Section 4
Response errors

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This page has been archived on the Web.

Response errors represent a lack of accuracy in responses to questions. They can be attributed to different factors, including a questionnaire that requires improvements, misinterpretation of questions by interviewers or respondents, and errors in respondents' statements.

In the Survey of Household Spending (SHS), there can be various reasons for errors in respondents' statements. First, there are recall errors that occur when a respondent forgets expenditures  made during the period covered by the survey (which corresponds to the calendar year), or when a respondent provides an erroneous value because of the time interval that has elapsed between the time of purchase and the date of the interview. Recall errors are probably the survey's largest source of response error, since the reference period is long (12 months) and a great variety of information is requested.

One of the main measures taken to minimize recall error in the SHS is to calculate the difference between receipts (income and other amounts received by the household) and disbursements (expenditures plus net change in assets and liabilities) for each household.  When the difference exceeds 15% of receipts or disbursements, with the higher amount being retained, respondents are contacted again in order to obtain additional information and to try to identify errors or omissions. Respondents are also encouraged to consult various documents (bills, bank statements, etc.) in order to provide more accurate information. To determine expenditures for small items purchased at regular intervals, interviewers generally suggest that respondents estimate the frequency of the purchases and the price generally paid in order to derive expenditures for a 12-month period.

A second source of error in respondents' reporting is telescopic error, which consists of including in the reference period events that occurred before or after it. In the SHS, the use of the calendar year is considered to provide a good marker for the start of the reference period. Furthermore, since the reference period is a long one, telescopic error has less impact.

Responses by proxy can also contribute to response error. The household member who made an expenditure is generally best able to report it accurately. This is definitely the case with, say, personal purchases.  Expenditures reported by an intermediary are more likely to be tainted by response error, and this type of error tends to have a greater effect on certain types of expenditures.

Among other sources of response error, the extent of the respondent's co-operation should not be overlooked. For personal reasons, the respondent may decide not to mention particular expenditures or decide to twist the facts.

In the SHS, another factor is the response burden, owing to the length of the interview and the great variety of items to be reported, as well as the pace of the interview. This can lead to respondent fatigue and affect the quality of the responses obtained. The interview time varies greatly from one household to another, depending on household size, income and various other characteristics.

While response errors are a major source of error in a historical interview, they are the aspect of data quality that is the hardest to measure. Generally, it is necessary to conduct quite costly special studies in an attempt to measure them. Efforts are made to combat response errors by using survey techniques designed to reduce them.