Labour Force Survey Response Rates, September 2023

Some stakeholders, including the media, have asked questions on the potential impact on the quality of the LFS estimates resulting from declined response rates. This brief document provides some background and key steps undertaken by Statistics Canada to ensure the LFS remains a high-quality survey and fit for use by policy makers and Canadians.

The COVID-19 pandemic and associated measures to counter its impact on citizens resulted in response rates declining for household surveys – including the LFS in Canada and in many other countries.

Canadians understand the value of participation and continue to support our efforts in ensuring high quality data. We understand the challenges faced by Canadians and are encouraged by steadily increased participation.

Statistics Canada has implemented a number of operational and methodological enhancements to counter the impact of reduced response rates and to ensure that the LFS remains a high-quality measure of the labour market situation in Canada.

Canadians can continue to rely on the high quality and timely labour market information system, and we encourage the support of all our stakeholders and the media to convey the importance of participating in the LFS.

In our commitment to remain transparent and responsive to the questions of our stakeholders and Canadians, we encourage those with questions or concerns to contact us.

LFS Response Rates

In 2019, prior to the COVID-19 pandemic, approximately 48,500 interviews LFS were completed each month. One in five (19.5%) of these was completed in-person or face-to-face, i.e. with a Statistics Canada interviewer visiting the respondent’s home. Just over half of interviews (51.8%) were completed from one of Statistics Canada’s five call centers and the remainder (28.7%) were completed by the respondent online.

The COVID-19 pandemic resulted in two significant changes to LFS data collection procedures. First, from March 2020 to October 2022, all in-person visits to respondents’ homes were suspended. Second, plans to virtualize call centres were accelerated and interviewers who had previously worked from a call centre transitioned to working from home.

As was the case in many other countries, changes in Canada’s LFS data collection operations largely caused by the pandemic contributed directly to a decline in response rates. Whereas the average monthly response rate for LFS was 87.0% in 2019, it fell to 69.8% in the second half of 2020 and averaged 69.5% in 2021.

In November 2022, following the lifting of most public health restrictions, face-to-face interviewing was reintroduced. Following this change, the response rate reached a high of 74.1% in February 2023. As of September 2023, the LFS response rate was 70.4%

As a result of investments in LFS collection through the Disaggregated Data Action Plan, the number of interviews completed in September 2023 (48,822) was slightly higher than the 2019 level (48,500).

Additional information on final response rates and the number of interviews completed can be found in Appendix A.

LFS Data Quality

Based on rigorous ongoing evaluation of two aspects of data quality – precision and sample composition – Statistics Canada is fully confident that LFS estimates continue to be of very high quality. In combination with other data sources – particularly the Survey of Employment Payroll and Hours (SEPH) and the Job Vacancy and Wage Survey (JVWS) – the LFS provides an accurate portrait of labour market conditions.

Precision

Precision refers to the extent to which the design of a given survey, including its sample size, results in statistical variability; that is, the extent to which a theoretical set of estimates produced by the same survey would differ from each other.

The LFS sample size is designed to ensure the precision of key estimates, particularly unemployment rates. Precision is measured using coefficients of variation (CVs) and the sample size is designed to ensure that CVs are below defined targets for provinces, census metropolitan areas (CMAs), economic regions (ERs) and Employment Insurance economic regions (EIERs). 

Each month, Statistics Canada provides information on whether these CV targets are met. In September 2023, for example, the CV of the estimate of unemployment exceeded the target in four provinces: Newfoundland and Labrador (8.0%; target 7.0%), Prince Edward Island (12.3%, CV target 9.7%), Nova Scotia (8.0%; target 7.0%) and New Brunswick (8.1%; target 7.0%).

Sample composition and bias

Bias refers to the risk that survey results are misleading or inaccurate as a result of the part of the population that is missing due to nonresponse being systematically different from the part that responds. To evaluate the risk that declines in LFS response rates have introduced bias, Statistics Canada evaluates the composition of responses received according to a range of socio-demographic characteristics, including age, education and immigration; as well as characteristics which may be correlated with labour force status, including previous month’s labour force status and industry of employment, where applicable.

In addition to evaluations of the composition of the collected sample, steps are taken in the edit and imputation, weighting and calibration processes to mitigate the impact of any changes in sample composition to the greatest extent possible. For example, survey results are weighted to ensure that they match the known distribution of the population by age and sex.

Further information

For further information on LFS, please contact:

Media:
Media Relations (statcan.mediahotline-ligneinfomedias.statcan@statcan.gc.ca)

Other stakeholders:
Labour / Travail (statcan.labour-travail.statcan@statcan.gc.ca)

Appendix A