and situations.
  • Training should be developed and delivered in collaboration with community groups.
• There was an emphasis on the need to reassure officers that data or the mistaken identification of an individual's Indigenous or racialized identity (officer perception) would not lead to reprimands or impact performance measure indicators.
• Concerns were raised related to the amount of time and resources required to deliver training to officers.
  • Smaller-scale and remote police services may be unable to provide in-person training sessions to their staff.
  • There could be time and schedule constraints for overburdened staff.
• Awareness and sensitivity must be an integral part of training, ensuring that experts are consulted to form the communication strategy and the training that is required for officers.
• Additional human rights, anti-sexism and anti-racism training was emphasized as necessary, along with additional awareness of trauma-informed approaches.
• There was support for information being presented by Statistics Canada to the boards, and this information will then trickle down to police services in a top-down approach.

Methods of training

Respondents were asked about what methods are recommended regarding the sensitive collection of these data. Various suggestions were made by police respondents, including the following:

• Training should be offered in person and online, through platforms such as the Canadian Police Knowledge Network, which allows for progress tracking and flexibility in completing the training across all police services.
• In-person training should be incorporated for topics that require human explanation and interaction.
• A hybrid approach should be used, with a combination of online and in-person training.
• Training should be scenario-based, emphasizing the use of discretion and judgment in the data collection, based on the situation, environment and individual history.
• A standardized script for officers to address questions from community members should be included in officer notebooks and training.
• Training should be in person, but to expedite the process, electronic training may be required.
• If training is held in person, there are opportunities to ask questions and test out the data collection cycle, receive feedback, and troubleshoot in real time.

Awareness

Respondents emphasized the need for a comprehensive public education campaign to raise awareness of the initiative. When asked about proposed methods for raising awareness of this initiative, police respondents made the following recommendations:

• Use modern communication tools, broadcast short and informative advertisements on TV and the radio, and use various forms of social media to raise awareness of the data collection initiative and reach a wider and more diverse audience.
• Collaborate with stakeholder organizations and community agencies to develop accessible communication materials in different formats and languages to deliver the message to communities.
• Spread information related to the data collection through provincial and municipal governments.

Community engagement

Police respondents highlighted the importance of community engagement and provided feedback related to raising awareness of the purpose of the data collection and associated concerns. The main input and considerations included the following:

• The "why" of data collection needs to be clearly communicated. Communities and police officers seek understanding, which requires transparent messaging.
• Collaboration with Indigenous and racialized community groups and advisory committees, along with a massive public information campaign, can help rebuild trust.
• There is fear of potential misuse and misinterpretation of data and a need for transparent dissemination of information.
• Community engagement should be extended to the training realm; respondents advocated for collaboration with community groups in developing and delivering training, where possible.

Findings from engagement with community organizations

The engagement with community organizations focused on feedback related to the initiative as a whole and on the September report. Key aspects of the discussions with the organizations were how to ensure successful collection and use of the data and manage and report self-identification and officer perception data. Questions were also asked about training needs for police regarding data collection and the types of data analyses to conduct once data are collected. The engagement guide used to facilitate some of the discussions with community organizations can be found in Appendix A. The following was the most prevalent feedback from community organizations.

Data collection and data standards

Most representatives of community organizations agreed that both methods, officer perception and self-identification, are essential and should be implemented. This would enable the most accurate collection of data, allowing for the best quality. The following summarizes the feedback and recommendations received from respondents, categorized into several subtopics.

Officer perception data collection

Community organizations were asked for their views on this method of collection. R espondents stated that officer perception involves assumptions based on physical characteristics and may not be correct and, as such, could cause issues with data quality and accuracy. There were also concerns about police bias impacting identification. Some respondents expressed concern that initially using the officer perception method of collecting these data could be seen as a form of "carding." ^{Footnote 4}

To avoid negative interactions or the retraumatizing of communities, respondents indicated that a thorough understanding of how practices such as street checks have disproportionately harmed racialized communities in old and recent history should be provided in police training. Training should also include cultural sensitivity and give an understanding of the Indigenous history of the justice system in Canada to help ensure these data are collected with respect. In tandem, respondents emphasized the importance of collecting self-identification data whenever possible.

Self-identification data collection

When asked about the collection of self-identification data, community organizations gave the following prevalent input:

• Individuals may choose not to self-identify out of mistrust and fear of discrimination and any impact on the case.
• There were concerns that self-identification could lead to discriminatory police interactions with Indigenous and racialized individuals, causing harm.
• For police services to ask self-identification questions, they need to have cultural sensitivity and data collection training, and to understand power dynamics.

Moreover, many community organizations shared thoughts on how to proceed with operationalizing the initiative, and much of the feedback was related to raising awareness within communities. The suggestions are listed below:

• It is important to clearly communicate the reason why these data are being collected.
• An education campaign should be proactively implemented for the initiative, with topics such as informing individuals of their right to consent or refuse to self-identify. This could be done through infographics shared on social media, hashtags, community groups, information cards, brochures, flyers, etc.

Furthermore, respondents also shared feedback on how information should be communicated about the project during interactions when data are to be collected:

• Communication should be in plain language, avoiding technical jargon, and in individuals' own language to ensure full comprehension (for example, by offering a multilingual sheet with a definition and examples of consent).
• Different ways should be used to communicate and ensure communication is accessible. The explanation and consent should be in different formats, including verbal, electronic and written on paper.

Other feedback related to the collection of self-identification data included the following:

• It is important to put safeguards in place to protect data from potential misuse.
• The process of determining how to ask self-identification questions with sensitivity, and how these questions are asked, should be co-developed with Indigenous and racialized communities.
• It was suggested to consult community groups by hosting townhalls and information sessions, to engage members and create discussion.
• Some respondents also stated that police services should not be the ones to collect this information. Rather, questions on self-identification should be asked by other professionals or third-party individuals, such as social workers, mental health workers or addiction specialists.

Community respondents were also asked to identify the circumstances in which police officers should refrain from collecting these data. The following were some of the most commonly reported scenarios in which self-identification data should not be collected:

• when individuals are in distress
• when individuals are inebriated
• when individuals are traumatized, have had previous negative experiences with the police or feel unsafe
• during volatile interactions
• during mental health crises.

"Data should be collected for criminal and non-criminal interactions"

Most respondents emphasized the importance of collecting Indigenous and racialized identity information for both criminal and non-criminal interactions. This was consistent with findings in the first phase of Statistics Canada's engagement (the September report)

Respondents highlighted that focusing only on criminal interactions would not fully address systemic issues. By analyzing data from both types of interactions, law enforcement agencies can address biases and disparities that may occur in a broader range of situations. Most respondents stated that through the process, individuals being questioned by police must be assured ("by actions and evidence, not just words") of the benefits to themselves and their families or communities.

Respondents mentioned benefits of collecting this type of data for both non-criminal and criminal interactions, including

• to reveal disproportionate treatment in situations such as traffic stops
• to demonstrate any evidence of over-policing and over-surveillance of racialized and Indigenous communities that is not captured by an exclusive focus on criminal incidents
• to support more welcoming, inclusive and safer communities
• to help identify systemic issues that may extend beyond criminal cases, enabling more effective policy changes.

Data analysis and dissemination

Overall, the collection of the data was regarded by community organizations as beneficial and needed. However, several recommendations were also made related to the analysis and dissemination of these data.

Respondents were asked what types of analysis they would like to see applied. Listed below are several of the prioritized analyses:

• Intersectional analysis: This is a framework that helps analyze how individuals and communities are affected by various social identity factors as they interact with each other and with
  • society (in the context of group membership)
  • organizations (in the context of institutional power, such as policies and practices)
  • systems of power (including prejudice and discrimination).^{Footnote 5}
  Examples of intersectionality are Indigenous identity; gender identity, especially for women, girls and Two-Spirit individuals; disability status; immigrant status; religion; age; socioeconomic status; and geographic location. Individuals have mixed identities.
• Descriptive analysis: In this type of analysis, basic statistics such as frequencies, proportions and percentages are calculated to provide an overview of the distribution of those with Indigenous and racialized identities among accused persons and victims, showing, for example, overrepresentations, disproportionality and disparities. It can be used to measure trends from raw data to get a clear picture of what is happening.
• Year-over-year comparison: Data from consecutive years are compared to understand short-term fluctuations and changes in the representation of Indigenous and racialized individuals in crime incidents.
• Comparative analysis: This type of analysis compares the representation of those with Indigenous and racialized identities in different types of crimes and differential outcomes across racialized identities, geographic regions or time periods to identify patterns and disparities.
• Geospatial analysis: This type of analysis uses geographic information systems to map the spatial distribution of crime incidents involving Indigenous and racialized individuals. This can reveal geographical patterns and disparities.
• Geographic comparisons: Comparing the representation of individuals with Indigenous and racialized identities in crime incidents across different geographic regions can identify regional trends and disparities.

Additionally, the need to make data accessible to community organizations, while respecting individuals' privacy, was emphasized throughout the engagement. It was articulated that access to the data can empower communities to assess the extent of criminal behaviour within their communities, identify barriers to justice and use this information to develop interventions. As summarized by one respondent,

"Communities have an inherent right to the information collected for the exact purposes they intend to use it for, such as identifying trends, allocating resources, and developing programs; and continuing to develop and draw down responsibilities as negotiated in self-governing agreements."

Some community organizations also suggested that one way of making data accessible could be the creation and provision of a data "dashboard" that allows different communities to pull reports specific to their populations or community members, in keeping with principles such as OCAP, ^{Footnote 6} FAIR ^{Footnote 7} and CARE. ^{Footnote 8}

Regarding the dissemination of findings from the data, respondents from community organizations

• cautioned about the potential negative reactions to results pointing to inequity, particularly from the police
• had concerns about accountability and transparency from law enforcement
• expressed worries about the public interpretation of findings, which could lead to varied reactions or misunderstanding of marginalized groups
• had preoccupations about the released data potentially perpetuating colonialism, racism, racial discrimination or community profiling, or creating further harm to cultures, communities and individuals.

Education and awareness

Training police officers and other police service personnel (e.g., data analysts) on the why, how and when of data collection was a significant component of the feedback received from community organizations. Respondents indicated that this training should be developed in partnership with diverse groups, including members of or experts from Indigenous and racialized communities. Further, there were strong suggestions that such training be conducted by community members to support relevant and accurate perspectives, history sharing and experiences.

Common suggestions related to training included the following:

• Co-designed workshops: Organize workshops where community members, law enforcement personnel and trainers work together to co-design the training materials. This ensures that the training is relevant, is accurate and resonates with all stakeholders.
• Indigenous-led training: Collaborate with Indigenous organizations or experts to lead training sessions specific to Indigenous perspectives, history and experiences.
• Continuous learning: Ensure that training is an ongoing process. Offer refresher courses and opportunities for law enforcement personnel to engage with communities beyond the initial training.
• De-escalation and conflict resolution: Learn de-escalation strategies to manage tense situations, and practise conflict resolution techniques to ensure safety.
• Trauma-informed practices: Recognize trauma and its potential impact on interactions, and use trauma-informed approaches to minimize re-traumatization.
• Data collection purpose: Explain the purpose and benefits of the initiative, and detail when, what and how data should be collected. Emphasize the importance of accurate and unbiased data.
• Intersectionality: Recognize the intersection of multiple identities and experiences, and address the complexities of individuals' identities.
• Cultural awareness and sensitivity: Provide training on developing cultural competency, addressing biases and stereotypes, understanding cultural dynamics and sensitivities, understanding the fears of communities, and demonstrating acceptable body language and eye contact.
• Feedback and continuous learning: Create opportunities for police to welcome feedback from community members and colleagues, and commit to ongoing learning and improvement.
• Bias: Provide training on implicit and explicit biases and their impact on data collection.
• Power imbalance: Provide training that includes discussions on power imbalances between police and victims or offenders and their impact on data collection.
• Data misuse: Provide information about how data have historically been used to harm.

Respondents also voiced concerns related to training, including, most prevalently, the following:

• If done in a superficial manner, without addressing underlying systemic racism or bias, training may further contribute to power imbalances and negative stereotypes.
• There is a need to ensure compliance and accountability in the delivery of training.
• Change management and best practices should be applied for police to welcome training and commit.

The following suggestions were proposed regarding effective training delivery:

• Comprehensive: Consider a multi-day format to allow for more interactive sessions, case studies and group activities.
• Ongoing: Encourage continuous learning.
• Interactive: Incorporate a mix of presentations, discussions, group activities, case studies, role-plays and simulations to keep participants engaged and encourage active learning.
• Accessibility: Include various components. Provide flexibility in the format, allowing for different learning styles and preferences.
• Testing component: Include methods to determine the effectiveness of training to ensure that concepts are properly understood. Consider the possibility of testing all participants after each module, with a minimum score required to pass. If the passing score is not achieved, participants should have to retake the module.
• Evolving: Continuously gather feedback from participants and community representatives to refine and improve the training content and approach. Training materials should continue to evolve over time based on new information.
• Feedback loops: Collect feedback from the community through, for example, surveys, complaint submission mechanisms and engagement (open conversations or townhalls). Collecting feedback, reporting on the level of engagement and collaboration between law enforcement and communities, and regularly assessing outcomes can support improved training. These can also lead to building better relationships in the communities.

Community engagement

Community respondents were asked about how communities can be encouraged to be involved with the data collection initiative. Most respondents spoke of the need for a new culture of trust between police and Indigenous and racialized individuals. Discussions revealed the following commonly noted approaches for engaging with communities:

• Explain clearly and succinctly why data are being collected and how they will be used and benefit communities. Explain what is in place to ensure the data will not be used to harm communities.
• Ensure there is collaboration between police and local cultural leaders or community partners and organizations to raise awareness and enhance engagement with community members.
  • The importance of taking in feedback and implementing it was also highlighted.
• Raise awareness of the data collection and educate the public (examples included through social media campaigns, information sessions in community spaces, community townhalls, public libraries, public transit advertising, electronic roadside billboards, podcasts, webinars and partnerships in schools).
  • Use multiple digital communication channels (e.g., social media, email, podcasts and commercials).
  • Produce plain language handouts, information packages and brochures that contain information in multiple languages and that are culturally specific.
• Create personal contact in the communities to build relationships and earn trust. This could include, for example, building reciprocal relationships and making respectful visits to local cultural or religious centres to introduce the initiative to community members in an environment where they are comfortable.

Findings from engagement with academics

The engagement with academics focused on feedback related to the initiative as a whole and on the September report. Key aspects of the discussions with the academics were how to ensure successful collection and use of the data, as well as manage and report self-identification and officer perception data. Questions were also asked about training needs for police regarding data collection and the types of data analyses to conduct once data are collected.

Throughout the engagement, academics were also presented with questions on developing guidelines that incorporate Indigenous and racialized communities' perspectives while ensuring transparency and context in data publications.

In this second phase of engagement, academics reiterated their acknowledgment of the importance of this initiative and provided the following feedback in terms of operationalizing the collection.

Data collection and data standards

All academic participants saw the value of collecting these data using both the officer perception and self-identification methods for accused persons and victims to ensure all necessary data are captured. Academics also stressed the importance of differentiating between the two methods of collection.

Academics also raised some cautionary notes related to data collection:

data quality and the risk of intentional misclassification of individuals (and related accountability) by police (specifically, where both officer perceptions and self-reported identity are collected, there are concerns that officers may be able to change their perception entry after collecting the self-identification information)

• overreliance on "catch-all" categories (for example, "unknown identity") for officer perception data collection, and its impact on data quality
• resistance of individuals to self-identify, and its impact on data quality and misleading results
• the impact of power dynamics that may affect how the victim or accused responds to questions pertaining to their Indigenous or racialized identity
• the potential for officers, victims and accused to conflate nationality, racialized identity and ethnicity
• the impact of refusing to provide self-identification information on the results.

Some academics also indicated that self-identification data should not be collected if officers perceive the interaction is dangerous. Some suggestions were made that individuals be offered the opportunity to self-report through a confidential third party.

Standards and guidelines

Academics were asked to provide feedback on how to ensure the data are collected in a standardized way (e.g., uniform record layouts) and on guidelines (e.g., operating manuals and best practice documentation) related to the technical aspects of data collection and what operational considerations need to be addressed. Numerous recommendations were provided by academics, including the following:

They suggested using insights from the United Kingdom's protocols for collecting information about race and ethnicity. The United Kingdom uses racialized categories broken down into many subgroups, which also include options for mixed race, and this is important for respondents who need to select multiple categories. Academics agreed that capturing intersectional identities is valuable.

Several academics proposed the development of clear guidelines and definitions to help respondents and officers differentiate between Indigenous and racialized identity categories.

Data analysis and dissemination

Academics were asked whether they had any feedback regarding the use of the data after they have been collected (e.g., analysis and dissemination). While there was broad support for this initiative, respondents identified some areas warranting caution and provided some recommendations:

• There is a need to develop a robust framework to address privacy concerns related to data collection and sharing.
• There is a need for contextual information related to the findings and how the data relate to social issues.
• Leveraging record linkages, whenever possible, would be a way to maximize data to enhance analysis.
• It is important to develop approaches to measuring concepts such as overrepresentation, differential outcomes and systemic bias that can be unified and consistent across jurisdictions, provinces and organizations to ensure meaningful comparisons.
• There is a need to commit to transparent public reporting, including recommendations for actioning results from the data, to better the lives of Canadians.

Academics underscored the importance of ensuring that data are not collected or used

• to reinforce negative stereotypes
• to justify additional policing and surveillance of communities
• to support social and economic policies that can harm Indigenous and racialized communities
• if the accused person or victim refuses to self-identify
• if the data collection is not accompanied by an educational and awareness campaign to inform Indigenous and racialized communities of the purpose and value of this initiative.

Finally, many academic respondents reported the importance of acknowledging limitations when using sensitive data such as PIRID. Key limitations include those of the Indigenous and racialized identity categories used, as well as the scope of the UCR Survey being limited to criminal incidents only and not all police interactions.

Education and awareness

Most academic respondents advised that clearly articulating the purpose of the data collection would be a significant factor for gaining both community and police personnel support. This was a key recommendation for training police services and raising awareness among communities.

Addressing power dynamics in police interactions was identified as another crucial component, given the authoritative position of police officers. The importance of creating training programs that foster a nuanced understanding of power dynamics was identified as a key contributing factor to fair and unbiased data collection.

The following were key takeaways from the discussions with academics, when they were asked for specifics related to officer training:

• Training should be mandatory and could be part of police officers' yearly reviews. Training should also be ongoing.
• Training should not be delivered by a police officer, or a person involved in policing, but a third party.
• Communities should be consulted on training development and delivery.
• Establishing training standards at the provincial, territorial and federal levels could support police services in ensuring delivery of training.
• Training should include many examples and scenarios.

Community engagement

Community engagement emerged as a key component of the engagement with academics. The majority of academic participants indicated that police services and the public both need to be informed about the importance of collecting Indigenous and racialized identity data. This involves disseminating information about the purpose of data collection, dispelling misconceptions, and ensuring the public understands the "why" behind the initiative and how the data will or will not be used.

Furthermore, academics highlighted the significance of acknowledging historical issues (e.g., carding practices) that continue to impact police–community relations and educating the public and affected communities on the benefits of the new data collection initiative. The consensus is that historical context plays a crucial role in shaping public perception and acceptance:

"…this will be the biggest hurdle—attempting to re-educate the public on the fact that this data is for addressing systemic racism and not for carding purposes. The need to address the history of systemic racism experienced by communities at the hand of Canadian institutions is necessary to ensure the community's acceptance and trust. This will also be imperative to understanding how the communities want the information and data to be collected."

Finally, academics emphasized the active role that communities should play in shaping data collection. The importance of consulting communities during the creation of training programs was stressed to ensure inclusivity and transparency. This theme underlines the need for a collaborative approach, recognizing the impact of data collection on the communities being represented.

Indigenous data sovereignty at Statistics Canada

The Government of Canada recognizes the unique rights, interests and circumstances of First Nations persons, Métis and Inuit and implements a distinctions-based approach to Indigenous statistics in support of self-determination and reconciliation.

Statistics Canada carries out its mandate of providing high-quality statistics that matter. It strives to carry out its mission in a way that is ethical, respectful and responsive to First Nations, Métis and Inuit needs and concerns, and to collaborate with First Nations, Métis and Inuit governments, communities and organizations. The agency's reputation, and ultimately its ability to produce high-quality Indigenous statistics, depends on appropriate engagement and relationship building.

As the national statistical agency, Statistics Canada carries out its work under theStatistics Act(the act). Under this federal legislation, Statistics Canada collects, compiles and publishes statistical information that is used by governments, businesses, researchers and the general public to understand demographic, social and economic realities across Canada. Any data collected under the authority of the act are kept confidential and are used only for statistical purposes. This includes data related to First Nations, Métis and Inuit communities.

Statistics Canada understands the importance of the ownership, control, access and possession (OCAP) principles that apply specifically to First Nations data, and the issues that underlay their development regarding data collection and research in First Nations communities. As a federal agency, Statistics Canada has been given the responsibility of acting as a responsible steward of its data holdings and works under a governance structure outlined by the act. The OCAP principles are a data governance structure that was developed by the First Nations Information Governance Centre and allow individual First Nations communities to determine whether projects or processes are OCAP-compliant. Both data governance regimes allow for the respective organizations to act as responsible data stewards and to ensure that privacy and confidentiality are protected while maintaining work and projects that are relevant, are valuable and uphold the policy to "do no harm." Métis and Inuit are also developing their own data governance strategies and frameworks.

Final recommendations

Based on the feedback received throughout the engagement for the PIRID initiative, the following final recommendations were developed to support the implementation of a national data collection strategy.

Recommendation 1

The collection of information on the Indigenous and racialized identity of accused persons and victims of crimes through the Uniform Crime Reporting Survey should be conducted through both the "officer perception" method and the "self-identification" method.

Recommendation 2

The collection of information on the Indigenous and racialized identity of accused persons and victims of crimes through the Uniform Crime Reporting Survey should be conducted using Statistics Canada's standardized population group categories for both the "self-identification" method and the "officer perception" method.

Recommendation 3

The Canadian Association of Chiefs of Police should work together with Statistics Canada and other parties of interest to establish national collection standards and guidelines that will integrate with police procedures, processes and workflow.

Recommendation 4

Any training delivered by Statistics Canada, or the police community, should emphasize the importance of the data collection initiative and the benefits for the Canadian population, policy makers and the police.

Recommendation 5

In developing or delivering any additional training related to the collection of Indigenous and racialized identity data, police services should consider including components related to systemic racism, the purposes of collecting these data, power differentials, the importance of informed consent without reprisal when collecting self-identification data, cultural competency, sensitivity training, and ongoing training and evaluation to address evolving needs and best practices.

Recommendation 6

Police services should consider how the voices of local community members can be incorporated in the development and implementation of a data collection initiative and related training through meaningful engagement and collaboration throughout the entire process, from initial planning to implementation and evaluation. Community advisory boards or working groups can provide ongoing input and guidance.

Recommendation 7

The analysis and use of information on the Indigenous and racialized identity of accused persons and victims of crimes should be done in a manner that reflects the realities experienced by Indigenous and racialized communities through the inclusion of context (e.g., colonialism, ongoing systemic barriers, the social determinants of health and inequities for Indigenous and racialized peoples, etc.) in all publications and related dissemination products.

Recommendation 8

To ensure consistency, the standards developed in the context of this initiative should be considered for future data collection within the justice and community safety sectors.

Recommendation 9

Police services should develop plans for implementing the standards and guidelines co-developed by Statistics Canada and the Canadian Association of Chiefs of Police in their data collection initiatives or processes, considering their local contexts and the need for flexibility and adaptability. Leveraging technology and data analytics can improve data collection and analysis.

Recommendation 10

The Canadian Association of Chiefs of Police should develop mechanisms that discourage police services from using any part of the self-identification data collection process as a performance metric and recommend systems of reassurance for police service members and the communities they serve. Performance metrics should focus on outcomes and impact, rather than on the quantity of data collected. Consideration should be given to developing a code of ethics for the collection and use of Indigenous and racialized identity data.

Recommendation 11

Statistics Canada should develop guardrails to ensure the responsible use of Indigenous and racialized identity data collected through the Uniform Crime Reporting Survey. Transparency, accountability and a data governance framework should be prioritized to oversee the collection, storage and use of these data.

Recommendation 12

Police services should consider expanding the collection of Indigenous and racialized identity information for specific types of police incidents beyond criminal incidents, such as traffic stops, use of force incidents and calls for service, and for other involved persons, such as persons of interest and subjects of various interactions, to provide a more comprehensive picture of policing interactions.

Translating PIRID Recommendations into Action

System updates at Statistics Canada

Updates to the UCR Survey system at Statistics Canada began at the end of 2022, and, as of February 13, 2024, the system has been updated to version 2.5 to enable collection of data reported by police services. Updates to the system were based on feedback received through the engagements, specifically Recommendations 1 and 2:

• Recommendation 1: The collection of information on the Indigenous and racialized identity of accused persons and victims of crimes through the Uniform Crime Reporting Survey should be conducted through both the "officer perception" method and the "self-identification" method.
• Recommendation 2: The collection of information on the Indigenous and racialized identity of accused persons and victims of crimes through the Uniform Crime Reporting Survey should be conducted using Statistics Canada's standardized population group categories for both the "self-identification" method and the "officer perception" method.

Recognizing that police services have flexibility and discretion with the collection method (i.e., officer perception, self-identification or both), the UCR 2.5 allows for police to report Indigenous and racialized identity regardless of the method of collection. ^{Footnote 9}

Data will become available at Statistics Canada upon police services' adoption of the updated UCR Survey system. ^{Footnote 10} Adoption of UCR 2.5 is not mandatory.

Development of Key Documents: Operational Guidelines and Analytical Framework

The recommendations emerging from the engagement feedback were also used to inform the development of two critical documents: (i) Operational Guidelines for aiding police services in the implementation of PIRID (i.e., the collection and use of the data); and (ii) an Analytical Framework to support analysts and researchers in the rigorous, ethical, and responsible analysis, interpretation and dissemination of the data.

These two documents were created by a new joint StatCan-CACP Special Purpose Committee, which waslaunched in September 2023 in response to Recommendation 3, namely, for the CACP to work together with Statistics Canada and other parties of interest to establish national collection standards and guidelines that will integrate with police procedures, processes and workflow.

For developing the PIRID Operational Guidelines for police services, five expert working groups were formed under the SPC to elaborate the following main themes emerging from the consultative engagement feedback:

• community engagement;
• legislative and regulatory considerations;
• education and awareness, including training;
• data standards, including system changes; and
• data analysis and dissemination.

Moreover, the working group tasked with developing the operational guidelines around data analysis and dissemination for police services was also responsible for the creation of a PIRID Analytical Framework for researchers and data analysts more broadly, which delivers on Recommendation 7:

• Recommendation 7: The analysis and use of information on the Indigenous and racialized identity of accused persons and victims of crimes should be done in a manner that reflects the realities experienced by Indigenous and racialized communities through the inclusion of context (e.g., colonialism, ongoing systemic barriers, the social determinants of health and inequities for Indigenous and racialized peoples, etc.) in all publications and related dissemination products.

Thus, the Framework aims to:

• support the responsible and ethical use of PIRID by proposing guiding principles to help avoid further stigmatization and marginalization of communities as a result of data use
• equip data users with tools and guidance for the careful, robust and culturally competent interpretation of the data.

Both the Operational Guidelines and Analytical Framework will be made publicly available in Summer 2025.

References

• Canadian Race Relations Foundation. (2015). A National Policy on the Collection of Race-Based Statistics.
• Commission on Systemic Racism in the Ontario Criminal Justice System. (1995). Report of the Commission on Systemic Racism in the Ontario Criminal Justice System.
• Clark, S. (2019). Overrepresentation of Indigenous People in the Canadian Criminal Justice System: Causes and Responses (PDF). Department of Justice Canada.
• Cotter, A. (2022). Experiences of discrimination among the Black and Indigenous populations in Canada, 2019. Statistics Canada.
• David, J., & Mitchell, M. (2021). Contacts with the Police and the Over-Representation of Indigenous Peoples in The Canadian Criminal Justice System. Canadian Journal of Criminology and Criminal Justice, 63(2).
• Foster, L., Park, S., McCague, H., Fletcher, M., & Sikdar, J. (2023). Black Canadian National Survey – Final Report 2023 (PDF). Institute for Social Research, York University.
• Millar, P., & Owusu-Bempah, A. (2011). Whitewashing Criminal Justice in Canada: Preventing Research through Data Suppression (PDF). Canadian Journal of Law and Society, 26(3), 653-661.
• Missing and Murdered Indigenous Women, Girls, and 2SLGBTQQIA+ People National Action Plan. (2021).
• National Inquiry into Missing and Murdered Indigenous Women and Girls. (2019). Reclaiming power and place: final report of the National Inquiry into Missing and Murdered Indigenous Women and Girls
• Ontario Human Rights Commission. (2003). Paying the Price: The Human Cost of Racial Profiling.
• Ontario Human Rights Commission. (2005). Policy and guidelines on racism and racial discrimination (PDF).
• Ontario Human Rights Commission. (2018). Under Suspicion: Issues Raised by Indigenous Peoples.
• Owusu-Bempah, A., & Jones, Z. (2023). Canada's Black Justice Strategy: Framework (PDF). Department of Justice Canada.
• Owusu-Bempah, A., Jung, M., Sbaï, F., Wilton, A. S., & Kouyoumdjian, F. (2023). Race and Incarceration: The Representation and Characteristics of Black People in Provincial Correctional Facilities in Ontario, Canada. Race and Justice, 13(4), 530-542.
• Owusu-Bempah, A., & Wortley, S. (2014).  Race, Crime, and Criminal Justice in Canada, Sandra M. Bucerius, and Michael Tonry (eds). The Oxford Handbook of Ethnicity, Crime, and Immigration.
• Research and Statistics Division. (2017). Indigenous overrepresentation in the criminal justice system. Department of Justice Canada.
• Royal Commission on Aboriginal Peoples. (1996). Report of the Royal Commission on Aboriginal Peoples Vol. 1-5.
• Saghbini, C., Bressan, A., & Paquin-Marseille, L. (2021). Indigenous People in Criminal Court in Canada. Department of Justice Canada.
• Samuels-Wortley, K. (2021). To Serve and Protect Whom? Using Composite Counter-Storytelling to Explore Black and Indigenous Youth Experiences and Perceptions of the Police in Canada. Crime & Delinquency, 67(8), 1137-1164.
• Statistics Canada. (2022). Report and Draft Recommendations: Police-Reported Indigenous and Racialized Identity Statistics via the Uniform Crime Reporting Survey.
• Statistics Canada. (2020) Joint Statement. The Daily — Collection of data on Indigenous and ethnocultural groups in Canada's official police-reported crime statistics
• Truth and Reconciliation Commission of Canada. (2015a).
• Willick, F. (2021). New African Nova Scotian institute to address racism in justice system. CBC.

Appendix: Discussion Guides

A: Discussion guide for community organizations

Discussion guide for community organizations on Police-Reported Indigenous and Racialized Identity Statistics via the Uniform Crime Reporting Survey

Background

On July 15, 2020, Statistics Canada and the Canadian Association of Chiefs of Police (CACP) released a joint statement announcing their commitment to working on the collection of data on the Indigenous and racialized identity of all victims and accused persons as it pertains to criminal incidents through the Uniform Crime Reporting Survey (UCR).

As the first step in this initiative, Statistics Canada embarked on a collaborative engagement process from July 2021 to February 2022 to acquire feedback from diverse perspectives on the collection of Indigenous and racialized identity data through the UCR Survey. This mobilization process involved collaborative efforts among various community organizations, academics, police services, the public and other parties of interest including government bodies at the national, provincial/territorial, municipal, and local levels. Perspectives and opinions heard throughout this engagement led to the release of a Report and Draft Recommendations on September 1, 2022, which can be found in the following link.

Report and Draft Recommendations: Police-Reported Indigenous and Racialized Identity Statistics via the Uniform Crime Reporting Survey

The recommendations developed from the results of the initial engagements are informing the next steps of the initiative (see Appendix A for the list of recommendations).

Objectives of the current engagement (agenda item):

Information for this initiative will be collected through the Uniform Crime Reporting (UCR) Survey. The UCR Survey, introduced in 1962, is Canada's primary source of information on police-reported data. The UCR Survey collects annual statistics on all federal crimes and offences recorded by police services to measure the incidence of crime in Canadian society and its characteristics.

To implement the recommendations developed from the initial engagements, and operationalize the initiative for police services across Canada to collect Indigenous and racialized identity data from victims and persons accused in criminal incidents, we are:

• Seeking feedback on the recommendations (particularly recommendations 1, 3, 4 and 5; see Appendix A) from Indigenous and racialized community organizations, in order to bring further refinements to the approach and ensure that we take any additional considerations into account.
• Gauging any further concerns related to operationalizing the initiative and next steps.

Engagement participant information

The following information will help us to compile and analyze the results of the engagement.

Please note that all individual and group responses will remain confidential. If you have any questions regarding the Engagement Document, please contact the Canadian Centre for Justice and Community Safety Statistics (CCJCSS) at statcan.ucrengagementmobilisationduc.statcan@statcan.gc.ca.

Please send your response to CCJCSS at statcan.ucrengagementmobilisationduc.statcan@statcan.gc.ca

If multiple representatives from your organization are responding, please combine your answers into one document.

If responding as an organization, please provide the following information:

• Full name of Organization:
• Address:
• Name and position of contact for the purpose of this engagement:
• Contact email address:
• Contact phone number:

List of contributors to this engagement (optional):

• Name
• Position

May we contact you for follow up information and discussion if necessary? Y / N

If responding as an individual:

• Name:
• Position/title:
• Contact email address:
• Contact phone number:
• Province:

May we contact you for follow up information and discussion if necessary? Y / N

Discussion Questions

Recommendation 1

The collection of information on the Indigenous and racialized identity of accused and victims of crimes through the Uniform Crime Reporting Survey should be conducted through both the "officer perception" method and the "self-identification" method.

The UCR Survey will be updated to include 2 new fields for victims and accused persons in criminal incidents (Indigenous and racialized identity through self-identification and Indigenous and racialized identity through officer perception).

Considerations

1. Self-Identification and officer perception have their own distinct advantages:
   1. The "self-identification" approach entails that the information on Indigenous or racialized identities of the people involved in a criminal incident is voluntarily provided by accused and victims. This method is said to provide the most accurate data and can help empower communities by providing data representing the needs of communities.
   2. The "officer perception" approach entails that the Indigenous or racialized identities of the people involved in a criminal incident (i.e., accused and victims) are recorded based on a police officer's subjective assessment of these individuals. This method is deemed to be important because perceptions could influence an officer's decision-making in an interaction and lead to disparities in outcomes.
2. Much of the feedback highlights the importance of self-identification to Indigenous and racialized community organizations.

Question 1.1

Do you have any concerns about police collection, management and reporting of: a. self-identification and b. officer perception?

a. Any concerns about collecting either one or the other?

Question 1.2

How can we encourage communities to be involved in the initiative to ensure that it is a collective (i.e., collaborative and participatory) endeavor?

Question 1.3

For the self-identification collection method, individuals are provided with the option to decline or refuse to respond. What are some ways that we can clearly ensure that the data are collected through an individual's consent?

a. How do we highlight and communicate the importance of consent?

Question 1.4

How can our organization raise awareness for this initiative and the collection of Indigenous and racialized identity data within your community?

a. How should this information be disseminated to ensure that all persons involved are aware of the intended objectives for this initiative?

Recommendation 3

The Canadian Association of Chiefs of Police work together with Statistics Canada and other parties of interest to establish national collection standards and guidelines that fit police procedures, processes, and workflow.

Procedures for collecting this information should be developed to reflect the various scenarios that officers are likely to encounter when asking these questions. Nevertheless, any approach should be standardized to ensure consistency across all police services. It should, among other things, include a standardized explanation of purpose so that police officers can clearly explain the motivation behind this data collection to both victims and accused. It is also clear from the feedback received in the initial set of engagements that there are situations in which it might be inappropriate, impractical or impossible to collect data on the Indigenous and racialized identity of accused and victims. The approach needs to include specific collection standards and guideline to account for these situations.

Considerations

1. Police officers are in positions of authority when interacting with the public. As such, clearly and respectfully communicating consent as it relates to self-reported data is of paramount importance.
2. Ensure that the data optimally reflects the realities of Indigenous and racialized individuals by allowing the selection of multiple Indigenous and/or racialized identities.

Question 2.1

How should the police communicate the purpose behind the collection of self-identification data when interacting with victims or accused?

Question 2.2

How should this explanation take place? (e.g., police explaining to the individual, providing an information card etc.)

a. How should the police collect the data during an interaction?

Question 2.3

Are there situations where it would not be appropriate to collect this information (using either method)?

a. Are their situations where it would not be appropriate for police officers to ask people directly for their Indigenous or racialized identities?

Question 2.4

Do you have any safety or security concerns for victims, accused persons or their communities in relation to this data collection?

Question 2.5

What are your thoughts on collecting this data for criminal and non-criminal interactions?

Question 2.6

How should your community members be informed about this initiative?

a. What role should police services play in informing communities about the initiative?

Recommendation 4

Any training delivered by Statistics Canada, or the police community should emphasize the importance of the data collection initiative and the benefits for the Canadian population, policy-makers, and the police.

Several respondents highlighted that the training should provide information on when data should be collected, what data should be collected, how they should collect it and why it is being collected. There was an emphasis on the importance of clearly explaining the purpose of this data collection to help facilitate collection by police officers. When developing the training, respondents also highlighted the importance of ensuring that the content and delivery are informed by the perspectives of Indigenous and racialized communities, thus ensuring that the training is culturally appropriate and that the initiative is a collective endeavor. Regarding training components relating to the experiences of First Nations, Inuit, and Métis people, respondents suggested drawing on the expertise of Indigenous Elders and prioritizing Indigenous-led training models.

Considerations

a. The training guidelines should incorporate Indigenous and racialized community perspectives and organizational needs.

Question 3.1

Do you have any suggestions on how an inclusive approach to training should be conducted?

Question 3.2

Current recommendations for training include the following: information on why the data is being collected, when it should be collected, what data should be collected, how they should collect it, and how the data will be used. Is there anything else that you think should be included?

Question 3.3

What topics should be included in the training of police?

Question 3.4

Do you have any concerns about the training on the collection of this data?

Question 3.5

Who do you think should be involved in providing training?

1. How long should the training be and what should it include to maximize its effectiveness ?
2. Howcan we determine the effectiveness of training?

Recommendation 5

The analysis and use of information on the Indigenous and racialized identity of accused persons and victims of crimes be done in a manner that reflects the realities experienced by Indigenous and racialized communities through the inclusion of context to all its publications and related dissemination products.

Respondents were asked to provide advice on the analysis and use of data on the Indigenous or racialized identity of accused persons and victims of criminal incidents. They highlighted the importance of leveraging perspectives of Indigenous and racialized communities to ensure that the data that are analyzed and used provide a comprehensive picture of their experiences.

Considerations

1. Potential for data quality/reliability issues.
2. Ensure that the data collected are accessible and presented as transparently and as completely as possible.
3. Necessity to include context during analysis, and in the interpretation and reporting of findings, to help reduce any potential misrepresentation.

Question 4.1

How should we engage diverse Indigenous and racialized communities/organizations in the development of analytical guidelines related to these data, to ensure that their input informs how the data is eventually interpreted and how the results are presented?

Question 4.2

What types of data analysis should be conducted with the data collected?

1. What types of data comparisons should be made to measure change over time?
2. Do you have any concerns related to data quality or reliability of results?
   1. If so, how can these concerns be addressed?

Question 4.3

Providing context to accompany the publication and dissemination of Indigenous and racialized identity data may include, but is not limited to, historical context and their continuing impacts on the current experiences of individuals and communities.

a. What contextual information should be included when publishing data on Indigenous and racialized peoples?

Final thoughts

Do you have any final thoughts on this initiative?

Key questions

• How would you like to be notified regarding the next steps of the initiative?
• What are potential positive/negative impacts do you think the initiative will have on Indigenous and racialized communities?
• How can we improve the initiative to ensure that it has a meaningful impact on your community?
• Any other thoughts, comments, or concerns about this initiative (not already shared above)?

B: Discussion guide for police organizations

Engagement Document
Police-Reported Indigenous and Racialized Identity Datathrough the Uniform Crime Reporting Survey
(Phase II)

Overview

Scope of the Data Collection Initiative:

Statistics Canada, in partnership with the Canadian Association of Chiefs of Police (CACP), is working toward determining the best method for collecting data on the Indigenous and racialized identity of all accused and victims of criminal incidents through its Uniform Crime Reporting (UCR) Survey. The UCR Survey collects information on all criminal incidents reported by Canadian police services to monitor the nature and extent of police-reported crime in Canada. Every police service in Canada reports to the UCR Survey. Only information on criminal incidents related to federal statutes are collected, which accounts for a minority of police-citizen interactions.

Detailed description of the Uniform Crime Reporting (UCR) Survey:
Uniform Crime Reporting Survey (UCR)

Detailed description of the joint partnership with CACP:
Resolution - Canadian Association of Chiefs of Police

In the first phase of the initiative, Statistics Canada embarked on an engagement process from July 2021 to February 2022 to seek feedback on the collection of Indigenous and racialized identity data through the UCR Survey. This engagement sought advice on the value of collecting this sensitive information, but also input on how the police should collect and report the data, what information should be reported by the police, how the data should be used and accessed, as well as other related concerns. This led to the development of an interim report and recommendations published on the Statistics Canada website: Police-Reported Racialized and Indigenous Identity Statistics via the Uniform Crime Reporting Survey: Report and Draft Recommendations

There was broad support amongst respondents for this initiative across all sectors canvassed, including community organizations and police services. As such, the feedback from respondents has led to the development of the following recommendations on the best way to move forward with this initiative so that the collection of data on the Indigenous and racialized identity of accused persons and victims of crime fulfill the data needs of communities, the police, policymakers, and the Canadian population broadly.

Recommendation 1

The collection of information on the Indigenous and racialized identity of accused persons and victims of crimes through the Uniform Crime Reporting Survey should be conducted through both the "officer perception" method and the "self-identification" method.

Recommendation 2

The collection of information on the Indigenous and racialized identity of accused persons and victims of crimes through the Uniform Crime Reporting Survey be conducted using Statistics Canada's standardized population group categories for both the "self-identification" method and "officer perception" method.

Recommendation 3

The Canadian Association of Chiefs of Police work together with Statistics Canada and other parties of interest to establish national collection standards and guidelines that will integrate with police procedures, processes, and workflow.

Recommendation 4

Any training delivered by Statistics Canada or the police community should emphasize the importance of the data collection initiative and the benefits for the Canadian population, policy-makers, and the police.

Recommendation 5

The analysis and use of information on the Indigenous and racialized identity of accused persons and victims of crimes be done in a manner that reflects the realities experienced by Indigenous and racialized communities through the inclusion of context to all its publications and related dissemination products.

Recommendation 6

To ensure consistency, the standards developed in the context of this initiative should be considered for future data collection within justice and community safety sectors.

Objectives of the Phase II engagement:

Statistics Canada is currently conducting a second phase of engagements with various partners from diverse perspectives, including community organizations, academics, police services and other parties of interest at the national, provincial/territorial, municipal, and local government level. This engagement seeks feedback on operationalizing data collection and analysis. Specifically, the objectives of this document are to:

• Walk through Recommendations 1 and 3 to obtain any further refinements or considerations from police services.
• Understand police services' requirements and concerns.
• Obtain feedback and endorsement on the work plan and next steps.

Engagement participant information

The following information will help us to compile and analyze the results of the engagement.

Please note that all individual and group responses will remain confidential. If you have any questions regarding the Engagement Document, please contact the Canadian Centre for Justice and Community Safety Statistics (CCJCSS) at statcan.ucrengagementmobilisationduc.statcan@statcan.gc.ca.

Please send your response to CCJCSS at statcan.ucrengagementmobilisationduc.statcan@statcan.gc.ca

If multiple representatives from your organization are responding, please combine your answers into one document.

If responding as an organization, please provide the following information:

• Full name of Organization:
• Address:
• Name and position of contact for the purpose of this engagement:
• Contact email address:
• Contact phone number:

List of contributors to this engagement (optional):

• Name
• Position

May we contact you for follow up information and discussion if necessary? Y / N

If responding as an individual:

• Name:
• Position/title:
• Contact email address:
• Contact phone number:
• Province:

May we contact you for follow up information and discussion if necessary? Y / N

Discussion Questions

Recommendation 1

The collection of information on the Indigenous and racialized identity of accused and victims of crimes through the Uniform Crime Reporting Survey should be conducted through both the "officer perception" method and the "self-identification" method.

1. CCJCSS will adapt the UCR survey to add 2 separate fields (self-identification and officer perception)
2. While the system undergoes technical changes, CACP & CCJCSS will work on developing guidelines on how to best collect the information, including the timing of collection

Considerations

1. Self-Identification and police perception have their own advantages:
   1. The "self-identification" approach entails that the information on the Indigenous or racialized identities of the people involved in a criminal incident is volunteered by accused and victims. This method is said to provide the most accurate data and can help empower communities by providing data representing the needs of communities.
   2. The "officer perception" approach entails that the Indigenous or racialized identities of the people involved in a criminal incident (i.e., accused and victims) are recorded based on a police officer's subjective assessment of these individuals. This method is deemed to be important because perceptions could influence an officer's decision-making in an interaction and lead to disparities in outcomes.
2. The UCR solution should be evergreen and also allow for future refinements to data collection.
3. Much of the feedback highlights the importance of self-identification to communities.

Question 1.1

Are there any current policies in your service or province stating that police officers should not collect Indigenous and racialized identity data?

a. If yes, what are your guidelines on collecting these data?

Question 1.2

Does your police service currently collect any data (via the self-identification or officer perception methods) on Indigenous and racialized identity?

a. If yes, what type of data do you currently collect?

Question 1.3

Do you have any concerns about the collection of both types of data (i.e., self-identification and officer perception)?

b. Do you have any concerns regarding the use of the data after it has been collected (e.g., analysis and publication)?

Question 1.4

What are your thoughts on making the officer perception field a mandatory field?

Question 1.5

How should the data collection be navigated while prioritizing consent for self-identification data?

Recommendation 3

The Canadian Association of Chiefs of Police work together with Statistics Canada and other parties of interest to establish national collection standards and guidelines that fit police procedures, processes, and workflow.

Procedures for collecting this information should be developed to reflect the various scenarios in which officers are likely to encounter when asking these questions. Nevertheless, any approach should be standardized to ensure consistency across all police services. It should, among other things, include a standardized explanation of purpose so that police officers can clearly explain the motives behind this data collection to victims and accused. It is also clear from the feedback received that there are situations in which it might be inappropriate, impractical or impossible to collect data on the Indigenous and racialized identity of accused and victims. The approach needs to include specific collection standards and guideline to account for these situations.

Considerations

1. For those already collecting data /or soon to be collecting, how do we consider standards/guidelines in place?
2. Privacy legislation and the Statistics Act.
3. Varying relationships between police and different communities.
4. Records Management Software (RMS): timing of the updates and adoption of new version by police services.

Question 2.1

What operational considerations need to be addressed?

Question 2.2

How should standards and guidelines for police officers collecting this information be developed?

Question 2.3

What are some ways that the CCJCSS can help police officers in explaining to victims and accused persons the motivation for this data collection?

a. At what point should the motivation be explained?

Question 2.4

When would be the ideal time during an interaction related to a criminal incident to collect self-identification data?

a. How should data be collected in these interactions (e.g., self-filled via iPad, collect at the time of arrest, initial introduction, after trust has been established?)

Question 2.5

In which circumstances should police officers refrain from collecting these data?

Question 2.6

Do you have any concerns about training officers on how to collect these data ?

a. Do you have any suggestions on how the training should be conducted?

General Comments

Question

Do you have any other comments, questions, concerns or recommendations regarding operationalizing the collection of information on the Indigenous and racialized identity of victims and accused persons as it relates to criminal incidents through the UCR Survey?

Next steps

Statistics Canada is committed to working with the policing community and key organizations to enable police to report statistics on the Indigenous and racialized identity of all victims and accused persons. Your feedback marks an important step in this engagement process for collecting more disaggregated data.

As a next step, Statistics Canada is currently working with the Canadian Association of Chiefs of Police (CACP) to launch a CACP special purpose committee to guide the next steps of the initiative, particularly as they relate to operationalizing the September recommendations, and the production of guidelines for police services to implement data collection. Thus, feedback received through these engagements will be presented to the committee to collaboratively identify next steps in establishing data collection standards and guidelines.

From a national standpoint, this initiative aims to develop evergreen national standards and guidelines for the data collection and analysis that build on expertise, established frameworks, lessons learned and best practices. We look forward to continuing to work together in this critical phase of operationalizing the data collection and analysis of police-reported identity information and finding a balance that will result in sound and meaningful data for all jurisdictions.

Thank you for your valuable input and participation.

C: Discussion guide for police associations

Engagement Document

Police-Reported Indigenous and Racialized Identity Datathrough the Uniform Crime Reporting Survey

(Phase II)

Overview

On July 15^th, 2020, Statistics Canada and the Canadian Association of Chiefs of Police (CACP) released a joint statement committing to working with Canada's policing community and organizations to collect police-reported information on the Indigenous and racialized identity of all victims and accused persons as it pertains to criminal incidents through its Uniform Crime Reporting (UCR) Survey. The UCR Survey collects information on all criminal incidents reported by Canadian police services to monitor the nature and extent of police-reported crime in Canada. Every police service in Canada reports to the UCR Survey. Only information on criminal incidents related to federal statutes are collected, which accounts for a minority of police-citizen interactions. Accordingly, incidents outside this purview are not reported to the UCR Survey. Incidents not covered by the UCR Survey include provincial statute offences, by-law infractions, use of force interactions, traffic stops and any other non-criminal incident or police interaction.

Whereas this initiative is focused on Statistics Canada's collection of race-based data through the UCR Survey, police services may choose to expand the scope of this data collection to include other type of interactions for their own internal use. For instance, police services in the province of Ontario are mandated to collect race-based and Indigenous data on persons involved in use of force incidents. Furthermore, Statistics Canada's present focus on UCR-based data collection does not preclude future expansion of the scope to other areas of interest. The current initiative is a first step toward gathering data that will help shed light on the experiences of Indigenous and racialized groups as they relate to policing and the criminal justice system more broadly.

Detailed description of the Uniform Crime Reporting (UCR) Survey:
Uniform Crime Reporting Survey (UCR)

Detailed description of the joint partnership with CACP:
Resolution - Canadian Association of Chiefs of Police

Phase I: Statistical Engagement:

Statistics Canada embarked on an engagement process from July 2021 to February 2022 to seek feedback on the collection of Indigenous and racialized identity data through the UCR Survey. This engagement sought advice not only on the value of collecting this sensitive information, but also input on how the police should collect and report the data, what information should be reported by the police, how the data should be used and accessed, as well as other related concerns. This led to the development of an interim report and recommendations published on the Statistics Canada website: Police-Reported Racialized and Indigenous Identity Statistics via the Uniform Crime Reporting Survey: Report and Draft Recommendations

There was broad support amongst respondents for this initiative across all sectors canvassed, including community organizations and police services. As such, the feedback from respondents has led to the development of the following recommendations on the best way to move forward with this initiative, so that the collection of data on the Indigenous and racialized identity of accused persons and victims of crime fulfill the data needs of communities, the police, policymakers, and the Canadian population broadly.

Recommendation 1

The collection of information on the Indigenous and racialized identity of accused persons and victims of crimes through the Uniform Crime Reporting Survey should be conducted through both the "officer perception" method and the "self-identification" method.

Recommendation 2

The collection of information on the Indigenous and racialized identity of accused persons and victims of crimes through the Uniform Crime Reporting Survey be conducted using Statistics Canada's standardized population group categories for both the "self-identification" method and "officer perception" method.

Recommendation 3

The Canadian Association of Chiefs of Police work together with Statistics Canada and other parties of interest to establish national collection standards and guidelines that will integrate with police procedures, processes, and workflow.

Recommendation 4

Any training delivered by Statistics Canada or the police community should emphasize the importance of the data collection initiative and the benefits for the Canadian population, policy-makers, and the police.

Recommendation 5

The analysis and use of information on the Indigenous and racialized identity of accused persons and victims of crimes be done in a manner that reflects the realities experienced by Indigenous and racialized communities through the inclusion of context to all its publications and related dissemination products.

Recommendation 6

To ensure consistency, the standards developed in the context of this initiative should be considered for future data collection within justice and community safety sectors.

Phase II: Operationalization

Statistics Canada is currently conducting a second phase of engagements with various partners from diverse perspectives, including community organizations, academics, police services and other parties of interest at the national, provincial/territorial, municipal, and local government level. The purpose of these engagements is to seek feedback on operationalizing data collection and analysis.

Engaging police associations at this juncture is a crucial step in operationalizing this initiative. With the interest of your members in mind, we are seeking your input on next steps. Specifically, the objectives of this document are to:

• Seek your feedback on the recommendations (particularly recommendations 1, 3 and 4), to bring further refinements to the approach and ensure that we take any additional considerations into account; and
• Identify and document concerns from police associations related to operationalizing the initiative or next steps so that we can properly address any challenges or concerns moving forward.

Engagement participant information

The following information will help us to compile and analyze the results of the engagement.

Please note that all individual and group responses will remain confidential. If you have any questions regarding the Engagement Document, please contact the Canadian Centre for Justice and Community Safety Statistics (CCJCSS) at statcan.ucrengagementmobilisationduc.statcan@statcan.gc.ca.

Please send your response to CCJCSS at statcan.ucrengagementmobilisationduc.statcan@statcan.gc.ca

If multiple representatives from your organization are responding, please combine your answers into one document.

If responding as an organization, please provide the following information:

• Full name of Organization:
• Address:
• Name and position of contact for the purpose of this engagement:
• Contact email address:
• Contact phone number:

List of contributors to this engagement (optional):

• Name
• Position

May we contact you for follow up information and discussion if necessary? Y / N

If responding as an individual:

• Name:
• Position/title:
• Contact email address:
• Contact phone number:
• Province:

May we contact you for follow up information and discussion if necessary? Y / N

Discussion Questions

Recommendation 1

The collection of information on the Indigenous and racialized identity of accused and victims of crimes through the Uniform Crime Reporting Survey should be conducted through both the "officer perception" method and the "self-identification" method.

The UCR Survey will be updated to include 2 new fields for victims and accused persons in criminal incidents (Indigenous and racialized identity through self-identification and Indigenous and racialized identity through officer perception).

Considerations

1. Self-Identification and police perception have their own advantages:
   1. The "self-identification" approach entails that the information on Indigenous or racialized identities of the people involved in a criminal incident is voluntarily provided by accused and victims. This method is said to provide the most accurate data and can help empower communities by providing data representing the needs of communities.
   2. The "officer perception" approach entails that the Indigenous or racialized identities of the people involved in a criminal incident (i.e., accused and victims) are recorded based on a police officer's subjective assessment of these individuals. This method is deemed to be important because perceptions could influence an officer's decision-making in an interaction and lead to disparities in outcomes.
2. Much of the feedback highlights the importance of self-identification to Indigenous and racialized community organizations.

Question 1.1

Are there any concerns about the collection of both types of data (i.e., self-identification and officer perception)?

1. Any concerns about collecting either one but not the other?
2. Does your association have any concerns regarding the use of the data after it has been collected (e.g., analysis and publication)?
   1. Do you have any suggestions or guidance on how these concerns can be addressed?

Question 1.2

Do you have concerns with making the officer perception field a mandatory field?  If so, how do we ensure this data is collected?

Question 1.3

How can CCJCSS assist your association in raising awareness for this initiative among officers on the collection of Indigenous and racialized identity data?

a. How should this information be disseminated to your membership that effectively ensures all officers are aware of the intended objectives for this initiative?

Question 1.4

What are some ways your members could benefit from the data collection (please specify if your response applies to officer-perception, self-identification, or both)?

Question 1.5

What types of reassurances do you believe your members would need to encourage cooperation related to the data collection (e.g., reassurances related to fears of being reprimanded)?

1. Would these be the same for officer-perception and self-identification?
2. How can the CCJCSS support in the provision of these reassurances?

Recommendation 3

The Canadian Association of Chiefs of Police work together with Statistics Canada and other parties of interest to establish national collection standards and guidelines that fit police procedures, processes, and workflow.

Procedures for collecting this information should be developed to reflect the various scenarios that officers are likely to encounter. Regardless of collection method, any approach should be standardized to ensure consistency across all police services. It should, among other things, include a standardized explanation of purpose so that police officers and the general public can clearly understand the motivation behind this data collection. It is also clear from the feedback received in the initial set of engagements that there are situations in which it might be inappropriate, impractical or impossible to collect both self-identified and officer perception data on the Indigenous and racialized identity of accused and victims. The approach needs to include specific collection standards and guidelines to account for these situations.

Considerations

1. The importance of acknowledging police officers' positions of authority when interacting with the public that may affect the willingness of some victims and accused persons to truly consent to the collection of self-identification information.
2. Recognizing the varying relationships and experiences different communities have with police.
3. Ensuring that the data analyses and reporting reflect the realities of policing, reporting and public experiences of police.

Question 2.1

Who should collect officer-perception information (e.g. first officer interacting with individual, arresting officer, etc.)?

b. When should the officer-perception data be collected?

Question 2.2

When collecting Indigenous and racialized identity information using the self-identification method, how should police services communicate the purpose behind the collection of the data when interacting with victims or accused?

a. What are some ways that CCJCSS can help police officers in explaining the motives for this data collection?

Question 2.3

For the self-identification collection method, individuals are provided with the option to decline or refuse to respond.

1. What are some ways that CCJCSS can support officers to clearly ensure that the data are collected through an individual's consent?
2. How do we highlight and communicate the importance of consent?

Question 2.4

When would be the ideal time during a criminal interaction to collect self-identification data (at the time of arrest, during initial introduction/identification, after trust has been established)?

1. For Accused?
2. For Victims?
3. Do you have any suggestions for how the data should be collected by police services in these criminal interactions (e.g., police personnel reading out the question)?

Question 2.5

Are there situations where it would not be appropriate to collect this information?

Question 2.6

Do you have any safety or security concerns for your membership and colleagues in relation to this data collection?

Question 2.7

Do you have any suggestions for how police officers can balance the need to collect these data while respecting communities and the individual right to refuse providing self-identification data?

Recommendation 4

Any training delivered by Statistics Canada, or the police community should emphasize the importance of the data collection initiative and the benefits for the Canadian population, policy-makers, and the police.

Several respondents highlighted that the training should provide information on when data should be collected, what data should be collected, how they should collect it and why it is being collected. There was an emphasis on the importance of clearly explaining the purpose of this data collection to help facilitate collection by police officers. When developing the training, respondents also highlighted the importance of ensuring that the content and delivery are informed by the perspectives of Indigenous and racialized communities, thus ensuring that the training is culturally appropriate and that the initiative is a collective endeavour. Regarding training components relating to the experiences of First Nations, Inuit, and Métis people, respondents suggested drawing on the expertise of Indigenous Elders and prioritizing Indigenous-led training models.

Considerations

a) The training guidelines should incorporate Indigenous and racialized community perspectives and organizational needs.

Question 3.1

Do you have any suggestions on how to incorporate feedback and advice from diverse communities in police training on the collection of data for:

1. Officer perception?
2. self-identification?

Question 3.2

Current recommendations for training include the following: information on when data should be collected, what data should be collected, how they should collect it and why it is being collected. Is there anything else that you think should be included?

Question 3.3

What topics should be included in the training for police?

Question 3.4

Do you have any concerns about the training on the collection of this data?

Question 3.5

Are there situations where it would not be appropriate to collect this information?

Question 3.6

Who do you think should be involved in developing and providing training?

Question 3.7

Are there any other training, resources or supports that your members could benefit from?

General Comments

Do you have any final thoughts on this initiative?

Examples:

• What resources/supports do you need to support your members as it relates to this initiative?
• How would you like to be notified regarding the next steps of the initiative?
• What impact do you think the initiative will have on the members of your association?
• How can we improve the initiative to ensure that it has a meaningful impact on policing and the public perception of police services and officers?
• What parts of this initiative do you think should be applied to other types of police interactions (non-criminal)?

Next steps

The September Report and draft recommendations was an interim report and is being used to inform the way forward, with room for further consideration, elaboration and development in light of emerging feedback and information. Feedback received through the engagements will further inform best practice recommendations, alignment opportunities and guidelines that consider these various perspectives.

In addition to these engagements, Statistics Canada is currently working on launching a CACP Special Purpose Committee to guide next steps. Issues, concerns and considerations raised throughout the engagements will be presented to the committee in order to ensure that they are considered in the development of national standards and guidelines for the collection and analysis of Indigenous and racialized identity information.

By Stany Nzobonimpa, Mohamed Abou Hamed, Ilia Korotkine, Tiffany Gao, Housing, Infrastructure and Communities Canada

Acknowledgements

The authors would like to acknowledge the following individuals for their contributions and support of the project: Kate Burnett-Isaacs, Director of Data Science and Matt Steeves, Senior Director of Engineering, Technical and Project Operations - Major Bridges and Projects.

Introduction

The identification and flagging of events such as data points that significantly deviate from standard or expected behaviour is known as anomaly detection in the field of data science. Anomalies are of special interest to data scientists because their presence signals underlying issues that prompt suspicions and usually warrants investigations. Various statistical techniques have historically been applied to answer anomaly-related questions and advances in Artificial Intelligence (AI) have made the exercise one of the most known applications of ML frameworks and approaches.

In this article, we present and discuss key findings of a project applying anomaly detection to sensor data. The project was carried out at HICC by a multi-disciplinary team of data scientists and engineers. While the project is multi-phased, this article reports on methods and results of the first phase in which sensor data from a federal bridge was used for early detection of anomalous readings.

The article is structured as follows: we start by a short review of relevant and recent literature on anomaly detection and its applications where we show that methods for data anomaly detection have evolved and continue to do so. After reviewing the literature, we present the problem, the methods, and results of this first phase of the project. In particular, we show that the approach enabled us to proactively, in a timelier and less laborious manner, flag anomalies in the generated sensor data that could have otherwise been missed. Indeed, our results show that for the period of April 2020 to September 2024, a number of sensor readings were successfully flagged as anomalies using the methods that we present in this paper. We conclude by stating the next steps of the project.

Literature Review

The application of machine learning to anomaly detection

Anomaly detection has long been a focus in data mining, which laid the groundwork for many modern ML approaches. Foundational works in data mining have explored anomaly detection techniques extensively, including clustering-based, distance-based, and density-based methods. Thus, anomaly detection is one of the core applications of ML technology that has been applied by AI practitioners for decades (Nassif et al., 2021). In its simplest form, anomaly detection is known as outlier detection and consists of flagging data points that significantly deviate from or do not conform with the majority of data.  The multiple applications of anomaly detection from fields as diverse as medical research, cybersecurity, financial fraud, and law enforcement have attracted both researchers and practitioners’ interests and resulted in a flourishing literature on the topic (Chandola et al. 2009).

Traditionally, anomaly detection could be completed manually by examining and filtering data and using context knowledge to provide insights on the detected behaviours. As datasets get larger, statistical approaches have been useful in detecting anomalies. For example, Soule et al., (2005) proposed a combination of filtering using Kalman filters and other statistical techniques to detect volume anomalies in large networks. The authors argue that anomaly detection can always be viewed as a problem of statistical hypothesis testing and they use behaviours, residual means, and variance changes over time to compare four different statistical approaches to anomaly detection (pp.333-338).

Using receiver operating characteristic (ROC) curves for binary normal vs anomaly illustrations, Soule et al., (2005, p. 334) argue that “ROC curves are grounded in statistical hypothesis testing” and that “[…] any anomaly detection method will at some point use a statistical test to verify whether or not a hypothesis (e.g., there was an anomaly) is true or false.” Another technique that has been applied to anomaly detection is what was called “robust statistics” by Rousseeuw & Hubert (2017). These authors revisited the techniques using fitting the majority of data and flagging anomalies once fitting normal data is completed. For example, the authors suggest estimating univariate locations and scales through means, medians, and standard deviations.

While the techniques such as those proposed by Soule et al. (2005) and Rousseeuw & Hubert (2017) for anomaly detection are popular and have been applied to various contexts and problems, the advent of advanced ML models and computing capabilities makes the exercise easier and scalable to large amounts of data. Additionally, some anomalies may be harder to detect and require more advanced techniques (Pang et al., 2021). These authors note that some unique complexities can prove more challenging and may require advanced approaches. They argue that anomaly detection is more problematic and complex because of its focus on minority data that are generally rare, diverse, unpredictable, and uncertain. For example, the authors note the following challenge related to the complexity of anomalies (Pang et al., 2021):

Most of existing methods are for point anomalies, which cannot be used for conditional anomaly and group anomaly, since they exhibit completely different behaviors from point anomalies. One main challenge here is to incorporate the concept of conditional/group anomalies into anomaly measures/models. Also, current methods mainly focus on detecting anomalies from single data sources, while many applications require the detection of anomalies with multiple heterogeneous data sources, e.g., multidimensional data, graph, image, text, and audio data. One main challenge is that some anomalies can be detected only when considering two or more data sources (p. 4).

Hence, the authors distinguish between what they call “traditional” anomaly detection and “deep” anomaly detection (p. 5) and argue that the latter is concerned with more complex situations and enables “end-to-end optimization of the whole anomaly detection pipeline, and they also enable the learning of representations specifically tailored for anomaly detection” (p. 5). Conditional anomaly detection, also known as contextual anomaly detection, refers to the method in where unusual values are identified in a subset of variables while taking into account the values of other variables. Group anomaly detection focuses on identifying anomalies within groups of data. Instead of looking at individual data points, this approach examines patterns and behaviors of groups to detect any deviations from the norm.

As noted by authors such as Chandola et al. (2009); Pang et al. (2021) and Nassif et al., (2021), among others, ML has made anomaly detection exercises more robust and seamless. Indeed, multiple techniques leveraging ML technologies have been suggested and applied to anomaly detection and have proved to be practical and handy in automating the process (Liu et al, 2015). For example, Nassif et al. (2021) split ML based anomaly detection into three broad categories: supervised anomaly detection where the process involves labelling a dataset and training a model to recognize anomalous points based on the labels, semi-supervised anomaly detection where the training set is only partly labelled and unsupervised anomaly detection where no training sets are needed, and a model automatically detects and flags abnormal patterns.

In this project, we leveraged the category of semi-supervised ML where sensor data are partly labelled. We show that this approach, combined with deep understanding of sensor readings, can achieve robust results, and presents the non-negligeable advantage of model and statistical parameter finetuning while keeping the automation power of machine learning.

Research Context: The Problem and Data Pipeline

The Problem

Physical structural health monitoring (SHM) sensors are installed at strategic locations along a given infrastructure such as a bridge or a dam to record its behaviour. Most sensors work by recording electronic signals, which are then converted into forces, movements, vibrations, inclinations, and other information that is critical in understanding the performance and the structural health of the infrastructure. These sensors send large amounts of data on an ongoing basis. In order for this SHM data to become useable by bridge engineers and operators for adequate asset management, the data must first be cleaned of anomalies such as outliers and noise.

Owner and operators of critical infrastructure perform regular due diligence activities to ensure the structural health of their assets. As a first step in identifying anomalies, data which exceeds predefined maximum or minimum thresholds, as set by the bridge designers, is identified as an outlier. However, Structural Health Monitoring System (SHMS) data is often more complex and requires further analysis to remove the outliers and noise from the data that falls within these predefined thresholds.

Prior to developing the ML approach, the Major Bridges and Projects (MBP) team had been conducting analyses which required manual anomaly detection of large amounts of sensor data which proved complex, time consuming, less effective, and prone to human error. The use of ML techniques has enabled the teams to test a faster and more accurate anomaly detection mechanism.

This project was conducted as a partnership between the Data Science (DS) team and the Major Bridges and Projects (MBP) team at HICC.

The general objective of the project is to improve the monitoring of the structural health of bridge infrastructure by outputting data with less noise. The specific objectives include detecting sensor data anomalies (counts) of various types such as those caused by irregular fluctuations or the lack of fluctuations and to start identifying sensor data anomaly trends. Additionally, the project aims to inform the decision-making process regarding the implementation of a new anomaly detection technology.

The Data Pipeline

The data pipeline of the project follows a typical data science process: the analytical team receives sensor readings, cleaning and pre-processing is done using python, training sets are created, labelled, and tested, machine learning models are applied using the scikit-learn library of Python and resulting predictions are visualized using MS Power BI. Figure 1 below summarizes the data pipeline for the project.

Methods

The project uses a semi-supervised ML classification approach similar to techniques discussed in our literature review section. This technique uses a combination of labelled and unlabelled data where we partially label a training set and apply statistical parameters in the anomaly detection process as described in the following sections. The choice of the method was motivated by multiple factors related to ensuring control over training data, the types of anomalies, cost-effectiveness, replicability, and performance, among others. Authors such as Soule et al. (2005) and Rousseeuw & Hubert (2017) proposed robust methods for using statistical techniques in anomaly detection and demonstrated the combination of such techniques with existing machine learning approaches yields robust results while the use of ML alone can generally miss hard-to-detect anomaly types (Spanos et al., 2019; Jasra et al., 2022). This indeed proved to be true in this project: when we applied an unsupervised ML model during the exploration phase, we realized that only outliers, i.e. data that were significantly above or below the centres, were being flagged as anomalies. Yet, we needed a method that can capture the anomalies that are not easily detected.

To successfully implement the approach, we partially labelled a sample set and trained a ML model to flag those anomalies. Prior to running the supervised model, we start by prepopulating anomaly predictions in an unsupervised and rule-based approach with pre-defined parameters for standard deviations and time series window sizes described in the following sections. With a partially labelled training set, ML classification is applied, and we chose the Gradient Boosting algorithm summarized by equations (1) and (2) below:

(1) $\hat{y}=F_{\mathrm{K}}\left(x\right)$

(2) $F_{\mathrm{K}}\left(x\right)=F_0\left(x\right)+\underset{i=1}{\overset{K}{\sum }}{\gamma }_i{\lambda }_i\left(x\right)$

Where:

• $\hat{y}$ is the predicted binary data label for a normal or an abnormal sensor reading;
• $X$ is a set of predictors for time series data with sensor reading, date and time;
• $\mathrm{F(}x)$ is the function learned by the classifier from the training (labeled) data;
• $F_0$ is a constant initial value for the target prediction;
• $K$ is the number of boosting stages, also called “weak learners”;
• ${\gamma }_i$ are coefficients also known as the learning rate;
• ${\lambda }_i$ are the weak learners or the decision trees

The expression in (1) represent the model’s predicted output y for input features x where F is the aggregated function combining all individual weak learners. The expression in (2) illustrates how the model is built iteratively over K boosting rounds: starting with an initial constant, each subsequent iteration adds a scaled weak learner to progressively minimize the residual errors.

It is important to note that the Gradient Boosting algorithm usually outperforms other ensemble learner models (Ebrahimi et al., 2019) and, in some instances, authors have found a gradient boosted decision tree was superior by a large margin to some neural models (Qin et al., 2021). Prior to employing the semi-supervised approach through the Gradient Boosting algorithm, we tested an Isolation Forest model as a fully unsupervised anomaly detection technique. Results of the Isolation Forests method did not meet our expectations in terms of accuracy, which is not surprising in such cases with conditional anomalies as shown by Pang et al. (2021) cited above.

This approach allowed full control over training data and gave room to possibilities of finetuning multiple parameters including model parameters such as testing and training sizes, random states as well as statistical parameters for standard deviations and time series window sizes. Indeed, we defined anomaly types by existing formulas defined by bridge engineers and by labelling training sets manually using an iterative process. For example, we were able to isolate the following type of anomalies called flatlines that were not flagged by the alarm system because they remained within the predefined thresholds previously mentioned and that would be time-consuming and not practical to identify manually:

• Absolute flatline anomalies: when a sensor returns exactly the same reading four times consecutively with standard deviation equal to 0
• Consecutive but relative flatline anomalies: when a sensor returns 12 consecutive readings in a day with very little fluctuation set to absolute standard deviation value inferior or equal to 0.15 

The use of this semi-supervised approach also ensured that our method could be applied to similar problems without the need to rewrite the algorithm, hence ensuring cost effectiveness and replicability. Finally, the choice of this method was also motivated by the well documented compute efficiencies and performances for supervised machine learning models (Akinsola et al., 2019; Aboueata et al., 2019; Ma et al., 1999). It is worth noting that while the results were satisfactory, this approach had some limitations and challenges. In particular, the manual labelling of training data required significant efforts. Additionally, the combination of techniques, while allowing flexibility, had a risk for model overfitting which we are mitigating by carefully inspecting the predicted results with analysts’ knowledge.

Results and Discussion

In its first phase, the project led to the identification of 714,185 data readings flagged as anomalies. This represented roughly 4.6% of all sensor readings and fell in one of the following pre-determined categories:

• Anomalies caused by irregular fluctuations
• Anomalies related to the lack of normal fluctuations including absolute and relative flatlines
• Anomalies caused by readings being outside of the expected range for sensors

The proportion of data points identified as anomalies was over 4% larger than what a non-supervised Isolation Forest model had returned in an earlier exploration, confirming the conclusions by Pang et al. (2021) that more complex anomaly types require more advanced combination of methods for accurate detection. We also obtained an average internal accuracy score of 0.988 (or over 98%), which, while not an absolute indication of success, was a signal for the model’s performance.

Predictive Accuracy

During the proof-of-concept phase of this project, we computed an F1 score for model accuracy and obtained a score of 0.956 (96%) accuracy. The F1 score is the harmonic mean of a model's precision and recall scores and is often used to evaluate classification models (for example, see Silva et al., 2024; Zhang et al. 2015). In the following section, we report on the internal model accuracy computed using the cross-validation approach from the scikit-learn library. To avoid overfitting, we held on to a small sample of data as a test set and computed the accuracy score using the `train_test_split` method from scikit-learn. The method splits datasets into training and testing subsets for ML tasks. It has various parameters such as the ‘test_size’ and ‘train_size’ which determine, randomly or in controlled manner, the fraction of number of samples for the test and train set. This approach allows better control, especially in the cases of imbalanced training data.

This process is well documented by scikit-learn as shown in figure 2 below:

Table 1 below shows accuracy scores obtained for a sample of 9 sensors out of 44 sensors in the scope of the first phase of the project. This score is calculated by dividing the number of correct predictions by the number of total predictions and thus allows comparison of the predicted values with the actual values. The score was computed using the y_val_split, y_val_pred methods from the sklearn.metrics sub-library. It is worth noting that the exercise of partially labelling training data was done on a full cycle of the sensor readings, covering a sample year. This helped overcome the problem of imbalanced training set and resulted in comparable shares of anomalies in training data and anomalies identified.

Table 1: Sample accuracy scores for the first 9 sensors using the test-split helper function

Sensor ID	Accuracy score
Sensor ID 1	0.998463902
Sensor ID 2	0.998932764
Sensor ID 3	0.990174672
Sensor ID 4	0.98579235
Sensor ID 5	0.99592668
Sensor ID 6	0.994401679
Sensor ID 7	0.998294486
Sensor ID 8	0.998385361
Sensor ID 9	0.99452954

While this score served as a good benchmark, it is worth noting that because of the specificity of anomalies being detected, more accuracy scoring metrics are being researched to ensure the best representation of the model’s performance.

Using the results: From Machine Learning to Business Value

The predicted anomalies need to be consumable to the bridge engineers so they can action on the insights produced by the ML model. The solution was to visualize the output in a simple and easy dashboard that summarizes the findings while providing enough information that can inform decision-making. We used Microsoft’s Power BI to produce a simplistic visualization that contain the following information:

1. Anomaly percentage by day using a stacked column chart, which can be drilled up to anomaly percentage by month
2. The sensor readings using a line chart and differentiating between normal readings and anomalies
3. A table with the detailed results provided by the algorithm for the selected sensor and year/month.
4. The performance indicator for the selected sensor for the selected time range, which is the percentage of the number of normal readings over the total number of readings
5. The share of normal sensor readings vs anomalies (donut chart) and a filter for comparison

Anomaly percentage by day using a stacked column chart, which can be drilled up to anomaly percentage by month

This type of dashboard is used to determined how well a specific sensor is performing and if it needs calibration, repair, or replacement. It provides limited visibility on the behaviour of the structure that the sensor is monitoring.

A snapshot of the dashboard showing the above is provided in Figure 3 below.

We also ensured the results can be visualized together for all sensors which allowed comparison and readability. The three figures below show the steps performed in order to attain this comparison and readability.

Unlike the previous dashboard, this type of dashboard, once anomalies such as noise and outliers are removed, allows for better visibility of the trends and the behaviour of the structures that the sensors are monitoring, as opposed to the behaviour of the sensors themselves.

Figure 4 shows the initial raw data of multiple sensors on the same graph, averaged monthly to show the results over extended periods of time (i.e. trend evaluation), along with temperature, which is distinguished by being dashed, as temperature is the main influencing reading predictor.

In Figure 5, the offsets between the sensor data lines are removed to ensure overlapping, and temperature is inverted so that it can also follow the upward and downward trend of the sensor data lines.

Finally, Figure 6. displays the final results with the removal of anomalies detected by our model. Once more, the overlap of temperature with the sensor data allows to qualitatively evaluate the behaviour of the structures at a glance.

The results obtained in this project have demonstrated the potential for using ML approaches to improve the anomaly detection exercise and to ensure a more efficient and accurate monitoring of sensor data. Indeed, the project has contributed to understanding historical data on bridge sensors and provided insights on the overall performance of the sensors. Moreover, the results assisted the Bridge Team in evaluating the performance of sensors, ultimately flagging a number of them for further analysis. Once the sensors are investigated and their functionality confirmed, the now noise-less data allows for a better appreciation of the behaviour of the structure, its performance trends, and its structural health over time.

Our approach has proven to have the potential to assist during the bridge operation, maintenance, and rehabilitation phase by helping to contribute towards early detection and inspection planning for infrastructure. Additionally, by exploring and manipulating historical data, the project has contributed to the overall objective of bridge health monitoring by checking data quality and reliability. The data quality checks that come with anomaly detection will help improve the accuracy of historical bridge SHMS data and facilitate the planning and management of future work. The control over anomaly types reduces the occurrence of false alarms while opening avenue for detailed inspections and checks. The resulting visualization will also be handy for the technical team and will provide insights on the performance of the sensors, as well as potential structural issues.

Conclusion and Next Steps

This project sought to leverage advanced techniques combining semi-supervised ML and statistical approaches for anomaly detection. As demonstrated in the literature section, anomaly detection is a complex problem that requires expertise and is usually well achieved by combining a host of methods. Indeed, we have shown that the exercise has the potential to flag sensor data anomalies in a controlled and replicable manner. In the next phases of the project, a multi-class model is being researched to ensure that not only anomalies are detected but they are also labelled and clustered into specific anomaly types. This granular approach will improve diagnostic clarity and help engineers to better understand sensor anomaly causes and ultimately support bridge maintenance.

It is worth noting, in this concluding paragraph, that with the advent of Generative AI (GenAI), future research should explore its potential for anomaly detection. GenAI has the ability to learn patterns and distributions of normal data and therefore could, if provided with examples, identify deviations that may signify anomalous data readings. Additionally, by training models like Generative Adversarial Networks (GANs) on large datasets, researchers can generate realistic representations of typical behavior and enhance anomaly detection exercises. The use of GenAI was outside of the scope of this project.

If you have any questions about my article or would like to discuss this further, I invite you to Meet the Data Scientist, an event where authors meet the readers, present their topic, and discuss their findings.

Register for the Meet the Data Scientist event. We hope to see you there!

Subscribe to the Data Science Network for the Federal Public Service newsletter to keep up with the latest data science news.

References

• Aboueata, N., Alrasbi, S., Erbad, A., Kassler, A., & Bhamare, D. (2019). Supervised Machine Learning Techniques for Efficient Network Intrusion Detection. 2019 28th International Conference on Computer Communication and Networks (ICCCN), 1–8. https://doi.org/10.1109/ICCCN.2019.8847179
• A insola, J., Awodele, O., Kuyoro, S., & Kasali, F. (2019). Performance Evaluation of Supervised Machine Learning Algorithms Using Multi-Criteria Decision Making Techniques. https://www.researchgate.net/publication/343833347
• Chandola, V., Banerjee, A., & Kumar, V. (2009). Anomaly detection: A survey. ACM Computing Surveys, 41(3), 1–58. https://doi.org/10.1145/1541880.1541882
• Ebrahimi, M., Mohammadi-Dehcheshmeh, M., Ebrahimie, E., & Petrovski, K. R. (2019). Comprehensive analysis of machine learning models for prediction of sub-clinical mastitis: Deep Learning and Gradient-Boosted Trees outperform other models. Computers in Biology and Medicine, 114, 103456. https://doi.org/10.1016/j.compbiomed.2019.103456
• Jasra, S. K., Valentino, G., Muscat, A., & Camilleri, R. (2022). Hybrid Machine Learning–Statistical Method for Anomaly Detection in Flight Data. Applied Sciences, 12(20), 10261. https://doi.org/10.3390/app122010261
• Nassif, A. B., Talib, M. A., Nasir, Q., & Dakalbab, F. M. (2021). Machine Learning for Anomaly Detection: A Systematic Review. IEEE Access, 9, 78658–78700. https://doi.org/10.1109/ACCESS.2021.3083060
• Pang, G., Shen, C., Cao, L., & Hengel, A. V. D. (2021). Deep Learning for Anomaly Detection: A Review. ACM Computing Surveys, 54(2), 1–38. https://doi.org/10.1145/3439950
• Qin, Z., Yan, L., Zhuang, H., Pasumarthi, R. K., Wang, X., Bendersky, M., & Najork, M. (2021). Are Neural Rankers still Outperformed by Gradient Boosted Decision Trees? International Conference on Learning Representations (ICLR). https://research.google/pubs/are-neural-rankers-still-outperformed-by-gradient-boosted-decision-trees/
• Rousseeuw, P. J., & Hubert, M. (2018). Anomaly detection by robust statistics. WIREs Data Mining and Knowledge Discovery, 8(2), e1236. https://doi.org/10.1002/widm.1236
• Scikit Learn. (2024). User Guide - 3.1. Cross-validation: evaluating estimator performance. https://scikit-learn.org/stable/modules/cross_validation.html
• Sheng Ma, & Chuanyi Ji. (1999). Performance and efficiency: recent advances in supervised learning. Proceedings of the IEEE, 87(9), 1519–1535. https://doi.org/10.1109/5.784228
• Silva, P., Baye, G., Broggi, A., Bastian, N. D., Kul, G., & Fiondella, L. (2024). Predicting F1-Scores of Classifiers in Network Intrusion Detection Systems. 2024 33rd International Conference on Computer Communications and Networks (ICCCN), 1–6. https://doi.org/10.1109/ICCCN61486.2024.10637544
• Soule, A., Salamatian, K., & Taft, N. (2005). Combining Filtering and Statistical Methods for Anomaly Detection. Proceedings of the 5th ACM SIGCOMM Conference on Internet Measurement. https://www.usenix.org/legacy/events/imc05/tech/full_papers/soule/soule.pdf
• Spanos, G., Giannoutakis, K. M., Votis, K., & Tzovaras, D. (2019). Combining Statistical and Machine Learning Techniques in IoT Anomaly Detection for Smart Homes. 2019 IEEE 24th International Workshop on Computer Aided Modeling and Design of Communication Links and Networks (CAMAD), 1–6. https://doi.org/10.1109/CAMAD.2019.8858490
• Zhang, D., Wang, J., & Zhao, X. (2015). Estimating the Uncertainty of Average F1 Scores. Proceedings of the 2015 International Conference on The Theory of Information Retrieval, 317–320. https://doi.org/10.1145/2808194.2809488.

Presentations

Presentation #1: A data dive into important trends and impacts of immigration on Newfoundland and Labrador

9:30 am-11:30 am NST

Just like Canada, Newfoundland and Labrador experienced some recent shifts in its demographics, notably in the number of incoming permanent and temporary immigrants, and in interprovincial migrations. On top of these shifts, Newfoundland and Labrador is also experiencing population aging, changes in ethnocultural and linguistic diversity, and increasing regional differences, notably between urban and rural areas.

The Chief Statistician of Canada will present a statistical portrait that will describe these recent shifts and trends, will put them in perspective compared to national and other provincial trends, and will also discuss some possible future trends on the basis of population projections. There will be a panel of experts who will present and engage the audience in the discussion on social and economic implications.

Presentation #2: Can data save lives in response to climate change and natural disasters?

1:00 pm-3:00 pm NST

Newfoundland is no stranger to extreme weather events and natural disasters. How can data provide actionable insights to understand the potential impact of natural disasters on the province’s people and its economy? The Chief Statistician of Canada will present a statistical portrait that will draw on recent events to demonstrate how data can inform the impact on the economy and will show how data can be used to understand the profiles of communities in order to prepare, respond and recover from natural disasters or emergency events.

In partnership with Memorial University’s Harris Centre and RAnLab, an expert panel discussion and question period will follow immediately after the presentations. You are encouraged to provide your input on these very important topics.

We sincerely hope you will join us in what will undoubtedly be thought-provoking and fruitful discussions.

General information

When: Wednesday, January 24, 2024 – 9:30 am to 3:00 pm NST

Where: Emera Innovation Exchange, Signal Hill Campus, Memorial University
100 Signal Hill Road, St. John's NL A1A 1B3

Cost: Free

If you are registered and your plans have changed and you can no longer attend, please cancel your registration.

Presenter

Anil Arora, Chief Statistician of Canada, Statistics Canada

Emcees

• Samuel Dupéré, Data Service Centre Manager (Eastern), Statistics Canada
• Jamie Ward, Manager, Regional Analytics Lab (RAnLab), Harris Centre, Emera Innovation Exchange, Signal Hill Campus, Memorial University

Moderator

Tony Labillois, Director General, Justice, Diversity and Population Statistics Branch, Social, Health and Labour Statistics Field, Statistics Canada

Panellists

Presentation #1

• Welcoming remarks, Neil Bose, President, Memorial University of Newfoundland
• Tony Fang, Professor and Stephen Jarislowsky Chair in Economic and Cultural Transformation, Memorial University of Newfoundland
• Jim Murphy, Director of Employment Services, Association for New Canadians (ANC)
• Constanza Safatle, Entrepreneur and advocate, Newbornlander

Presentation #2

• Welcoming remarks, Lisa Browne, Vice-President (Administration, Finance and Advancement), Memorial University of Newfoundland
• Emma Power, Consultant, Fundamental Inc.
• Joel Finnis, Climatologist, Department of Geography, Memorial University of Newfoundland
• Willa Neilsen, Labrador Climate Services Specialist, CLIMAtlantic

---

Please note that you will receive a confirmation email after registering.

Simultaneous interpretation (English, French) will be available during the event.

Registration

Registration is now closed.