Data in Social Inequalities

In Table 4.3 of the Norris book, there is an increase in percent on-line for each occupational status. This sample includes students. I would imagine that some of the people included in the 1996 student category are in the other categories in 1999. If this is indeed the case, should Norris control for this? Does this fact change any implications or conclusions drawn from the results? For example, the +30 results for managers be due in part to the graduation of students who were already using the Internet in 1996?