ABSTRACT
In a probability sample drawn from appropriate rural and urban populations, within probability limits, can be legitimately established with a test of significance. The uncontrolled variables or “correlated biases,” such as education or income, which may “explain” an observed rural-urban difference or lack of difference in political interest do not vitiate the legitimacy of a test of significance for the existence of a difference. Thus, the injunction cannot hold that one cannot legitimately test for significance when there is lack of randomization or correlated biases have not been controlled. An important weakness of much analysis in current social research is the failure of the analyst to consider the distinction between statistical significance and substantive importance. Since, at a given level of significance, statistical significance demands a greater degree of relationship from a small sample than a large sample, it might appear that the researcher can more easily treat substantively important differences by selecting small samples rather than large samples.
