Survey of Postsecondary Faculty and Researchers (SPFR)

Detailed information for 2019

Status:

Active

Frequency:

One Time

Record number:

5299

The objective of the survey is to fill data gaps on equity, diversity, and inclusion (gender, visible minority status, Indigenous identity, self-reported disability, sexual orientation, use of official language) among those who teach or conduct research in Canada's postsecondary sector and providing an overview of career experiences and barriers to career advancement.

Data release - September 22, 2020

Description

The objective of the survey is to fill data gaps on equity, diversity, and inclusion (gender, visible minority status, Indigenous identity, self-reported disability, sexual orientation, use of official language) among those who teach or conduct research in Canada's postsecondary. It covers faculty and researchers at publicly-funded Canadian colleges and universities, including full- and part-time university faculty, college instructors, postdoctoral fellows and doctoral students across Canada. It examine various topics such as equity and inclusion, employment security, job duties and other employment, learning and development opportunities, access to research funding and harassment and discrimination.

The Survey of Postsecondary Faculty and Researchers is a survey sponsored by Innovation, Science and Economic Development Canada (ISED), in collaboration with the Canadian Association of University Teachers (CAUT), Colleges and Institutes Canada (CICan), Polytechnics Canada, Universities Canada, the Canadian Institutes of Health Research (CIHR), the Natural Sciences and Engineering Research Council (NSERC), and the Social Sciences and Humanities Research Council (SSHRC).

Statistics Canada recognizes that some of the terms used to examine the above topics may be considered out of date by scholars and other experts who study the socioeconomic characteristics of the Canadian population. These terms include variables that refer to sex, gender, racialized groups, and Indigenous identity. The variables used in the Survey of Postsecondary Faculty and Researchers are consistent with other surveys delivered by Statistics Canada. Revised terms are currently under review for the 2021 Census.

Data from the survey will contribute to the development of federal policies and programs to support faculty and researchers working and studying at colleges and universities across Canada.

Subjects

  • Education, training and learning
  • Equity and inclusion
  • Human resources in science and technology
  • Labour
  • Research and development
  • Science and technology
  • Society and community
  • Students
  • Teachers and educators

Data sources and methodology

Target population

The target population consists of faculty and researchers at publicly funded Canadian colleges and universities at the time of the survey, including full-time and part-time university faculty, college instructors, postdoctoral fellows, and doctoral students. Public Canadian institution staff who do not teach or research, such as administrative staff, janitorial staff, housing staff are excluded from the survey.

The observed population, also called survey population, is slightly different than the target population and consists of faculty and researchers at publicly funded Canadian colleges and universities, including full-time and part-time university faculty, college instructors, postdoctoral fellows, and doctoral students, who were listed on the various components of the sampling frame. Since there exists no up-to-date list of the members of the population, the sampling frame was built using the three following sources: 1) the 2016 Census of Population, 2) the 2017 T4 slips that are issued by an in-scope institution and 3) the Post-secondary Information System (PSIS) data.

Instrument design

Not available

Sampling

This is a sample survey with a cross-sectional design.

Sampling unit

The SPFR has a targeted response sample, where the person is the sampling unit.

Stratification method

The SPFR strata were formed in multiple steps by cross-tabulating some of the variables on the frame. First, the data sources (i.e., 2016 Census of Population, the 2017 T4 and PSIS data) were used to create seven strata. Those strata were divided into smaller substrata using the following variables: province, minority indicator (source: administrative data linked to the frame), researcher indicator (source: administrative data linked to the frame) and T4 Slip in 2017.

Sampling and sub-sampling

A few allocation methods including the optimum allocation and the square root of the population size were used between the substrata. For calculating the sample size, different types of expected sample size loss, such as expected non-response and expected out-of-scope rate were taken into consideration. A simple random sampling method is use within each stratum. Total sample size was 100,406 people.

Data sources

Data collection for this reference period: 2019-10-07 to 2019-12-06

Responding to this survey is voluntary.

Data are collected directly from survey respondents.

The SPFR used the Electronic Questionnaire (EQ) to collect data. The survey was available in the two official languages, English and French. An introductory letter was mailed out to the entire sample, containing a Secure Access Code, inviting respondents to participate online. It was expected that the questionnaire would take approximately 10 to 15 minutes to complete. In the event respondents didn't participate in a timely manner, three reminder letters were sent. If an email address was available on file, an e-invitation and a number of e-reminders were also sent. If a cell phone number was available on file, some text reminders may also have been sent.

Proxy response was not allowed. In the event that respondents have moved, or cannot be reached at the contact information available on file, various tracing methods were used to locate them. If the respondent was associated with a particular institution, invitations or reminders could have been sent to the address of the institution. No tracing was done if respondents were living outside of Canada at the time of the interview.

View the Questionnaire(s) and reporting guide(s) .

Error detection

Data editing

The electronic questionnaire (EQ) included automated logical flows to control the questions that were asked to respondents and a limited number of soft edits that checked for unusual values or inconsistencies. The collected data was processed using the Social Survey Processing Environment (SSPE) that was developed at Statistics Canada. This included: i) coding of open-ended text entries to the appropriate standard classification system; ii) a series of detailed consistency edits to check for logical inconsistencies or extreme values; and iii) creating an extensive set of derived variables.

Imputation

This methodology type does not apply to this statistical program.

Estimation

In order for estimates produced from survey data to be representative of the target population, and not just of the sample itself, users must incorporate the survey weights into their calculations. A survey weight is given to each person included in the final sample, that is, the sample of persons who responded to the survey questions. This weight corresponds to the number of persons within the survey population represented by the respondent. If the frame used was perfect (covering exactly the population of interest) and all selected units were traced, contacted and completed the survey, then the design weight assigned to each unit, given by the inverse of the probability of selection of each unit in the sample, would represent accurately and exactly the number of people in the target population. In this situation, using this weight would yield unbiased estimates. However, this is not the case when surveys are faced with non-response and imperfect frames. The process of computing survey weights for each survey respondent involves several steps:

1) Each sampled unit is given an initial weight equal to the inverse of the probability of selection multiplied by the final 2016 long-form Census weight (when applicable).

2) The weights are adjusted to account for the unresolved units (i.e., persons with unknown scope status) by using propensity scores.

3) The weights for the out-of-scope units were set to 0.

4) The weights of respondents are then adjusted to represent the persons which did not respond to the survey.

It should be noted that SPFR weights are not calibrated for two reasons: 1) there are no reason to believe that the not covered population act the same way as the covered population; 2) There are no good quality known totals that are in line with the population of interest.

The SPFR uses the bootstrap method, a replicate-based method, for calculating variance. The bootstrap method involves taking subsamples with replacement from the sample and weighting them. 500 sets of bootstrap weights were generated using Rao-Wu method.

Estimates for SPFR were calculated using the Generalized Tabulation System (GTAB). GTAB is a common generalized tabulation tool, developed at Statistics Canada, which standardizes the process of tabulating data and applying confidentiality to survey and administrative data sources for the purpose of data dissemination.

Quality evaluation

Quality assurance measures were implemented at every collection, processing and development survey step. Measures included procedures to ensure that coding errors were minimized, and edit quality checks to verify the processing logic. Data were verified to ensure internal consistency and were also compared to other sources when available. Consultations were held with subject-matter experts within Statistics Canada in relation to the content of the survey.

In order to validate the data output by the Generalized Tabulation System (GTAB) for the CODR tables, each table was recreated externally in SAS using BOOTVAR, this allowed for the calculation of the table statistics, the coefficient of variance (CV), and the 95% confidence interval (CI). The CVs were also used to calculate the quality indicators (QI) for average and median statistics. These outputs were then compared with the CODR table, downloaded in full from the working copy on the website. Some initial discrepancies were found in the statistics estimates across tables. These were easily attributed to a difference in rounding rules and the SAS/BOOTVAR programs were updated to align with GTAB rounding rules. A complete and detail data quality validation process have been used to assess this survey data and have demonstrated that the data is deemed acceptable for an official release.

Disclosure control

Statistics Canada is prohibited by law from releasing any information it collects that could identify any person, business, or organization, unless consent has been given by the respondent or as permitted by the Statistics Act. Various confidentiality rules are applied to all data that are released or published to prevent the publication or disclosure of any information deemed confidential. If necessary, data are suppressed to prevent direct or residual disclosure of identifiable data.

The survey outputs disseminated through aggregate-level estimate tables incorporate suitable rules to suppress estimates that do not meet minimum thresholds. Rounding and cell suppression are used to protect confidentiality of individuals.

In the absence of a survey frame that would list all post-secondary institutions faculty and researchers, several files were linked--including tax data (T1 and T4), Census, Postsecondary Information System (PSIS), immigration, and research funding data sets. Survey weights were adjusted to account for non-respondents and were not further calibrated due to the complexity of the survey frame design and the unavailability of external control totals that were aligned with the concepts and the coverage of the survey population. The survey population was derived from the list of faculty and researchers found in the databases used to create the survey frame. Therefore, the survey results are only representative of the surveyed population, not necessarily the targeted population. Due to the methodology used, survey results cannot be released for individual post-secondary institutions.

Revisions and seasonal adjustment

This methodology type does not apply to this statistical program.

Data accuracy

While considerable effort is made to ensure high standards throughout all stages of collection and processing, the resulting estimates are inevitably subject to a certain degree of error. These errors can be broken down into two major types: sampling and non-sampling. Frame imperfection and non-response are important sources of non-sampling error.

The sample is designed to yield estimates at the provincial level for full-time and part-time university faculty, college instructors, postdoctoral fellows, and doctoral students, as well as several minority groups of interest.

Response rates:
The targeted response rate for the 2019 SPFR is 30%.

The overall response rate was 42.4%.

Newfoundland and Labrador 32.3%
Prince Edward Island 32.7%
Nova Scotia 45.2%
New Brunswick 47.2%
Quebec 45.8%
Ontario 40.4%
Manitoba 43.0%
Saskatchewan 44.6%
Alberta 41.4%
British Columbia 46.0%
Yukon Territory 35.6%
Northwest Territories 35.7%
Nunavut 14.6%

Non-sampling error:
There are many sources of non-sampling errors that are not related to sampling but may occur at almost any phase of a survey operation. Respondents may misunderstand survey instructions, or make a mistake in answering the questions, responses may be recorded in the questionnaire incorrectly, or errors may be made in the processing or tabulating of the data. For the 2019 SPFR, quality assurance measures are implemented at each phase of the data collection process to monitor the quality of the data.

Non-response, if not appropriately corrected, is a type of error that can lead to bias in the survey estimates. Biased estimates can occur when unusable units have characteristics that are significantly different from the usable ones.

Non-response bias:
To limit the effect of non-response, survey weights were adjusted to account for unresolved units and non-respondents.

Coverage error:
In the case of SPRF, the main challenge was to build a quality sampling frame. Since there was no list of members of the target population available, the SPFR was built by combining three sources of information to get a better coverage of the population of interest. Since all these sources were dated from at least 2017, the survey does not reach any people that became in-scope after 2017. No weighting adjustments were done to compensate for under-coverage, as no information is available on the characteristics of the population that is not covered.
The SPFR estimates may be subject to over-coverage error. For example, administrative staff in an in-scope institution could be surveyed if they received a T4 in 2017, although they would not be in-scope. To adjust for the over-coverage, the out-of-scope cases are dropped during the weighting strategy after adjusting for the unresolved cases.
The final sampling frame is prone to both undercoverage and overcoverage (the out-of-scope rate was 36.5%). The imperfection of the sampling frame may lead to some bias in the estimates.

Other non-sampling errors:
Not applicable

Date modified: