The Youth in Transition Survey (YITS) is a longitudinal survey undertaken jointly by Statistics Canada and Human Resources and Skills Development Canada. This survey is designed to examine the major transitions in the lives of youth, particularly between education, training and work.
Data release – April 5, 2004
The Youth in Transition Survey (YITS) is designed to examine the patterns of, and influences on, major transitions in young people's lives, particularly with respect to education, training and work. Human Resources and Skills Development Canada and Statistics Canada have been developing the YITS in consultation with provincial and territorial ministries and departments of labour and education. Content includes measurement of major transitions in young people's lives including virtually all formal educational experiences and most labour market experiences, achievement, aspirations and expectations, and employment experiences. The implementation plan encompasses a longitudinal survey of each of two cohorts, ages 15 and 18-20, to be surveyed every two years.
The results from the Youth in Transition Survey will have many uses. Human Resources and Skills Development Canada will use them to aid policy and program development. Other users of the results include educators, social and policy analysts, and advocacy groups. The information will show how young adults are making their critical transitions into their adult years.
These researchers and analysts will have access to important information that can be used in developing programs to deal with both short-term and long-term problems or barriers that young adults may face in their pursuit of higher education or in gaining work experience. Information from the survey will help to evaluate the effectiveness of existing programs and practices, to determine the most appropriate age at which to introduce programs, and to better target programs to those most in need.
Young adults themselves will be able to see the impact of decisions relating to education or work experiences. They will be able to see how their own experiences compare to those of other young adults.
The survey population for the 18 to 20 year-old cohort includes persons born in the years 1979 to 1981. Geographically, the target population excludes the northern territories, Indian reserves, Canadian Forces bases and some remote areas.
The survey population for the Reading Cohort (15 year-olds) comprises persons who were born in 1984 and were attending any form of schooling in the ten provinces of Canada. Schools on Indian reserves were excluded, as were various types of schools for which it would be infeasible to administer the survey, such as home schooling and special needs schools. These exclusions represent less than 4% of 15-year-olds in Canada.
As comparability with the 1999 survey results was an important objective of YITS 2002, only minimal modifications were made to the wording of the questions that were asked of 18-20 year-olds in 1999. Integration of the sample of 15 year-olds (Reading Cohort) from 1999 required verifications and edits to ensure that this cohort answered questions already asked of the 18-20 year-olds in Cycle 1. Check questions, flags and edits were used to verify historical information brought forward from Cycle 1 for the 18-20 year-old cohort.
In Cycle 2, the survey application was reprogrammed from the software «CASES »to «BLAISE» software application. Although the layout and style of the questionnaire was different to meet BLAISE standards the questionnaire structure (modules, flows, edits, coding) was similar to the original version.
New questions were focus group tested and a pilot survey was held in one of the regions to determine interviewers' ability to use the application and understand the questionnaire concepts. As well, respondent acceptance of the survey was monitored.
This is a sample survey with a longitudinal design.
18-20 year-old Cohort
Factors such as the high mobility rate of the 18-20 year-old cohort and its relatively low incidence at the household level led to a stratified multi-stage sample design based on the use of the Labour Force Survey sample, drawing from currently active and rotate-out households. Within each household, one person in the target population was pre-selected for YITS. The initial sample size was 29,000 persons.
Reading Cohort (15 year-olds)
The sample design for the Reading Cohort (15 year-olds) entails two-stage probability sampling, with a stratified sample of 1,200 schools selected at the first stage and a sample of eligible students selected within each sampled school. The initial student sample size for the reading cohort which was conducted in 2000 was 38,000.
Among the Reading Cohort (15 year-olds) and the 18 to 20 year-old cohort, only those who responded in Cycle 1 were recontacted in Cycle 2. The resulting sample size was 29,660 for the Reading Cohort (15 year-olds) and 22,352 for the 18 to 20 year-old cohort.
Data collection for this reference period: 2002-02-14 – 2002-06-15
Responding to this survey is voluntary.
Data are collected directly from survey respondents.
Collection for Cycle 2 took place from mid-February to mid-June 2002 using computer-assisted telephone interview (CATI). The response rate for the 18-20 year-old cohort was 85%. The response rate for the Reading Cohort (15 year-olds) was 90.5%. The combined response rate for both cohorts in Cycle 2 was 88.1%.
A series of verifications took place to ensure that the records were consistent and that collection and capture of the data did not introduce errors. Reported data were examined for completeness and consistency using automated edits coupled with manual review. Some responses reporting uncommon values or characteristics were processed manually.
For quantitative variables such as income and earnings, imputation was performed to treat missing responses and/or outliers. The first step in the imputation process was a within-record imputation where missing information was replaced with values derived from the respondent's answer to other questions in the questionnaire using deterministic edit rules. The remaining missing data were imputed using nearest-neighbour donor imputation. Impudon was used for donor imputation.
Impudon is a generalized system for donor imputation built as a SAS macro using the version number 8 of SAS. It manages both numeric (positive, zero, negative) and categorical (or character) variables; uses matching fields that can be numeric and/or categorical; can use weights for relativizing the matching fields in the distance function between a recipient and a donor. There are many criteria of distance that can be used, and the user can specify more than one in an Impudon execution. The imputation groups are automatically generated from a variable list. The post-imputation edit rules can be of any type (linear, non linear, conditional, applying boolean operators, on numeric and/or categorical variables).
The estimation of population characteristics from a survey is based on the premise that each sampled unit represents, in addition to itself, a certain number of non-sampled units in the population. A basic survey weight is attached to each record to indicate the number of units in the population that are represented by that unit in the sample. This basic weight is derived from the sample design.
18-20 year-old Cohort:
Because the sample for the YITS 18-20 year-old cohort is derived from households that were in the Labour Force Survey (LFS) sample, part of the LFS weighting procedure is applicable to the YITS sample. More precisely, the LFS sub-weight is used as the initial weight for the YITS sample units. This weight represents the inverse of the probability of dwelling selection for the LFS, adjusted for LFS non-response.
To reflect the YITS sample design and take into account the non-response for this survey, several adjustments must be applied to the LFS sub-weight to derive the YITS final weight. In total, the following five adjustments are required to obtain the final Cycle 1 weights and compensate for:
1. including 36 groups from the LFS
2. a priori non-response at the household level
3. selecting one person per household
4. non-response of the YITS selected persons
5. differences between post-censal demographic estimates and the weighted counts derived from the first four adjustments.
Reading Cohort (15 year-olds):
An initial weight was derived based on the two-stage sample design used for this survey. Components to this survey are PISA/YITS student and YITS parent. Weights were calculated on these components and on Math and Science. A number of non-response adjustments were applied in order to obtain final weights. More than one adjustment was required because non-response can occur at various levels (e.g., schools, students, parents, non-consent of parents for students) for these respondents. BRR (balanced repeated replication) and Bootstrap replicate weights were derived to allow users to estimate CVs and standard errors for estimates.
For both cohorts the final Cycle 1 weight was adjusted to take into account non- response in Cycle 2.
All 15 year-olds from Cycle 1 Reading Cohort were recontacted in Cycle 2. For further quality evaluation, refer to Chapter 11 of the Cycle 1, 15 year-old User Guide (under record number 5058)
In Cycle 1, the sample was chosen from LFS rotation groups, both active and inactive. The older the rotation group, the harder it was to find the respondents which would have created a lower response rate. For further quality evaluation, refer to Chapter 5 of the Cycle 1, 18-20 year-old User Guide (under record number 4435).
In Cycle 2 tracing of both groups of respondents was conducted first at Head Office then during interviewing in the regional offices. Many forms of tracing were utilized to lower the rate. Unsuccessful tracing accounted for attrition of 2,649 cases nationally (5.01%). Due to the reporting structure, it is not possible to identify the number of cases for each cohort. The attrition is due to the high mobility rate of the older age group.
Calibrated weights for province, age and gender were updated against Census data. Imputation rates were used for income.
Statistics Canada is prohibited by law from releasing any information it collects which 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.
Data quality is affected by both sampling and non-sampling errors. Non-sampling errors were minimized through testing (focus group, pilot survey and main survey); training of regional office staff; observation by head office personnel; tabulations of initial data; and adjustment of questionnaire specifications for future cycles. Quality assurance measures were implemented at each step of the data collection and processing cycle to monitor data quality. For sampling error, data reliability guidelines were established based on coefficient of variations (CV). It is recommended that any estimate based on fewer than 30 observations or with a CV greater than 33.3% not be released.