2015 National Youth Survey
Appendix A: Methodology
The 2015 National Youth Survey (NYS) used a mixed-mode methodology, which allowed Elections Canada to reach its target populations. This section outlines the overall approach for this research as well as the resulting sample. Nielsen proposed to complete:
- A total of 3,000 20-minute surveys among Canadians in all provinces and territories in the official language of the respondents' choice. The final sample had a total of 3,009 respondents, who completed a 22-minute survey. Respondents were divided into three groups: the core sample, the oversample and the older adult group.
- The core group was to comprise a representative sample of 1,250 young Canadians aged 18 to 34, segmented by region. Respondents were to be divided equally into three age categories as follows:
- 18-22-year-olds (34% of respondents)
- 23-29-year-olds (33% of respondents)
- 30-34-year-olds (33% of respondents)
- The oversample was an overrepresentation of five groups of interest among young Canadians (aged 18–34). The additional 1,250 respondents would include the five subgroups below.
- Youth residing in rural areas
- Youth with disabilities
- Unemployed youth (not in school)
- While planning fieldwork and after discussions with the project manager and the sample specialist, it was decided to increase the core sample to allow for the natural fallout of the oversample populations. The remaining oversample would specifically target Aboriginal youth, unemployed youth and youth with self-identified disabilities. This strategy would produce a more representative sample, while still achieving the minimum in each target population.
- The final sample had a total core youth sample of 1,752 respondents and an oversample of 754. The core sample followed a proportional distribution of age, gender and region according to the 2011 census. Respondents in the oversample were screened only for the target characteristic (Aboriginal, unemployed or self-identified disability).
- The control group was composed of 503 Canadians aged 35 and over.
- We targeted 40% of the respondents to be of non-voters, defined as those who did not vote in the 2015 federal election. However, we faced challenges in achieving this goal. The challenges of talking to non-voters are well documented and particularly affect phone interviews. Our youth sample had 33% of non-voters, while our oversample had 20% of non-voters. This could be due to a higher turnout in the last election, particularly among Aboriginal populations. The sample was weighted to reflect actual turnout.
For the purpose of this research, Elections Canada proposed the following definitions of the five sub-samples used. It should be noted that not all of these categories are mutually exclusive.
- Aboriginal: As defined by Statistics Canada – First Nations both on and off reserve, Métis and Inuit.
- Ethnocultural: Those who identify themselves as a visible minority or those who were born outside Canada but who do not consider themselves a visible minority and for whom English, French or an Aboriginal language is not their first language.
- Youth residing in rural areas: Defined as those living in a rural community with a population of less than 10,000.
- Youth with disabilities: Defined as those identifying themselves as having a physical, mental or learning disability.
- Unemployed youth (not in school): Those who do not work either full time or part time, whether at home or for an employer, and are not a student, homemaker or retired person.
The methodology consisted of 1,503 interviews conducted by cell phone (1,000 youth and 503 people over 35) and 1,506 surveys completed online, including the remaining 752 youth of the core sample and the 754 youth of the oversample.
The table below shows the breakdowns of both methodologies.
|Methodology||Telephone (CATI)*||Online (panel)||Total|
|Core sample||Augments (convenience sample)||Gen pop (35+)||Core sample||Augments (convenience sample)|
|No. of interviews||1,000||–||503||752||754||3,009|
*Computer-assisted telephone interviewing.
We used our online panel combined with Edge Interactive's online panel to complete the oversample. Particularly, finding unemployed youth was a bigger challenge than initially anticipated. Most panels do not target for unemployed respondents since this is a situation that is constantly changing. In the end, after reaching out to several partner panels, this was the one quota we could not meet in the time frame given.
The table below outlines the sampling strategy, which was developed to be regionally representative, based on the latest census counts. The table shows the distribution as it actually happened. Please note that, as mentioned, our sampling strategy was changed to include respondents who were previously part of the oversample.
|Age||18–22 (34%)||23–29 (33%)||30–34 (33%)||35+||Total|
|Gender||M (50%)||F (50%)||M (50%)||F (50%)||M (50%)||F (50%)||M (50%)||F (50%)|
*Max. n=500 respondents who are 35 or older.
|Age||18–22 (34%)||23–29 (33%)||30–34 (32%)||35+||Total|
|Gender||M (52%)||F (49%)||M (47%)||F (53%)||M (47%)||F (53%)||M (53%)||F (47%)|
As stated above, of the 2,255 respondents in the core sample, 1,503 (1,000 youth respondents and 503 adults) were randomly selected and contacted by cell phone, and the remaining 752 completed the survey using our online panel.
The remaining sample (754) was divided into five sub-samples. Based on our previous experience, the table below is how we estimated the results would fall by target group. Please note that these groups are not mutually exclusive; therefore, the quotas in each group are filled starting with the ones with lower incidence.
|Rural – Northern Canada||225||100||325|
Filling all the quotas required constant monitoring of fieldwork as well as having several versions of the survey open at the same time. We started with the core sample using computer-assisted telephone interviewing and, once we had a good understanding of how distribution would fall, we opened the online core and the augment. The table below presents the actual distribution of the oversample groups.
|Rural – Northern Canada||575||240||815|