 |
|
|
|
|
|
|
|
|
|
|
|
|
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
|
|
||||||||||||||||||||||||||
|
|
Find more Kappan articles in the Subscribe today to access complete current issues online! RESEARCH: Dropping In on Dropouts AS WE come to the end of another school year, it seems appropriate to take another look at students who have brought their school careers to a premature end -- dropouts. Dropouts have become big news lately. The National Governors Association made them an issue in 2005. The 9 April 2006 cover of Time reads, "Dropout Nation." And a methodologically pathetic Gates Foundation-funded report from Peter Hart, The Silent Epidemic, interviewed dropouts to see why they left school. The current conventional wisdom holds that the high school graduation rate is about 70%, but only around 50% for blacks and Hispanics and 25% for minorities in the cities. The December 2005 Research column summarized Paul Barton's report of the conventional wisdom, One Third of a Nation: Dropouts Rising, Opportunities Declining. I've always had difficulty accepting these numbers. First, anyone who's ever worked in a school district knows how fuzzy dropout data are. And in several state accountability systems, the numbers have had high stakes attached to them, which has led to their further corruption. The most notorious instance of this, of course, was Sharpstown High School in Houston, which had more than 1,000 ninth-graders, fewer than 300 seniors, and zero dropouts. When assistant principal Robert Kimball called public attention to this anomaly, he was shunned and had to seek employment elsewhere. (He sued and received $90,000 from the district.) Second, the Digest of Education Statistics 2004 states that in 2003 there were 19.24 million Americans aged 14-17. If only 70% of that number graduate from high school, it would mean that every year we are dumping 1,443,000 kids onto the streets without diplomas. Over a decade, I would think these millions would have shown up in reports on social indicators: employment rates, crime rates, etc. But I've seen none of that. Now come Lawrence Mishel and Joydeep Roy of the Economic Policy Institute (EPI) to say that the conventional wisdom is wrong. Mishel first made his case in Education Week on March 8 of this year. Then he backed that up in a book with co-author Roy titled Rethinking High School Graduation Rates and Trends, published in April 2006 by EPI. The usual method of calculating a graduation rate or a dropout rate is to compare the size of a ninth grade to the number of diplomas granted to that cohort four years later. One immediate problem with this method is that, as a matter of policy, schools have been increasing the number of students they hold for two years in the ninth grade. For example, ninth-grade retention rates in Massachusetts rose from 6% in 1994 to almost 9% in 2001. Graduation rates would be significantly higher, says Mishel, if the ratio were based on the number of entering ninth-graders in a particular year, but no one calculates this. Jay Greene, Marcus Winters, and Christopher Swanson attempted to compensate for this ninth-grade bulge by using an average of eighth, ninth, and 10th grades. Their rejoinder to Mishel's essay appeared in the March 29 Education Week. Mishel and Roy contend that, while this modification attenuates the bias, it doesn't eliminate it. Using longitudinal data from Florida and New York City, Mishel and Roy obtain much higher graduation rates than Greene and his colleagues. Mishel and Roy think the usual means of calculating dropouts is fundamentally flawed. They contend that, at the very least, the method has never been checked for accuracy. Schools send figures to the district, the district sends numbers to the state, and the state forwards them on to the National Center for Education Statistics, which dutifully reports them in its Common Core of Data (CCD). There are lots of places for inaccuracy to slip in. For instance, the April 9 issue of Time reported that for years Shelbyville, Indiana, counted as graduates everyone who left school and promised to take the GED. Mishel and Roy prefer to look at data sources whose characteristics are better known. They rely largely on data from the National Education Longitudinal Study (NELS:88) and from the Census Bureau's Current Population Survey (CPS). Greene rejects the CPS as a source of data because it relies on self-reports. He fears that people would be reluctant to say that they dropped out or that their children were dropouts. In his Education Week essay, Mishel rejected this objection with the -- to me not compelling -- argument that the CPS is used successfully for other purposes. I would worry more about lack of knowledge than about lying as a source of error in the CPS figures. The lone question about educational attainment that the CPS asks is if the respondent got a high school diploma or the equivalent, with only a GED specifically mentioned as an equivalent. Just one person per household responds for all people in the household. Would that person know if a "certificate of completion" (given in some states to those who fail a high school exit test) is "equivalent"? Is a diploma awarded to a special education student "equivalent" to a regular diploma? (The problems with GED inclusion/exclusion are dealt with below.) Greene also finds a problem in the CPS data because the CPS excludes institutionalized groups and the military. Mishel and Roy agree that those in prisons and mental hospitals have lower educational attainment, but they feel that this is balanced by the exclusion of the military, which contains few dropouts. A larger problem is controlling for undercoverage of certain groups. Those that the CPS survey has trouble locating -- especially low-income young black males -- are likely to be less well educated, although no good empirical data exist about rates. Greene declares that dropout expert Phillip Kaufman "estimated that if we made the reasonable assumption that 50% of those undercovered by the CPS were dropouts, we would end up with a completion rate of 80.4%. If we then further excluded GED recipients from that estimate, we would get much closer to the estimate of 70%," which is Greene's estimate. Actually, Kaufman did not say that assuming a 50% dropout rate was "reasonable." He said only that it was reasonable to assume that the undercovered are more likely to be dropouts. He then set forth two scenarios, one depicting the impact on overall dropout rates if 50% of the undercovered were dropouts; the other, if all of the undercovered were dropouts. It is a bit curious that Greene seeks Kaufman as an ally. In the same paper, Kaufman accepts estimates by the CPS over those of the CCD because they align so closely to those of NELS:88. The CCD estimates stand alone in their lower completion rates. Mishel and Roy deal with these two problems with a census data subset called Microdata, which does include institutionalized people and those in the military. They conclude that the two exclusions in the CPS data mostly offset each other. The Microdata lead to a small downward adjustment for black males. Mishel and Roy find that the greatest bias in the usual reporting techniques comes from a failure to exclude recent immigrants. The exclusion of recent immigrants from the calculations has little impact on black or white completion rates, but an enormous effect for Hispanics. For civilian, noninstitutionalized Hispanics, the graduation rate jumps from 56.9% to 73.4% when recent immigrants are excluded. "Recent immigrants" for this purpose are defined as people aged 25-29 who have come to the U.S. less than 15 years ago, meaning they would have been 10 to 14 years old on arrival. While partial to the CPS, Mishel and Roy hold up NELS:88 as the gold standard for estimating graduation rates, although they also use data from other longitudinal studies: High School and Beyond, the National Longitudinal Study of the High School Class of 1972, and the National Longitudinal Surveys from the Bureau of Labor Statistics. NELS:88 avoids many problems of the CPS data because it actually tracked individual students and conducted transcript verification studies. NELS data show the percentage of students graduating on time, the percentage who had graduated two years later, and a "final status" rate when the students were 26 years old. The data reveal an on-time graduation rate of 82.4% for whites, 63.2% for blacks, 66.1% for Hispanics, and 93.4% for Asians. Two years later, those rates had risen to 84.9%, 73.5%, 72.4%, and 94.7% respectively. In 2000, when the initial cohort was 26 years old, the figures looked like this:
The column headed "GED" is problematic. People differ over how to describe students who obtain the General Education Development certificate. In an earlier Manhattan Institute paper, Greene and Winters reject it, holding that it is "inappropriate to count GED recipients as graduates in graduation rate calculations because doing so credits the very schools that failed to graduate these students with their successes. The primary reason we calculate graduation rates is to evaluate the performance of schools. But GED recipients are not truly 'graduates' of any particular school. They are high school dropouts who later in life took it upon themselves to earn an alternative certificate." This statement strikes me as both simplistic and fatuous. As Mishel and Roy note, dropping GED completers would be valid only if no student opting for a GED would get a diploma in a GED-less world. In addition, nonschool factors (pregnancy, jobs, the need to help support the family) often play a role in a student's decision to leave school and obtain a GED. Mishel and Roy point out that GEDs have market value -- those with them earn less than graduates but more than dropouts. And, with students under 21 now permitted to take the GED, they have become for some people something of an alternative diploma. This is particularly the case under NCLB, where a low-achieving student might threaten a school's ability to make AYP (adequate yearly progress). In terms of trends, different researchers find different trend lines. Walt Haney and his colleagues at Boston College find graduation rates declining. Greene finds them stagnant. Mishel and Roy find them rising. From 1962 to 1979, rates for blacks and whites improved, with blacks gaining on whites. After 1979 the white rate inches up past 90% (a figure Greene finds impossible to believe), while blacks continue to gain, landing in the upper 80s by 2003. These different findings largely reflect different databases and different methods. Haney has mostly used ninth-grade class size compared to graduating class size. Greene uses an average of grades 8, 9, and 10. Both rely on CCD information. Mishel and Roy rely on the CPS data and various longitudinal data, especially NELS:88. All of this is not to say that Mishel and Roy are Pollyannas about graduation rates, especially rates for blacks and Hispanics. And when they delve into rates for cities, some statistics are stark indeed. In Chicago, even by age 19, the graduation rate for black males is under 40%. For black females, it is 57%. Only for Asian students of both genders and for white females does the rate exceed 70%. Whatever the true dropout rate, the recent attendant publicity guarantees that the topic will be news for the foreseeable future. GERALD W. BRACEY is an associate for the High/Scope Foundation, Ypsilanti, Mich. His most recent book is Reading Education Research: How to Avoid Getting Statistically Snookered (Heinemann, 2006). He can be reached at gbracey1@verizon.net. |
||||||||||||||||||||||||
|
|
