The Canadian Student Tobacco, Alcohol and Drugs Survey (CSTADS; renamed from the Youth Smoking Survey (YSS) in 2014-15) is funded by Health Canada, and since 2004-05 has been implemented biennially by the Propel Centre for Population Health Impact at the University of Waterloo, in partnership with researchers across Canada. CSTADS/YSS provides timely and reliable data on tobacco, alcohol and drug use and related issues among Canadian students (grades 6-9/10-12) attending generalizable samples of private, public, and Catholic schools. CSTADS/YSS uses a repeated cross-sectional survey design to collect data through classroom-based surveys. Eight cycles of the CSTADS/YSS have been conducted: 1994, 2002, 2004-05, 2006-07, 2008-09, 2010-11, 2012-13, and 2014-15. Sampling methodology varied by survey cycle, as outlined below.
In 1994 and 2002, schools were stratified by census metropolitan area (CMA) versus non-CMA. Grades within schools were then sampled by two levels of stratification in 1994 (province and grade) and three levels of stratification in 2002 (province, school and grade). Students in grades 5-9 were included in the target sample. In 1994 and 2002, one class was randomly selected in the desired grade in the selected schools and all students within each selected class were invited to participate in the survey. In 1994, all grade 5-9 students completed the same questionnaire. In 2002, grade 5 and 6 students completed a version of the questionnaire that excluded the drug and alcohol questions.
In 2004-05, the sampling of schools was conducted in two stages. At stage 1, school boards were sampled within each province using a stratified sampling design. The school boards were rank ordered based on their adult smoking rates, and each board was assigned to one of two strata (low or high smoking rate), so that approximately half of the total student enrolment in any province was assigned to each stratum. At stage 2, schools were sampled from the selected school boards, according to two strata: senior (senior elementary or high school grades) and junior (junior elementary grades). From 2004-05 onward, all eligible students within participating schools/classrooms were invited to participate in the survey. Similar to 2002, grade 5 and 6 students completed a version of the questionnaire that excluded the drug and alcohol questions.
Beginning in 2006-07, schools became the sampling units, and grade 10-12 students were also included in the target population. In 2006-07, a stratified single-stage cluster design with strata based on health region smoking rate and type of school was employed. In each province, two health region smoking rate strata (low and high) and two school-level strata (elementary and secondary) were defined. Similar to 2002 and 2004-05, grade 5 and 6 students completed a version of the questionnaire that excluded the drug and alcohol questions. However, grade 7-12 students were randomly selected to receive one of two versions (modules) of the questionnaire that included the alcohol and drug questions. For this reason, the 2006-07 YSS data has two survey weights (rather than one, as in other years), the use of which depends on which questionnaire module(s) included the variables being analysed (see 2006-2007 YSS Main Microdata User Guide).
For the 2008-09 to 2014-15 cycles, a third health region smoking rate stratum (urban) was introduced, and grade 5 students were removed from the target population. Similar to 2006-07, a stratified single-stage cluster design with strata based on health region smoking rate and type of school was employed. In each province, two or three health region smoking rate strata (low, high, urban) and two school-level strata (elementary and secondary) were defined. The urban stratum was introduced in Ontario in 2008-09, Alberta in 2010-11, Quebec in 2012-13 and Nova Scotia in 2014-15. With the removal of the grade 5 population, only the grade 6 students responded to a version of the questionnaire that excluded the drug and alcohol questions.
Once the sampling frame was defined, schools were randomly selected within each stratum to ensure a generalizable sample of schools for each province and for Canada. The following details some variations in the sampling methodology since 2006-07.
- In 2010-11, the province of New Brunswick did not participate in the survey. Based on comparative analysis conducted using 2008-09 survey data, there were no statistically significant differences in national estimates for current or ever smoking with and without New Brunswick. A census sample was conducted in the province of Quebec in partnership with another project (see 2010/2011 YSS Microdata User Guide).
- In 2012-13, the province of Manitoba did not participate in the survey. Based on comparative analysis conducted using 2010-11 survey data, there were no statistically significant differences in national estimates for current or ever smoking with and without Manitoba.
- In 2014-15, a provincially generalizable sample was not obtained in the province of New Brunswick; therefore provincial estimates are not available, but data from the participating New Brunswick students was included in the estimates for Canada.
In all years, the sample excluded residents of the Yukon, Nunavut and Northwest Territories, residents of institutions, and those attending schools on First Nations reserves, special needs schools (e.g., schools for visually- or hearing-impaired individuals) or schools located on military bases; schools that did not have at least 20 students enrolled in at least one eligible grade were also excluded.