Academic journal article Monthly Labor Review

Reconciling Educational Attainment Questions in the CPS and the Census

Academic journal article Monthly Labor Review

Reconciling Educational Attainment Questions in the CPS and the Census

Article excerpt

The decennial census and the Current Population Survey (CPS) are two of the most widely used household surveys of the U.S. population. Both surveys measure the educational attainment of their respondents. From the 1940s to the 1990s both surveys asked the respondents to report their highest grade attended and whether they had completed that grade, permitting the creation of the measure highest grade completed. Beginning with the 1990 census and the January 1992 CPS, both surveys have asked instead about individuals' highest degree received. An earlier research summary in the Review documented both the motivation for the change and the specific wording of the old and new questions.(1)

Researchers who previously used the continuous variable highest grade completed created from responses to the questions on education in the CPS and the census must now find new ways to represent educational attainment. Two possible methods of doing so are to impute a measure of the highest grade completed, using responses to the new question, and to recode responses to both questions, to provide comparable aggregate categories.(2) This report presents a summary account of these methods. The proposed methods are most useful to researchers using CPS and census microdata who wish to bridge the break in the question, but they also provide a benchmark for the magnitude of the differences in aggregate measurements of educational attainment using the old and the new questions. In general, the measure of the highest grade completed imputed from the new question is comparable to the highest grade completed as measured by the old question. And the overall average imputed highest grade completed is essentially the same as the average actual highest grade completed. However, the imputed highest grade completed overstates the highest grade completed for individuals at the lowest end of the educational distribution and slightly understates the highest grade completed for individuals at the highest end of the distribution. In addition, one can reconcile the old and new questions into four educational categories and make the distributions of educational attainment almost identical with the old and new questions.


The data used were a matched sample of the 1991 and 1992 March CPS'S The sample design of the CPS allows one to match a portion of individuals in the sample across those years and create a new sample that contains responses to both the old and new CPS questions.(3) Individuals in rotation groups one through four of the March 1991 CPS were asked the old question on educational attainment in 1991. These individuals appear as rotation groups five through eight in the March 1992 CPS and were asked the new question in 1992. To reduce the likelihood that the sample contains individuals whose true level of schooling changed between 1991 and 1992, the sample is limited to individuals 25 to 64 years old who were not enrolled in school in either year.(4) The question asked in the 1990 census is essentially the same as the new question in the CPS. The census question provides more detail for very low levels of education (that is, "no school completed," "kindergarten," and "nursery school" in the census, compared with "less than 1st grade" in the CPS) and combines "5th or 6th grade" and "7th or 8th grade" into one category. Table 1 lists the response categories for the new question in both surveys, along with the codes used in the microdata.

Table 1. Imputations of highest grade completed for new
education questions


   New question category                 CPS    Census

No school completed                      ...      01

Nursery school                           ...      02
Kindergarten                             ...      03

Less than 1st grade                       31      ...
First, 2nd, 3rd, or 4th grade             32      04
Fifth or 6th grade                        33      . … 
Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.