BOX 3: Measuring the likelihood of being an intense computer user
Data for this section come from a logistic regression which is designed to measure the odds of being a “highintensity” user of computers for task-oriented purposes. We identify “high-intensity” users as those respondents who are in the top quartile (highest 25%) of the scale measuring the use of computers for task-oriented purposes.
While the previous section measured variables in isolation, this section uses a logistic regression model that incorporates several variables in the analysis. The model was used to study the influence of age, gender, educational attainment, employment status, household income and literacy skills on the use of computers for task-oriented purposes. Using this technique, it is possible to isolate the influence of each variable by controlling for all other variables in the model. For example, the influence of education can be examined among those with the same age, gender, employment status and so on.
The results varied substantially by country. In some countries gender exerted a particularly strong influence on the use of computers for taskoriented purposes. While controlling for other factors, men in Italy, Norway and Switzerland were still more likely to be high-intensity computer users. In contrast, and consistent with findings reported earlier in this paper, gender differences with respect to ICTs were smaller in North America. In fact, in Bermuda there was no difference in the odds between males and females of being a high-intensity computer user.
The analysis also confirmed that education is strongly associated with computer use. In the United States and Italy, adults with upper secondary education had more than two times the odds of being high-intensity computer users compared to those with less education. In the remaining countries, the odds were approximately twice as high for adults with post-secondary education compared to those with less than upper secondary educational attainment - even while controlling for other variables.
The results also reaffirmed that those with high levels of household income were more likely to be intense computer users. In most countries, respondents whose income falls in the top income quartile had approximately two times the odds of being high-intensity users of computers for task-oriented purposes compared to those with lower income.
Literacy skills proved highly correlated with computer use. As literacy skill levels6 increased, the odds of being a high-intensity computer user increased. For example, in the United States and Switzerland, a respondent with high prose literacy skills (levels 4 and 5) had nearly twice the odds of being a high-intensity user compared to respondents with low literacy (levels 1 and 2). The relationship between literacy skills and computer use was even stronger in Canada, Bermuda, and Norway, where respondents with high literacy skill levels had between two to more-than-three times the odds of being intense computer users compared to those with low literacy skills. Furthermore, in all countries except Norway, the gaps between low and average literacy groups were greater than the gaps in the odds ratios between those with average and high literacy. Table 11 depicts the odds ratios of being a highintensity computer user by different literacy levels.
6 For a detailed description of literacy levels, see Statistics Canada and OECD (2005), Learning a Living: First Results of the Adult Literacy and Life Skills Survey, Ottawa and Paris.