Methodology of the Canadian Labour Force Survey
Chapter 1 Introduction and Overview
1.0 Introduction
This publication is a reference guide to the methodology of the Labour Force Survey (LFS). This guide will primarily focus on the methodology used for the ten provinces of Canada, though the LFS also covers the three territories. It describes all current survey steps and highlights the changes made during the most recent sample redesign.
A separate document called the Guide to the Labour Force Survey (Catalogue No. 71-543-G, available online) is a complement to this report and describes the concepts, definitions and data produced in the LFS.
1.1 Background
The LFS was created after the Second World War to meet an urgent need for reliable and timely data on the labour market that reflected the transition from a war-time economy to a peace-time economy. The survey was designed to produce estimates on employment and unemployment at the regional and national levels.
Conducted quarterly when it began in 1945, the LFS became a monthly survey in 1952. In 1960, the Interdepartmental Committee on Unemployment Statistics recommended that the LFS become the official tool for measuring unemployment in Canada. Once this recommendation was adopted, the demand for data increased, since users wanted a broader range of labour market statistics and, in particular, more detailed regional data. The range of estimates produced by this survey has grown considerably over the years, and today it provides a detailed portrait of the Canadian labour market.
1.2 LFS concepts and products
The LFS is the official source of monthly estimates of total employment and unemployment. The main monthly indicators published include the unemployment rate, the employment rate and the participation rate. The LFS is also one of the main sources of information on socio-demographic characteristics of the working-age population such as age, marital status, level of education and family status.
Employment estimates are produced at various levels such as by sex, age group, industry, occupation, educational attainment, and immigrant status. Statistics are also produced on characteristics such as length of job tenure, usual and actual hours worked, and employee wages. The questions asked by the survey make it possible to examine a wide variety of topical employment issues such as involuntary part-time employment, multiple job-holding, and absence from work.
Unemployment estimates are produced by industry, occupation , duration of unemployment, type of work sought, and activity before looking for work. Supplementary measures of unemployment are also produced annually to shed further light on the degree of labour market slack and the extent of hardship associated with joblessness. Information is also available on the recent labour market activity of persons currently not in the labour market. The Guide to the Labour Force Survey provides a complete description of the LFS questionnaire content.
In addition to providing national and provincial estimates, the LFS produces data for sub-provincial regions, such as Economic Regions (ERs), Employment Insurance Economic Regions (EIERs) and Census Metropolitan Areas (CMAs). The federal and provincial governments use LFS data to distribute funding and other resources to the different political and administrative jurisdictions.
The LFS standard estimates are published every month in Labour Force Information (Catalogue No. 71-001-X, available online). A wide variety of labour market data are also available through CANSIM, Statistics Canada’s key socioeconomic database and electronic extraction system. There are more than 100 CANSIM tables representing several thousand chronological series that are updated either monthly or annually with new LFS data.
The LFS can produce much more information than what is regularly published. Specific tabulations can be produced on a cost-recovery basis. For more information about available survey products and services, please see Section 9 of the Guide to the Labour Force Survey.
1.3 General survey overview and document structure
In the provinces, the LFS is a monthly household survey providing a sample of individuals who are representative of the civilian, non-institutionalized population, 15 years of age or older. Excluded from the survey’s coverage are: persons living on reserves and other Aboriginal settlements, residents of institutions, full-time members of the Canadian Forces and residents of regions that are extremely remote or of extremely low population density. These groups together represent an exclusion of approximately 2% of the population aged 15 and over.
These groups are excluded from the survey target population due to specific operational challenges or for conceptual reasons. For example, it would be difficult to interview members of the Canadian Forces who live in locations that are inaccessible to LFS interviewers (e.g., aboard warships or in military camps and barracks). Residents of institutions (for example, inmates of penal institutions, patients in hospitals or nursing home residents) are excluded because the LFS is designed to measure the labour force participation in the current labour market and residents of institutions are for the most part not able to participate in the current labour market and are not economically active.
The survey uses a two-stage sample design. In the first stage, a sample of primary sampling units (PSUs) corresponding to geographical regions is selected. In each selected PSU, a sample of dwellings is drawn at the second stage. Households are identified within the selected dwellings and all individuals in the household who are part of the target population are selected for the survey. The dwellings selected remain in the sample for a period of six months. Outgoing dwellings are replaced by dwellings from the same PSU, or from a similar PSU if the previous PSU is retired and replaced. This sample design results in a five-sixths month-to-month sample overlap, which makes the design efficient for estimating month-to-month changes. The rotation of dwellings after six months prevents undue respondent burden for households that are selected for the survey. The high proportion of PSUs in common between samples twelve months apart makes the design efficient for estimating year-to-year changes. Chapters 2 and 3 provide more information on the sample design.
Data collection for the LFS is carried out in the week following the LFS reference week. Usually, the reference week contains the 15th day of the month. In 2015, about 88% of sampled households responded to the LFS questionnaire each month. The LFS interview is mandatory and takes an average of eight minutes. The data are collected using a computer-assisted interviewing system. Several collection methods are used, including in-person and telephone interviews, and an Internet questionnaire. More information on the collection strategy is presented in Chapter 4.
In the days following collection, the data are processed. Editing, imputation and weighting are performed, and quality indicators are derived. These steps are described in Chapters 5, 6, 7 and 8. Despite the large amount of data to process every month, Statistics Canada publishes the LFS estimates only ten days after the end of collection.
The LFS sample and frame are also used for many of Statistics Canada’s other social surveys. This is described in Chapter 9. Several appendices covering special topics and survey reference materials are included at the end of this guide.
1.4 Changes introduced in 2015
Every ten years, after the decennial census of population, the LFS undergoes a sample redesign to account for the evolution of the population and labour market characteristics, to adjust to the current and expected needs of data users (in terms of statistical analyses), and to update the geographical information used to carry out the survey.
The most recent sample design was gradually introduced in January 2015 and was fully implemented by June 2015.
The 2015 redesign introduced a number of major changes to the methodology of the survey. These changes were introduced to reduce survey costs, use updated collection methods, and allow data users to compute and report design-based variance estimates on their own.
In this survey redesign, the primary sampling units were constructed from the Dissemination Areas defined for the 2011 Census. In addition to streamlining the work involved with the sample redesign, this change makes the LFS geography more standard, which helps in the comparison of estimates across surveys and in analysis involving multi-level modeling. The sample allocation strategy was modified to use quality targets that prevent the allocation algorithms from automatically increasing sample sizes in areas of low unemployment. The changes made to the PSUs, allocation, and stratification are detailed in Chapter 2.
An innovation that was introduced with the 2005 design, the use of existing lists of addresses, has been expanded significantly in the 2015 design. Statistics Canada’s residential address register (AR) has been incorporated into a new household survey frame service. The Dwelling Universe File (DUF) is an extraction of addresses from the AR which is now being used to produce the list of addresses for over 90% of the PSUs in the LFS sample. This reduces the work of field interviewers who would otherwise have to create the list of addresses by directly observing the neighbourhoods / PSUs in the LFS sample. The frame service also supplies telephone numbers that will help interviewers establish contact with sampled households. More information about the address register is given in Chapters 2 and 3.
The LFS has added a third collection method in 2015: eligible respondents can now complete the questionnaire using the Internet. This new strategy is discussed in Chapter 4.
The overall imputation strategy did not change, but the list of variables used to create the imputation groups for donor imputation was reviewed and updated to include industry. This change and other changes made to edit and imputation are discussed in Chapter 5.
Last but not least, a significant change was made in 2015 to variance estimation. Starting in January 2015, the bootstrap method has replaced jackknife as the variance estimation technique for the LFS. This allows users to compute and report design-based variance estimates for state-of-the-art analyses on their own. This important change is presented in Chapter 7.
1.5 Changes coming beyond 2015
A few additional changes are already on the agenda for the coming years.
In January 2016, the classification systems used to categorize industry and occupation on LFS data will be updated to more recent standards. Specifically, the currently used North American Industry Classification System 2007 (NAICS 2007) will be updated to the NAICS 2012 standard, and the currently used National Occupational Classification–Statistics 2006 (NOC-S 2006) will be updated to the NOC 2011 standard.
In the coming years, the systems used for collection, processing, estimation and tabulation of LFS data will be migrated to new corporate business processes to achieve cost savings while maintaining the highest standards of quality and timeliness.
- Date modified: