The research conducted for the Survey of the Candidates of the 43rd Federal General Election consisted of an attempted census of candidates registered for that election, using a dual-mode telephone-online survey.
First attempts to reach candidates were by telephone. When there was no response, voice messages were left when possible explaining the study and email invitations were automatically sent with the link to the survey. In cases where candidates expressed a preference for answering the survey online, email invitations were also sent and participation was monitored. When there was no online response within one week, subsequent calls and emails were initiated. In total, 1,172 interviews were completed, with 396 interviews completed by phone, and 776 interviews completed online.
The population of 2,146 candidates was provided by Elections Canada. Information included in the sample included: full name, address, phone numbers, email address, party affiliation, gender, federal electoral district, whether the candidate won, incumbency status.
Before any fieldwork commenced, Elections Canada sent a letter from the Chief Electoral Officer, Stéphane Perrault, to all candidates. The purpose of the letter was to:
A 1-800 number was provided to candidates to schedule an interview at a time that was convenient for them, if desired.
Elections Canada provided a revised questionnaire based on the one used for the 2015 Candidates' Survey. The survey was programmed and EKOS Research Associates conducted a thorough review, provided written comments, and feedback to Elections Canada. Test interviews were conducted between October 22 and 23, with 26 interviews completed (13 in French and 13 in English) at an average time of 26.5 minutes per interview. The survey data and audio recordings were carefully reviewed and a few minor changes were implemented. Some changes were made to increase clarity and provide additional instructions to interviewers. Wording was streamlined in a few questions. The survey was launched on October 25, 2019. The pretest cases were included in the final sample with some minor adjustments based on changes as a result of the test.
All candidates with telephone numbers were contacted initially by phone. Cases with a missing telephone number were individually researched online, using a general search engine. Records where a telephone number was found to be invalid (i.e. not the correct number or a number no longer in service) were also subsequently researched online. When it was not possible to obtain a viable telephone number but a valid email address was available, email invitations to the survey (including the survey link) were sent. All candidates from one party (n=8) were listed with a common telephone number and email address for the party office. In those cases, the party office was contacted a number of times with no contact made. In a few cases no telephone number was available, but an email address was available. Email invitations were sent in all those cases.
Interviews were recorded to ensure data quality. Callback times were made at various times of the day and appointments scheduled throughout the day and evening, across the week to ensure a greater chance of successfully contacting candidates. Up to 10 contacts per candidate were made, including both telephone contact and email invitations/reminder invitations. Candidates contacted by telephone who declined to take part in the research were offered the opportunity to complete the survey online. Further, multiple email invitations were sent to candidates who did not complete the survey over the telephone, including those not reached and those whose telephone numbers were found to be missing or invalid. This was also the case with those who agreed to participate online and subsequently did not. Candidates were informed of the sponsor of the survey and provided with the opportunity to leave an email address to be informed once the results of the survey are released.
The survey collection period spanned October 25 to November 27, 2019. Throughout the course of fieldwork, EKOS provided Elections Canada with weekly updates on data collection progress with details of the numbers of calls made and interviews completed, as well as procedures used to contact candidates, along with response rates.
The survey was registered with the National Survey Registration System and the study sponsor was provided in all contact attempts. Copies of the Elections Canada advance letter were also offered by email or fax to anyone wanting to receive it. The response rate for the survey based on the complete sample out of the remaining valid sample (2,109) is 55.6%. Following are the call results at the end of the fieldwork, on November 27:
| Disposition | Total |
|---|---|
| Invalid number, no viable email address | 20 |
| Fax/unmanned riding office, no viable email address | 11 |
| Unavailable for survey period | 6 |
| Remaining valid sample | 2,109 |
| Answering machine – multiple messages left, emails sent | 27 |
| General callback – multiple calls made, emails sent | 8 |
| Incomplete refusal | 1 |
| Refusal | 190 |
| Contact made, agreed to go online – multiple calls made, emails sent | 711 |
| Complete | 1,172 |
| Total | 2,146 |
Over the course of the survey collection, open-ended responses were reviewed and assigned to existing or new category codes. Data quality was also closely monitored through the collection period, starting with a close review of the results from each round of the pretest, and the first day after the survey launch. A database was created with full variable and value labels, and the non-response rate was assessed based on administrative information in the initial list of candidates provided. The final data was weighted by party, age, incumbent/not, and elected/not. No segment was more than three per cent lower than the population distribution in the same segment. Data tables were produced providing results overall and by region, age, gender, language, and whether the candidate was born in Canada or not, and whether their party is represented in the House of Commons, whether they were running for the first time or not, elected or not, and as an incumbent or not.