Summary and implications

View the most recent version.

Young children differ substantially in their language and literacy skills when they enter school at age 5, in large measure due to their exposure to language and their interactions with parents and siblings (Hart and Risley, 1995). During the primary school years, the rate of acquiring literacy skills continues to vary considerably. Family experiences continue to play a prominent role, but the quality of schooling contributes also. The transition from "learning-to-read" to "reading-to-learn", which for most children should occur during the second or third grade, is critical, as children who fail to make this transition tend to be poor readers throughout their school career, and are prone to leaving secondary school before graduating. Different groups of people leave the educational system at different points, and have much different skill levels.

However, literacy is not a static commodity that is acquired in youth and maintained throughout life. Some groups of individuals find a way to add to their level of skill over their adult lives, while others manage to maintain their skill level and others lose skill. The data from the IALS and ALL provide a rare opportunity to compare the distributions of literacy skills for nationally representative samples of adults in 1994 and 2003. The two studies used comparable assessment tools and sampling designs such that one can construct "synthetic cohorts" of people who were born in the same time period but assessed at different ages. This provides a means of untangling the effects on literacy skills associated with aging versus cohort and period effects. The analyses in this paper examine the synthetic cohorts for Canada, with attention to levels of skill gains or losses for different sub-populations.

The results provide compelling evidence that on average people lose skills after the period of formal schooling, but the amount of skill loss differs considerably from group to group. Skill loss in Canada appears to be a gradual process, which begins at about age 25, peaks at around age 40, and tapers off during late middle age. For example, adults who were age 40 in 1994 had average scores on the IALS literacy test of about 288, while the same cohort, when tested 9 years later at age 49, had average scores of about 275. A skill loss of about 13 points is roughly equivalent to nearly half a year of additional schooling over the nine-year period.¹  Taking into account that skill loss appears to be less for young and late middle age adults, we estimate that on average most Canadian adults experience a skill loss over their lifetime of about one grade level.

The analyses also provide an account of the factors related to skill acquisition, based on data from the combined IALS and ALL data sets. Exposure to education appears to have a positive impact on skill change. Individuals with university completion have average scores that are about 30 points higher than secondary school graduates, and even those with some post-secondary education do better, by about 15 points. Those who have not completed secondary school do considerably worse, scoring nearly 50 points lower than their counterparts who finish secondary school, a drop equivalent to the average learning gain associated with over a year and a half of additional education. These are the effects associated with level of education across all adults, after controlling for labour market participation, further education and training, and engagement in literacy practices.

The level of general reading engagement at work also has a positive effect: individuals who read more frequently and a wider range of materials scored about 11 points higher than those with low levels of engagement. Similarly, those who had participated in further education or training scored about 16 points higher than those who had not participated. The combined effect of these two factors is equivalent to the positive effect associated with completing university. We cannot conclude that increasing further education and training or opportunities for work engagement would cause higher literacy skills. Clearly there are selection effects at play that are not captured by the other variables in the model. However, as descriptive data, they do suggest that there are likely certain occupations and firms that create a culture that supports and values the acquisition and maintenance of literacy skills.

Engagement in technical literacy practices at work increased over the study period, but this does not seem to have as strong an impact on people's literacy skills. In contrast, the amount and range of what people choose to read at home, away from the job, seems to have a strong influence on skill development. In fact, the increase in skill observed for individuals with higher levels of reading at home is about the same as the skill loss observed on average over the ten-year period when skill loss is greatest. Clearly, lifestyle and individual choice matter.

Finally, employment seems to have a very positive effect on literacy skills – those individuals who were employed scored about 12 points higher than those who were not in the labour force. As was noted with regards to the effects associated with further education and engagement at in literacy activities at work, this analysis does not allow one to determine if this loss is the result of unemployment or simply a symptom, but the effect is real and socially significant to the individuals involved and to the economy.

Our analysis also examined differences among the provinces in their average levels literacy and their skill loss. We found that provinces and regions varied substantially in their average levels of skills. A small proportion of this difference is simply attributable to differences in the demographic age and sex distributions of the provinces. But even when this is taken account, there remains considerable variation, ranging from about 13 points below the national average to 13 points above it, or about one grade level. The levels of education attained by adults in each province explain about 40% of this variation. A finding that is more salient though is that the other factors – employment, further education and training, and engagement in literacy practices at home and at work – account for over one-half of the inter-provincial variation. The variance explained by these factors overlaps that associated with level of education, such that the two sets of factors together explain nearly 60% of the inter-provincial variation. 

The statistical modeling allows us to obtain more accurate measurement of the skill loss, essentially by equating the two synthetic cohorts in terms of their sex and age distribution. However, the design does not allow us to estimate the effects of changes in levels of further education or engagement on skill loss. When we examined the changes from 1994 to 2003 in the various factors that drive literacy scores, we found that while levels of both formal and informal education and training had increased, levels of engagement at work had fallen slightly, and levels of home engagement had fallen considerably.

The results presented in this report are troubling and hold several important messages for policy makers. First, the existence of literacy skill loss is confirmed. Second, the magnitude of skill loss is high when judged in educational terms, eliminating literacy acquisition that took months, or even years, to acquire on average. Third, given the relationship of literacy skill to individual economic and social outcomes, and to macro economic performance, it is reasonable to assume that the economy is paying a price for skill loss. Fourth, the probability of whether a group will gain or lose skills appears to depend on a variety of factors over which both individuals and governments can exert some degree of control. Post-secondary education, the amount of reading on the job and off, and stable employment all appear to have had a positive impact on the stock of literacy skills and reduce the magnitude of skill loss.

One interpretation of these findings is that Canadians are over-educated, that our education systems are producing individuals with literacy skills that the economy is not able to absorb. If this is true then one might consider reducing investments in post-secondary education where the losses appear to be the greatest.

A second interpretation of the findings is that governments need to do something to increase the social and economic demand for literacy skill. Literacy skills drive economic growth. The fact that one observes skill loss implies either that some individuals are adopting lifestyles that do not support and maintain skill levels, or that some employers are employing economically sub-optimal strategies which, while profitable, are constraining the performance of the overall economy. Both forces could be at play. If this is true then it could be taken as evidence of a market failure of the sort that only governments can correct. If so, investments by government might induce employers to make full use of the available human capital to the benefit of all.

A third interpretation of these findings is that Canada's education system is failing to impart durable skills, or at least the attitudes, values and behaviours that would allow their graduates to retain the literacy skills they learned. If this is true, then Canadian educators need to take a hard look at the content and delivery of instruction to see what might be improved.

This report does not pretend to provide answers to these questions, or to say which of these interpretations is correct.  This is a matter for careful and thoughtful analysis and debate.  The report does show that something is wrong and that the economic and social consequences of ignoring it are likely to large to bear.

Notes

1. In earlier work based on the IALS, Willms (1999) found that for the full international youth cohort aged 16 to 25, an effect size of 0.15 of a standard deviation, or 30 points on the literacy scale, was associated with about one additional year of schooling. This was determined by regressing the standardized literacy scores on the "years of education" reported by the respondent, and by examining the relationship between respondents' standardized literacy scores and their level of education (for example, completed secondary school, some college or university, university graduate).