domly selected from the electoral registries of six cities in Great Britain. The cities were selected to include those that had experienced both higher and lower than average levels of unemployment. Respondents were asked a wide range of questions concerning their employment experience, including whether they felt the skills they used on the job had increased, decreased, or stayed the same over the preceding five years. Overall, 52 percent of respondents felt that their skills had increased, whereas only 9 percent reported that their skills had decreased. And 60 percent reported that levels of responsibility in their job had grown, whereas only 7 percent reported a decrease (Gallie, 1994). The pattern of results was robust across occupational categories and industries and was similar for people who did and did not change jobs, but evidence of upgrading was least pronounced among low-skilled manual workers in service industries.
Furthermore, the SCELI data clearly show that upgrading was associated with the use of "automated or computerized equipment" (Gallie, 1994:63). The data show that 39 percent of the respondents reported using digital technologies. Of these, 67 percent reported an increase in skill and 74 percent reported an increase in responsibility. Among those who did not use digital technologies, the percentage reporting an increase in skill and responsibility were 39 and 49 percent, respectively. These results were also robust across occupational categories.
The SCELI data on the impact of digital technologies are consistent with attempts to estimate financial returns from computer use. Analyzing data from 1984 and 1989 Current Population Survey, Krueger (1993) found that using a computer in one's job led to a 10–15 percent premium in wages after controlling for obvious covariates, such as years of education, job tenure, industry, and occupation. DiNardo and Pischke (1997) found returns to computer use of similar magnitude using German data. However, they also found that use of other office technologies, including pencils and paper, also increased wages after controlling for computer use. DiNardo and Pischke therefore argue that economic returns to computer use potentially reflect an unmeasured phenomenon, since it is hard to believe that the ability to use pencil and paper is in short supply. Gallie reaches much the same conclusion regarding the use of computers in the SCELI data: "It is