The 22% and the 29% in the private sector are the gross unadjusted rates. It's virtually all women, all men, with no attempt to make comparisons among those who have comparable levels of income.
Now, with this data source, the question is on limitations of the data. For example, regarding the data source with the 22% and the 29% that we have here, we don't know anything about those individuals aside from whether they're male or female. We don't know their earnings or their ages. To get a comparison of what you're talking about, one would hope we could go to tax data, for example, and there we have very good information on the earnings and pension coverage. The problem is that for prior to 2000 we have no information about the industries in which they're working. I think in more recent years we do have that, so we could look at private sector coverage, in a sense controlling for earnings, as you're suggesting. We could go back and do that, using the taxation data or the survey data to make the comparison you're asking for.