A Comparison of Alternative Indicators for the Risk of Nonresponse Bias