Demographic Forecasting is a nifty new book by Federico Girosi and Gary King that describes ways to forecast mortality when time series data are, as the authors describe, "noisy and sparse." In other words, they have pioneered a smart technique for getting a handle on mortality in developing countries, where data series tend to be rare and lower-quality.
What is a little surprising, given the specificity of the title, but less so when you consider who the authors are, is that the statistical methodology they suggest is much more broadly applicable, basically in any social science setting where researchers have priors about the variable they're modeling but doubts about the data quality. For example, the authors reveal, comparative political science, where "standard" metrics might take on vastly different qualitative meanings across political boundaries.
