the presumption of doing such weighting is that we will reduce nonresponse bias. but weights are effective in addressing nonresponse bias only if the respondents within population subgroups do nit differ significantly from the nonrespondents in those groups.