Highlights

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This page has been archived on the Web.

Sampling errors
Nonresponse
Coverage errors
Response errors
Processing errors related to imputation

Sampling errors

• The coefficients of variation (CVs) of the average estimates of total expenditure per household vary between 1.3% and 2.7% for the provinces. The CV at the national level is 0.7%.

• The coefficients of variation (CVs) of the average estimates for the different summary level expenditure categories are in most cases less than or equal to 1.9% at the national level and are generally lower than 5.0% at the provincial level. The results for the different categories of dwelling and household equipment characteristics are similar. Since the sample size is smaller in Prince Edward Island, the CVs tend to be higher than those of the other provinces.

Nonresponse

• The final response rate is 69.2%. Provincial response rates range from 58.7% in Ontario to 78.1% in Saskatchewan.

• The nonresponse rate is 30.8%. Nonresponse is due to refusals (18.4%), to households that could not be contacted (8.6%), and to households with data that were considered unusable (3.8%). 

• The nonresponse rate tends to increase with the urbanization level. The nonresponse rate is 24.3% in rural areas and 33.7% in urban centres with a population of one million or more.

• According to an analysis of nonresponse rates in the strata consisting of the high-income geographic areas created under the sample design, the rate in high-income strata (38.9%) is higher than the rate observed in other strata (29.9%). The refusal rate for high-income strata (25.3%) is higher than the rate observed for other strata (17.6%).

Coverage errors

• Undercoverage of households is 8.3% at the national level. Undercoverage of households is observed for all provinces, with rates varying from 4.1% to 12.1%. British Columbia has the highest undercoverage of households.

• Undercoverage of persons is 9.3% at the national level. Undercoverage of persons is observed for all provinces and territories, with rates varying from 4.1% to 14.2%. British Columbia has the highest undercoverage of persons.

• The national slippage rate for children (aged 0 to 17) is very different from the rates for other age groups. For children, the undercoverage rate is 0.1%, whereas for adults, undercoverage is much higher at 11.9%.

Response errors

• Response errors include recall errors, telescopic error and errors due to proxy response. Because the Survey of Household Spending (SHS) interview is lengthy, the response burden can lead to respondent fatigue and have an impact on data quality. Total interview time varies according to household characteristics, and it can be substantial for some households. The average interview time was one hour and forty-five minutes.

Processing errors related to imputation

a) Expenditure variables

• It was necessary to impute at least one expenditure variable¹ for 31.1% of households nationally. This higher rate in 2004 is due to the change made to the questionnaire regarding expenditures related to communication services in the home (telephone, cell phone and Internet access), cable television services and satellite distribution services. For these services, it is now accepted that a respondent household can provide only total expenditure for a package (bundled services), indicating which services are included. As a result, 21.5% of usable households required imputation for at least one of these five services.

• The overall imputation rate excluding these five services is 13.2% at the national level, which is comparable to the rates obtained in previous years. For these expenditures, most households required imputation for one or two of the 240 expenditure variables. Provincially, the imputation rates range from a low of 6.2% for Quebec to a high of 18.8% for Nova Scotia.

• Imputed values account for 20.4% of the estimate of total expenditure on cable television services and 32.8% of the estimate of total expenditure on Internet access services. Imputation rates for households reporting expenditures on one or the other of these two services are respectively 20.5% and 40.3%. These higher rates are mainly due to the fact that among households that reported paying for a package, a large proportion of packages included these two services.

• About 20% of individuals required imputation for clothing variables, but in the vast majority of such cases, the respondents provided the totals and only the components were imputed.

• About 4% of individuals aged 15 and over required imputation on at least one variable in the Personal Taxes, Security and Money Gifts section.

b) Income variables

• Imputation for at least one income variable was required for 5% of individuals aged 15 and over. For 80% of them, total income was provided by the respondent but all components (income sources) were imputed.

c) Categorical variables

• For 7.6% of households, it was necessary to impute at least one categorical variable. Approximately 72% of them had only one variable imputed. Provincially, imputation rates range from a low of 2.2% for New Brunswick to a high of 12.9% for Alberta. Categorical variables that required imputation can be found in the following sections of the questionnaire: Dwelling Characteristics; Facilities Associated with the Dwelling; Tenure; and Tobacco and Miscellaneous, for variables pertaining to purchases through direct sales.

---

Note

1. Excluding the Clothing section and Personal Taxes, Security and Money Gifts section, which are collected at the individual level and not the household level.