New Research Memorandum (ROA-RM-2017/6)
Estimating literacy levels at a detailed regional level: An application using Dutch data by Ineke Bijlsma, Jan van den Brakel, Rolf van der Velden and Jim Allen
Policy measures to combat low literacy are often targeted at the level of municipalities or regions with an above-average population with low literacy levels. However, current surveys on literacy do not contain enough respondents at this level to allow for reliable estimates, at least when using only direct estimation techniques. To provide more reliable results at a detailed regional level, alternative methods must be used.
The aim of this paper is to obtain literacy estimates at the municipality level using model-based small area estimation techniques in a hierarchical Bayesian framework. To do so, we link Dutch Labour Force Survey data to the most recent literacy survey available, that of the Programme for the International Assessment of Adult Competencies (PIAAC). We estimate the average score, as well as the percentage of people with a low literacy level.
Additional complications arise, as the PIAAC framework assumes that test scores reflect an underlying latent construct. Moreover, as an adaptive design has been used with rotating modules, not all respondents are assigned the same test items. This is why an item response model is used with multiple imputation resulting in 10 so-called plausible values for the literacy proficiency level per respondent. Variance estimators for our small area predictions explicitly account for this imputation uncertainty.
The average literacy score is estimated with a unit-level model, while the percentage of low literates is estimated using an area-level model utilizing pooled variance. Optimal models are selected using a conditional Akaike information criterion score. Municipalities with less than 40,000 inhabitants were clustered with neighboring municipalities to ensure sufficiently large sample sizes.
The PIAAC survey is currently carried out in 36 countries. Most of these countries also have labor force surveys that contain similar information as the one used in this analysis. This opens up the possibility of applying the same method in other countries.