Who Counts in Canada

Which outliers are excluded in order to clean data? Often these outliers are aboriginal populations who are consistently and systematically discriminated against by the Canadian government, and federal statistics agencies are no exception. What are the implications of being a statistical outlier in Canada? If proper methodology and sampling techniques do not take you into account then social policy research will not take into consideration your community, your livelihood, and your lived reality. Although data on aboriginal populations has been collected throughout the years, there are some historical changes that have occurred to the Canadian census that will have serious implications for First Nations populations. In addition to these misrepresentations, a grosser statistical and social crime has been committed. In one of the most crucial areas of research that could be used to create social policy that can aid in revitalizing Aboriginal communities, First Nations data has been completely removed, or not collected in the Labour Force Survey. (LFS) There are no records of unemployment statistics for First Nations living on reserves under the guise of the Federal LFS. Why is this happening? What are the implications of this purposeful act of ignorance?

Historically, Aboriginal populations were included in the long-form census surveys through the option of self-identifying as either "North American Indian," "Metis," or "Inuit" since 1996 (AANDC 2013). Previous to this, Aboriginal populations could participate in national surveys based on official Indian status, and ancestry (AANDC 2013). Since 2011, the long-form census has been replaced by the National Household Survey (NHS). The last long-form Canadian census in 2011 had a total of 31 Indian reserves and settlements that were 'incompletely enumerated' (StatsCan Appendix 3 2011). Enumeration was either not permitted, interrupted, or not possible due to extenuating circumstances such as the Ontario wildfires (StatsCan Appendix 3 2011). Despite, these discrepancies the long-form census remained a reliable resource for representative samples of Canada, and could be used to combat against smaller, but still potent reports, such as the Labour Force Survey. However, under the Conservative government in 2010, the decision was made to eliminate the mandatory census under the guise of 'privacy complaints,' and the voluntary National Household Survey took its place.

The content of these two forms is virtually the same, all households in First Nations and Inuit communities receive the questionnaire, and participation in First Nations and Inuit communities was reported as higher than in other communities in Canada (AANDC 2013). Within this survey, information about immigration, ethnocultural diversity, education, labour, mobility, migration, income, housing, and family arrangements has been collected (National Household Survey Dictionary 2011). 

However, despite these assurances outlined on official Canadian government sites, there are some immediate general issues with the new NHS that need to be addressed; in addition, to some ominous implications specifically for Aboriginal populations. Statistics Canada warns about the usability, comparability, and validity of the NHS by noting that this methodology has been introduced rapidly with very limited testing, and will not provide the same level of quality information on Canada's status (StatsCan 2011). In addition, such changes can lead to a decrease in response rates from marginalized groups, and aboriginal communities (Green and Milligan 2010). With a decrease in response rates from particular populations, and subsequently an increase in non-response bias, the government could try to justify a reallocation of money away from programs that aid the under-represented. Specifically, Aboriginal funding grant decisions and policies based on statistical data may not be accurate to the population and their needs. NHS methodologies and sampling techniques also do not provide as strong of counter to the already innumerable other surveys and data collection endeavours employed by the government.

One of the most infamous of these sub-divisions of Statistics Canada is the Labour Force Survey. This survey covers research areas such as employment and unemployment rates, labour market sectors, and wealth disparities. This information is used in order to inform policies about funding for welfare, economy stimulating projects, infrastructure, and much more. Typically, these statistics are compared or complimented by the long-form census; however, economists warn that with the methodological problems of the NHS, such comparisons may not yield fruitful anymore (Darroch 2010). This is due, in part, to the non-response bias of lower-income families and higher-income families because of the voluntary nature of the survey. If this is the case, the NHS results will tend to be skewed towards a middle class, and paint a rosier picture of Canada’s income gap, wealth equality, and employment rates (Darroch 2010). The problematic result of this, is that biases in other surveys, such as the Labour Force Survey will be exaggerated and not reduced. For example, to make matters worse, the complimentary Labour Force Survey contains a startling misrepresentation of Canada; in so much as, First Nations data has been completely removed. This removal of First Nations data is a powerful tool in making one of Canada’s most under represented and economically suffering populations invisible on two national levels of data collection; thus, by exaggerating the false results of Canada’s equal economy, there is a systematic exacerbation of social and economic discrimination against First Nations populations.

Every month, Statistics Canada releases unemployment figures using the Labour Force Survey (LFS), based on a sample of approximately 54,000 households that rotate (no household stays in the sample for more than six months) (LFS 2015). According to AANDC, in 2014 there were 901,053 registered Indians, and based on the 2011 NHS 1,836,035 people in Canada who report having Aboriginal ancestry (5.6% of the population) (AANDC 2014). Furthermore, 52.6% of the Aboriginal population in Canada lives on reserves or on communities on ‘Crown’ land (AANDC 2014), yet none of these populations are included in the LFS and in national unemployment rates.

Statistics Canada’s rationalization for the removal of Aboriginal data from the LFS is that it presents serious methodological challenges for contacting and interviewing potential respondents, and the cost and time it would take to travel to every location given the short month to collect data in (StatsCan 2012). However, there seems to be a major discrepancy between this statement and the actual methods used, according to the Methodology of the Canadian Labour Force Survey (2008):

"LFS interviews are conducted using two collection methods, computer-assisted personal interviewing (CAPI) and computer-assisted telephone interviewing (CATI). Historically, CAPI has been used for households in their first month of the survey, with interviewers visiting in person to conduct the interview. [Subsequent interviews] … are normally conducted using CATI…"

Since 2004, households have been brought in the first time via a telephone call, not an in-person interview (StatsCan 2008). Yet, somehow according to both the 2009 and more recent 2012 edition of the LFS’s justifications, it would just be too strenuous for data collection to “travel” (I thought that was not necessary?), and too high of cost and time taken to interview on-reserve families (because apparently a computer-assisted phone call to a reserve would be infinitely more expensive and time-consuming).

This is a poor excuse. Computer-assisted telephone interviews have been used to collect on-reserve data in government surveys before, and there is no reason, other than a political agenda, that it should not happen for the LFS as well. Federal government commissioned phone surveys have occurred frequently throughout the years. According to the LFS’s most recent report based on the May 2015 results, Canada’s unemployment rate is decreasing, but how would this rate change if Aboriginal data were actually to be included?

There are many social implications of removing Aboriginal data from unemployment statistics. The foremost of these reasons is that it will maintain the status quo; therefore, enforcing the systematic discrimination of First Nations in Canada, and enforcing systems that foster high social stratification and impoverished areas.

The pro-natalist nature of statistics is undeniable. Through the collection of census information and statistical data, countries are able to not only inform federal policies, but also purport an international presence. By removing unemployment statistical data that would heavily decrease Canada's rate, Canada can remain higher on the list for international reports. For example, Canada ranks seventh on the international OCED development index for income (OCED 2014). However, according to my approximations, the inclusion of Aboriginal unemployment data compared to other nation's unemployment rates would decrease Canada's ranking. This change in ranking affects Canada's international standing as "one of the most liveable countries,” and calls into question international development qualifications, when there are serious development problems within the nation itself.

Even within national statistical reconfigurations, the unemployment rate would no doubt be affected by the inclusion of reserves data. The most reliable resource for Aboriginal statistics would be derived from the long-form census, and according to the 2006 report (before the transition from the mandatory original census to the flexible, small-scale NHS), there was 2,600 First Nations people who reported being unemployed (StatsCan 2006). On the 35 reserves recorded, the average unemployment rate was 30% (StatsCan 2006). If a region employed 100,000 people (the average regional area size used to calculate the minister of finance's 6% unemployment rate indicator in order to create appropriate policy), the unemployment rates would increase by at least two percentage points in the corresponding regional areas. This pushes most regional area unemployment scores over the maximum 6% rate.

How would such a reconfiguration affect First Nations populations in Canada? One concrete example is the local effect of the temporary foreign worker program on reserve populations. Although Kenney, Canada's minister of finance, banned the implementation of this program in regional areas where unemployment rates supersede 6%, with these new reconfigurations, areas such as Northern Saskatchewan, which host the Ermineskin Cree Nation, at a 5.7% unemployment rate, would not be allowed to hire temporary foreign workers on reserve businesses, and surrounding neighbourhoods. Instead, employees from reserves would be encouraged to work, hopefully decreasing the 60% unemployment rate on this particular reserve (StatsCan 2006, Globe and Mail).

These problems are complicated even further by the recent removal of the long form census in Canada under the Harper government (Darroch 2010, Green and Milligan 2010). By removing census data, only statistical information from the early 2000s will be able to be used to represent Canadians, reliance on the incomplete representation of the NHS, or the adoption of recent American statistics, all of which are a gross misrepresentation. Aside from the obvious problems of not having recent information on the state of the nation, this means that recent statistical data that does remove Aboriginal populations from analysis will be continually recycled and presented as the "current" data of Canadian populations. Thus by being rendered invisible during this step of statistics, there is a potential to be infinitely rendered invisible under Canadian federal statistics.

How can this cycle of systematic mathematical discrimination be combatted? The onus for ethical and thorough collection of data now seems to fall on the shoulders of NGOs, and private researchers and collectives. Naturally, this seems to be a product of the very neoliberal agenda that the Harper government is pushing on Canadians. This pursuit will be a challenging one, as major funding for social research can be scarce to begin with; moreover, as the values of the free market and competition entrench the applications for funding, those that are applying for accurate representation of marginalized populations may not privileged over those researching subjects that fit more nicely under capitalist ideals. Will aboriginal unemployment rate research be chosen over health policy research, or family planning research? 

Despite these potential difficulties in conducting research privately, there is a potential to create impactful research collectives that share a vision of social justice, and priority for Indigenous rights. Perhaps this shared vision could not be accomplished within Canadian statistical collection because of the existing prejudice within the polis of policy making (Westhues & Wharf 2012). Of course to truly solve the root of the problem, these prejudices within Canada need to be addressed, but if by necessity, revolutionary statistics through smaller groups could be achieved through pure grit and determination. By recognizing a systematic discrimination in statistics, it allows for the root issues and prejudices to be addressed, and by providing alternative quantitative research these existing prejudices will be illuminated until the discrepancies are unavoidably obvious.

Another issue that needs to be addressed is the translation process of statistical analysis within the government and social research sectors, to the popular media centres. How accurate is this information? Does the misrepresentation or elimination of Aboriginal statistics within popular media discourse, necessarily predicate a misrepresentation or elimination of Aboriginal statistics within other, more officialized, spheres? I argue that this is not always necessarily true. There are a few important aspects to remember when dealing with any representation of statistics. First, all statistics are a manipulation of abstract mathematical concepts; they are inherently constructions (Sismondo 2010). For example this is how you can create a normal curve distribution with an abnormal population. In addition data collection can differ so greatly; you cannot separate human/research manipulation from the process. Second, when statistical information is presented, (especially in media) it is often presented under false pretences. These false pretences may include not citing the source, the confidence level or degree of error, the proper context, or the population. By manipulating the representation of statistics like this, information (such as unemployment), can be assumed to be a projection of entire nation-state, rather than a small group of people that (most likely) isn't fully representative.

Even aside from the issues of the misrepresentation of First Nations through statistical manoeuvres made by media monopolies, the issues of misrepresentation in the National Household Survey and elimination in the Labour Force Survey remain to be a prioritized concern for social scientists, activists, and economists alike. There is a need for more accurate statistical methodologies and sampling techniques, a return to the long-form census, and an inclusion of ALL Canadians in data collection and analysis stages. In addition, the implicit prejudices in the bureaucracies that handle statistics and social policies needs to be addressed. It is a serious problem when rationalizations for the exclusion of  an entire group of people, even more so, a founding people of our very nation, are skimmed over in a short paragraph that does not even truthfully explain itself and the methods being employed. How is it, that Statistics Canada, and the federal government think that this is acceptable?

 photo SimpleSignature_zps5a059e3e.png


Aboriginal Affairs and Northern Development Canada. 2014. First Nations People in Canada. https://www.aadnc-aandc.gc.ca/eng/1303134042666/1303134337338

Aboriginal Affairs and Northern Development Canada. 2013. Aboriginal Data as a Result of Changes to the 2011 Census of Population. (Catalogue R3-196/3013E-PDF)

Darroch, M., & Darroch, G. (2010). Commentary Losing Our Census. Canadian Journal of Communication, 35(4), 609.

Globe and Mail. 2014. Statistical black hole opens door to foreign workers Statistical black hole opens door to foreign workers. http://www.theglobeandmail.com/news/national/canadas-skewed-labour-data-tips-balance-in-favour-of-foreign-workers/article21158372/

Green, D. A., & Milligan, K. (2010). The importance of the long form census to Canada. Canadian Public Policy, 36(3), 383-388.

Labour Force Survey. 2015. The Daily: Labour Force Survey, May 2015. http://www.statcan.gc.ca/daily-quotidien/150605/dq150605a-eng.htm?HPA

National Household Survey Dictionary. 2011. (Catalogue 99-000-X20111001). http://www12.statcan.gc.ca/nhs-enm/2011/ref/dict/99-000-x2011001-eng.pdf

OCED. 2014. Canada. http://www.oecdbetterlifeindex.org/countries/canada/

Sismondo, S. (2010). An introduction to science and technology studies. Chichester, West Sussex, U.K.: Wiley-Blackwell.

Statistics Canada. 2012. A Guide to the Labour Force Survey. (Catalogue 71-543-G). http://www.statcan.gc.ca/pub/71-543-g/71-543-g2012001-eng.pdf

Statistics Canada. Appendix 3 for 2011 Canadian Census. 2011. http://www12.statcan.gc.ca/census-recensement/2011/ref/irr-app-ann-3-eng.cfm

Statistics Canada. NHS: Data Quality. 2011. http://www12.statcan.gc.ca/NHS-ENM/2011/ref/about-apropos/nhs-enm_r005-eng.cfm

Statistics Canada.Methodology of the Canadian Labour Force Survey. 2008. http://www.statcan.gc.ca/pub/71-526-x/71-526-x2007001-eng.htm

Statistics Canada. Aboriginal Peoples in Canada in 2006: Inuit, Métis and First Nations, 2006 Census (Catalogue 97-558-XIE) Ottawa: Minister of Industry, 2008a.

Warren, J.W. (2001). Racial revolutions: Antiracism and Indian resurgence in Brazil. Durham: Duke University Press.

Westhues, A. & Wharf, B. (2012). Canadian Social Policy: Issues and Perspectives, 5th Edition. Waterloo, ON: Wilfred Laurier University Press.

Appendix of Tables

Table 1: Population Estimates of Incompletely Enumerated Indian Reserves by Province and Territory, 2011 Census

Table 2: LFS May 2015 Unemployment Rate Chart

Table 3: Sample Federally Commissioned Phone Surveys





*Largest sample size was 2,002 First Nations residents. Quite an “expensive and time consuming” sample to say the least.*
Table 4: OCED Canadian Income/Employment at a Glance
© Miss Lauren Kyle
Maira Gall