If We Want to Conduct a Representative National Survey with Approximately

Essay

If we want to conduct a representative national survey with approximately 2,000 respondents, with the goal of obtaining a relatively accurate idea of people's opinions regarding a certain national issue, why might it be necessary to over-sample certain groups? For the purposes of our survey, why would we want to necessarily find more Caucasian women, who are in a committed same-sex relationship, with children, compared to any other combinations of demographic characteristics? Why wouldn't we want to find a number of respondents who fit this group that is proportional to the size of the larger population (e.g., if 0.1 percent of people in the population fit this category, why would we want to recruit more than two respondents from this group)?