Ethnicity, Language and Immigration Thematic Series
Recent immigrants and non-permanent residents missed in the 2011 Census

by Julien Bérard-Chagnon, Stacey Hallman and Geneviève Caron

Release date: May 22, 2019

Skip to text

Text begins

Acknowledgements

The authors would like to thank a number of Statistics Canada employees for their contribution to this project. First, we would like to thank Mélanie Meunier and David Binet of the Demography Division (DEM) and Martin St-Pierre of the Social Survey Methods Division (SSMD) for their invaluable help in manipulating the data from Immigration, Refugees and Citizenship Canada (IRCC) and from the Reverse Record Check (RRC). The first versions of this document also benefitted from the comments of Éric Caron Malenfant, Tristan Cayn, Martin Turcotte, Kathryn Spence, and Scott McLeish of the Social and Aboriginal Statistics Division (SASD), and Dave Dolson (SSMD).

Start of text box

Highlights

  • Recent immigrants, or those who landed in the country in the five years preceding the census, and non-permanent residents (NPRs) were much more likely to be missed in the 2011 Census than the rest of the population.
    • The missed rates for these two groups, estimated at 17.8% and 43.2% respectively, were much higher than that for the total population (8.3%).
  • Several characteristics of recent immigrants were associated with being missed in the 2011 Census.
    • Nearly 20% of recent immigrants from Asia or Oceania and over 30% of those whose mother tongue was Punjabi were missed.
    • More than one-third of recent immigrants who landed in 2011 and about a quarter of those who landed in 2010 were missed.
    • About 12% of recent immigrants admitted in the refugee categories were missed.
    • Knowledge of official languages, age and marital status were also correlated with the propensity for being missed in the 2011 Census.
  • Some characteristics of NPRs were also related to the propensity for being missed in the 2011 Census.
    • Almost half of NPRs who were not in a couple were missed.
    • More than 50% of NPRs who were granted temporary residence status in Canada no more than six months before the census were missed.
    • Age and having a temporary residence permit for the first time were also relatively correlated with the propensity for being missed in the 2011 Census.
End of text box

Introduction

International migration is playing an increasingly important role in Canada’s demographic dynamics. Since 1995/1996, international migration has always been the main driver of population growth (Statistics Canada 2016a: 16). If recent demographic trends continue, international migration may become the driver of almost all the country’s population growth over the next few decades (Statistics Canada 2015a: 10).

The two major groups that make up international migration are immigrants and non-permanent residents (NPRs). An immigrant is a person who has been granted the right to live in Canada permanently by immigration authorities. An NPR is a person who has legally been granted the right to live in Canada on a temporary basis under the authority of a temporary resident permit, such as a work permit (Statistics Canada 2016b).

These two demographic groups are growing in Canada. In 2011, immigrants made up 20% of the Canadian population and could represent more than a quarter of the Canadian population in 2036 (Statistics Canada 2017a). Moreover, the number of NPRs more than tripled between 1997 and 2014, from 234,400 to 770,600 (Martel and D’Aoust 2016).

In parallel to that fast demographic growth, these two populations face a number of significant socioeconomic issues. For example, immigrants, particularly recent immigrants, are more likely to be overqualified (Uppal and Larochelle-Côté 2014), to have less success on literacy and numeracy proficiency tests (Statistics Canada 2013) and to live in a low-income situation (Picot and Hou 2014). The sharp increase in temporary resident permits in the last few years also raises some issues with respect to the actual need for temporary labour to support the Canadian economy (McQuillan 2013). The Canadian government acknowledges many of these barriers and is developing various policies to promote the integration of immigrants into Canadian society and the entry of temporary residents (Immigration, Refugees and Citizenship Canada 2016).

Many issues specific to international migration are studied using census data. However, while the census strives to provide comprehensive coverage of the population, some groups are less likely to be enumerated. This is known as undercoverage. Not only has it increased in recent decades, but it is especially pronounced among recent immigrantsNote 1 and NPRs (Statistics Canada 2010; Statistics Canada 2015b). The relatively higher undercoverage observed for these groups affects the quality of census data by making them less representative of the population and potentially diminishing the scope of the studies conducted based on these data.

Despite the growing proportion of immigrants and NPRs in the Canadian population, the magnitude of the socioeconomic issues faced by these two groups and their undercoverage in censuses, the specific mechanisms associated with being missed in censuses remain largely unknown. The purpose of this paper is to examine the characteristics associated with being missed in the 2011 Census, for recent immigrants and NPRs, using data from the Reverse Record Check (RRC).

The next section presents issues related to census undercoverage. The data used are described in the second section. The last two sections present the 2011 omissions rates for recent immigrants and NPRs, respectively, by different characteristics available in the RRC and data from Immigration, Refugees and Citizenship Canada (IRCC).

1. Census coverage issues

Censuses are the cornerstone of a country’s demographic measure. In contrast to surveys, censuses strive to provide a comprehensive enumeration of the population and its main demographic, economic and social characteristics (Bryan 2004). However, in reality, despite considerable efforts, censuses fail to count the entire population. According to the United Nations Economic Commission for Europe (UNECE):

“Census designers and administrators must keep in mind that no matter how much effort is expended, complete coverage and accuracy in the census data are unattainable goals.” (UNECE 2015: §366).

UNECE identifies two main types of errors that can occur during a census. In general, content errors are due to inaccurate statements or incorrect records, while coverage errors result from omissions or duplication during enumeration.

Coverage errors are the ones we are interested in for this study. As mentioned above, they can be divided into two categories. Undercoverage occurs when people who should have been counted were not, while overcoverage counts individuals more than once or counts those who should not have been counted (Statistics Canada 2015b). Coverage errors can occur for a variety of reasons, such as refusals to respond to the census, individuals who are counted in the wrong place, natural disasters that affect collection operations, or collection errors in the field.

Undercoverage is generally greater than overcoverage, making it a prevalent issue for most statistical agencies and data users. If a segment of the population is undercounted, the census figures are less representative of this population, which could decrease the relevance of the census data and affect the results obtained as well as the policies developed using these data.

1.1. Undercoverage in Canadian censuses

Due to the potential consequences of coverage errors, the UNECE (2015: §373) recommends assessing the completeness and accuracy of [census] data and publishing the findings, as much as possible: "along with the initial census results, including a detailed description of the methods used. Additional results can be released at a later date."

Several countries are therefore using different statistical operations to assess the coverage of their censuses. For example, the U.S. Census Bureau uses a Demographic Analysis (DA) and a Census Coverage Measurement (CCM) study (U.S. Census Bureau 2014), while the Office for National Statistics (ONS) in the United Kingdom conducts a Census Coverage Survey (Office for National Statistics 2012).

In Canada, Statistics Canada has performed coverage studies to measure the coverage of the Canadian censuses since 1961 (Dolson 2010). The results of the coverage studies are published on the Statistics Canada website approximately 28 months after each census and the technical report is published a few months later. The technical report provides information on the methods used to estimate coverage and the level of coverage error.

The issues relative to undercoverage in the censuses notably attracted attention in the 1980s. Some studies at the time shed light on the possible effects of census coverage errors on different demographic indicators (Malo 1981; Bourbeau and Robitaille 1980) and demographic estimatesNote 2 used to apply many laws based on population counts (Fellegi 1980; Keyfitz 1989). These studies led Statistics Canada to consider adjusting census data for coverage in order to calculate demographic estimates. The series of demographic estimates based on the 1991 Census was the first series adjusted for net undercoverage. This was done by combining coverage studies data with statistical models (Dick 1995). Moreover, the 1991 Census was the first census to include NPRs in its universe.

More recently, the results of the coverage studies have also been used to assess the quality of demographic estimates (Morissette and Bérard-Chagnon 2014), to produce population projections and to improve collection operations for harder-to-reach groups.

1.1.1. Extent of undercoverage

While the coverage of Canadian censuses is very high, undercoverage is not insignificant. The chart below illustrates how coverage errors have changed since the 1971 Census.Note 3

Chart 1

Data table for Chart 1 
Data table for Chart 1
Table summary
This table displays the results of Data table for Chart 1. The information is grouped by Year (appearing as row headers), Undercoverage, Overcoverage and Net undercoverage, calculated using percent units of measure (appearing as column headers).
Year Undercoverage Overcoverage Net undercoverage
percent
1971 1.9 Note ..: not available for a specific reference period Note ..: not available for a specific reference period
1976 2.0 Note ..: not available for a specific reference period Note ..: not available for a specific reference period
1981 2.0 Note ..: not available for a specific reference period Note ..: not available for a specific reference period
1986 3.2 Note ..: not available for a specific reference period Note ..: not available for a specific reference period
1991 3.4 0.6 2.9
1996 3.2 0.7 2.4
2001 4.0 1.0 3.0
2006 4.3 1.6 2.7
2011 4.1 1.9 2.2

In 2011, undercoverage was estimated at 4.1%, or twice the rate estimated in 1981. However, because overcoverage also rose, net undercoverageNote 4—estimated at 2.2% in 2011—has remained constant at between 2% and 3% over time.

Technical reports on census coverage provide descriptive results for some basic demographic characteristics.Note 5 Undercoverage in censuses is notably higher among young adults (above 10% among men aged 25 to 34 in 2011), men, people who are not in a couple, and people whose mother tongue is not an official language.

1.1.2. Undercoverage of recent immigrants and non-permanent residents

While some basic characteristics related to undercoverage are relatively well known, the reasons for it are generally less well understood. This is especially true for recent immigrants and NPRs. However, a number of signals indicate that these two groups are more likely to be missed in Canadian censuses.

A comparison of the numbers of recent immigrants in the National Household Survey (NHS)Note 6 and from IRCC data reveals significant differences. The NHS counted approximately 1,162,900 immigrants who landed between January 1, 2006 and May 10, 2011, while IRCC data show that more than 1,345,500 permanent resident permits were granted during this period.Note 7 This is a difference of about 15%. While some recent immigrants may have died or left Canada (a non-negligible phenomenon), these two demographic events cannot account for the entire gap observed between the two sources.

Undercoverage of NPRs is believed to be even greater than that of recent immigrants. The NHS enumerated 356,385 NPRs, while the Demographic Estimates Program (DEP) estimated this number on May 1, 2011 at just over 615,700, based on temporary resident permits.Note 8

As shown later in this study, the results of the RRC echo the results for these two demographic groups, in that they have significantly higher rates of omission than the general population.

This situation does not seem to be unique to Canada. A number of international studies have also revealed higher undercoverage of immigrants. Net undercoverage for the 2006 and 2011 Australian censuses was higher for several immigrant groups than for the general population, especially for those born in China and India (Australian Bureau of Statistics 2007, 2012).Note 9

In the United States, immigrants are also more likely to not respond to the American Community Survey (ACS)Note 10 (Jensen et al. 2015). This is truer for undocumented immigrants and recent immigrants (Gonzalez-Barrera 2017). Recent immigrants are also thought to be more likely to be missed in an enumerated household (Fein and West 1988).

Integration into a new country is a gradual, multidimensional process that can take several years. This is why several factors can contribute to the relatively higher undercoverage of recent immigrants and NPRs.

The factors associated with responding to the census and to surveys have some similarities. Thus, the characteristics associated with survey non-response could also be related to undercoverage in censuses. In this regard, some studies indicate that immigrants are less likely to respond to social surveys and panels than the rest of the population (Ahlmark et al. 2014; Swain and Dolson 1998; Bérard-Chagnon 2007). An analysis of data from the New Immigrant Survey (NIS), an American longitudinal survey that followed several cohorts of immigrants, found that among immigrants, those with a lower level of education, who are men, who rent their home, who have fewer children in the household, and who come from the Middle East, North Africa or East Asia are less likely to respond to the survey (Massey et al. 2017). In Canada, the Longitudinal Survey of Immigrants to Canada (LSIC) team used information such as age, sex, country of birth and immigration category in the poststratification of the survey weights. This strategy reveals the relationship between these characteristics and non-response to this survey (Statistics Canada 2003).Note 11

Language barriers faced by many recent immigrants to Canada may also contribute to higher undercoverage of this group. IRCC data show that annually, between 24.8% and 33.1% of immigrants who landed between 2006 and 2011 did not know English or French when landing (Citizenship and Immigration Canada 2012). While many immigrants learn an official language shortly after they land in Canada, their proficiency in the official languages may only be moderate. In this regard, data from the Programme for the International Assessment of Adult Competencies (PIAAC) indicate that many recent immigrants seem to have significant difficulty speaking English or French. More than one-third of immigrants who landed between 2002 and 2012 reported having, at best, fair proficiency in the language in which they took the survey’s proficiency tests.Note 12 As a result, immigrants whose mother tongue is not an official language may have higher levels of undercoverage than the rest of the population.

Since recent immigrants and NPRs are new to the country, they may also be less aware of the Canadian census tradition, the obligation to complete the questionnaire and the importance of the results. In this sense, participating in the census could be seen as a marker of social participation in Canada. Some studies reveal that recent immigrants face certain barriers to social integration. They are thought to be less likely to vote (Uppal and Larochelle-Côté 2012) or to volunteer (Thomas 2012).

In addition to the above factors, recent immigrants are in a transition period, and this could contribute to their likelihood of being missed in the census. It can take immigrants several years to settle, including finding a stable job or a place to reside permanently. In that regard, recent immigrants are much more mobile than the rest of the population, especially in the first few months after landing in Canada (Houle 2007; Dion 2010). Interprovincial migration is generally associated with higher undercoverage levels (Burgess 1988). This “hypermobility” of recent immigrants could therefore increase their propensity of omission from the census because they may maintain residential ties with more than one place at the same time, or they may not be associated with a particular dwelling.

For NPRs, some limitations to the residency rules for including NPRs in the census universe can be added to the aforementioned issues. As the following figure shows, refugee claimants, work or study permit holders and their families must be included in the census (rule 2). However, if they are in Canada temporarily, only individuals whose main residence is not elsewhere must be enumerated (rule 3). Due to the temporary nature of their permit, many NPRs may believe that their main residence is in another country, especially if their temporary residence permit in Canada is short-term or was obtained shortly before census day. As a result, some NPRs might think they are not part of the census universe, and so they do not have to complete the questionnaire. Moreover, some NPRs might not fully understand the residency rules, such as passages that deal with different types of permits, or they might not read the rules at all.

Figure 1

Description for Figure 1

This figure presents the section of the 2011 Census Form 2A on residency rules. Three rules describe the population that must be enumerated in a dwelling. These rules are:

  • All persons who have their main residence at this address on May 10, 2011, including newborn babies, room-mates and persons who are temporarily away;
  • Canadian citizens, permanent residents (landed immigrants) persons asking for refugee status (refugee claimants), persons from another country with a work or study permit and family members living here with them;
  • Persons staying at this address temporarily on May 10, 2011 who have no main residence elsewhere.

Source: Statistics Canada, 2011 Census.

2. Data

This study is based on data from the 2011 RRC and IRCC. This section presents these two data sources and characteristics examined in this analysis.

2.1. Reverse Record Check

Since 1961, the Reverse Record Check (RRC) has measured the number of people missed by Canadian censuses. The 2011 RRC sample was drawn from the following six sampling frames:

These frames provide a sample that is not only independent of the 2011 Census, but that also almost completely replicates the census universe.Note 13

The RRC then classifies the sampled individuals according to whether they were enumerated, missed or out of scope.Note 14 To do this, the sample is first linked with the census response database.Note 15 Then, it locates and interviews sampled individuals who could not be linked in order to collect information to determine their status. Linkages are also made with other sources, such as death records or tax data, to support tracing.

The following figure summarizes the major steps in the development of the 2011 RRC.Note 16

Figure 2

Description for Figure 2

This figure presents a flowchart of the 2011 RRC. The RRC sample is drawn from six survey frames: the 2006 Census, the 2006 RRC, births that occurred between 2006 and 2011, immigrants who landed between 2006 and 2011, NPRs with valid permit on census day and health insurance files (territories only). After, using record linkages and data collection, the selected persons are classified as being enumerated, missed or out of scope.

Source: Statistics Canada, Reverse Record Check, 2011.

The following table shows the sample size of each 2011 RRC frame.


Table 1
Sample size and classification, by sampling frame, 2011
Table summary
This table displays the results of Sample size and classification. The information is grouped by Survey frame (appearing as row headers), Sample size, Classification (unweighted), Enumerated individuals, Missed individuals, Out of scope individuals and Non-respondents, calculated using number units of measure (appearing as column headers).
Survey frame Sample size Classification (unweighted)
Enumerated individuals Missed individuals Out of scope individuals Non-respondents
number
2006 Census 54,772 47,854 3,152 2,343 1,423
Missed individuals from 2006 5,431 3,787 737 452 455
Births 3,619 3,223 226 55 115
Immigrants 2,548 1,900 332 110 206
Non-permanent residents 1,470 670 298 82 420
Health insurance files 76,711 74,889 901 591 330

For the purposes of this study, the immigrant and non-permanent resident frames were used. Therefore, recent immigrants are defined as those who obtained a permanent residence permit between May 16, 2006 and May 10, 2011. These dates correspond to the 2006 and 2011 census days. In addition to the operational benefits of using the immigrant frame exclusively, this definition of recent immigrants is also often used in the literature (Vézina and Houle 2017; Hudon 2015).

The immigrant and NPR frames comprise 2,548 respondents and 1,470 respondents, respectively. In addition, among these two samples, 2,232 immigrants and 968 NPRs were classified as enumerated or missed. These are sufficient numbers for conducting descriptive analyses and certain multivariate analyses. It should be noted that the number of non-respondents in the NPR frame is relatively high. While the weighting was adjusted to account for non-response, it must still be taken into account when interpreting the results.

Furthermore, it is important to note that the immigrant and NPR frames do not completely represent the number of immigrants who landed between 2006 and 2011 or the number of NPRs with a valid permit on census day. For example, some immigrants may have been living in Canada in 2006 as NPRs, and may have already been covered by the 2006 Census and missed frames. At the same time, NPRs may have had a valid permit on 2011 census day and 2006 census day, and thereby also be covered by those two frames. To avoid sampling these people more than once, the coverage studies team identified and removed them from the immigrant and NPR frames.Note 17 This affected just over 120,000 immigrants and about 50,000 NPRs in 2011.

The bootstrap weights developed by the coverage studies team was used to ensure that the variance estimates accounted for the complex survey design.

2.1.1. Components of undercoverage

A key aspect of using the RRC for this study is that the concept of a missed person as measured by the RRC is not identical to the concept of undercoverage. According to the RRC, a person is missed if he or she is in the census target population and has not been enumerated. However, the RRC cannot determine the status of three subpopulations that are included in the census.

The first group consists of the individuals imputed by the Dwelling Classification Study (DCS). This survey estimates the number of non-responding dwellings that were occupied on census day. Following this survey, census data are adjusted for non-responding dwellings and for occupied dwellings incorrectly classified as unoccupied using whole household imputation (WHI). The people added through this statistical operation are missed since they did not complete the census questionnaire, but they are not undercovered because of the correction made by the WHI.

The second group consists of late enumerations, which cannot be included in the RRC for operational reasons.

The last group is made up of enumerations that were considered too incomplete to be used by the RRC to determine whether or not a person was enumerated. The RRC needs basic information from the sampled individuals, such as date of birth, in order to trace them through linkages or field collection.

Undercoverage is calculated by subtracting these three groups from the missed population.

The following table provides the counts for each component of undercoverage for 2011.


Table 2
Components of the estimation of undercoverage error, 2011
Table summary
This table displays the results of Components of the estimation of undercoverage error. The information is grouped by Elements (appearing as row headers), Components of undercoverage and Number (appearing as column headers).
Elements Components of undercoverage Number
(1) Missed individuals 2,828,228
(2) = (3) + (4) + (5) Census population that cannot be identified in the RRC as being enumeratedTable 2 Note 1 1,436,257
(3)   Persons imputed by the Dwelling Classification StudyTable 2 Note 1 780,737
(4)   Late enumerationsTable 2 Note 1 95,757
(5)   Incomplete enumerations according to the RRCTable 2 Note 1 559,763
(6) = (1) – (4) – (5) Collection undercoverage 2,172,708
(7) = (1) – (2) Undercoverage 1,391,971

The total number of missed people is approximately twice the undercoverage. The difference between the two groups stems mainly from DCS imputations and incomplete enumerations.

This study examines the population missed from the census and not undercoverage because it is based on the RRC and because it is impossible to determine the immigrant or NPR status of many individuals in the three groups mentioned above. For example, the fact that some enumerations are incomplete greatly limits the possibility of using record linkages with IRCC data to identify immigrants and NPRs. The results presented here represent to some extent an upper bound relative to the undercoverage of recent immigrants and NPRs.

The missed rate was calculated using a ratio of the number of missed to the sum of this population and the number of enumerated individuals estimated by the RRC. Out-of-scope individuals are excluded from the analysis as they were no longer in the census universe on census day.

2.1.2. Missed rates by survey frame

Examining the missed rates of the different survey frames reveals that recent immigrants and NPRs are more likely to be missed in the 2011 Census than the rest of the population. The following chart illustrates this.

Chart 2

Data table for Chart 2 
Data table for Chart 2
Table summary
This table displays the results of Data table for Chart 2 Percent and 95% confidence interval (appearing as column headers).
Percent 95% confidence interval
Lower Upper
Health insurance files 18.2 15.2 21.2
Non-permanent residents 43.2 39.1 47.3
Immigrants 17.9 16.1 19.7
Births 7.1 6.1 8.1
2006 missed individuals 21.0 19.2 22.8
2006 Census 6.3 6.0 6.6
Total 8.3 8.0 8.6

The overall missed rate in the 2011 Census was just over 8%.Note 18 However, respondents in some frames had much higher missed rates, with the highest rate in the NPR frame, at over 40%. The missed frame and immigrant frame came next, with the missed rate from the two hovering around 20%. One consequence of the higher missed rates from these frames is that 13.6% of persons missed came from the immigrant and NPR frame, though these two frames accounted for only 3.6% of the RRC population. These results are consistent with data from the 2006 RRC (Statistics Canada 2010).

2.2. Immigration, Refugees and Citizenship Canada data

The IRCCNote 19 is the department responsible for issuing permanent and temporary residence permits under the Department of Citizenship and Immigration Act. The IRCC sends files to Statistics Canada every month essentially for the computation of demographic estimates. As previously mentioned, these data are used to create the RRC immigrant and NPR frames. Some of the information in the IRCC data is relevant for this study.

2.3. Characteristics examined

The RRC data contain a limited number of demographic characteristics. As a result, the majority of the characteristics studied here come from IRCC data. The characteristics examined in this study are largely based on those identified in the literature. It should be noted that some characteristics are only available for only one of the two studied groups.

The list below presents the characteristics examined in this study.

Start of text box

Characteristics examined in this study:

Recent immigrants

  • Immigrant category
  • Sex
  • Age on census day
  • Level of education at landing
  • Year of landing
  • Region of birth
  • Province of residence on census dayNote 1
  • NPR status prior to landing
  • De facto marital status on census dayNote 1
  • Mother tongue at landing
  • Knowledge of official languages ​​at landing
  • Interprovincial migration since landing

Non-permanent residents

  • Type of temporary residence permit
  • Length of stay in the country
  • Sex
  • Age on census day
  • Region of birth
  • Province of residence on the most recent permit
  • Holder of several temporary residence permits
  • First permit held
  • Mother tongue on census dayNote 1
  • De facto marital status on census dayNote 1
Note 1

Characteristics from the RRC for missed persons and from the census for enumerated individuals.

Return to note 1 referrer

End of text box

Some characteristics were derived indirectly.

NPR status prior to landing refers to whether the immigrant already held one or more temporary resident permits before moving to the country as an immigrant.

Information on interprovincial migration of immigrants is obtained indirectly by comparing the intended province of residence in the IRCC data with the province of residence in 2011. This approach has two limitations. First, the intended province of residence does not always reflect the actual province where immigrants landed (Bonikowska et al. 2015). Just under 10% of immigrants admitted between 2011 and 2015 who completed a tax return in the year that they landed did not have the same province of residence in both sources (Bérard-Chagnon 2018). If this phenomenon could be the consequence of the high interprovincial mobility of recent immigrants, it could also result from a conceptual difference between the intended province and the actual landing province. This situation could contribute to slightly overestimating interprovincial migration by introducing a “false” migration between the intended province and the landing province. Second, the period during which immigrants can migrate depends on the year of landing. Immigrants who arrived in 2006 had more time to change provinces than those who arrived a few months before the census.

Length of stay is the length of time that the NPR has a temporary permit on census day without a continuous break of 30 days, regardless of the type of permit. This period corresponds to the period used by the DEP to calculate the net NPR estimates.

Just over one-quarter of NPRs in the RRC sample had more than one valid permit on census day. In this situation, if they are different types of permits, the type of permit is determined by following the DEP hierarchy: refugee claimant, work permit, study permit or minister’s permit.

The characteristics available in the IRCC files correspond to the characteristics at the time the permit was issued. Some of them could have changed between then and census day. For this reason, the results for characteristics that can change rapidly over time, such as knowledge of official languages, must be interpreted with some caution.

The data on mother tongue compiled by the RRC combines non-official languages ​​into one category. However, given the very diverse origins of immigrants, this represents a significant limitation for the analysis. For this reason, information from the IRCC data was used for immigrants even if it is also available in the RRC. However, for NPRs, the mother tongue identified in the RRC data and the census was used because this information is not available in the IRCC files currently provided to Statistics Canada.

Finally, some characteristics were coded differently for the immigrant and NPR analysis because of the sometimes very different distributions between the two groups and the number of observations available in the two frames.

3. Recent immigrants missed in the 2011 Census

This section presents the results of the evaluation of the missed rates for recent immigrants by different characteristics. It begins with a presentation of the missed rates for the characteristics examined in the study. Then, the results of regression models built to isolate the effect of each factor are presented.

3.1. Missed rates by characteristic

The table below presents the missed rates of recent immigrants based on their characteristics.


Table 3
Distribution of immigrants in the RRC immigrant frame and missed rates, Canada, 2011
Table summary
This table displays the results of Distribution of immigrants in the RRC immigrant frame and missed rates. The information is grouped by Characteristic (appearing as row headers), Distribution and Missed rates , calculated using percent units of measure (appearing as column headers).
Characteristic Distribution Missed rates
percent
2011 RRC Note ...: not applicable 8.3
Immigrant frame 100 17.8
Province of residenceTable 3 Note 1
Atlantic 1.9 12.6
Quebec 18.6 12.5
Ontario (ref.) 43.2 19.4
Prairies 19.1 18.1
British Columbia 17.2 20.0
Age group on census day
0 to 9 years 11.2 14.6
10 to 19 years 14.2 14.1
20 to 29 years 18.7 21.0
30 to 39 years (ref.) 27.0 19.4
40 to 49 years 16.0 16.2
50 years and over 12.9 18.9
Sex
Men 48.0 19.0
Women (ref.) 52.0 16.7
Region of birth
Europe (ref.) 16.0 11.8
Africa 13.7 17.2
Asia or Oceania 56.4 20.2 Note ***
North America, Central America or South America 13.9 15.8
De facto marital status on census day
Single, widowed, separated or divorced 42.9 18.7
Married or common law (ref.) 57.1 16.9
Mother tongue at landing
English or French (ref.) 13.0 14.8
Arabic 9.8 22.7
Mandarin 9.4 18.0
Tagalog 9.1 15.0
Punjabi 7.2 31.1 Note **
Spanish 6.4 12.2
Other languages 45.4 16.9
Knowledge of official languages at landing
Yes (ref.) 69.0 19.2
No 31.0 14.9
Level of education at landing
High school or less 53.3 16.4
Postsecondary studies 14.3 19.8
Undergraduate degree 22.1 18.5
Postgraduate studies (ref.) 10.4 21.0
Interprovincial migration
No (ref.) 90.2 17.4
Yes 9.8 22.2
Year of landing
2006 (ref.) 12.0 11.4
2007 17.7 15.6
2008 19.2 14.6
2009 20.3 13.8
2010 23.5 23.4 Note **
2011 7.4 35.6 Note ***
Immigrant category
Economic (ref.) 57.3 19.7
Family class 28.3 17.0
Refugee 14.5 12.3 Note **
Non-permanent status prior to landing
Has never been an NPR (ref.) 85.2 18.6
Has already been an NPR 14.8 13.6

Some characteristics of recent immigrants are statistically associated with being omitted.

The region of origin and mother tongue are closely related to being missed in the 2011 Census. Close to 20% of immigrants born in Asia or Oceania were missed in the 2011 Census, a statistically higher proportion than that of immigrants born in Europe (11.8%). This is mostly due to immigrants born in India (28.2%). At the same time, immigrants whose mother tongue is Punjabi, a language spoken in India and Pakistan, were much more likely to be missed (31.1%) than immigrants whose mother tongue was an official language. This is not unique to Canada. People born in Asia were also more likely than the rest of the population to be missed in the 2006 and 2011 Australian censuses (Australian Bureau of Statistics 2007, 2012). Two factors may account for this in Canada. First, mastering Canada’s official languages could be a considerable barrier for these immigrants. PIAAC data suggest that about 40% of immigrants who landed between 2002 and 2012 and whose mother tongue is Punjabi reported having, at best, fair proficiency in the language of the interview. This compares with just under 30% of other immigrants.Note 20 Although census questionnaires are translated into multiple languages, respondents must request one, which may be harder for people whose mother tongue is neither English nor French who have lower proficiency in an official language. In addition to language skills, cultural factors may also account for the higher probability for these immigrant groups of being missed. The social integration in Canada of Sikhs, Buddhists and Hindus—groups generally associated with Punjabi—is thought to be slower than that of other immigrant groups (Reitz et al. 2009). This study examined different social integration markers, such as the sense of belonging to Canada, reporting Canadian identity, and voting in the most recent federal election, using data from the Ethnic Diversity Survey (EDS). One of the highlights of the analysis is that  “Muslims, Sikhs, Buddhists and Hindus are slower to integrate socially, mainly because they are mostly racial minorities.”

The likelihood to be missed in the 2011 Census follows a clear gradient based on the year of landing. Immigrants who settled in 2011, i.e., just a few months before census day, had a missed rate of 35.6%. This proportion is double that for the entire immigrant frame (17.8%) and four times higher than the entire RRC sample (8.3%). Immigrants who settled in Canada in 2010 had a missed rate of 23.4%. At the other end of the spectrum, 11.4% of immigrants who landed in 2006 were missed. This proportion is relatively close to that of the entire population. Immigrants who landed shortly before census are usually less advanced in their settlement process in Canada. The higher missed rates for these immigrants could be a reflection of that situation.

Simultaneously analyzing the year of landing and having been an NPR in the past sheds new light on the link between the year of landing and the propensity for being missed. In general, immigrants who were NPRs before getting their permanent residence permit were slightly less likely to be missed than immigrants who had never had an NPR permit.Note 21 However, when we also consider the year of landing, immigrants who landed in 2008 or 2009 and who had previously been temporary residents had a missed rate of 6.9%, compared with 15.4% for immigrants who had never been an NPR. Although the differences are only statistically significant for two years, they tend to support the assumption mentioned in the previous paragraph.

Interestingly, refugees (12.3%) were less likely than immigrants from the economic categories (19.7%) to be missed in the 2011 Census. Yet, these immigrants tend to have different characteristics from immigrants in the other categories, who are selected based on criteria likely to foster their integration into the country, such as level of education, knowledge of the official languages, ​​or having relatives in Canada. Two reasons could explain the lower likelihood for refugees to be missed. First, refugees tend to be in contact with the Canadian government more often than immigrants of other categories. Refugees usually maintain this contact after getting permanent residence because the IRCC has several programs to help them integrate into Canadian society. Some refugees are also sponsored by individuals or organizations that can also help them settle in Canada. Second, refugees land in Canada as immigrants because of a well-founded fear of returning to their home country, which could foster a different sense of belonging to Canada from that among other immigrant groups. This sense of belonging to Canada could be a factor in reducing their likelihood to be missed.

Some dynamics associated with being missed in the 2011 Census are different between recent immigrants and the general population.Note 22 This is notably the case for age where the missed rates for immigrants were relatively similar from one group to another. For example, they rose from 14.1% among immigrants aged 10 to 19 years to 21.0% among those aged 20 to 29. However, for the general population, the missed rates among adults in their twenties were much higher than in other age groups. They were above 15% for individuals aged 25 to 29, nearly twice that of the entire population. Young adults are in a phase in their lives marked by frequent transitions, such as moving away from their parents and joining the labour market. Such situations could make it more likely for the general population to be missed in the census. However, since recent immigrants are very often in a transition period, regardless of their age, this may have less of an influence on the missed rates of this subpopulation.

3.2. Multivariate analysis

The table below presents the results of logistic regression models that examine the relationship between the different characteristics of recent immigrants and the propensity for being missed in the 2011 Census. The odds ratios of univariate models only include the characteristic studied, while the odds ratios of multivariate models take other factors into account.


Table 4
Odds ratios of the propensity for being missed for the immigrant frame, 2011
Table summary
This table displays the results of Odds ratios of the propensity for being missed for the immigrant frame. The information is grouped by Characteristics (appearing as row headers), Univariate model and Multivariate model, calculated using odds ratios, percentage and number units of measure (appearing as column headers).
Characteristics Univariate model Multivariate model
odds ratios
Province of residenceTable 4 Note 1
Atlantic 0.60 0.55 Note *
Quebec 0.60 Note ** 0.49 Note **
Ontario (ref.) 1.00 1.00
Prairies 0.92 0.96
British Columbia 1.04 1.10
Age group on census day
0 to 9 years 0.71 0.43 Note **
10 to 19 years 0.68 0.43 Note ***
20 to 29 years 1.10 0.93
30 to 39 years (ref.) 1.00 1.00
40 to 49 years 0.80 0.78
50 years and over 0.97 0.98
Sex
Male 1.17 1.16
Female (ref.) 1.00 1.00
Region of birth
Europe (ref.) 1.00 1.00
Africa 1.55 1.33
Asia or Oceania 1.90 Note *** 1.69 Note **
North, Central or South America 1.40 2.04 Note *
De facto marital status on census day
Single, widowed, separated or divorced 1.11 1.93 Note ***
Married or common-law (ref.) 1.00 1.00
Mother tongue at landing
English or French (ref.) 1.00 1.00
Arabic 1.69 Note * 2.32 Note **
Mandarin 1.26 1.33
Tagalog 1.01 0.69
Punjabi 2.58 Note *** 2.37 Note **
Spanish 0.79 0.71
Other languages 1.16 1.30
Knowledge of official languages at landing
Yes (ref.) 1.00 1.00
No 0.74 Note ** 0.67 Note **
Level of education at landing
High school or less 0.74 0.99
Postsecondary studies 0.93 1.29
Undergraduate studies 0.85 0.97
Postgraduate studies (ref.) 1.00 1.00
Interprovincial migration
No (ref.) 1.00 1.00
Yes 1.36 1.33
Year of landing
2006 (ref.) 1.00 1.00
2007 1.44 1.51
2008 1.33 1.33
2009 1.24 1.36
2010 2.37 Note *** 2.74 Note ***
2011 4.31 Note *** 5.38 Note ***
Immigrant category
Economic (ref.) 1.00 1.00
Family class 0.84 0.79
Refugee 0.57 Note ** 0.59 Note **
Non-permanent status prior to landing
Has never been an NPR (ref.) 1.00 1.00
Has already been an NPR 0.69 Note * 0.56 Note **
percentage
R-squared (Cox and Snell) Note ...: not applicable 7.4%
number
Number of observations (unweighted) Note ...: not applicable 2,232

Overall, the results of the multivariate model and the descriptive analysis tend to be similar.

When accounting for the effect of other factors, including knowledge of official languages at landing, recent immigrants whose mother tongue is Punjabi and whose birth country is in Asia or Oceania were still more likely to be missed in the 2011 Census. Moreover, when taking other factors into account, recent immigrants with Arabic as their mother tongue were also more likely to be missed. Just under a quarter of these immigrants were missed in 2011. These results tend to support the assumption that these groups may face particular obstacles that could explain their higher rates of being missed.

Immigrants who landed in 2010 or 2011 were still much more likely to be missed than those who landed in 2006. In addition, when accounting for other factors, immigrants who were not NPRs before landing in the country as immigrants were also more likely to be missed. These results back the assumption that the time spent in the country as a temporary or permanent resident is an important correlate of being missed in the 2011 Census.

Refugees were still less likely to be missed than immigrants from the economic categories. Some potential factors that decrease this likelihood include the context in which refugees were admitted to Canada, their sense of belonging to the country, and the contact they have with the Canadian government and various organizations.

The multivariate analysis also reveals a correlation between knowledge of official languages and being missed among recent immigrants. However, the direction of this correlation is very interesting. When accounting for other factors, immigrants who did not know English or French at landing were less likely to be missed than those who knew English, French or both official languages.Note 23

Different factors may explain this result. First, because of the critical importance of learning an official language to integrate into Canada, recent immigrants are very likely to take language courses organized by the IRCC or by private, public or community organizations.Note 24 According to LSIC data, 45% of immigrants reported having taken language training in English and 10% in French in their first four years in the country (Grondin 2007). Immigrants who took these courses may be more exposed to information about life in Canada, such as the need to complete the census, or be more inclined to collaborate with the federal government.

Second, thanks in part to language training, recent immigrants usually learn one of the two official languages quickly. As a result, knowledge of the official languages at landing, as reported in IRCC data, may not be fully representative of the level of knowledge on census day, particularly for immigrants who claimed to not know English or French at landing. LSIC data showed that six months after arriving, i.e., in cycle 1, 58% of immigrants spoke English well or very well. This proportion rose to nearly 70% starting in the second cycle of the LSIC, i.e., two years after arrival (Grondin 2007). Data from the Refugee Resettlement Project also point to immigrants learning the official languages ​​quickly a few years after landing. According to these data, which are about refugees from Southeast Asia, 16.3% of the refugees in the survey did not speak English when they first completed the survey, i.e., no more than two years after landing. This proportion fell to 8.4% approximately four years after they landed (Hou and Beiser 2006).

The concept of knowledge of official languages and its measurement are different in the IRCC and census data. Not only does IRCC’s information include knowledge of official languages at landing, but immigrants from certain economic categories must also pass language proficiency tests by an IRCC-approved organization to demonstrate their proficiency in the official languages.Note 25 On the other hand, census information consists of a self-reported assessment of knowledge of official languages and is often obtained by proxy.Note 26 However, the perception recent immigrants have of knowledge of official languages on a dichotomous scale like that of census may not fully reflect the complexity of the language learning process. As mentioned earlier, PIAAC data have shown that many recent immigrants report a partial proficiency in the official languages. Some immigrants may think they can conduct a conversation in an official language for the census, but did not get a passing mark in the language tests requested by the IRCC.Note 27

Some methodological factors might also explain this result. For example, recent immigrants who did not know English or French at landing may have different characteristics than other immigrants, characteristics that cannot be measured using the data in this study. Often referred to as missed bias, this could result in certain statistical associations not moving in the expected direction (Schuit et al. 2013).

A similar situation was observed in the NIS (Massey et al. 2017). The American survey data revealed that, when accounting for a number of other factors, the probability of responding to the second wave of the survey was lower among immigrants who said they understood English very well than among immigrants who did not understand it at all. However, the study does not provide an explanation for this discrepancy.

While the results show different dynamics for recent immigrants, the results by age, marital status and province of residence also reflected the mechanisms observed for the entire population.

The results of the regression model report links between the probability of being missed and age, which were not as obvious in the descriptive analysis. Recent immigrants aged 19 or under were less likely than those between 30 and 39 years to be missed. These results tend to reflect the dynamics observed for the general population, where young adults generally had much higher missed rates than those in other age groups.

At the same time, immigrants who were not in a couple on census day were more likely to be missed than those who were in a couple. These results were also consistent with the undercoverage observed in Canadian censuses (Statistics Canada 2015b).

Finally, the regression analysis also illustrates that immigrants who were residing in Quebec were less likely to be missed than those residing in Ontario. Quebec generally has lower missed rates than the national average, which could also been seen among recent immigrants.

4. Non-permanent residents missed in the 2011 Census

This section presents the results of the evaluation of NPR missed rates for different characteristics. As in the previous section, it begins with a presentation of the missed rates for the characteristics examined in the study. Then, the results of regression models built to isolate the effect of each factor are presented. Due to the smaller sample size, the comments in this section not only focus on statistically significant results at the 95% confidence level, but also on those significant at the 90% level.

4.1. Missed rates by characteristic

The following table presents the missed rates of NPRs based on their characteristics.


Table 5
Distribution of NPRs in the RRC non-permanent resident frame and missed rates, Canada, 2011
Table summary
This table displays the results of Distribution of NPRs in the RRC non-permanent resident frame and missed rates. The information is grouped by Characteristics (appearing as row headers), Distribution and Missed rates, calculated using percent units of measure (appearing as column headers).
Characteristics Distribution Missed rates
percent
2011 RRC Note ...: not applicable 8.3
NPR frame 100 43.2
Province of residenceTable 5 Note 1
Atlantic 3.7 48.2
Quebec 17.9 36.4
Ontario (ref.) 40.0 45.0
Prairies 17.7 40.6
British Columbia 20.6 47.0
Age group
Under 20 years 15.9 46.7
20 to 24 years 24.1 49.4 Note *
25 to 29 years 20.1 48.8
30 to 34 years (ref.) 13.0 31.4
35 years and over 27.0 37.1
Sex
Male 49.3 44.6
Female (ref.) 50.7 41.9
Region of birth
Europe 21.8 46.4
Africa 9.6 40.7
Asia or Oceania 48.7 43.9
United States (ref.) 6.3 48.5
Central or South America 13.7 34.8
De facto marital status on census day
Single, widowed, separated or divorced 69.0 47.3 Note **
Married or common-law (ref.) 31.0 32.7
Mother tongue on census dayTable 5 Note 2
English (ref.) 21.7 47.0
French 9.2 37.0
Non-official language 69.1 42.8
First NPR permit
No (ref.) 63.7 41.7
Yes 36.3 46.0
Holds more than one permit
No (ref.) 71.0 45.0
Yes 29.0 39.1
Length of stay in Canada
0 to 6 months 22.4 53.1 Note *
6 to 12 months 20.1 50.4
12 to 24 months 24.4 38.0
24 months or more (ref.) 33.0 36.4
Type of permit
Refugee status claimant 20.5 31.5 Note *
Work permit (ref.) 52.4 46.2
Study or minister’s permitTable 5 Note 3 27.2 46.5

While the overall missed rate of NPRs was above 40% in 2011, rates sometimes fluctuate significantly depending on their characteristics.

As with the general population, age and marital status were correlated with being missed. Nearly half of NPRs aged 20 to 24 were missed in the 2011 Census, compared with just under one-third of NPRs 30 to 34 years. Furthermore, NPRs who were not in a couple had a missed rate of almost 50%, almost 15 percentage points higher than that of NPRs who were in a couple.

The likelihood of being missed followed a very clear gradient based on the length of stay in the country. More than half of the NPRs who were granted temporary residence less than six months before census day were missed. On the other hand, 36.4% of NPRs who were granted temporary residence in Canada two or more years before census day were missed. Due to the residency rules mentioned earlier, NPRs whose stay began shortly before the census might be more likely to consider their usual residence to be outside the country, and therefore not respond to the census. Conversely, NPRs whose residency permit has been valid for at least two years might be more likely to consider their usual place of residence to be in the country. NPRs who received their permit shortly before the census may also be in a transition period following their very recent arrival in the country, which could increase the propensity for being missed.

The type of permit also plays a role in the likelihood of being missed. Refugee status claimants were less likely to be missed (31.5%) than NPRs who held a work permit (46.2%). There could potentially be three reasons for this. First, refugee status claimants have a very different profile from other NPRs. They were older and more likely to have had the right to live in Canada on a temporary basis for a long time, two factors that decreased their likelihood to be missed. Second, refugee status claimants must keep contact with the Canadian government on a regular basis for the processing of their refugee claim. As a result, they may be more likely to collaborate with the government for other purposes, such as completing the census. Third, refugee claimants often come from countries where their situation was difficult, so they may be less likely to consider keeping a usual place of residence in their home country. Finally, the results observed for refugee status claimants and immigrants in the refugee category are consistent with those of the other groups of NPRs and immigrants. This result was expected because the majority of refugee status claimants who get a permanent resident permit obtain it in the refugee category.

4.2. Multivariate analysis

The following table presents the results of logistic regression models that examine the relationships between the characteristics of NPRs and the propensity for being missed in the 2011 Census. The odds ratios of the univariate models only include the characteristic studied, while the odds ratios of the multivariate model take other factors into account.


Table 6
Odds ratios of the propensity for being missed for the NPR frame, 2011
Table summary
This table displays the results of Odds ratios of the propensity for being missed for the NPR frame. The information is grouped by Characteristics (appearing as row headers), Univariate models and Multivariate model, calculated using odds ratios, percent and number units of measure (appearing as column headers).
Characteristics Univariate models Multivariate model
odds ratios
Province of residenceTable 6 Note 1
Atlantic 1.14 0.92
Quebec 0.70 0.80
Ontario (ref.) 1.00 1.00
Prairies 0.84 0.72
British Columbia 1.09 0.91
Age group
Under 20 years 1.91 Note * 1.77
20 to 24 years 2.14 Note ** 1.95 Note *
25 to 29 years 2.08 Note ** 1.88 Note *
30 to 34 years (ref.) 1.00 1.00
35 years and over 1.30 1.35
Sex
Male 1.12 1.06
Female (ref.) 1.00 1.00
Region of birth
Europe 0.92 0.69
Africa 0.73 0.86
Asia or Oceania 0.83 0.77
United States (ref.) 1.00 1.00
Central or South America 0.57 0.63
De facto marital status on census day
Single, widowed, separated or divorced 1.85 Note ** 1.64 Note *
Married or common-law (ref.) 1.00 1.00
Mother tongue on census dayTable 6 Note 2
English (ref.) 1.00 1.00
French 0.66 0.77
Non-official language 0.84 0.98
First NPR permit
No (ref.) 1.00 1.00
Yes 1.19 0.60 Note *
Holds more than one permit
No (ref.) 1.00 1.00
Yes 0.78 0.84
Length of stay in Canada
0 to 6 months 1.98 Note *** 2.24 Note **
6 to 12 months 1.78 Note * 2.09 Note *
12 to 24 months 1.07 1.11
24 months or more (ref.) 1.00 1.00
Type of permit
Refugee status claimant 0.54 Note * 0.53
Work permit (ref.) 1.00 1.00
Study or minister’s permitTable 6 Note 3 1.01 0.77
percent
R-squared (Cox and Snell) Note ...: not applicable 7.0%
number
Number of observations (unweighted) Note ...: not applicable 937

Overall, as evidenced by the odds ratios produced using the regression models, the factors associated with being missed in the 2011 Census are quite similar to those in the descriptive stage when the effect of other factors is taken into account.

Taking into account other factors, respondents aged 20 to 24 years and 25 to 29 years were still significantly more likely to be missed than those aged 30 to 34. This was also the case for NPRs who were not in a couple.

The links between the length of stay in the country and the missed rates observed at the descriptive stage remain statistically significant even when considering other characteristics. For example, NPRs who were granted temporary residence status in Canada less than one year before the 2011 Census were more likely to be missed than those who were granted temporary residence at least two years before census day. This result supports the assumption that NPRs who had the right to temporarily live in Canada for a shorter time would be more likely to consider their usual residence to be abroad, and therefore not respond to the census.

Considering several characteristics also sheds light on new associations between certain characteristics and the propensity for being missed in the 2011 Census.

NPRs whose temporary residence permit was their first permit were less likely to be missed than NPRs who had obtained other permits in the past. This differs from the observations in the descriptive stage, where 46.0% of NPRs with a first permit were missed, compared with 41.7% of those who have had a permit in the past. This difference was not statistically significant. NPRs with a first permit differed from other NPRs because they were more likely to have obtained the right to live in Canada temporarily not long before the census. When accounting for other factors, the association between having a first permit and the propensity for being missed in the 2011 Census is apparent. However, it is difficult to explain this association.Note 28

Finally, when taking the effect of various factors into account, refugee status claimants no longer had a statistically higher propensity for being missed than NPRs who held a work permit. Accounting for specific characteristics of NPRs in this category—such as age and length of stay in Canada—could therefore largely explain the lower missed rates observed for refugee status claimants.

Conclusion

Recent immigrants and NPRs are growing segments of the Canadian population. While censuses strive to provide comprehensive coverage of the population, these groups are less likely to be enumerated. The purpose of this analysis was to examine the factors associated with the propensity for being missed in the 2011 Census for recent immigrants and NPRs using RRC data.

According to the RRC, just under 20% of recent immigrants and more than 40% of NPRs were missed by the 2011 Census, compared with 8.3% of the total population. While missed rates are not a direct reflection of undercoverage but are rather one of the elements of undercoverage, they are still a clear sign that these two populations could have been less covered than the rest of the population in the 2011 Census.

Some characteristics of recent immigrants and NPRs are associated with the propensity for being missed.

First of all, this study highlighted the close links between the year at landing and the propensity of recent immigrants for being missed. More than one-third of immigrants who settled in 2011 and almost a quarter of those who settled in 2010 were missed in the 2011 Census. Immigrants who held a temporary residence permit before being admitted as immigrants were also slightly less likely to be missed, when the effect of other characteristics are accounted for.

About 30% of recent immigrants whose mother tongue was Punjabi were missed in the 2011 Census. The multivariate analysis also highlighted the higher likelihood for immigrants with an Arabic mother tongue to be missed. These results might stem from cultural factors specific to immigrants from certain countries, notably regarding social integration to Canada.

The context in which immigrants are admitted to the country might also affect the likelihood to be missed in the census. While a fifth of immigrants were missed in 2011, 12.3% of refugees were missed. These immigrants fled very difficult situations in their home country and usually maintain contacts with the Canadian government on a regular basis. For these reasons, they may have a better relationship with the government.

Multivariate analysis identified additional correlates of the likelihood for recent immigrants to be missed. Immigrants who were in a couple, who were living in Quebec and who were under the age of 20 were less likely to be missed. These results are similar to the ones observed for the entire Canadian population.

Knowledge of the official languages is a very important marker of integration into a new country. Recent immigrants who reported not speaking English or French at landing seem to be less likely to be missed. This could be because they take language training classes, which might introduce them to the topic of the census, because they learn an official language shortly after landing, and because of differences in concepts and measurement of concepts between census data and IRCC data. It would be very relevant to examine the 2016 RRC data when they become available to see if there is the same finding.

For NPRs, the duration of the permit held by NPRs played a role in being missed in the 2011 Census. For example, more than half of NPRs who received their temporary resident permit no more than six months before the census were missed in 2011. Because they arrived in the country very recently, these NPRs may consider their usual residence to still be in their country of origin, and therefore not consider themselves part of the census universe. Conversely, 36.4% of NPRs who were granted temporary residence two or more years before census day were missed.

Missed rates for NPRs were above 45% for NPRs who were not in a couple. NPRs in their twenties were also more likely to be missed. As with immigrants, these results tend to be similar to the results of the general population.

When accounting for the effect of other factors, NPRs who held their first temporary permit were less likely to be missed than those who already had a permit in the past. This is difficult to interpret and could be studied a second time when the 2016 RRC data become available. It should be noted that the sample from the NPR frame was increased in 2016; as a result, more precise analyses could be conducted for this subpopulation when the data become available.

Refugee status claimants were less likely to be missed than other NPRs. However, the multivariate analysis revealed that much of this difference could come from the specific characteristics of refugee claimants, including their length of stay in the country.

The results of this study have some analytical and methodological implications.

First, they indicate that census data on recent immigrants and NPRs have some coverage gaps. Users should therefore be aware of this when interpreting the data. For example, counts of groups with higher missed rates, such as very recent immigrants, could be underestimated in censuses. This could also cause some biases for data analysis, in particular when these analyses focus on specific population groups.

The results of this study also show that the period of arrival in the country is a major correlate of the propensity for being missed for both recent immigrants and NPRs. This finding reaffirms that social integration into a new country, which could include participation in the census, is a gradual process and that these two demographic groups face several challenges in this regard.

Coverage issues are especially significant for NPRs. Some findings from this study, such as the length of stay in Canada, support the assumption that higher missed rates among NPRs may be partly due to the census residency rules. It may be harder to apply these rules for this subpopulation. This grey area could be explored further to clarify the definition of the census universe for future censuses, for example, by changing the rules of residency or how to present them to respondents.

In addition, it may be worthwhile to conduct a study to examine the relationship between the probability for being missed and the permit expiry date of NPRs. It is possible that NPRs whose permit expires shortly after the census and who do not intend to obtain another permit may feel less concerned by the census, and as a result, may be more likely to be missed.

Census coverage errors are divided into two parts: undercoverage and overcoverage. This study examined missed rates, the main component of undercoverage. However, it cannot provide an adjusted census count for recent immigrants and NPRs. In this regard, Castonguay (2005) proposed an adjustment of language data for net undercoverage and suggested that the effect of coverage errors on Quebec’s linguistic composition could be significant. In addition, the Castonguay study indicated that coverage errors could also have a certain effect on other characteristics measured by censuses, such as level of education. A review of the mechanisms associated with overcoverage and with the other elements of undercoverage for these two groups would also be appropriate to examine both types of coverage errors.

Several statistical agencies, including Statistics Canada, are increasingly examining the potential of administrative data to support or replace certain census-related operations (Statistics Canada 2017b). Recent immigrants and NPRs are probably two of the segments of the population most likely to be missed in Canadian censuses for which relatively complete administrative data are available. The legal aspect of permanent and temporary residence permits ensures that they cover both populations very well. However, IRCC permits also have significant limitations in measuring the population. For example, data on temporary resident permits include information on permits issued, but not on the number of permit holders who may arrive in Canada after obtaining the permit or leave before the permit expires. On the other hand, permanent residence permits provide information on the immigrant’s intended destination, but not on the one where the immigrant actually landed. Moreover, at this time, these data do not contain information on the permit holder’s specific place of residence. Therefore, this raises the question of to what extent these administrative data could support census operations for these two segments of the population.

Finally, censuses are regularly used to obtain information on international migration that is comparable from one country to another (United Nations 2017: §84). However, the fact that recent immigrants and NPRs are missed more often than the rest of the population reaffirms the significance of the challenges associated with measuring international migration. In a context where these dynamics could become more important and more complex, an accurate measure of recent immigration and temporary immigration is increasingly crucial in order to inform public debate, develop appropriate policies and make international comparisons.

Appendices

2011 Census of Population universe

The population universe (target population) of the 2011 Census includes the following groups (Statistics Canada 2015b):

For census purposes, these last three groups of people are referred to as “non-permanent residents.” They have been included since 1991.

2011 Census missed rates


Table A.1
Missed rates for certain demographic characteristics, 2011
Table summary
This table displays the results of Missed rates for certain demographic characteristics. The information is grouped by Characteristics (appearing as row headers), Missed rate and Standard error, calculated using percent units of measure (appearing as column headers).
Characteristics Missed rate Standard error
percent
Provinces 8.3 0.1
Province of residence
Newfoundland and Labrador 7.1 0.5
Prince Edward Island 7.1 0.6
Nova Scotia 7.7 0.5
New Brunswick 6.6 0.4
Quebec 6.5 0.3
Ontario 8.4 0.3
Manitoba 8.2 0.4
Saskatchewan 10.4 0.5
Alberta 9.7 0.4
British Columbia 10.1 0.4
Age group
0 to 4 years 7.4 0.5
5 to 9 years 6.6 0.6
10 to 14 years 6.6 0.6
15 to 19 years 9.0 0.6
20 to 24 years 14.3 0.6
25 to 29 years 15.0 0.7
30 to 34 years 11.8 0.6
35 to 39 years 9.5 0.7
40 to 44 years 8.0 0.6
45 to 49 years 7.6 0.6
50 to 54 years 6.3 0.6
55 to 59 years 5.4 0.6
60 to 64 years 5.0 0.5
65 to 69 years 6.7 1.0
70 to 74 years 4.4 0.7
75 to 79 years 4.0 0.6
80 to 84 years 6.7 1.1
85 years or over 9.2 1.3
Sex
Male 9.4 0.2
Female 7.2 0.2

Knowledge of official languages

The following tables illustrate the differences in knowledge of official languages between RRC and 2011 Census data.

Chart A.1

Data table for Chart A.1 
Data table for Chart A.1
Table summary
This table displays the results of Data table for Chart A.1. The information is grouped by (appearing as row headers), IRCC data (enumerated immigrants), IRCC data (missed immigrants), 2011 Census, Percent and 95% confidence interval (appearing as column headers).
IRCC data (enumerated immigrants) IRCC data (missed immigrants) 2011 Census
Percent 95% confidence interval Percent 95% confidence interval Percent 95% confidence interval
Lower Upper Lower Upper Lower Upper
English 52.1 49.8 54.4 58.1 52.1 64.1 68.6 66.5 70.7
French 6.1 4.8 7.4 3.4 0.6 6.2 7.4 6.1 8.7
English and French 9.7 8.2 11.2 12.6 8.1 17.1 12.9 11.2 14.6
Neither English nor French 32.1 29.9 34.3 25.9 21 30.8 11.2 9.7 12.7

Chart A.2

Data table for Chart A.2 
Data table for Chart A.2
Table summary
This table displays the results of Data table for Chart A.2 English, French, English and French and Neither English nor French, calculated using percent units of measure (appearing as column headers).
English French English and French Neither English nor French
percent
English 90.3 0.4 4.5 4.9
French 2.1 54.6 43.3 0.0
English and French 17.4 13.1 66.5 3.0
Neither English nor French 61.5 8.0 4.4 26.1

Bibliography

Ahlmark, Nanna, Maria Holst Algren, Teresa Holmberg, Marie Louise Norredam, Signe Smith Nielsen and Astrid Benedikte Blom (2014). “Survey nonresponse among ethnic minorities in a national health survey – a mixed-method study of participation, barriers, and potentials”. Ethnicity & Health. Vol. 20, no 6. Pp 611-632.

Australian Bureau of Statistics (2007). Census of Population and Housing – Details of Undercount. June 2007. Catalogue no 2940.0. 32 p.

Australian Bureau of Statistics (2012). Census of Population and Housing – Details of Undercount. June 2012. Catalogue no 2940.0. 64 p.

Australian Bureau of Statistics (2017). Census of Population and Housing – Details of Undercount. June 2017. Catalogue no 2940.0. 64 p.

Bérard-Chagnon, Julien (2007). L’attrition dans les enquêtes sociale longitudinales : le cas de l’Étude longitudinale du développement des enfants du Québec (ÉLDEQ). Master thesis, Département de démographie, Université de Montréal. 129 p.

Bérard-Chagnon, Julien (2018). Evaluation of the 2016 errors of closure for Canada, provinces and territories. Preliminary report. Demography Division. Statistics Canada. 38 p.

Bonikowsa, Aneta, Feng Hou and Garnett Picot (2015). “Changes in the Regional Distribution of New Immigrants to Canada”. Analytical Research Studies Branch Paper Series. No. 366. Catalogue no 11F0019M. 27 p.

Bourbeau, Robert And Norbert Robitaille (1980). “L’effet du sous-dénombrement sur l’estimation des soldes migratoires par groupe quinquennal de génération et par sexe, Québec, 1971 et 1976”. Cahiers québécois de démographie. Vol. 9, no 1. April 1980. Pp 59-86.

Bryan, Thomas (2004). “Basic Sources of Statistics”. The Methods and Materials of Demography. Second Edition. Directed by Jacob S. Siegel and David A. Swanson. Pp. 9-39.

Burgess, R. D. (1988). “Evaluation of Reverse Record Check Estimates of Undercoverage in the Canadian Census of Population”. Survey Methodology. Vol. 14, no 2. Pp. 137-156.

Castonguay, Charles (2005). “Incidence du sous-dénombrement et des changements apportés aux questions de recensement sur l’évolution de la composition linguistique de la population du Québec entre 1991 et 2001”. Suivi de la situation linguistique. Office québécois de la langue française. 30 p.

Citizenship and Immigration Canada (2012). Facts and figures. Immigration Overview. Permanent and Temporary Residents. 2011. Ci1-8/2010E-PDF. 130 p.

Dick, Peter (1995). “Modelling net undercoverage in the 1991 Canadian census”. Survey Methodology. Vol. 21, no 1, June 1995. Pp 51-61.

Dion, Patrice (2010). “Migrations secondaires des nouveaux immigrants au cours de leurs quatre premières années au Canada : motivations et trajectoires”. Cahiers québécois de démographie. Vol. 39, no 2. Pp 243-273.

Dolson, Dave (2010). “Census Coverage Studies in Canada: A History with Emphasis on the 2011 Census”. JSM Proceedings 2010. Pp 441-455.

Fein, David and Kirsten West (1988). “The Sources of Census Undercount: Findings from the 1986 Los Angeles Test Census”. Survey Methodology. Vol. 14, no 2. December 1988. Pp 223-240.

Fellegi, I. P. (1980). “Should the Census Count Be Adjusted for Allocation Purposes-Equity Considerations”. Proceedings of the 1980 Conference on Census Undercount. Pp 193-203.

Gonzalez-Barrera, Ana (2017). Mexican Lawful Immigrants Among the Least Likely to Become U.S. Citizens. Pew Research Center. June 2017. 39 p.

Grondin, Chantal (2007). Knowledge of Official Languages Among New Immigrants: How Important Is It in the Labour Market? Special Surveys Division. Statistics Canada. Catalogue no 89-624-XIE. 92 p.

Hou, Feng and Morton Beiser (2006). “Learning the Language of a New Country: A Ten-year Study of English Acquisition by South-East Asian Refugees in Canada”. International Migration. Vol. 44, no 1. Pp. 135-165.

Houle, René (2007). “Secondary Migration of New Immigrants to Canada”. Our diverse cities. No. 3. Summer 2007. P. 16-24.

Hudon, Tamara (2015). “Immigrant Women”. Women in Canada: A Gender-based Statistical Report. October 21, 2015. Catalogue no 89-503-X. 39 p.

Immigration, Refugees and Citizenship Canada (2016). Report on Plans and Priorities 2016-2017.. Page consulted on June 12, 2017.

Jensen, Eric B., Renuka Bhaskar and Melissa Scopilliti (2015). “Demographic Analysis 2010: Estimates of Coverage of the Foreign-Born Population in the American Community Survey”. Working Paper No. 103, Population Division. U.S. Census Bureau. 34 p.

Kennedy, P. (2002). “Oh No! I Got The Wrong Sign! What Should I Do?”. Discussion Papers. Department of Economics, Simon Fraser University. 02-3. 14 p.

Keyfitz, Nathan (1989). “Faut-il ajuster les données de recensement?” Cahiers québécois de démographie. Vol. 18, no 2. Pp. 247-256.

Malo, Renée (1981). Sous-dénombrement et estimations résiduelles de l’émigration internationale, Canada, Canada moins Québec, Québec et régions, de 1971 à 1976. Master thesis. Département de démographie. Université de Montréal. 165 p.

Maheux, Hélène and René Houle. “150 years of immigration in Canada”. Canadian Megatrends. June 29, 2016. Statistics Canada catalog no 11-630-X.

Martel, Laurent and Carol D’Aoust (2016). “Permanent and temporary immigration to Canada from 2012 to 2014”. Report on the Demographic Situation in Canada. July 5, 2016. Statistics Canada catalog no 91-209-X. 14 p.

Massey, Douglas S., Guillermina Jasso and Minoca ESPINOZA (2017). Weighting for Nonresponse on Round Two of the New Immigrant Survey. 49 p.

Mcquillan, Kevin (2013). “All the Workers We Need: Debunking Canada’s Labour-shortage Fallacy”. SPP Research Papers. Vol. 6, no 16. May 2013. The School of Public Policy. University of Calgary. 30 p.

Morissette, Denis and Julien Bérard-Chagnon (2014). Evaluation of the 2011 Errors of Closure for Canada, Provinces and Territories. Demography Division. Statistics Canada. 65 p.

Office for National Statistics (2012). “The 2011 Census Coverage Assessment and Adjustment Process”. 2011 Census: Methods and Quality Report. July 2012. 18 p.

Picot, Garnett and Feng Hou (2014). “Immigration, Low Income and Income Inequality in Canada: What’s New in the 2000s?” Analytical Studies Branch Research Paper Series. December 15, 2014. Catalog no 364. No 11F0019M. 41 p.

Reitz, Jeffrey G., Rupa Banerjee, Mai Phan and Jordan Thompson (2009). ”Race, Religion, and the Social Integration of New Immigrant Minorities in Canada”. International Migration Review. Vol. 43, no 4. Winter 2009, Pp. 695-726.

Schuit, Ewoud, Rolf H. H. Greenwold, Frank. E. Harrell Jr., Wim L.A.M. De Kort, Anneke Kwee, Ben Willem J. Mol, Richard D. Riley and Karel G.M. Moons (2013). “Unexpected predictor–outcome associations in clinical prediction research: causes and solutions”. Canadian Medical Association Journal. Vol. 185, no 10, July 9, Pp 499-505.

Statistics Canada (2003). Microdata User Guide. Longitudinal Survey of Immigrants to Canada, Wave 1. Special Surveys Division. 102 p.

Statistics Canada (2010). 2006 Census Technical Report: Coverage. Census year 2006. March 2010. Catalog no 92-567-X. 130 p.

Statistics Canada (2013). Skills in Canada: First Results from the Programme for the International Assessment of Adult Competencies (PIAAC). Catalog no 89-555-X. 110 p.

Statistics Canada (2015a). Population Projections for Canada (2013 to 2063), Provinces and Territories (2013 to 2038). May 26, 2015. Catalog no 91-520-X. 58 p.

Statistics Canada (2015b). 2011 Census Technical Report: Coverage. Census year 2011. Catalog no 98-303-X. 152 p.

Statistics Canada (2016a). Annual Demographic Estimates: Canada, Provinces and Territories. 2016. September 2016. Catalog no 91-215-X. 192 p.

Statistics Canada (2016b). Population and Family Estimation Methods at Statistics Canada. March 3, 2016. Catalog no 91-528-X. 96 p.

Statistics Canada (2017a). Immigration and Diversity: Population Projections for Canada and its Regions, 2011 to 2036. January 25, 2017. Catalog no 91-551-X. 139 p.

Statistics Canada (2017b). “Researching a new approach to census-taking”. Updates on the Census Program Transformation Project. August 11, 2017. Catalog no 98-506-X. 5 p.

Swain, Larry and Dave Dolson (1998). “Current Issues in Household Survey Nonresponse at Statistics Canada”. Nonresponse in survey research: proceedings of the Eighth International Workshop on Household Survey Nonresponse. September 1997. Pp 1-21.

Thomas, Derrick (2010). “Foreign nationals working temporarily in Canada”. Canadian Social Trends. June 8, 2010. Winter 2010, no. 90. Catalog no 11-008. 17 p.

Thomas, Derrick (2012). “Giving and volunteering among Canada’s immigrants”. Canadian Social Trends. May 17, 2012. Summer 2012, no. 93. Catalog no 11-008. 15 p.

Unece (2015). Conference of European Statisticians Recommendations for the 2020 Censuses of Population and Housing. United Nations. 251 p.

United Nations (2017). Handbook on Measuring International Migration through Population Censuses. Department of Economic and Social Affairs. Statistics Division. March 1, 2017. 154 p.

Uppal, Sharanjit and Sébastien Larochelle-Côté (2012). “Factors associated with voting”. Perspectives on Labour and Income. February 24, 2012. Catalog no 75-001-X. 15 p.

Uppal, Sharanjit and Sébastien Larochelle-Côté (2014). “Overqualification among recent university graduates in Canada”. Insights on Canadian Society. April 2014. Catalog no 75-006-X. 13 p.

U.S. Census Bureau (2014). The Undercount of Young Children. February 2014. 29 p.

Vézina, Mireille and René Houle (2017). “Settlement patterns and social integration of the population with an immigrant background in the Montréal, Toronto and Vancouver metropolitan areas”. Ethnicity, Language and Immigration Thematic Series. May 8, 2017. Catalog no 89-657-X2016002. 62 p.

Notes


Date modified: