Declining trust in politicians and political institutions is one of the clearest trends in American public opinion. Confidence in religious, educational and other institutions has also waned, but emphasis has focused on political trust, both because it may summarize a wide range of diffuse grievances and because it might indicate increased potential for disruptive action, political violence and instability. In spite of investigations by numerous scholars in collecting longitudinal data on political trust over nearly three decades, there are serious doubts about the contribution of the concept to explaining political change. It is not clear what is measured by available indicators, and only weak and inconsistent evidence links expressed political trust with other political attitudes or with behavior. Research to date gives little guidance into whether the fault lies with theories about political trust or with current indicators. The goal of this research is to evaluate available longitudinal data, from the National Election Studies and other national surveys, against several validity criteria, and to derive proposals for selectively deleting and improving the items that comprise this core component of the National Election Studies time series. Criteria for construct validation include the relationships between political trust and : a) other aspects of the global concept of political alienation; b) representation, and especially the notions of fairness and efficiency intrinsic to procedures for holding governments publicly accountable; c) policy implementation, especially politicians claims that a few policy failures can initiate a "spiral of distrust," with citizens refusing to comply with new policies. Each criterion is specified as a distinctive covariance structure model or family of models, and parameter estimates will allow quantitative judgments of relative validity. Potential contributions of the research are three-fold. First, it will foster better-informed decisions about continuing or modifying data collection on political trust. Second, because the concept of political trust is imbedded in theories about alienation and political legitimacy, construct validation with multiple measures of multiple concepts are not well developed, and this research will provide an arena for formulating and testing procedures that adapt classical inference criteria to covariance structure analysis.
