试析人工神经网络运用于社会统计分析中的可能性

社会统计分析是社会研究方法中十分重要的一部分,因为定量研究所得的原始资料在经过审核、整理与汇总之后,需要进行科学系统的统计分析,才能揭示出研究资料所包含的众多信息,从而得出研究的结论。在多元统计分析方法中,研究者用得比较多的是多元线性回归分析、方差分析、log线性模型、probit模型及logistic回归分析,这些分析方法能够帮助我们对数据进行简化、对变量之间的相互依赖关系进行分析,以及根据模型进行预测。但上述方法也存在着一定的使用局限性,比如,方差分析法要假定进行均值比较的总体具有相同的方差和服从正态分布,如果不符合…

The Potential Application of Neural Network Models in Social Science Statistics

Social and political relationships are generally characterized by nonlinearity and complexity, and are usually of unknown functional forms. Many statistical methods employed in social science research, such as multi-linear regression, logistic regression, logit and probit models, always assume exact and, in general, linear functional forms for the underlying utility functions. Neural network models, on the other hand, are capable of approximating arbitrary functional forms under general conditions and can handle rich patterns of nonlinearity in the utility functions. This article suggests that neural network models can be widely applied to social science statistics.