The Prediction of Industrial Mail-survey Response Rates
研究了工业邮件调查回复率的预测问题，发现现有模型预测能力不足，基于线性logit模型开发了新模型，并用独立样本验证其优于简单线性模型，帮助研究者优化调查设计。
Mail-survey response rates and their prediction are important issues for researchers. The conceptual differences between industrial and most non-industrial populations raised doubt about the applicability of the Heberlein and Baumgartner model for the prediction of industrial mail-survey response rates. A test of their model using data from industrial mail-survey studies revealed a low level of predictive ability. A new model, based upon a linear logit specification and data from industrial populations, was developed and validated using a hold-out sample. The results suggested that the model based upon the new specification was superior to the simple linear alternative. The model identified a number of variables which were associated with response rates from industrial populations, and allows researchers to test and refine alternative survey designs.