Comparing Census Counts to Estimates, Dependency Ratios Projections

by William Warren Munroe, November 5, 2016

The Population Projections Project calculates several population projections scenarios directly from Censuses of Population available over the last 20 years, from the nation as a whole to small towns and villages.

However, the counts are not entirely accurate, some people are missed and some counted more than once. Statistics Canada offers an estimate of the amount of error, aka an undercount, of approximately 2 to 3 % for Canada as a whole. Some areas within Canada have much higher error. Statistics Canada provides annual population estimates, numbers based on the census counts but adjusted for the estimated undercount.

Let's look at projections calculated directly from counts using cohort change ratios (CCRs) compared with projections calculated directly from the estimates also using CCRs.

To do this, we'll use the census years since 1996 to calculate dependency ratios projections (average scenario) referring to counts (Figure 1) compared to STC's estimates (Figure 2). We understand right away that the counts will be lower in absolute numbers (by about 2 to 3%) than the estimates, but we want to see if there is much of a difference in the demographic trends.

Figure 1: Dependency Ratios Projections, based on Counts, Canada, 1996 to 2041¹

Figure 2: Dependency Ratios Projections, based on Estimates, Canada, 1996 to 2041¹

Although the ratio of people of working age (in this case 20 to 64 years of age) and elderly are slightly higher, the ratio of children (in this case 0 to 19 years of age) is almost the same.

The Population Projections Project does not over look nor by-pass the census counts, but rather starts with the census counts because they are raw, genuine and unadulterated. The estimates from Statistics Canada make sense and may be preferable to the counts; nonetheless the counts and the estimates create very similar demographic trends. One of the benefits of the Census of Population is that counts are available for small areas such as towns and villages. These come with an estimate of undercount.²

Let's not stop here though. In the next article we'll take a look at Broad Age Groups calculated referring to counts compared to estimates.

We look forward to the release of the 2016 census counts in 2017.

Endnotes

¹ Adapted from the Census of Population counts (1996, 2001, 2006, and 2011, Statistics Canada); and projections (2016 to 2041) created by William Warren Munroe.

To download census counts go to: Statistics Canada's website at http://www12.statcan.ca/census-recensement/index-eng.cfm. Be sure to read the footnotes and check for important information about possible changes from one census to the next. These are counts; therefore, everyone may not be counted and sometimes people are counted more than once.

² Go to Cansim online, Population Demography, Population estimates and projections, Estimates for Canada, provinces territories, 051-0001; Estimates to the census division level 051-0062.

³ Caution should be exercised when looking at small population areas as the error may make analysis less refined. I recommend looking at areas across scale, from Canada to the province or territory then to the Census Division (i.e. Region, Division, County etc.) then go to the CMA or CA then to the small population area such as cities, towns and villages.