![[Table of Contents]](/icons/up.gif){kind=icon}
1 For example, the headline for the June 2, 1997 ACE News reads "Record Numbers Complete GED Tests, Earn High School Diplomas; Most Plan Further Education" (American Council on Education 1997). Also see the announcement for GED Profiles: Adults in Transition (GED Testing Service n.d.). Another example at the local level is the fall 1997 continuing education course catalogue of the Anne Arundel Community College, which indicates that successful GED completers can earn their "Maryland State High School Diploma" (Anne Arundel Community College 1997). Many states have adopted policies designating GED certification as a high school diploma.
2 For example, see Malizio and Whitney (1982), p. 10.
3 In the years during and immediately after the war, there were two GED tests, a high school equivalency or "high school level" test and a more difficult advanced placement "college-level" test. The college test became the College-Level Examination Program (CLEP), currently administered by the College Board. The high school test became the GED that is used today and that is produced and administered by ACE's GED Testing Service.
4 Early studies of GED outcomes in the late 1940s and early 1950s focused on the performance of veterans--all males--in postsecondary education. Later studies focused on civilians, both male and female, in college and in the labor market.
5 Data provided by the National Center for Education Statistics (NCES). The increase is probably not due to immigration; growth was strongest in the midwest.
6 Completions include tests completed the first time they are taken plus those completed by individuals retaking the GED. Obviously completions do not include tests administered but not completed in that year. In 1995, 829,904 tests were administered. Of those, 615,132 tests (74 percent of the total) were first-time completions; 108,767 (13 percent) were completions by individuals retaking the GED; and 106,005 (13 percent) were tests taken for the first time but not completed.
7 Since not all examinees complete the GED, the percentage difference between these two years is not quite as large as it appears. When the denominator is all tests administered, as it was in 1958, the percentage of those retaking the test will be somewhat smaller than when the denominator is all tests completed, as it was in 1995.
8 Kroll and Qi's rate is based on tests administered in the United States and the Territories, while table 1 reports on all GED tests, including those in Canada.
9 The essay is the chief difference between the second and third versions of the test and the exception to the multiple-choice approach.
10 Note that the early pass rates of veterans without high school diplomas (e.g., 92 percent) exceeded those of the high school seniors whose performance was used to norm the tests. The reasons for this difference are not clear but may be related to military experience.
11 Probabilities have been calculated from data in Quinn 1997a and GED Testing Service 1993a. Quinn's (1997a) report provides the number of questions on each test and the number greater than chance required to meet the minimum passing score. It does not provide the number correct expected by chance or the required minimum number. The 1995 GED writing test includes an essay that does not involve multiple choice. Chance probabilities on this test cannot be computed from the data available.
12 On the entire five-test battery, the lowest possible standard score is 100 points (20 points on each of five tests). The Princeton Review (Martz 1995) calculates that in 1995, guessing on every question would have added 47 standard points to 100, making up a little over one-third of the difference between the lowest possible score and the minimum required for passing.
13 Seniors who do not graduate are excluded from the sample after the fact.
14 Of course, both groups answered substantially more questions correctly than did the GED examinees who failed the regular test battery.
15 Although percentages of correct answers and the means for the two groups are similar, the distributions are not, as Cameron and Heckman (1993a) have observed. The high school graduates' score distribution is approximately normal, while the GED recipients' distribution is truncated at the cut point for passing the test. The left tail of the distribution is lopped off, resulting in a concentration of GED recipients around the mean for all GED examinees. On another point, Quinn (1997a, 1997b) observes that, unlike high school seniors who take the tests, GED examinees can take the tests more than once and many are coached on the tests. However, the GED examinees in the Enger and Howerton study took the test only once, and many high school seniors are coached on other standardized tests such as the SAT.
16 Quinn (1997a) notes that Lindquist's data from early years of the test show that the majority of Iowa students entering the ninth grade were passing each subtest at the passing norms established for veterans. Moreover, the 1977 Educational Testing Service norming study found that 73 percent of ninth-graders could pass all five of the GED tests under the GED standards at the time--35 or 45.
17 For example, see Gary S. Becker's Human Capital (1964) and Jacob Mincer's Schooling, Experience, and Earnings (1974).
18 According to a major evaluation of adult literacy programs (Development Associates 1994), participants in English-as-a-Second Language (ESL) programs received a median 113 hours of instruction, compared to 35 hours for those in Adult Basic Education (ABE) and 28 hours in Adult Secondary Education (ASE). Moreover, many immigrants take some combination of ESL, ABE, and ASE classes before taking the GED test. In general, the role of the GED in the acculturation of immigrants is an important subject on which little systematic research has been conducted.
19 This rough estimate is derived from Cheryl M. Kane, Prisoners of Time (1994), the report of the National Education Commission on Time and Learning. The commission found that American students spend an average of 1,000 hours per year in school and an estimated 41 percent of this time was spent on core academic subjects. Elsewhere the commission estimated that the amount of time required by states to be spent on core subjects was 1,460 over 4 years, or an average of 365 per year.
20 Note that the data on preparation time apply to GED examinees; no comparable data on GED recipients are available.
21 There are some incentives to do well on the test, rather than merely passing it. Many colleges include GED scores among the factors they consider in admissions, and some states base scholarships in part on GED scores. Examinees wanting to enroll in these institutions or to win the scholarships do have incentives to do well on the tests.
22 The Armed Forces Qualifying Test (AFQT) is a subset of the Armed Services Vocational Aptitude Battery (ASVAB).
23 The GED data in this comparison reflect the mean NALS scores of the GED recipients in Baldwin's 1993 sample of GED test takers. A substantial number of these recipients, we can safely assume, later went on to college (Baldwin et al. 1995, p. 20). The NALS scores of all high school graduates are weighted averages for those with high school diplomas and above (Kirsch et al. 1992, pp. 119-121). The NALS scores for adults are weighted averages for all NALS examinees. These adults include GED recipients, but removing GEDs with no further education from the calculations results in minimal changes to the estimates. The data did not enable us to remove GEDs who received additional education, but their impact on the estimates for all adults is probably small or negligible. There may be a slight upward bias in the scores of the GED test takers, but we cannot tell from these data. Only GED examinees who completed the five-test battery at one sitting were included in the survey. It is possible that non-completers, who might have retaken and passed the test later, would have had lower scores than those who passed the first time.
24 Though the comparison with high school graduates in table 8 seems to contradict the first part of table 7, it does not. In table 8, many of the GEDs will go to college, but the high school graduates have no college experience, by definition. The GED examinees have not yet had a chance to go to college. The high school graduates have had a chance to go to college but have not done so.
25 Of course, the GEDs and high school graduates scored better on these tests than did high school dropouts (not shown in table 8).
26 Maloney's method of comparing each group with the whole sample yields smaller differences than one would find by comparing each group with the other.
27 To make this estimate, we calculated the weighted percentage of GED test takers who passed the test in the years 1981-86 inclusive, using the data in table 1. The result was 72.9 percent. We also used data from Kolstad and Kaufman (1989), who estimated that 30.7 percent of the dropouts in the High School and Beyond sophomore cohort obtained a GED between 1980 and 1986. This is very close to the estimate by Murnane, Willett, and Boudette (1994). Using NLSY data from the period 1979-91, they concluded that one-third of the dropouts in their cohort obtained a GED. If GED recipients equal 30.7 percent of dropouts and 72.9 percent of test takers, then test takers equal .307/.729 of dropouts, or 42.1 percent.
28 This estimate is in Baldwin's August 8, 1997 comments on the draft of this study. Obviously an adult dropout chosen at random is much less likely to have taken the test than a younger person without a diploma. One reason is that test administration was not widespread when many adults who are now older were in their teens and twenties.
29 Correspondence from Janet Baldwin, director of research, GED Testing Service. April 15, 1997.
30 A small percentage of postsecondary institutions accept nonhigh school graduates who do not have GED degrees. Many of these have other alternative credentials, such as adult high school diplomas or acceptable CLEP scores.
31 Class rank is one criterion for admission that 4-year colleges frequently use. The GED testing service has attempted to convert GED scores into class-rank equivalents.
32 Outside the United States, a survey of 15 postsecondary institutions conducted by the Department of Continuing Education in Saskatchewan found that 13 accepted the GED certificate. In 6 of the 13, the GED alone was sufficient for admission. (Jantzen and Quigley 1982).
33 Cited in Carson (1986).
34 Here and elsewhere in this paragraph, item nonresponse accounts for the difference between the percentages presented and 100 percent.
35 Murnane et al. also found that for females, GED attainment was associated with a slight increase in the probability of receiving company training and more substantial increases (6-13 percent) in the probability of receiving noncompany training (often in government-sponsored training programs). GED attainment was also associated with increased male participation in noncompany training, but not in company training.
36 Unpublished data available on the Internet (http://www.ed.gov/NCES/pubs/r94/r9412t23.html).
37 The "other outcomes" in table 12 include 5 percent of high school graduates and 10 percent of GED recipients who enter the military in Cameron and Heckman's study. It also includes participants in adult basic education and GED classes in Behal's study.
38 Altogether, the proportions receiving education and training across the categories in Cameron's study total 72 percent, but only 51 percent of the GED women in this study had either some college or some training after passing the tests.
39 The Cameron/Heckman data in table 12 reflect the first educational decisions after attainment of the GED. The Cameron/Heckman data in table 11 reflect all educational decisions after attainment of the GED.
40 Data are from U.S. Department of Education (1996a), table 10 and U.S. Department of Education (1995b), table 165.
41 Three technical points bear on the following analysis: (1) College grades are not standardized the way tests are, but within the last several decades most colleges have adopted a four-point scale: A=4, B=3, C=2, D=1, and F=0. Grades based on other metrics can often be converted to this scale; we have done so wherever possible. (2) The studies of grade point averages examined here are based on student records, and the treatment of course withdrawals in calculating grade point averages varies somewhat across institutions. As a rule, students may withdraw from a course for any reason without GPA penalty until fairly late (e.g., within the last two weeks or a month of the course's conclusion) and their performance in the course is not counted in their GPA. However, late withdrawal is usually treated the same as a failing grade and is counted as a zero. (3) For studies of institutions that use a four-point grading system (or one that can be converted to it) and that conform to the treatment of withdrawals described above, it is possible to provide summary statistics for comparative purposes by averaging GPAs.
42 Some of the studies reviewed aggregated statistics for more than one college. The data have been weighted to reflect the number of colleges in the studies.
43 Along the same lines, a study by Keller (1958) shows that GEDs and high school diploma holders who graduated from East Tennessee State had about the same average grades across four subjects: American history, American literature, composition and rhetoric, and mathematics. The mean grade for GEDs was 2.53, that for high school graduates, 2.49.
44 The Army is a good example. People who enlist in the Army have a variety of goals in mind. Some may want to leave after their goals have been met but before their term of service has been completed. Their early departure is counted as attrition, regardless of their individual goals or reasons for leaving.
45 It is reasonable to assume that most students who enroll in certificate-granting one-year vocational programs intend to complete the programs. The same is probably true of most students who initially enroll in degree-granting programs in 4-year colleges.
46 The current problem may not be quite as serious as these data indicate. If the Current Price Index (CPI) overstates inflation, as many economists believe, the earnings declines would be less marked than they appear. Moreover, the current full-employment labor market may bid up wages. Nevertheless, the long-term declines are probably real and a matter of concern.
47 They also reflect the wages at which employees are willing to work.
48 Averaging the data for time worked shows that male dropouts worked 0.73 percent more time (hours, weeks) per year than male GEDs in simple comparisons and 3 percent more in controlled comparisons. In effect, they work the same number of hours. At ages 25 and 28, male dropouts had 15.1 percent and 21.5 percent more work experience than GED males. Male dropouts at these ages spent 28.2 percent and 42.2 percent more time working on the current job than comparable GED males. The finding that male dropouts work about the same amount of time per year as GEDs is consistent with the finding of Murnane et al. (1995) that the GED did not have a significant impact on the number of hours worked per year.
49 Maloney (1993), p. 27.
50 This study used repeated cross-sectional analyses to show the earnings effects of the GED over time. It did not directly compare the earnings of GEDs (and of controls) before and after the GEDs attained their credential in 1990. Rather, it examined differences in the earnings of GEDs and controls in each of the 2 years before and 5 years after 1990. These cross-sectional comparisons provided a longitudinal assessment of earnings effects.
51 They find dropouts with 10 years of schooling earn significantly more (10 percent) than dropouts with only 9 years. Similarly dropouts with 11 years earn 18 percent more than dropouts with only 9 years. There is a similar pattern for GED recipients, although two of the four coefficients are statistically insignificant. See Cameron and Heckman (1993), table 14.
52 Interpretation of the ASVAB effect is clouded by the fact that the variables for marital status and children were added at the same time as the ASVAB variable.
53 To use ASVAB in this way, attainment of the ASVAB scores had to follow GED certification and high school graduation. However, the ASVAB tests were administered to NLSY participants of different ages (15-23) at different stages of their education in 1980. Maloney therefore "updated" the ASVAB scores of the respondents to the time of the followup survey, using multivariate techniques to predict what an individual's score would be in 1987. He then used the updated ASVAB score as the mediating variable in estimating indirect effects of the GED and the high school diploma. Dropouts were the omitted category.
54 Here the negative signs indicate wages lower than those of high school graduates.
55 Not in table B-7 because the coefficients were not reported.
56 Note that this specification assumes that the return to additional tenure and experience is the same for all three groups. Below we review the evidence from Murnane et al. (1995) that finds that the return to tenure increases for GED recipients after they receive certification.
57 The age dummies captured the increase in earnings with age across the three groups. The year variable (1979 or 1980) captured general economic conditions in that year.
58 One limitation of this approach is that the wages of some individuals are represented only after receipt of the GED, because they obtained it early in the sample period, or only before receiving the GED, because they obtained in late in the period.
59 Richard Murnane and John Willett collaborated in both of these studies.
60 The surveys have typically been conducted by mail, and the response rates have usually been low--between 20 percent and 40 percent. Some of the reports have included nonresponse analyses examining differences between respondents and nonrespondents in demographic and other characteristics. The subjective nature of the respondents' perceptions and judgments about whether they had benefitted occupationally from the GED is another limitation of this approach.
61 Estimates by Janice Laurence of the Human Resources Research Organization and Ed Schmitz of Navy Recruiting using GED and Bureau of Labor Statistics data.
62 In these studies, the category high school graduate also includes individuals who have some college in addition to a regular high school diploma.
63 Although AFQT scores may not be a useful tool for reducing attrition, a study by Lyle (1988) showed that GED scores and certain ASVAB subscores (auto and shop knowledge, mathematics knowledge, mechanical comprehension, electronic information) are negatively correlated with attrition. Among 3,979 army enlistees in the 1984 fiscal year holding GED certificates at the time of entry, the attrition group had significantly lower average GED scores than the nonattrition group (264.16 compare to 267.24). Although the difference in average GED scores was small (less than a 10th of a standard deviation), the author recommended that GED scores rather than the certificate alone become an integral part of the screening process of recruits for service in the Army.
64 Persistence, as used throughout the study, is an individual and organizational outcome, not a personal characteristic. It is partly a function of personal characteristics--such as perseverance, work ethic, and adaptability--that enable one to work well within an organization. It is also a function of factors, such as low income, single parent status, and having dependent children, that may adversely affect participation in an organization.
65 Quinn (1997b) cites a report of student resentment at being pulled from regular classes in a Wisconsin school during the two days of GED testing (p. 44). The quotation is from the same source and page.
66 Among other things, the survey sample was self-selecting, because it was drawn only from GED Test takers who agreed at the time of the test to participate in the planned survey. The survey response rate was only 24 percent. Nonresponse analysis showed that respondents were older, had higher GED test scores, had slightly less postsecondary education, and were more likely to be female than nonrespondents. The author concluded that both groups represented a single population but urged caution in generalizing from the data.
67 Two different calculations in the study estimated the "employed" rate at 59 percent and 61 percent.
68 As in Cervero and Peterson, the survey sample was drawn only from GED Test takers who agreed at the time of the test to participate in the future survey. Even so, the response rate was low (22 percent).
69 Behal's 1982 follow-up survey of GED examinees has no comparative data on labor market outcomes, but it does have data on employment status and occupation of respondents at the time of the survey. Despite a low response rate (35 percent), Behal's estimate of the GED examinees' unemployment rate (24 percent) is similar to Cervero and Peterson's (21 percent) and Kroll and Qi's (also 21 percent).
-###-
[Bibliography]