Statistics Canada
Symbol of the Government of Canada

Site navigation menu

Skill gains and losses for Canada and the provinces

Warning View the most recent version.

Archived Content

Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.

The overall levels of skill gain or loss by age for Canada are shown in Chart 2.  The chart shows the average relationship between document literacy scores and the age of the respondent for adults participating in the 1994 IALS and the 2003 ALL.1 The x-axis is scaled to include the "synthetic cohorts" of adults aged 16 to 65 in 1994 (IALS) and aged 25 to 74 in 2003. For example, adults who were born in 1954 were 40 years old in 1994, when IALS was conducted, and 49 years old in 2003 when ALL was conducted. For adults at this age, the ALL line is below the IALS line; the difference is 13.1 points, indicating a skill loss for 40-year old adults. Judged in educational terms, these are significant losses of skill, roughly equivalent to the average learning gain in literacy associated with an additional 3.5 months of schooling (Willms, 1999).

Chart 2 Document literacy scores versus age for Canada, 1994 and 2003. Opens a new browser window.

Chart 2 Document literacy scores versus age for Canada, 1994 and 2003

Chart 2 indicates that there was a skill loss, on average, for Canadian adults in the middle of the age 16 to 65 age range. The lines cross at the points when adults were 20 or 60 years old in 1994. For these cohorts there is virtually no skill loss or gain. Table 1 provides estimates of the skill gain or loss for adults at ages 25, 40, and 55 for all of Canada, for all Canadian males and females, and for each of the provinces.2 The table also shows the standard errors of the estimates3, which provide an indication of their accuracy. Skill gains or losses that were statistically significant (p < 0.05) are marked with an asterix.

Table 1 Estimates of skill gain (or loss) for adults at ages 25, 40, and 55, based on IALS (1994) and ALL (2003). Opens a new browser window.

Table 1 Estimates of skill gain (or loss) for adults at ages 25, 40, and 55, based on IALS (1994) and ALL (2003)

For all of Canada, as Chart 2 suggested, the skills losses are large for middle-age adults, those at about age 40. For them, the skill loss amounts to about 13 points. It is important to put these estimates of skill loss in perspective. Average years of schooling rose by almost a full year between 1994 and 2003; this is an amount that would normally be associated with skill gain in the order of 25 points. Comparison of results from the 2000 and 2003 cycles of OECD PISA assessment of 15 year olds reading literacy suggests that skill levels of secondary graduates have at least remained stable. Finally, public and private per capita expenditure on education rose during the period, an increase that should have had a positive effect on skill levels. Thus, it would be reasonable to expect that literacy skill levels should have risen between 1994 and 2003. The fact that skills actually appear to have fallen suggests that skill loss is real.

Chart 3 shows the skill versus age curves for males and females for Canada. The distributions are quite similar, except that the skill loss for females is more pronounced, especially among late middle age adults. The skill loss for females at age 40 is statistically significant, and for males it is not significant; however, the sexes do not differ significantly in the extent of their skill loss.

Chart 3 Document literacy scores versus age, by sex, 1994 and 2003. Opens a new browser window.

Chart 3 Document literacy scores versus age, by sex, 1994 and 2003


1. The results were estimated with a simple linear regression of document literacy skills on age and age-squared for the combined sample, with age-by-cohort interactions to capture the synthetic cohort differences:

2. The sample sizes in the 1994 IALS were much smaller than those realized in the 2003 ALL, which limits the amount of disaggregation of results at the provincial level. The multi-level, multivariate methods employed in this analysis make use of the full information and provide standard error estimates that permit one to judge the reliability of the results.

3. The research and sampling design for IALS and ALL entailed the use of multiple test booklets, and a stratified sample design. The estimation of the standard errors requires special programming that uses the replicate jackknife design weights and a set of plausible test scores.