A r c h i v e d I n f o r m a t i o n

Educational and Labor Market Performance of GED Recipients - February 1998

Appendix D: GED Follow-up Surveys

In Fall 1981, Cervero and Peterson (1982) conducted a national survey of successful GED candidates who had first taken the tests in the Spring of 1980. Like most of the GED follow-up surveys, this one had a low response rate and methodological shortcomings that caused the authors to urge caution in generalizing from the data.^[66] The study found that 18 months after taking the tests, approximately 60 percent of the respondents were employed full-time or part-time, and approximately 40 percent were out of the labor market or unemployed.^[67] Nevertheless, the proportion of GEDs working full-time for pay increased from 39 to 48 percent in those 18 months. All other employment status categories (e.g., unemployed, not in the labor force) fell between 1 and 3 percentage points. The perceived results of taking the GED were not as great as expected, but a significant proportion of respondents thought the tests had helped them. For example, 52 percent said it had helped them qualify for a job and 20 percent said it had helped them win a job promotion.

Kroll and Qi (1995) conducted another national follow-up survey, in this case of examinees (rather than just GED recipients). As in the previous study, there were methodological problems that caused the authors to advise against generalizing from the data.^[68] Among the strengths of this study was the fact that it provided both before and after measures and GED/non-GED comparisons. The results of the Kroll and Qi survey were rather similar to those of Cervero and Peterson. For example, they found that 2 years after taking the test, 61 percent of their respondents were employed full-time or part-time, and 39 percent were out of the labor market or unemployed. Their data also suggest some labor market benefits of GED test taking. Among GED recipients, those who had their current jobs before taking the tests reported post-test hourly wages that were 14 percent higher than pretest wages. Comparable GED examinees who did not pass the test had a 10 percent increase. GED recipients who got new jobs after the test (jobs that did not require a GED) had wages 17 percent higher than before, while their unsuccessful GED counterparts saw only a 3 percent increase. Finally, GED recipients who got new jobs that required GED certification reported wages that were 24 percent higher than their previous wages; for this group there could be no nonGED comparison.^[69]

Several state and local follow-up surveys used similar methods and also reported GED benefits. For example, the Iowa Department of Education (1992) surveyed GED recipients two, five, and ten years after their completion of the tests in 1980, 1985, and 1988. Controlling for inflation, reported personal income of GED graduates declined 6 percent between the time of passing the tests and the survey administration in 1990, while the mean income of all Iowans declined 10 percent between 1980 and 1990. (The comparison with all Iowans would have been better had it been limited to the GED 1980 cohort.)

In 1981, Moore (1982) surveyed both successful and unsuccessful candidates for the GED in Kentucky five years after they had taken the tests. Again, there were methodological problems, including a 28 percent response rate among successful GEDs in the sample and a very low 15 percent rate among the unsuccessful candidates. Moore's data show that successful GED candidates who responded to the survey were more likely to report before and after improvement in employment status (54 percent) than were unsuccessful candidates (30 percent).

Eight state or local follow-up surveys collected data on the employment status of GED recipients before and after they earned their credentials. In all eight of these studies we have data on two important status variables full-time employment and unemployment (table D-1). On average, across the 8 studies, the full-time employment rate of GEDs increased from 46.1 percent before certification to 57.0 percent afterward. In two of the studies (Darkenwald and Valentine 1985 and Iowa Department of Education 1992) GED full-time employment rates increased substantially more than statewide rates over the periods in question. The average GED unemployment rate across the 8 studies fell from 18.6 percent to 14.3 percent.

Table D-1.-Employment status of GED recipients

Percentage rate

Author,
GED employment status Before
At time of testing After
At time of survey

Reed 1985, Maryland

   Employed full-time 40 54

   Unemployed, looking 23 18

Carbol 1985, Alberta

   Employed full-time 53 58

   Employed part time 13 12

   Unemployed, looking 16 15

   Unemployed, not looking 8 3

Darkenwald & Valentine 1985, New Jersey

   Employed full-time 31 49

   Employed part time 21 16

   Unemployed, looking 35 22

   Unemployed, not looking 14 14

Goodwin 1991, New York

   Employed full-time 46 42

   Employed part time 13 14

   Unemployed, looking 12 16

   Not in labor force 27 28

Hayes 1991, Wisconsin

   Employed full-time 43 49

   Employed part time 16 18

   Unemployed, looking 15 13

   Unemployed, not looking 24 17

Iowa Department of Education, 1982

   Employed 54 71

   Unemployed, looking 19 9

Martin 1992, Wisconsin
1986 GEDs in 1990

   Employed full-time 53 73

   Employed part time 19 12

   Unemployed, looking 12 9

   Unemployed, not looking 16 6

Martin 1992, Wisconsin
1986 GEDs in 1990

   Employed full-time 49 60

   Employed part time 19 12

   Unemployed, looking 17 12

   Unemployed, not looking 16 15

NOTE: Martin's employment rates have been recalculated. Martin's rates for 1986 and 1989 used the same sample base, but the item nonresponse rate (the percentage of survey items not answered) seems to have been much higher for reports of employment status before the test than for current employment status. Using the entire sample as a base provided an inflated denominator for calculating employment rates before the test, thereby making pretest rates look much lower than they must have been. Here rates have been recalculated based on number of respondents to each question.
SOURCE: Studies in this table.

-###-

[Appendix C: Technical Issues]