number of states; they used the staggered implementation of exit exams to examine their effect on several different outcomes.
Study 4A looked at long-term trend NAEP results in reading and mathematics for eighth and twelfth grades from 1971 to 2004: it found no effect of the introduction of high school exit exams for either lower or higher performing students (Grodsky et al., 2009, Tables 3-4). Over four combinations of subject and grade, the average effect size was 0.00 standard deviations, evenly divided between small positive and negative effects, and none was statistically significant.16
Two studies looked at effects on graduation rates. Study 4B used state graduation rates from 1975 to 2002: it found that states adopting more difficult exit exams showed a statistically significant decrease in graduation rates of 2.1 percentage points (Warren et al., 2006, Table 2).17 This result came from an analysis using Common Core Data that distinguished a high school diploma from a GED (general education development) certificate. An alternate analysis based on census data that used a graduation measure that combined high school diplomas and GED certificates showed no effect of exit exams: this result suggests that the requirement may shift some students from a obtaining a diploma to obtaining a GED.18
Study 4C used individual census data for 2000 with state fixed effects that identified changes resulting from exit exam requirements: it found that the requirements for more difficult exams were associated with a decrease in high school graduation—including both diplomas and GED certificates—of about 0.6 percentage points (Dee and Jacob, 2007, Table 6-2).19 Over three different model specifications, all estimates were negative, and two of them were statistically significant. For the less difficult exit exams, Dee and Jacob (2007) found an average decrease of 0.3 percentage points, with only one of the three estimates statistically significant.
16We used the coefficients in the “HSEE” line of Table 3 of Grodsky et al. (2009, Table 2), dividing each by the standard deviation for reading and math scores, respectively.
17We used the estimates based on the Common Core Data with the model that distinguishes between minimum competency and more difficult exit exams (Warren et al., 2006, Table 2).
18Outcomes for high school graduates with a regular diploma are substantially better than those with a GED so it is better to distinguish the two outcomes (National Research Council and National Academy of Education, 2011).
19We averaged the three estimates in the “More difficult exit exam” line of Table 6-2 of Dee and Jacob (2007) for columns (3), (4), and (5).