View the most recent version.

Families Reference Guide, 2006 Census

Catalogue no. 97-553-GWE2006003

• Definitions and concepts
• Classification
• Data
• Questions
• Data quality

Definitions and concepts

The Families topic includes concepts such as marital status, family structure and whether related or unrelated individuals are living together as a household. Characteristics and concepts related to families appear in the population universe, the family universe, and the household universe.

Within the family universe, two definitions of families exist, one being a subset of the other. The census family is the narrower concept, defined by couples living together, with or without children, and lone parents living with their children. The economic family is broader, and refers to two or more people living together who are related to each other by blood, marriage, common-law union or adoption.

When data on counts of persons are presented according to their relationship to others in the family, such as spouse/common-law partner, lone parent, or child, the concepts of census family status or economic family status apply. Tabulations on counts of families make use of census family structure and of economic family structure.

Note that parent-child relationships, key to identifying the family status and family structure, are not restricted to sons and daughters below a certain age unless an age limit is stated in the table. When the family status or structure includes an age limit for children, such as under 25 years of age, the sons and daughters aged 25 years or older become persons not in census families and other relatives in economic families. This in turn affects the number of families (or persons) and their family structure.

It is also important to note the specific criteria regarding 'children' in census families as described in More information on Census family. Because of these criteria, small differences in the analysis of family characteristics may result depending on whether or not census family concepts are used to identify parents and children.

For example, in the analytical document of September 12, 2007 , the analysis of young adults aged 20 to 29 years living in the parental home considered all direct parent-child relationships and excluded any grandparent grandchild relationships. In contrast, when applying the census family concept as in the 'census family status' variable in the standard tables, grandchildren in skip-generation families, as well as sons and daughters, are included, provided they are not married, living common-law or parents themselves. According to the concepts used for the analytical document published on September 12, 2007, there were 43.5% of 20 to 29 year-olds living in the parental home. With the census family concept, there were 41.8% of 20 to 29 year-olds considered as children in census families.

Other family-related concepts include household living arrangements (population universe) and household type (household universe).

Classification

The relationship between households, economic families and census families is illustrated in Figure 14 Economic and census family membership and family status.

Data

Family characteristics, excluding marital status, are shown for the 20% sample population, for all private households. As a consequence, characteristics in tabulations which would normally include the population in private households plus non-institutional collective dwellings are restricted to private households when crossed with family characteristics (20% sample data). When marital status data are crossed with demographic variables only, such as age and sex, they are available for the total population (100% data), including persons in institutional and non-institutional collective dwellings. All children under age 15 years are treated as single (never married); therefore, data on marital status are sometimes shown only for the adult population (15 years and older).

Census 2006 website

Questions

Family data for private households are obtained from the combination of Questions 2 (sex), 3 (date of birth), 4 (marital status), 5 (common-law status) and 6 (relationship to Person 1).

The responses to all these questions are processed together to prevent inconsistencies. In addition, the demographic characteristics of all persons living in the same household are processed together, to prevent inconsistent relationships between household members.

The questions appear on page 4 of the 2A (short) questionnaire and on page 4 of the 2B (long) questionnaire. 2006 Census questionnaires

For more information on all the questions of the 2006 Census, please refer to 2006 Census questions and reasons why the questions are asked.

Data quality

Data quality verification in place for the 2006 Census

Imputation for non-response and inconsistent responses

During automated data processing, the quality of the data on age, sex, marital status and family relationships is maximized by checking the logical consistency of the different characteristics. Then, any inconsistent or missing responses are replaced with acceptable values. This is done by identifying households in the same geographical area that have similar, but complete and consistent characteristics and then copying their values to fill in the missing or erroneous data among the 'failed edit' households.

Very few changes were made to the edit and imputation methods for the 2006 Census regarding demographic and family characteristics. Changes were made to include same-sex married couples. As well, persons reported as foster children are now counted as part of the same economic family as their foster parents. They are not counted in the same census family. More information on edit and imputation is available in Age, Sex, Marital Status and Common-law Status: 2001 Census Technical Report. The Appendix B of the report lists the edit rules that apply to demographic and family characteristics.

In the 2006 Census, the imputation rates of the variables 'age' and 'sex' were 1.0% and 1.4% respectively, without taking into account total non-response households. The age confirmation question in the Internet version of the questionnaire has a positive impact on data quality.

Imputation was applied to 1.5% of all persons for Question 6, Relationship to Person 1. One-third of these imputed values were due to inconsistent reported relationships compared to the age, sex, or marital status of the person or to the relationships reported for other members in the household. This does not reflect all inconsistencies however; the age, sex, legal marital status or common-law status was at times imputed instead of the relationship to Person 1.

When added to total non-response by households in 2006 of just under 3%, the rates of imputation for age, sex and relationship to Person 1 were approximately 4.0% to 4.4%. Imputation rates for legal marital status and common-law status tend to be higher, but these include invalid or missing responses for people under age 15, for whom values of 'never married' and 'not in a common-law relationship' are applied in standard fashion. Also, non-response to the common-law question tends to be high because it overlaps conceptually with the legal marital status question, which precedes it.

Comparison with other data sources

As with every census, the quality of the 2006 Census information released in July 2007 (age and sex) and September 2007 (marital status, families and households) was evaluated internally prior to publication. The data were compared, to the extent possible, with alternative data sources. In the case of data by age, sex and legal marital status, the main source for comparison was the post-censal estimates produced by Demography Division, which are based on administrative records of births, deaths, migration and marriage. For data on families and households, the main sources of comparison were the Survey of Household Spending, the General Social Survey (family cycle), and the Survey of Labour and Income Dynamics. Only aggregate data were used; individual records were not linked.

Same-sex couples

The measurement of same-sex common-law couples was introduced in the 2001 Census, when the category 'Same-sex common-law partner of Person 1' was added to Question 6. The 2006 Census used this question to collect, for the first time, information on same-sex married couples. Same-sex married spouses could report their relationship either by writing in their relationship or by marking 'Husband or wife of Person 1'. Post-collection edits were applied to correct the misreporting of sex by opposite-sex spouses. Given their large numbers in comparison to same-sex married couples, a small percentage error in the reporting of sex by opposite-sex spouses can result in a large overcount of same-sex married couples. Probabilistic techniques using first names were used to correct for the misreporting of sex.

The data quality studies conducted by Statistics Canada revealed no problems with the final data on same-sex common-law couples or on same-sex married couples. More information can be found on the Statistics Canada Census web module by following the 'Reference Material' link under the 2006 Census header and selecting the '2006 Census information on same-sex common-law and married couples'.

Marital status of persons aged 15 to 19

When comparing the census results to other Statistics Canada data sources, it appears that there is some overestimation of persons aged 15, 16 and 17 who are counted as married, common-law, separated, divorced or widowed, rather than never married (single). Although it affects a relatively small population, it is best to apply caution when analysing the census data on the marital status of 15 to 17 year-olds. There may also be some impact on tabulations by 5-year age groups, namely the group of persons aged 15 to 19 years. Efforts are being made to reduce this problem in future censuses.

For more information on factors that may explain such variances in census data, such as response errors and processing errors, please refer to the 2006 Census Dictionary, Appendix B Data quality, sampling and weighting, confidentiality and random rounding.

Household size

Families and households are closely related concepts. Prior to the 2006 Census, the census questionnaire had traditionally included response categories for up to six people living in the same household. For the 2006 Census, the shorter questionnaire (2A), delivered to 80% of households, continued to include response categories for six persons. However, the long questionnaire (2B) delivered to 20% of households, included response categories for only five persons. Respondents reporting for more than six household members (2A) or for more than five household members (2B) had to obtain a second paper questionnaire to record answers for the additional persons.

A small impact was observed resulting from the change from six to five persons on the long paper questionnaire. The number of six-person households obtained by the long questionnaire (after weighting) was slightly lower, by about 16,000, than for both the short and long questionnaires combined. Conversely, the number of five-person households obtained by the long questionnaire alone was slightly higher, also by about 16,000 households. This impact is judged to be minimal in comparison to the nearly one million households that contained five or six persons.

Changes in the definition of 'census family' between 1996 and 2001

Significant changes were made to the definition of a census family in the 2001 Census, particularly as regards children in census families. Refer to More information on Census family for the details.

These changes resulted in 1.4% more census families in 2001, including 9.6% more lone-parent families, than would have been the case if the definitions had remained constant. Therefore, they could affect the analysis of trends covering the 1996 to 2001 period. Historical comparisons for families, particularly for lone-parent families, must be interpreted with caution as a result of these conceptual changes.