Beginning in Wave 2 of both the 2008 and 2004 Panels, the Core Educational Attainment variables (EEDUCATE and RGED), and the Educational Attainment Recode variable (RCOLLVOC) were incorrectly coded for some respondents. An instrument design flaw re-asked people with less than a high school education and no GED if they had earned a GED at each wave. Educational attainment should have only been updated for respondents under age 26 or those enrolled in school at the prior wave. Although a few of these respondents may actually have earned a GED between interview waves, the magnitude of the increase in GEDs over the Panels is inconsistent with other data sources, suggesting a large proportion of false positive responses. Since the instrument records GED holders as high school graduates, re-asking the GED question led to an increase in high school graduates in Wave 2 and subsequent waves of both Panels.

For additional details see: Core User Note “2008 Panel (Waves 2 - 5) and 2004 Panel
(Waves 2 - 12) Educational Attainment User Note, Concerning Variables EEDUCATE, RGED, RCOLLVOC” posted 8/25/2011.

As a result of the core educational attainment variable corrections explained above, both the 2008 Panel and the 2004 Panel Education and Training History Topical Module variables EGEDTM, EHSYR, EPUBHS, ECOURSE1-ECOURSE7, EPROGRAM, ELSTSCHL, and ALSTSCHL require corrections in order to make them consistent with the updated core attainment data. This user note addresses changes made to the 2004 Panel topical module.

A separate issue in the Education and Training History Topical module in the 2008 and 2004 Panels arose from setting universes for variables based on values of educational attainment only (EEDUCATE), and not the attainment recode that takes into consideration vocational certificate completion in combination with other degree completion (RCOLLVOC). Some respondents who completed a vocational certificate (EEDUCATE=41) had values for items such as year of high school completion (EHSYR) even though they had completed a vocational certificate without completing high school (RCOLLVOC=6). Such respondents should not have been in universe for these topical module items.

The 2004 Panel uncorrected data estimated there were 19.9 million adults with a high school education or higher who completed high school via a GED, but the corrected data show there are 16.0 million (Table 1). Table 2 reports the unweighted distributions and counts of corrected responses for selected items in the topical module affected by these changes.

The 2004 Panel Wave 2 topical module variables affected by the GED change (ELSTSCHL, EGEDTM, EHSYR, EPUBHS, ECOURSE1-ECOURSE7, and EPROGRAM) were corrected in order to make them consistent with the updated core 2004 Panel education variables.

For the year last attended regular elementary or high school (ELSTSCHL), a subset of respondents should have been in universe but were not asked the question and so their values were allocated by adding the modal age for ELSTSCHL to the respondent’s birth year. The flag for this item (ALSTSCHL) was set to 2 for affected cases, indicating cold deck imputation.

The values of the remaining variables for affected respondents, including EGEDTM, EHSYR, EPUBHS, ECOURSE1-ECOURSE7, and EPROGRAM, were set to “not in universe” as appropriate based on the corrected EEDUCATE and RCOLLVOC variables. To make the universes for topical module items consistent with both EEDUCATE and RCOLLVOC, respondents with less than a high school degree (EEDUCATE < 39) or who completed a vocational certificate but not high school (RCOLLVOC=6) were set to “not in universe” for EGEDTM and EHSYR. These respondents values were also set to “not in universe” for EPUBHS and EPROGRAM if they did not finish 9th grade and for ECOURSE1-ECOURSE7 if they did not finish 10th grade. The highest grade completed was determined from unedited data unavailable to the public.

If you have additional questions or would like information, please contact the Education and Social Stratification Branch at (301) 763-2464.