8 Your responses to the latest U.S. Census were used for two purposes. First, the Census Bureau tallied each response to produce an offi cial population count. Second, it produced a 1-in-20 sub-sample used for analysis by researchers. For those younger than 65, the estimates from the sample are similar to the full count. For those over age 65, the estimates disagree by as much as 15 percent. The sample data suggest that there are more very old men than very old women. And, the error jumbles the correlation between age and employment, age and marital status, and, possibly, other correlations as well. The Census Bureau has refused to correct the data.

a Should the data in the 1-in-20 micro-sample be used to study people aged 65 and over?

b What’s the source of the problem? Programming error, coding error, or manipulating the data to protect the identity of each individual?