A survey of Chicago, continued. The sample for the survey described in Exercise 8.28 included 477 non- Hispanic black adults, 304 non-Hispanic whites, and 263 Hispanics. Results for the entire sample are considered to be accurate within ±4%, the margin of error for non-Hispanic blacks was ±5%, and for non-Hispanic whites and Hispanics, the margin of error is ±7%.

a. Why is the accuracy greater for non-Hispanic blacks than for non-Hispanic whites and Hispanics?
b. The poll details indicate
For purposes of analysis, black and Hispanic respondents were oversampled. This involved (additional) random dialing in landline exchanges and cell phone areas with high proportions of minority residents. What would have been true of the margin of error for blacks and Hispanics if they were not oversampled? If we want to compare the responses of different ethnic groups, explain briefly the purpose of oversampling these groups.