Coverage

View the most recent version.

Archived Content

Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.

Archived

This page has been archived on the Web.

The coverage of a survey or administrative data source expresses the degree to which the program captures the concept intended to be measured. Many accuracy, cost and collection burden objectives have to be balanced when designing a survey. Administrative data programs face the difficulty that the data was not originally gathered to produce these statistics. How each survey responds to these different choices may impact comparisons across different sources.

Perhaps the most obvious place to start comparisons is by identifying differences in the population covered by these sources. This information is described in general terms below; additional detail is provided in Appendix A.

Survey of Labour and Income Dynamics (SLID)	The SLID is an important source for income data for Canadian families, households and individuals. Income data from SLID are collected from a sample of the population aged 16 years or over. The 2005 data were derived from approximately 30,000 households. Several groups are excluded from the sample, and from the estimates: residents of the territories, Canadians living outside Canada, persons living on Indian Reserves, residents of institutions and the military living in barracks. In total, they are estimated to represent less than 3% of the population. Weights are calibrated to match population estimates from Statistics Canada's Demography Division for different age/sex groups and  earnings groups in Canada Revenue Agency (CRA) administrative data (T4 file), to ensure income distributions match those of the population. Income data is retrieved from the T1 file for consenting respondents and from computer assisted telephone interviews for the others.
Annual Estimates for Census Families and Individuals (TIFF)	The intention of T1FF is to provide information on the income of all persons residing in Canada, or spouses living outside the country who filed tax returns. The information source is personal tax and benefit returns (T1), and other administrative sources such as Canada Child Tax Benefit files and the T4 file. Therefore some individuals, who did not file a tax return, or filed it quite late, may be missing, if information for them is not derived from the other administrative sources. T1FF data is estimated to cover about 95% of the population.
Census of Population (CP)	The Census provides information on all Canadian citizens, non-permanent residents living in Canada and those temporarily living abroad. Income data are reported by a one in five random sample of dwellings in Canada and total population estimates are derived using weights. (The initial weight of about five is adjusted to ensure closer agreement between sample estimates and the enumerated population counts for different variables, such as age and sex.) The income data for individuals 15 years of age or older is retrieved from the T1 file for consenting respondents and from a paper questionnaire and/or via Internet for the others.
System of National Accounts (SNA)	The SNA considers persons and households as part of the Persons and Unincorporated Business Sector (PUBS), and treats that sector as one unit. It includes, in addition to persons and unincorporated businesses, non-profit institutions serving households.  They are non-profit institutions, such as churches, labour unions and charitable organizations, plus credit unions, trusteed pension plans, life insurance companies, fraternal societies and mutual non-life insurance companies. This means, for example, that the investment income of pension funds that is not directly received by individuals, but held in trust for them, is included in the income for the SNA's Income and Expenditure Accounts (IEA) for the PUBS; that investment income  would not be reported as personal income. The income data is constructed partially from the T1 file but also from the information slips payers file with the CRA.

The impact of each of these specific decisions to include or exclude difficult-to-reach populations or selecting different age, is relatively marginal at the Canada-level. However, these conceptual coverage discrepancies are often discussed and may be crucial if one is interested in a particular population, region or income component.  Where it is possible to derive a Canada-wide estimate, the impact of coverage decisions on aggregates of employment income and total income are presented in Table 2.

Table 2 Estimate of Absolute and Relative Impact on Earnings and Total Income of Different Coverage Restrictions, 2005

There remain further coverage differences across the programs. Those included in this second group are of a less conceptual nature but rather more akin to unintentional coverage error or measurement error. Certain segments of the population may be more difficult to reach with some of the collection methods (for example, because they do not file a tax return or do not complete their Census form, etc.). Figure 1 illustrates the empirical differences in population coverage according to data source for the programs that are micro-record based by age characteristics.  Three points draw attention in figure 1. First, coverage of the 20 to 40 year olds on T1FF and Census is lower than for other segments of the population. SLID manages to partially compensate the difficulties of reaching this population during the weighting stage by calibrating the counts in specific age groups. Second, there is a spike of over-coverage on T1FF for young persons 15 to 19. This results from counting twice some persons that filed a tax return (possibly with income) and were listed as dependents of their parent(s) in another dwelling. Third, the population over 70 is less well covered due to the exclusions of persons in institutions or collectives.

Figure 1 Coverage profiles by age

Coverage profiles by age such as shown in Figure 1 that compare counts with official population estimates are relatively easy to create. Alternate profiles (for recent immigrants or Aboriginals for example) would be more difficult to generate because of data availability and sample size limitations though they probably would show even greater discrepancies across the sources.