Between 2014 and 2023, the dimensions of the allocation matrices for EEDUC and RGED were incorrectly specified, which means that distributions for imputed data may not align with expectations given respondent characteristics. The impact of these errors on unweighted in-universe data varies by year.

The error in the EEDUC matrix has downstream effects on RGED, which incorporates EEDUC values in its allocation matrix, and ECERT and EPROCERT, which are determined to be in-universe based on EEDUC values.

Data users should be aware of these issues when using EEDUC and RGED, particularly if comparing prior estimates with 2024 data.