What we can learn from selected, unmatched data: Measuring internet inequality in Chicago