Changes to The Conference Board of Canada’s Index of Consumer Confidence
The Index of Consumer Confidence is constructed using responses from a sample of Canadian households to four attitudinal questions. Every month, those surveyed are asked to give their views about their households’ current and expected financial positions and their outlook for employment over the short term. They are also asked to assess whether it is a good or bad time to make a major purchase, such as a house, a car, or other big-ticket item.
The Consumer Confidence Survey has been conducted on an ongoing basis since 1982—at first on a quarterly basis, but since 2001 on a monthly basis. Until December 2014, survey questions were posed by telephone to a random sample of Canadian households. However, in recent years we have been concerned about the ability of telephone surveys to ensure an accurate representation of Canadian households. More and more households no longer have landline telephones, and while we adapted by including cell phone numbers in the surveys, the drop rates on cell phone calls were very high. That led to concern about our ability to obtain survey responses from specific cohorts.1 As such, starting in January 2015, The Conference Board of Canada switched to an online survey conducted using an Internet panel and administered by Leger2. To ensure consistency with the results of our previous telephone surveys, the panel survey is conducted in two waves over the first weeks of each month. The response rate has also been increased to 3,000 respondents, up substantially from the 2,000 for the phone surveys.
For the month of December 2014, both the telephone survey and the online panel survey were run conjunctly, allowing us to compare differences in the answers generated by the two survey methods. Because respondents to an online survey have more time to reflect and are not answering to a person, the results tend to differ. To avoid a break in the data, we extended the new results back over history using the historical data from our telephone surveys. Specifically, the results for each individual component from the online survey were grown back using the percentage changes from monthly data collected from our past telephone surveys.3 The data have been extended back to 2002. (We are considering using the quarterly data to extend the data back to 1982.)
In addition, we have created a new index for Alberta and combined Saskatchewan and Manitoba into one index. The Alberta and Saskatchewan–Manitoba indexes both start in 2006.
Finally, the base year for all of our indexes has been updated from 2002 to 2014.
Download the concordance table (xls)
1 According to Leger’s Internet and Telephone Survey, December 2014, recent research tends to demonstrate that “probabilistic” telephone surveys suffer increasingly from coverage errors, as reflected by plummeting response rates over the phone and a declining number of households with a fixed phone line. This trend is leading to a systematic bias in data collection, rendering such telephone surveys unrepresentative.
2 Surveys with Internet panels do not use a sample that is strictly probabilistic in nature. In other words, not every Canadian has an equal chance of being picked for the survey. To ensure that its Internet samples are fully representative, Leger has created an Internet panel of over 315,000 respondents that is representative of the Canadian community. Panel results are weighted to be fully representative of the latest census. In addition, 25 per cent of the panel is replaced every year.
3 We grew back the share, or proportion, of answers associated with each question. In order to ensure that the sum of the proportions remains equal to 100, we constrained the sum of the proportions associated with each possible answer to 100, starting from November 2014 and going back to January 2002. This process altered the monthly changes in the historical data slightly.