This project introduces a framework for inference using qualitative and quantitative information. It is designed for applications where the researcher is faced with the problem of drawing inferences of an unobservable state variable in a time series context. Quantitative information is taken to be observable time series variables. Qualitative information is taken to be an indicator (dummy) variable that is coded by the researcher and that reflects the researcher's distillation of literary materials that is believed to contain information about the true unobserved state variable. The framework is based on an iterative algorithm that produces an inference oaf the unobservable for each sample period using the two types of information and presents a probabilistic assessment of how useful an indicator of the true unobserved state is the indicator variable proposed by the researcher. The framework's output fills a gap in the applied macroeconometric literature that was created by the recent revival of interest in the use of qualitative information in macroeconomic analysis. It also complements classical regression analysis in cases where qualitative events are important by providing a novel way of assessing dummy variables. Finally, theoretical applications of the approach to inference suggest that the algorithm could be used by policymakers when policy depends in part on an inference of an unobservable.
