The CPEP Survey

The Comparative Provincial Election Project survey is being administered online, using panels assembled by Abacus Data. As only one of three national polling firms to predict the results of the 2011 federal election within its margin of error, Abacus has established itself as a trusted research source for the academic, public, and private sectors. 

2011-15 Election Codebook and Dataset

2012-13 Elections Questionnaires

This Codebook contains questionnaires deployed in Prince Edward Island, Manitoba, Ontario, Newfoundland and Labrador, Saskatchewan, Alberta, British Columbia, and Quebec.  Nova Scotia, New Brunswick questionnaires will be added once available.  

To date, the survey has been completed by over 7500 Canadians, including:

Prov.

Sample

Survey Dates

PE

509

October 4-25, 2011

MB

775

October 5-31, 2011

ON

1044

October 7-31, 2011

NL

851

October 12-30, 2011

SK

821

November 8-21, 2011

AB

897

April 25-May 15, 2012

QC

1009

Sept. 5-Oct. 11, 2012

BC

803

May 15-29, 2013    

NS

797

October 9-30, 2013

NB  662 October 6-November 1, 2014
AB  805 May 6-28, 2015

Data was collected using a mix-mode survey.  Respondents were recruited using two methods.  First, respondents were recruited from the Probit online research panel which was randomly generated using random-digit dialing (RDD) live telephone recruitment.  If the online panel was insufficient to complete all the survey, Interactive Voice Response (IVR) to Web technology was used to randomly recruit participants.

The IVR-to-Web methodology is based on a random digit dial sample (RDD) that is drawn from a dual land-mobile frame.  The sample is dialed by an Interactive Voice Response (IVR) system where an announcement is made to the responding household indicating that an online survey that is available for this randomly selected household.  The IVR sample is rotated and replacements are made during the course of field collection.  IVR employs a rule of no more than four (4) call-backs.  The automated message follows all of the rules laid out by the Canadian Radio-Television and Telecommunications Commission (CRTC) with respect to Automatic Dialing-Announcing Device (ADAD).  The main body of the announcement follows the recommended survey introduction standards laid out by the Market Research and Intelligence Association (MRIA).   In addition to these standards the message indicates the survey site URL and a unique password to access the survey website.  The respondent then goes online at their convenience, inputs his or her password and completes the survey.  All of the geo-encoded information from the RDD sample is merged back to the unique passwords, allowing for greater depth of analysis.

Survey respondents are recruited from across Canadian society, to obtain a representative sample of all eligible voters. Specifically, we seek representativeness in terms of gender, age, education, community size (urban/rural), number in household, immigration status, religion, religiosity, income, and occupation. Geographically, we have also oversampled residents of northern regions in each province.  Unless indicated otherwise, findings are produced by weighting cases according to gender, age, education, and region, using census data.

The results of the survey are being analyzed by members of the research team and their graduate student assistants.