Category: Uncategorized

Description: This is a bulk open-access dataset in JSON, parquet and Hugging Face dataset formats with the full text of Tax Court of Canada decisions. The process through which data is processed and code snippets for loading the data are available in a repository on the Refugee Law Lab Github.

Data: https://github.com/Refugee-Law-Lab/tcc_bulk_data/tree/master/DATA/YEARLY

GitHub Repository: https://github.com/Refugee-Law-Lab/tcc_bulk_data

Hugging Face Repository: https://huggingface.co/datasets/refugee-law-lab/canadian-legal-data

Current Coverage: 2003 – 2023 (* to July 24 in 2023)

Number of Decisions: ~15,000

Languages: English & French

Formats: JSON (yearly files), Parquet, Hugging Face Dataset

License: Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0). NOTE: Users must also comply with upstream licensing for the TCC source data, as well as requests on source urls not to allow indexing of the documents by search engines to protect privacy. As a result, users must not make the data available in formats or locations that can be indexed by search engines.

Citation: Sean Rehaag, “Tax Court of Canada Bulk Decisions Dataset” (2023), online: Refugee Law Laboratory https://refugeelab.ca/bulk-data/tcc

Data Fields:

• citation1 (string): Legal citation for the document (neutral citation where available)
• citation2 (string): For some documents multiple citations are available (e.g. for some periods the Supreme Court of Canada provided both official reported citation and neutral citation)
• dataset (string): The name of the dataset (in this case “TCC”)
• year (int32): Year of the document date, which can be useful for filtering
• name (string): Name of the document, typically the style of cause of a case
• language (string): Language of the document, “en” for English, “fr” for French, “” for no language specified
• document_date (string): Date of the document, typically the date of a decision (yyyy-mm-dd)
• source_url (string): URL where the document was scraped and where the official version can be found
• scraped_timestamp (string): Date the document was scraped (yyyy-mm-dd)
• unofficial_text (string): Full text of the document (unofficial version, for official version see source_url)
• other (string): Field for additional metadata in JSON format, currently a blank string for most datasets

Programmatic Access in Python (via Hugging Face Datasets):

from datasets import load_dataset
import pandas as pd

dataset = load_dataset("refugee-law-lab/canadian-legal-data", split="train", data_dir="TCC")

# convert to dataframe
df = pd.DataFrame(dataset)
df

Programmatic Access to in Python (via Parquet):

import pandas as pd
import requests
from io import BytesIO

url = 'https://huggingface.co/datasets/refugee-law-lab/canadian-legal-data/resolve/main/TCC/train.parquet'

# load data
results = requests.get(url)

# convert to dataframe
df = pd.read_parquet(BytesIO(results.content))
df

Programmatic Access in Python (via JSON):

import pandas as pd
import json
import requests

# Set variables
start_year = 2003  # First year of data sought (2003 +)
end_year = 2023  # Last year of data sought (2023 -)

# load data
base_ulr = 'https://raw.githubusercontent.com/Refugee-Law-Lab/tcc_bulk_data/master/DATA/YEARLY/'
results = []
for year in range(start_year, end_year+1):
    url = base_ulr + f'{year}.json'
    results.extend(requests.get(url).json())

# convert to dataframe
df = pd.DataFrame(results)
df

NOTES

(1) Data Source: Tax Court of Canada.

(2) Unofficial Data: The data are unofficial reproductions of materials on the Tax Court of Canada website. Links to official versions are included in the dataset.

(3) Non-Affiliation / Endorsement: The data has been collected and reproduced without any affiliation or endorsement from the Tax Court of Canada.

(4) Non-Commerical Use: As indicated in the license, data may be used for non-commercial use (with attribution) only. For commercial use, see the Tax Court of Canada website’s Terms of Use.

(5) Accuracy: Data was collected and processed programmatically for the purposes of academic research. While we make best efforts to ensure accuracy, data gathering of this kind inevitably involves errors. As such the data should be viewed as preliminary information aimed to prompt further research and discussion, rather than as definitive information.

(6) Limitation: Only includes cases with neutral citation, which began to be used in 2003

(7) Delay: Decisions may take many months to be translated (sometimes over a year). As a result, in the most recent years, decisions may only be available in one language.

The Refugee Law Lab has part-time research assistant positions available for current law students at any Canadian faculty of law (JD or graduate level) who are considering careers at the intersection of AI and legal practice and who have advanced professional coding or data science skills.

The Lab has several ongoing projects related to open-source legal analytics in the refugee law field. The projects use data about Canada’s refugee determination process scraped from online sources, apply various NLP and LLM tools to extract useful information and analyze the data, and present insights from that data in accessible formats to assist refugee lawyers appearing before the Immigration and Refugee Board and the Federal Court.

If you have advanced professional coding / data science skills that could assist with these projects and you would like opportunities to explore leveraging those skills to help legal practitioners in advancing the rights and interests of refugees and other non-citizens, let’s chat about how we might work together.

Examples of sought-after expertise:

• Data scraping / data wrangling using Python (Pandas, BeautifulSoup, Selenium, Playwright, Scrapy, etc).
• Working with LLMs using Python (Langchain, OpenAI, HuggingFace, PyTorch, etc)
• Natural Language Processing using Python (RegEx, SpaCy, Scikit-learn, etc)
• Data labeling using Python (Prodigy, etc)
• Data analysis and visualization using Python (Pandas, Numpy, Matplotlib, Seaborn, etc)
• Back-end development and databases (RestAPIs, PostgreSQL, MongoDB, etc)
• MLOps and DevOps
• Experience with Advanced Research Computing or Research Data Management platforms provided by the Digital Research Alliance of Canada
• Lived experience with forced migration or being a member of an equity seeking group is an asset

Context: The Refugee Law Lab, based at York University’s Centre for Refugee Studies and Osgoode Hall Law School, is devoted to research and advocacy related to new legal technologies and their impact on refugees, other displaced communities, and people on the move. We develop datasets and legal analytics that enhance transparency in refugee law processes. We study and critique the use of artificial intelligence and other technologies by governments and private actors in the migration field. And we produce legal technology that advances the rights and interests of refugees and other marginalized people on the move. We are committed to social justice, to interdisciplinarity, to evidence-informed policy, and to ensuring that the data, research and technologies that we produce are freely accessible to the public. We strive to work from a community-based perspective, foregrounding the lived experiences of people on the move and their interactions with technology.

Place of work: remote

Salary: Current JD student: $25/hr; Current graduate student: $40/hr (both hourly salaries inclusive of vacation pay)

Time Commitment: Negotiable.

Eligibility: Current (or incoming for Fall 2023) JD or graduate student at a Canadian faculty of law. Must have professional coding / data science skills.

Application Deadline: Open until filled.

Application process: Apply online with a CV, electronic copy of unofficial transcripts, and a brief cover note describing your interest in the position.

Description: This is a bulk open-access dataset in JSON, parquet and Hugging Face dataset formats with the full text of Federal Court (Canada) decisions. The process through which data is processed and code snippets for loading the data are available in a repository on the Refugee Law Lab GitHub.

Data: https://github.com/Refugee-Law-Lab/fc_bulk_data/tree/master/DATA/YEARLY

GitHub Repository: https://github.com/Refugee-Law-Lab/fc_bulk_data

Hugging Face Repository: https://huggingface.co/datasets/refugee-law-lab/canadian-legal-data

Current Coverage: 2001 – 2023 (* cases with neutral citation; to July 1 in 2023)

Number of Decisions: ~60,000

Languages: English & French

Format: JSON (yearly files for each language), Parquet, Hugging Face Dataset

License: Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0). NOTE: Users must also comply with upstream licensing from the FC data source, as well as requests on source urls not to allow indexing of the documents by search engines to protect privacy. As a result, users must not make the data available in formats or locations that can be indexed by search engines.

Citation: Sean Rehaag, “Federal Court Bulk Decisions Dataset” (2023), online: Refugee Law Laboratory https://refugeelab.ca/bulk-data/fc

Data Fields:

• citation1 (string): Legal citation for the document (neutral citation where available)
• citation2 (string): For some documents multiple citations are available (e.g. for some periods the Supreme Court of Canada provided both official reported citation and neutral citation)
• dataset (string): The name of the dataset (in this case “FC”)
• year (int32): Year of the document date, which can be useful for filtering
• name (string): Name of the document, typically the style of cause of a case
• language (string): Language of the document, “en” for English, “fr” for French, “” for no language specified
• document_date (string): Date of the document, typically the date of a decision (yyyy-mm-dd)
• source_url (string): URL where the document was scraped and where the official version can be found
• scraped_timestamp (string): Date the document was scraped (yyyy-mm-dd)
• unofficial_text (string): Full text of the document (unofficial version, for official version see source_url)
• other (string): Field for additional metadata in JSON format, currently a blank string for most datasets

Programmatic Access in Python (via Hugging Face Datasets):

from datasets import load_dataset
import pandas as pd

dataset = load_dataset("refugee-law-lab/canadian-legal-data", split="train", data_dir="FC")

# convert to dataframe
df = pd.DataFrame(dataset)
df

Programmatic Access in Python (via Parquet):

import pandas as pd
import requests
from io import BytesIO

url = 'https://huggingface.co/datasets/refugee-law-lab/canadian-legal-data/resolve/main/FC/train.parquet'

# load data
results = requests.get(url)

# convert to dataframe
df = pd.read_parquet(BytesIO(results.content))
df

Programmatic Access in Python (JSON):

import pandas as pd
import json
import requests

# Set variables
start_year = 2001  # First year of data sought (1877 +)
end_year = 2023  # Last year of data sought (2023 -)
languages_sought = ['en', 'fr']  # languages in list 


base_ulr = 'https://raw.githubusercontent.com/Refugee-Law-Lab/fc_bulk_data/master/DATA/YEARLY/'

# load data
results = []
for year in range(start_year, end_year+1):
    for language in languages_sought:
        url = base_ulr + f'{year}_{language}.json'
        results.extend(requests.get(url).json())

# convert to dataframe
df = pd.DataFrame(results)
df

Description: This is a bulk open-access dataset in JSON, parquet and Hugging Face dataset formats with the full text of Federal Court of Appeal (Canada) decisions. The process through which data is processed and code snippets for loading the data are available in a repository on the Refugee Law Lab GitHub.

Data: https://github.com/Refugee-Law-Lab/fca_bulk_data/tree/master/DATA/YEARLY

GitHub Repository: https://github.com/Refugee-Law-Lab/fca_bulk_data

Hugging Face Repository: https://huggingface.co/datasets/refugee-law-lab/canadian-legal-data

Current Coverage: 2001 – 2023 (* cases with neutral citation; to July 1 in 2023)

Number of Decisions: ~14,000

Languages: English & French

Format: JSON (yearly files), Parquet, Hugging Face Dataset

License: Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0). NOTE: Users must also comply with upstream licensing from the FCA data source, as well as requests on source urls not to allow indexing of the documents by search engines to protect privacy.

Citation: Sean Rehaag, “Federal Court of Appeal Bulk Decisions Dataset” (2023), online: Refugee Law Laboratory https://refugeelab.ca/bulk-data/fca

Data Fields:

• citation1 (string): Legal citation for the document (neutral citation where available)
• citation2 (string): For some documents multiple citations are available (e.g. for some periods the Supreme Court of Canada provided both official reported citation and neutral citation)
• dataset (string): The name of the dataset (in this case “FCA”)
• year (int32): Year of the document date, which can be useful for filtering
• name (string): Name of the document, typically the style of cause of a case
• language (string): Language of the document, “en” for English, “fr” for French, “” for no language specified
• document_date (string): Date of the document, typically the date of a decision (yyyy-mm-dd)
• source_url (string): URL where the document was scraped and where the official version can be found
• scraped_timestamp (string): Date the document was scraped (yyyy-mm-dd)
• unofficial_text (string): Full text of the document (unofficial version, for official version see source_url)
• other (string): Field for additional metadata in JSON format, currently a blank string for most datasets

Programmatic Access in Python (via Hugging Face Datasets):

from datasets import load_dataset
import pandas as pd

dataset = load_dataset("refugee-law-lab/canadian-legal-data", split="train", data_dir="FCA")

# convert to dataframe
df = pd.DataFrame(dataset)
df

Programmatic Access in Python (via Parquet):

import pandas as pd
import requests
from io import BytesIO

url = 'https://huggingface.co/datasets/refugee-law-lab/canadian-legal-data/resolve/main/SCC/train.parquet'

# load data
results = requests.get(url)

# convert to dataframe
df = pd.read_parquet(BytesIO(results.content))
df

Programmatic Access in Python (via JSON):

import pandas as pd
import json
import requests

# Set variables
start_year = 1877  # First year of data sought (1877 +)
end_year = 2023  # Last year of data sought (2023 -)

# load data
base_ulr = 'https://raw.githubusercontent.com/Refugee-Law-Lab/scc_bulk_data/master/DATA/YEARLY/'
results = []
for year in range(start_year, end_year+1):
    url = base_ulr + f'{year}.json'
    results.extend(requests.get(url).json())

# convert to dataframe
df = pd.DataFrame(results)
df

Description: This is a bulk open-access dataset in JSON format with the full text of all Immigration and Refugee Board (Refugee Protection Division) cases included in the Refugee Law Lab Reporter. The process through which data is collected and updated, as well as code snippets for loading the data, are available in a repository on the Refugee Law Lab Github.

Data: https://github.com/Refugee-Law-Lab/rllr_bulk_data/blob/master/DATA/yearly

Code Repository: https://github.com/Refugee-Law-Lab/rllr_bulk_data

Current Coverage: 2019-Present

Number of Decisions: ~500

Languages: English

Format: JSON

License: Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0)

Citation: Sean Rehaag, “RLLR Bulk Decisions Dataset” (2023), online: Refugee Law Laboratory https://refugeelab.ca/bulk-data/rllr

Programmatic Access in Python (JSON via GitHub):

import pandas as pd
import requests

# Set variables
start_year = 2019  # First year of data sought (2019 +)
end_year = 2022  # Last year of data sought (2022 -)

base_ulr = 'https://raw.githubusercontent.com/Refugee-Law-Lab/rllr_bulk_data/master/DATA/YEARLY/'

# load data
results = []
for year in range(start_year, end_year+1):
        url = base_ulr + f'{year}.json'
        results.extend(requests.get(url).json())

# convert to dataframe
df = pd.DataFrame(results)

Description: This is a bulk open-access dataset in JSON, parquet and Hugging Face dataset formats with the full text of Supreme Court of Canada decisions. The process through which data is processed and code snippets for loading the data are available in a repository on the Refugee Law Lab Github.

Data: https://github.com/Refugee-Law-Lab/scc_bulk_data/tree/master/DATA/YEARLY

GitHub Repository: https://github.com/Refugee-Law-Lab/scc_bulk_data

Hugging Face Repository: https://huggingface.co/datasets/refugee-law-lab/canadian-legal-data

Current Coverage: 1877 – 2023 (* to July 1 in 2023)

Number of Decisions: ~15,500

Languages: English & French

Formats: JSON (yearly files), Parquet, Hugging Face Dataset

License: Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0). NOTE: Users must also comply with upstream licensing from the SCC data source, as well as requests on source urls not to allow indexing of the documents by search engines to protect privacy. As a result, users must not make the data available in formats or locations that can be indexed by search engines.

Citation: Sean Rehaag, “Supreme Court of Canada Bulk Decisions Dataset” (2023), online: Refugee Law Laboratory https://refugeelab.ca/bulk-data/scc

Data Fields:

• citation1 (string): Legal citation for the document (neutral citation where available)
• citation2 (string): For some documents multiple citations are available (e.g. for some periods the Supreme Court of Canada provided both official reported citation and neutral citation)
• dataset (string): The name of the dataset (in this case “SCC”)
• year (int32): Year of the document date, which can be useful for filtering
• name (string): Name of the document, typically the style of cause of a case
• language (string): Language of the document, “en” for English, “fr” for French, “” for no language specified
• document_date (string): Date of the document, typically the date of a decision (yyyy-mm-dd)
• source_url (string): URL where the document was scraped and where the official version can be found
• scraped_timestamp (string): Date the document was scraped (yyyy-mm-dd)
• unofficial_text (string): Full text of the document (unofficial version, for official version see source_url)
• other (string): Field for additional metadata in JSON format, currently a blank string for most datasets

Programmatic Access in Python (via Hugging Face Datasets):

from datasets import load_dataset
import pandas as pd

dataset = load_dataset("refugee-law-lab/canadian-legal-data", split="train", data_dir="SCC")

# convert to dataframe
df = pd.DataFrame(dataset)
df

Programmatic Access to in Python (via Parquet):

import pandas as pd
import requests
from io import BytesIO

url = 'https://huggingface.co/datasets/refugee-law-lab/canadian-legal-data/resolve/main/SCC/train.parquet'

# load data
results = requests.get(url)

# convert to dataframe
df = pd.read_parquet(BytesIO(results.content))
df

Programmatic Access in Python (via JSON):

import pandas as pd
import json
import requests

# Set variables
start_year = 1877  # First year of data sought (1877 +)
end_year = 2023  # Last year of data sought (2023 -)

# load data
base_ulr = 'https://raw.githubusercontent.com/Refugee-Law-Lab/scc_bulk_data/master/DATA/YEARLY/'
results = []
for year in range(start_year, end_year+1):
    url = base_ulr + f'{year}.json'
    results.extend(requests.get(url).json())

# convert to dataframe
df = pd.DataFrame(results)
df

NOTES:

(1) Data Source: Supreme Court of Canada.

(2) Unofficial Data: The data are unofficial reproductions of materials on the Supreme Court of Canada website. Links to official versions are included in the dataset.

(3) Non-Affiliation / Endorsement: The data has been collected and reproduced without any affiliation or endorsement from the Supreme Court of Canada

(4) Non-Commerical Use: As indicated in the license, data may be used for non-commercial use (with attribution) only. For commercial use, see the Supreme Court of Canada website’s Terms of Use.

(5) Accuracy: Data was collected and processed programmatically for the purposes of academic research. While we make best efforts to ensure accuracy, data gathering of this kind inevitably involves errors. As such the data should be viewed as preliminary information aimed to prompt further research and discussion, rather than as definitive information.

The Refugee Law Lab supports bulk open-access to legal data to facilitate research and advocacy. Bulk open-access also helps avoid asymmetrical access-to-justice and amplification of marginalization that results when commercial actors leverage proprietary legal datasets for profit.

Unless otherwise indicated, the data linked below is subject to a Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0). Users must also comply with upstream data licenses, including requests in source urls to prevent search engine indexation to protect privacy (see details in each dataset).

Full Dataset

Refugee Law Lab: Canadian Legal Data (~105,000 cases)

Court Decisions (Full Text)

Supreme Court of Canada Bulk Decisions Dataset (1877-2023, ~15,500 cases)

Federal Court of Appeal Bulk Decisions Dataset (2001-2023, ~14,000 cases)

Federal Court Bulk Decisions Dataset (2001-2023, ~60,000 cases)

Tax Court of Canada Bulk Decisions Dataset (2003-2023, ~15,000 cases)

Tribunal Decisions (Full Text)

Refugee Law Lab Reporter Bulk Decisions Dataset (2019-present, ~500 cases)

2022 Refugee Claim Data and IRB Member Recognition Rates

The following note and the accompanying data are provided by Sean Rehaag, Director of the Refugee Law Laboratory, Director of the Centre for Refugee Studies, and Associate Professor at Osgoode Hall Law School, York University.

26 March 2023

Data obtained from the Immigration and Refugee Board (IRB) through Access to Information Requests and a data sharing agreement reveals vast disparities in refugee claim recognition rates across decision-makers in 2022. This is consistent with similar findings from prior years for Canada’s previous and new refugee determination systems.

Note that in addition to the description and tables below, interactive visualizations of this data (as well as data for prior years) are available on the new Refugee Law Lab Portal.

In 2022, some Refugee Protection Division (RPD) decision-makers who decided over 30 cases granted refugee status in all the cases they heard, including R. Bafaro (100.0%), K. Bugby (100.0%), M. Robinson (100.0%), H. Savage (100.0%) and D. Simonian (100.0%). Others who decided over 30 cases granted refugee protection much less frequently, including V. Ike (8.6%), G. Brien (9.8%), R. Nirman (11.4%) and P. Panajotov (13.4%).

The tables also show substantial variance for some decision-makers between the recognition rates that would be predicted based solely on the average recognition rates for the countries of origins in the cases they decided, and their actual recognition rates. For instance, L. Parker (predicted 57.7%; actual 95.8%), A. Guarino (predicted 58.9%; actual 94.7%) and S. Yorgun (predicted 64.3%; actual 96.1%) had much higher recognition rates than predicted, whereas B. Guillaume (predicted 59.7%, actual 9.8%), V. Ike (predicted 53.1%; actual 8.6%) and P. Panajotov (predicted 56.3%; actual: 13.4%) had much lower recognition rates than predicted. All decided at least 30 cases.

Some of the recognition rate variation observed in the data is due to specialization in particular types of cases. For example, some decision-makers specialize in geographic regions with especially high or low refugee claim recognition rates. It should also be kept in mind that to enhance efficiency the RPD has recently placed increased emphasis on streaming cases into different categories, including expedited cases that are granted based on paper reviews rather than hearings. The proportion of such claims heard by particular decision-makers may affect their recognition rates. For further possible explanations for variations in recognition rates, please see an explanatory note (also available in French) that was provided by the IRB.

The data for 2022 also includes information about outcomes on appeals at the IRB’s Refugee Appeal Division (RAD). Like the variations seen in RPD decision-making, RAD decision-makers have very different rates at which they grant appeals. For example, in RAD cases decided on the merits, claimants were much more likely to succeed in their appeals before T. Acton (80.4%), R. Liberman (73.7%) or I. Singh (68.1%) than before M. Lamani (2.9%), R. Al-Salem (5.2%) or A.G. Poole (9.0%). All decided at least 30 cases.

A few implications of this year’s data are worth highlighting:

• The persistence of variations in recognition rates across adjudicators, combined with the devastating potential impact of false negative refugee decisions (i.e. refugees being returned to face persecution), make robust oversight mechanisms essential. Unfortunately, some refugee claimants continue to be denied access to the appeal at the IRB and are ineligible for automatic stays on removal pending judicial review at the Federal Court. This includes claimants who transited to Canada via the United States under an exception to the Safe Third Country Agreement – even though one’s route to Canada has little to do with whether one has a well-founded fear of persecution. For further analysis, see: https://ssrn.com/abstract=2647638
• The overall success rate in RAD appeals is remarkably high. Indeed, appeals brought by claimants and decided on the merits in 2022 were granted in around a third of cases (31.9%). The fact that the RAD so frequently finds either that the claim was wrongly denied at the RPD or that new evidence demonstrates that the claimant is entitled to refugee protection emphasizes the importance of this form of oversight. And it is yet another reason why all claimants, including those who have transited to Canada via the United States, must be entitled to a full appeal on the merits. For further analysis, see: https://ssrn.com/abstract=2647638

For a discussion of the methodology used to obtain the data and to calculate the statistics, as well as an analysis of the implications of similar data for a previous year, see https://ssrn.com/abstract=1468717

For additional breakdowns of data by claim type, see: https://ssrn.com/abstract_id=4341740

This data was provided by the IRB through a data sharing agreement that requires compliance with the Treasury Board Privacy Implementation Notice and the IRB’s Small Value Suppression Policy. While we are grateful to the IRB for engaging with researchers in this way – and while we share the view that it is important to protect the privacy interests of refugee claimants – the agreement limits how much data we are able to publicly share. Thus, for example, unlike in many past years we are not posting the raw data we obtained. We are also only posting statistics relating to outcomes in cases decided on the merits by Board Members (leaving hundreds of claims decided on other grounds out of the reported figures). We are also only posting recognition rates, rather than counts of cases and breakdowns of outcomes. If you are interested in general statistics on refugee claim outcomes, including breakdowns for all claims from particular countries, we encourage you to consult the statistics webpages of the IRB and the United Nations High Commissioner for Refugees.

Tables for RPD Cases:

1.1. Outcomes by Board Member (Alphabetical)

1.1a. Outcomes by Board Member (Organized by Recognition Rate, 30+ Decisions)

1.1b. Outcomes by Board Member (Organized by RR Nominal Variance, 30+ Decisions)

1.2. Outcomes by Board Member (Excluding Paper Review & Expedited Positive Decisions) (Alphabetical)

1.2a. Outcomes by Board Member (Excluding Paper Review & Expedited Positive Decisions) (Alphabetical) (Organized by Recognition Rate, 30+ Decisions)

1.2b. Outcomes by Board Member (Excluding Paper Review & Expedited Positive Decisions) (Alphabetical) (Organized by RR Nominal Variance, 30+ Decisions)

1.3. Outcomes by Board Member and Country

1.4. Outcomes by Country and Board Member

Tables for RAD Cases:

2.1. RAD Outcomes by Board Member (Alphabetical)

2.2 RAD Outcomes by Board Member (Organized by Allowal Rate, 30+ Decisions)

To be cited as: Sean Rehaag, “2022 Refugee Claim Data and IRB Member Recognition Rates” (26 March 2023), online: https://refugeelab.ca/refugee-claim-data-2022.

NOTES:

• The data was obtained through Access to Information Request A-2022-01644 and a data sharing agreement with the IRB dated 8 July 2022.
• Tables 1.1, 1.1a, 1.1b, 1.3 and 1.4 include only cases resulting in positive (including expedited positive and paper review positive) or negative (including no credible basis) decisions, excluding cases that were abandoned, withdrawn or otherwise decided. Tables 1.2, 1.2a and 1.2b, include only cases resulting in positive (excluding expedited positive and paper review positive) or negative (including no credible basis) decisions, excluding cases that were abandoned, withdrawn or otherwise decided.
• All statistics (including recognition rates) include only principal applicant claims (i.e. excluding associated claims by family members of principal applicants) made under the new refugee determination system that came into effect on 15 December 2012.
• A small number of cases may be decided by panels of Board Members. Only the first listed Board Member is included in the statistics.
• Country of origin averages and predicted recognition rates are calculated separately for tables that include (i.e. Tables 1.1, 1.1a, 1.1b, 1.3 and 1.4) and exclude (i.e.  Tables 1.2, 1.2a and 1.2b) expedited positive and paper review positive decisions.
• The data refers to “recognition rates”. The term “recognition rate” is used to mean the proportion, expressed as a percentage, of positive (including expedited positive) decisions relative to the total number of positive (including expedited positive and paper review positive where applicable) and negative (including no credible basis) decisions, excluding cases that are abandoned, withdrawn or otherwise resolved. This is the standard practice for reporting outcomes by the United Nations High Commissioner for Refugees (https://www.unhcr.org/statistics), and it is the way that both “recognition rates” and “grant rates” were reported for data obtained for prior years (see links below).
• Tables 2.1 and 2.1a only include principal applicant RAD appeals brought by claimants (i.e. excluding appeals brought by the minister) that are decided on the merits (i.e. excluding appeals that are abandoned, withdrawn, not perfected, denied on jurisdictional grounds, or otherwise resolved).
• This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License (CC BY-NC-SA 4.0)

Sean Rehaag

Director, Refugee Law Laboratory
Director, Centre for Refugee Studies (on sabbatical leave in 2022-23)
Associate Professor, Osgoode Hall Law School
York University

Data from previous years via Refugee Law Lab:

2021

2020

2019

Data from previous years via Canadian Council for Refugees:

2018
2017
2016
2015
2014
2013
2012
2011 (Updated)
2011 (Original)
2010
2009
2008
2007
2006

2021 Refugee Claim Data and IRB Member Recognition Rates

The following note and the accompanying data are provided by Sean Rehaag, Director of the Refugee Law Laboratory, Director of the Centre for Refugee Studies, and Associate Professor at Osgoode Hall Law School, York University.

21 October 2022

Data obtained from the Immigration and Refugee Board (IRB) through Access to Information Requests and a data sharing agreement reveals vast disparities in refugee claim recognition rates across decision-makers in 2021. This is consistent with similar findings from prior years for Canada’s previous and new refugee determination systems.

In 2021, some Refugee Protection Division (RPD) decision-makers who decided over 30 cases granted refugee status in all the cases they heard, including K. Bugby (100.0%), D. Coyne (100.0%), H Savage (100.0%) and D. Simonian (100.0%). Others who decided over 30 cases granted refugee protection much less frequently, including M. Bourassa (9.3%), P Panajotov (3.0%) and D. Mungovan (0.0%).

The tables also show substantial variance for some decision-makers between the recognition rates that would be predicted based solely on the average recognition rates for the countries of origins in the cases they decided, and their actual recognition rates. For instance, J. Berkovits (predicted 69.3%; actual 95.9%), C. Côté (predicted 63.2%; actual 88.9%) and A. Guarino (predicted 64.2%; actual 89.4%) had much higher recognition rates than predicted, whereas M. Bourassa (predicted 54.0%, actual 9.3%), K. Gibson (predicted 73.4%; actual 23.6%) and D. Mungovan (predicted 59.5%; actual: 0.0%) had much lower recognition rates than predicted. All decided at least 30 cases.

Some of the recognition rate variation observed in the data is due to specialization in particular types of cases. For example, some decision-makers specialize in geographic regions with especially high or low refugee claim recognition rates. It should also be kept in mind that to enhance efficiency the RPD has recently placed increased emphasis on streaming cases into different categories, including expedited cases that are granted based on paper reviews rather than hearings. The proportion of such claims heard by particular decision-makers may affect their recognition rates. For further possible explanations for variations in recognition rates, please see an explanatory note (also available in French) that was provided by the IRB.

The data for 2021 also includes information about outcomes on appeals at the IRB’s Refugee Appeal Division (RAD). Like the variations seen in RPD decision-making, RAD decision-makers have very different rates at which they grant appeals. For example, in RAD cases decided on the merits, claimants were much more likely to succeed in their appeals before J. Corry (71.7%), H. Shepherd (70.8%) or J. Pollock (69.2%) than before G. Guerrier (5.4%), E. Rose (4.1%) or M. Lamani (1.1%). All decided at least 30 cases.

A few implications of this year’s data are worth highlighting:

• The persistence of variations in recognition rates across adjudicators, combined with the devastating potential impact of false negative refugee decisions (i.e. refugees being returned to face persecution), make robust oversight mechanisms essential. Unfortunately, some refugee claimants continue to be denied access to the appeal at the IRB and are ineligible for automatic stays on removal pending judicial review at the Federal Court. This includes claimants who transited to Canada via the United States under an exception to the Safe Third Country Agreement – even though one’s route to Canada has little to do with whether one has a well-founded fear of persecution. For further analysis, see: https://ssrn.com/abstract=2647638
• The overall success rate in RAD appeals is remarkably high. Indeed, appeals brought by claimants and decided on the merits in 2021 were granted in around a third of cases (34.9%). The fact that the RAD so frequently finds either that the claim was wrongly denied at the RPD or that new evidence demonstrates that the claimant is entitled to refugee protection emphasizes the importance of this form of oversight. And it is yet another reason why all claimants, including those who have transited to Canada via the United States, must be entitled to a full appeal on the merits. For further analysis, see: https://ssrn.com/abstract=2647638

For a discussion of the methodology used to obtain the data and to calculate the statistics, as well as an analysis of the implications of similar data for a previous year, see https://ssrn.com/abstract=1468717

Note that this data was provided by the IRB through a data sharing agreement that requires compliance with the Treasury Board Privacy Implementation Notice and the IRB’s Small Value Suppression Policy. While we are grateful to the IRB for engaging with researchers in this way – and while we share the view that it is important to protect the privacy interests of refugee claimants – the agreement limits how much data we are able to publicly share. Thus, for example, unlike in many past years we are not posting the raw data we obtained. We are also only posting statistics relating to outcomes in cases decided on the merits by Board Members (leaving hundreds of claims decided on other grounds out of the reported figures). We are also only posting recognition rates, rather than counts of cases and breakdowns of outcomes. If you are interested in general statistics on refugee claim outcomes, including breakdowns for all claims from particular countries, we encourage you to consult the statistics webpages of the IRB and the United Nations High Commissioner for Refugees.

Tables for RPD Cases:

1.1. Outcomes by Board Member (Alphabetical)

1.1a. Outcomes by Board Member (Organized by Recognition Rate, 30+ Decisions)

1.1b. Outcomes by Board Member(Organized by RR Nominal Variance, 30+ Decisions)

1.2. Outcomes by Board Member (Excluding Paper Review & Expedited Positive Decisions) (Alphabetical)

1.2a. Outcomes by Board Member (Excluding Paper Review & Expedited Positive Decisions) (Alphabetical) (Organized by Recognition Rate, 30+ Decisions)

1.2b. Outcomes by Board Member (Excluding Paper Review & Expedited Positive Decisions) (Alphabetical) (Organized by RR Nominal Variance, 30+ Decisions)

1.3. Outcomes by Board Member and Country

1.4. Outcomes by Country and Board Member

Tables for RAD Cases:

2.1. RAD Outcomes by Board Member (Alphabetical)

2.2. RAD Outcomes by Board Member (Organized by Allowal Rate, 30+ Decisions)

To be cited as: Sean Rehaag, “2021 Refugee Claim Data and IRB Member Recognition Rates” (21 October 2022), online: https://refugeelab.ca/refugee-claim-data-2021.

NOTES:

• The data was obtained through Access to Information Request A-2021-01792 and a data sharing agreement with the IRB dated 8 July 2022.
• Tables 1.1, 1.1a, 1.1b, 1.3 and 1.4 include only cases resulting in positive (including expedited positive and paper review positive) or negative (including no credible basis) decisions, excluding cases that were abandoned, withdrawn or otherwise decided. Tables 1.2, 1.2a and 1.2b, include only cases resulting in positive (excluding expedited positive and paper review positive) or negative (including no credible basis) decisions, excluding cases that were abandoned, withdrawn or otherwise decided.
• All statistics (including recognition rates) include only principal applicant claims (i.e. excluding associated claims by family members of principal applicants).
• A small number of cases were decided by panels of Board Members. Only the first listed Board Member is included in the statistics.
• Country of origin averages and predicted recognition rates are calculated separately for tables that include (i.e. Tables 1.1, 1.1a, 1.1b, 1.3 and 1.4) and exclude (i.e.  Tables 1.2, 1.2a and 1.2b) expedited positive and paper review positive decisions.
• The data refers to “recognition rates”. The term “recognition rate” is used to mean the proportion, expressed as a percentage, of positive (including expedited positive) decisions relative to the total number of positive (including expedited positive and paper review positive where applicable) and negative (including no credible basis) decisions, excluding cases that are abandoned, withdrawn or otherwise resolved. This is the standard practice for reporting outcomes by the United Nations High Commissioner for Refugees (https://www.unhcr.org/statistics), and it is the way that both “recognition rates” and “grant rates” were reported for data obtained for prior years (see links below).
• Tables 2.1 and 2.1a only include principal applicant RAD appeals brought by claimants (i.e. excluding appeals brought by the minister) that are decided on the merits (i.e. excluding appeals that are abandoned, withdrawn, not perfected, denied on jurisdictional grounds, or otherwise resolved).
• This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License (CC BY-NC-SA 4.0)

Sean Rehaag

Director, Refugee Law Laboratory
Director, Centre for Refugee Studies (on sabbatical leave in 2022-23)
Associate Professor, Osgoode Hall Law School
York University

Data from previous years via Refugee Law Lab:

2020
2019

Data from previous years via Canadian Council for Refugees:

2018
2017
2016
2015
2014
2013
2012
2011 (Updated)
2011 (Original)
2010
2009
2008
2007
2006

Statistics and Documents Relating to Deportation of People with a Refused Refugee Claim

Principal Researcher: Kathryn Tomko Dennler

Last updated: November 8, 2022

The data repository contains data received from Access to Information and Privacy (ATIP) requests to obtain data and corporate documents related to the removal of people with a refused refugee claim, as well as records from relevant ATIP requests made by others. Most documents are from Canada Border Services Agency (CBSA), with some from Immigration Refugees and Citizenship Canada (IRCC).

All data refers to people with a refused refugee claim unless otherwise specified in the table of contents or on the data itself.

Contents:

• Removals statistics
• Removal inventory
• Outcomes of immigration applications
• CBSA internal guidance documents
• Select CBSA job descriptions

The ATIP requests were made between September 2020 and May 2022. Updates to the data repository will be infrequent.

You can read the public report from the research: https://romerohouse.org/wp-content/uploads/2022/08/Report-on-deportation.pdf

Data repository to be cited as: Dennler, Kathryn Tomko, “Deportation Data Repository: Statistics and Documents Relating to Deportation of People with a Refused Refugee Claim” (October 24, 2022), online: https://refugeelab.ca/projects/deportation-data.

Useful links:

• CBSA corporate documents (public): https://www.cbsa-asfc.gc.ca/agency-agence/reports-rapports/menu-eng.html
• Access to ATIP requests made by other people: https://open.canada.ca/en/search/ati
• Make an ATIP request: https://atip-aiprp.tbs-sct.gc.ca/

If you have questions about the data and research, or if you have documents to add to the data repository, please contact the principal researcher, Kathryn Tomko Dennler by email: Kathryn dot Dennler at gmail dot com.