You asked (1) for information on the relationship between high-stakes educational testing and school dropout rates and (2) whether increased testing in Connecticut required by the No Child Left Behind Act has affected the dropout rate here.

SUMMARY

High-stakes education tests require a student to pass the test to graduate from high school or be promoted to a higher grade in school. Such tests are becoming more common around the country, with 20 states, including Massachusetts and New York, currently requiring students to pass a test to receive a high school diploma. Connecticut does not currently have such high-stakes graduation or school promotion tests.

One concern frequently raised about the use of high-stakes in public schools is that they lead more students, especially low achieving and minority students, to drop out of school. Many studies of the relationship between high-stakes tests and dropout rates have produced a mixed picture, with equal numbers of researchers finding a relationship and finding no relationship. Even researchers who find a relationship agree that, while such tests may be correlated with increased dropout rates, there is no evidence that the tests cause the higher rates.

Research on the issue is hampered by inconsistent state and school district methods of counting school dropouts, which makes state comparisons difficult. Another problem is the difficulty of isolating one factor in a student’s decision to leave school.

The statewide average four-year cumulative dropout rate in Connecticut fell from 17. 4% to 10. 8% from 1995 to 2002 (the latest year for which data is available). In Educational Reference Group (ERG) I, the seven districts that consistently have the lowest achieving students and the highest concentrations of low-income and minority students, rates dropped more than 40% between 1995 and 2001 before rising slightly for the class of 2002. The ERG I dropout rate held steady at more than double the statewide average rate from 1995 through 2002.

Since Connecticut does not have high-stakes testing and because the additional testing required under the federal No Child Left Behind (NCLB) Act does not begin until the 2005-06 school year, changes in the state’s dropout rate cannot be tied to any increase in testing.

HIGH-STAKES TESTS

A “high-stakes test” is a test that carries major consequences for the person taking it. Many states require public school students to pass a standardized test to get a high school diploma or be promoted to the next grade in school. Such tests are considered “high stakes” because a student who fails to pass cannot graduate from high school or is required to repeat a grade.

According to Education Week, 20 states currently require high school graduation tests, also known as “exit tests” or minimum competency tests (MCTs). Four states and the District of Columbia have established test requirements that take effect in future years. For example, California and Arizona require exit tests starting with the class of 2006. Washington’s test is effective with the class of 2008. Among northeastern states, New Jersey, New York, and Massachusetts have high school exit tests.

States with high school exit tests typically require students to take the tests in the 10^th grade. Students who fail are allowed to retake the tests in the 11^th and 12^th grades. Students can graduate from high school only if they pass.

Connecticut requires students in the 4^th, 6^th, 8^th, and 10^th grades to take state mastery exams. In addition, the federal NCLB Act requires student testing in the 3^rd, 5^th, and 7^th grades. The new testing is scheduled to start in the spring of 2006. Connecticut’s mastery tests are not high-stakes tests. They are used to evaluate schools and school districts and to determine which students need additional help, but state law expressly prohibits school districts from using mastery test scores as the sole criterion for promotion or graduation (CGS §10-14n).

STUDIES OF HIGH-STAKES TESTING AND DROPOUT RATES

The question of the whether there is any relationship between high-stakes tests and student dropout rates has been extensively researched, but the issue is far from resolved. Our survey of nine recent studies of this issue shows a mixed picture, with some researchers finding a relationship and others finding none. Even researchers who find a relationship between high-stakes tests and an increased dropout rate acknowledge that they can only show a correlation between the two. They cannot show that the tests cause increased dropout rates.

Our survey shows a consensus that research into the effect of high-stakes testing on dropout rates is hampered by inconsistent and inaccurate tabulations of dropout rates by states and school districts, which makes comparisons among states difficult. In addition, researchers agree that many factors contribute to a student’s decision to drop out of school and that it is difficult to isolate any one of them.

The conclusions of the studies we surveyed are briefly summarized below. Study summaries are listed in chronological order by year of publication. Full citations appear at the end of this report.

Griffin and Heidorn, Georgia Southern University and Florida Department of Education, 1996

This study looked at a random sample of 76,664 high school students from 14 school districts in Florida.

Although many have argued that competency tests are likely to have adverse affects on academically challenged, at-risk, and minority students, this study does not support these hypotheses. The researchers found that neither low achieving nor minority students showed any significantly increased likelihood of dropping out after an MCT failure. But failure on an MCT did seem to increase the likelihood that a student who is doing well academically will drop out.

The study could only establish linkages, not causal relationships, between MCT performances and dropping out.

Clarke, Haney, and Madaus, Boston College, 2000

This study of the 10 states with the highest and the 10 with the lowest 1986 dropout rates showed a strong link between attrition or dropout rates and use of high-stakes MCT programs.

The study found that, in schools with proportionately more students of low socio-economic status that used high-stakes MCTs, early dropout rates – between the 8^th and 10^th grades – were four to six percentage points higher that similar schools without high-stakes tests. The evidence was divided on whether there is any differential effect by race. Although the evidence is mainly correlational, the researchers believe it is suggestive enough to warrant further research to clarify the role of high-stakes testing in dropout decisions.

Schiller and Muller, State University of New York at Albany and University of Texas at Austin, 2000

This study was based on an analysis of data from the National Educational Longitudinal Study (NELS), a nationally representative longitudinal survey of students who were in the 8^th grade in 1988. Students were resurveyed in 1990, 1992, and 1994. The researchers also looked at state accountability policies in the 1990s.

They found that, overall, the greatest influence on high school completion rates is the extensiveness of testing and that the effect of the testing was positive. In those states with greater levels of testing, students identified by their teachers as at-risk of dropping out were more likely to graduate from high school than those at-risk students from states with less testing. The authors propose that this result is due to the fact that assessments provide students and teachers with performance benchmarks and with data on how well these are being met.

Rabinowitz, Zimmerman, and Sherman, WestEd, 2001

After surveying previous studies, the authors conclude that (1) there is not enough good data or research to settle the debate over whether high-stakes tests drive up dropout rates and (2) if we really want to understand dropout trends, the current focus on high-stakes testing is too narrow.

They note that national high school completion rates have remained relatively flat for the past 25 years, rising only 2% since 1973 and that dropout rates have remained consistently high in certain student groups, especially poor and low-income students.

It is almost impossible to demonstrate a causal connection between any single factor and a decision to quit school. Several factors correlate with dropping out, including being held back and being too old for a grade and having a history of poor academic achievement as reflected in grades and test scores. While a test may be the immediate trigger of a student’s behavior, the test results may simply aggravate or reflect a history of low performance.

Jacob, Harvard, 2001

This study is based on an examination of NELS data along with such school and student variables as median income, location, and single parent household.

Jacob found that dropout rates are roughly 6. 5% greater among students in the bottom-scoring achievement test quintile in states with high school graduation tests than in states without. He concluded that “students who live in testing states and who are low-achievers in math and reading appear to be significantly more likely to drop out than their peers in nontest states. ”

Reardon and Galindo, Pennsylvania State University, 2002

This study is also based on NELS data. The authors found that the presence of an 8^th grade promotion test requirement is strongly associated with increased probability of dropping out prior to the 10^th grade. However, they concluded that it is not clear if the relationship between testing and dropping out is causal.

Heubert, Columbia University Teachers’ College, 2002

Heubert’s article surveys other studies of the effects of high-stakes testing on dropout rates. The author notes that there is considerable debate about whether graduation testing causes increased dropout rates. The effect of testing is at best uncertain. On the one hand, it appears that many low achievers start to disengage from school well before graduation tests loom, reducing the effect of the test itself. On the other hand, failing a graduation test can increase the likelihood that low-achievers will leave school.

In addition, the current climate of accountability places new pressures on schools to increase student pass rates, which can lead to increased pressure on students to drop out or even to understated dropout rates. Study of this critical issue is complicated by a lack of uniformity among states in defining and counting dropouts.

Rankin, New York Community Service Society, 2003

New York State’s new requirement that all students pass the Regents examinations to earn a high school diploma was phased in over six years starting in 1996. Rankin links the requirement to higher dropout rates. New York City’s dropout rate reached 20. 3% in 2002, up from 15. 6% for the class of 1998. Dropout rates are much higher for some groups and individual schools: 37% for William H. Taft High School in the Bronx, 31% for Franklin K. Lane in Queens, and 36% for Bushwick in Brooklyn.

In addition to these New York City dropout statistics, Rankin cites a Cornell University survey that found that about 45% of the superintendents of low-performing state school districts outside the city reported an increase in dropouts two years after the stiffer Regents test requirements were introduced. (“NY Regents Standards resulting in more dropouts, school system administrators report in CU survey,” Blaine P. Friedlander, Jr. , Cornell Chronicle, June 27, 2002. )

Rankin found that the strongest predictor of student achievement in high school is preparation before entering high school. “It becomes increasingly harder to boost high school achievement for students who have fallen seriously behind by the 8^th grade,” she concluded.

Green and Winters, Manhattan Institute, 2004

This report compares graduation rates in states with exit exams with rates in those without. Authors also examined whether there was a change in graduation rate when a state adopted its exam and examined the pattern among states.

They conclude that high school exit exams do not significantly affect dropout rates. While some students may be unable to get diplomas because they fail to pass exit exams, other students are motivated by the exams and work harder. Some schools also serve their students better, thus increasing graduation rates. The groups offset each other, keeping rates steady.

CONNECTICUT’S DROPOUT RATES

Because Connecticut does not have high-stakes testing and because the additional testing required under the federal NCLB does not start until next year, it is not possible to correlate changes in the state’s dropout rate to the existence of high-stakes or extra NCLB tests. Connecticut’s four-year cumulative dropout rate fell between 1995 and 2002 both on a statewide basis and in the seven ERG I districts (see Table 1). (2002 is the latest year for which dropout data is available. ) ERG I, which comprises the state’s seven largest and poorest cities, consistently has the lowest overall academic achievement. It is here that any effects of high-stakes testing on school dropout rates would be felt the most.

Table 1: Cumulative Dropout Rates (%) for Classes of 1995 through 2002 by ERG

STUDY CITATIONS