Can we create accurate synthetic adult literacy estimates using more recent population data than the 1990 Census?
Although this could be tried, it seems unlikely to produce reliable estimates. Both the Census and NALS were designed as “snapshots” of a population at a fixed point in time. The synthetic estimation model depends strongly on a very tight linkage between these two sets of data. Such a linkage would not be present if the predictive model based on combining Census and NALS were applied to a different source of population data.
