The TCRN has made some methodological advances related to assessing disclosure risk and imputing missing data.  TCRN investigators Daniel Manrique-Vallier and Jerry Reiter developed a way to estimate whether records that are unique in a sample are in fact unique in a population.  Their approach, based on grade of membership models, outperforms state-of-the-art methods based on log-linear modeling.  Manrique-Vallier and Reiter are developing other approaches that are computationally fast and can handle structural zeros, the latter of which has not been dealt with satisfactorily in the disclosure risk estimation literature.  Check the White Papers section of this site soon for more details.