Joshua Goodman

Probabilistic Feature Grammars

We present a new formalism, probabilistic feature grammar (PFG). PFGs combine most of the best properties of several other formalisms, including those of Collins, Magerman, and Charniak, and in experiments have comparable or better performance. PFGs generate features one at a time, probabilistically, conditioning the probabilities of each feature on other features in a local context. Because the conditioning is local, efficient polynomial time parsing algorithms exist for computing inside, outside, and Viterbi parses. PFGs can produce probabilities of strings, making them potentially useful for language modeling. Precision and recall results are comparable to the state of the art with words, and the best reported without words.

Parental Socioeconomic Status, Child Health, and Human Capital

Abstract Investments in education pay off in the form of higher future earnings, and differences in educational attainments explain a significant fraction of the adult variation in wages, incomes, and other outcomes. But what determines a childs educational success? Most studies point to family background as the primary factor. But why does background matter? While many aspects are no doubt important, research increasingly implicates health as a potentially major factor. The importance of health for education and earnings suggests that if family background affects child health, then poor child health may in turn affect education and future economic status.

The Wages of Sinistrality: Handedness, Brain Structure and Human Capital Accumulation

Left- and right-handed individuals have different brain structures, particularly in relation to language processing. Using five data sets from the US and UK, I show that poor infant health increases the likelihood of a child being left-handed. I argue that handedness can thus be used to explore the long-run impacts of differential brain structure generated in part by poor infant health. Even conditional on infant health and family background, lefties exhibit economically and statistically significant human capital deficits relative to righties. Compared to righties, lefties score a tenth of a standard deviation lower on measures of cognitive skill and, contrary to popular wisdom, are not over-represented at the high end of the distribution. Lefties have more emotional and behavioral problems, have more learning disabilities such as dyslexia, complete less schooling, and work in less cognitively intensive occupations. Differences between left- and right-handed siblings are similar in magnitude. Most strikingly, lefties have six percent lower annual earnings than righties, a gap that can largely be explained by these differences in cognitive skill, disabilities, schooling and occupational choice. Lefties work in more manually intensive occupations than do righties, further suggesting that lefties’ primary labor market disadvantage is cognitive rather than physical. Those likely be left-handed due to genetics show smaller or no deficits relative to righties, suggesting the importance of environmental shocks as the source of disadvantage. Handedness provides parents and schools a costlessly observable characteristic with which to identify young children whose cognitive and behavioral development may warrant additional attention.

Middle School Math Acceleration and Equitable Access to 8th Grade Algebra: Evidence from the Wake County Public School System

Taking algebra by eighth grade is considered an important milestone on the pathway to college readiness. We highlight a collaboration to investigate one district’s effort to increase middle school algebra course-taking. In 2010, the Wake County Public Schools began assigning middle school students to accelerated math and eighth-grade algebra based on a defined prior achievement metric. This policy reduced the relationship between course assignment and student characteristics such as income and race/ethnicity, while increasing its relationship to academic skill. The policy increased the share of students on track for algebra by eighth grade. Students placed in accelerated math were exposed to higher-skilled peers but larger classes. Future work will assess impacts on subsequent achievement and course-taking outcomes.School districts across the country have struggled to increase the proportion of students taking algebra by 8th grade, thought to be an important milestone on the pathway to college preparedness. We highlight key features of a research collaboration between the Wake County Public School System and Harvard University that have enabled investigation of one such effort to solve this problem. In 2010, the district began assigning middle school students to accelerated math coursework leading to 8th grade algebra on the basis of a clearly defined measured of prior academic skill. We document two important facts. First, use of this new rule greatly reduced the relationship between course assignment and student factors such as income and race while increasing the relationship between course assignment and academic skill. Second, using a regression discontinuity analytic strategy, we show that the assignment rule had strong impacts on the fraction of students on track to complete algebra by 8th grade. Students placed in accelerated math were exposed to higher-skilled peers but larger class sizes. We describe future plans for assessing impacts on achievement and high school course-taking outcomes.

Skills, Schools and Credit Constraints: Evidence from Massachusetts

Low college enrollment rates among low-income students may stem from a combination of credit constraints, low academic skill, and low-quality schools. Recent Massachusetts data allow the first use of school district fixed effects in the analysis of credit constraints, leading to four findings. First, low-income students in Massachusetts have lower intended college enrollment rates than higher income students but also have dramatically lower skills and attend lower-quality school districts. Second, inclusion of skill controls greatly reduces but does not eliminate this intended enrollment gap. Third, inclusion of school district fixed effects has little further impact, with low-income students eight percentage points less likely to intend enrollment than higher income students of the same skill and from the same school district. Fourth, medium- and high-skilled low-income students appear the most constrained. State governments could use the methods employed here to target financial aid more efficiently.

Access to 4-Year Public Colleges and Degree Completion

Does access to 4-year colleges affect degree completion for students who would otherwise attend 2-year colleges? Admission to Georgia’s 4-year public sector requires minimum SAT scores. Regression discontinuity estimates show that access to this sector increases 4-year college enrollment and college quality, largely by diverting students from 2-year colleges. Access substantially increases bachelor’s degree completion rates for these relatively low-skilled students. SAT-retaking behavior suggests students value access to 4-year public colleges, though perhaps less than they should. Our results imply that absolute college quality matters more than match quality, and they suggest potential unintended consequences of free community college proposals.

Bankruptcy Law and the Cost of Credit: The Impact of Cramdown on Mortgage Interest Rates

AbstractRecent proposals to address housing market troubles through principal modification could increase the cost of credit in the mortgage market. We explore this possibility using historical variation in federal judicial rulings regarding whether Chapter 13 bankruptcy filers could reduce the principal owed on a home loan to the home’s market value. The practice, known as cramdown, was definitively prohibited by the Supreme Court in 1993. We find that home loans closed during the time when cramdown was allowed had interest rates 12–16 basis points higher than loans closed in the same state when cramdown was not allowed, which translates to a roughly 1 percent increase in monthly payments. Consistent with the theory that lenders are pricing in the risk of principal modification, interest rate increases are higher for the riskiest borrowers and zero for the least risky and higher in states where Chapter 13 filing is more common.

Can States Take Over and Turn Around School Districts? Evidence From Lawrence, Massachusetts

The Every Student Succeeds Act (ESSA) requires states to identify and turn around struggling schools, with federal school improvement money required to fund evidence-based policies. Most research on turnarounds has focused on individual schools, whereas studies of district-wide turnarounds have come from relatively exceptional settings and interventions. We study a district-wide turnaround of a type that may become more common under ESSA, an accountability-driven state takeover of Massachusetts’s Lawrence Public Schools (LPS). A differences-in-differences framework comparing LPS to demographically similar districts not subject to state takeover shows that the turnaround’s first 2 years produced sizable achievement gains in math and modest gains in reading. We also find no evidence that the turnaround resulted in slippage on nontest score outcomes and suggestive evidence of positive effects on grade progression among high school students. Intensive small-group instruction over vacation breaks may have led to particularly large achievement gains for participating students.

Intensive College Counseling and the Enrollment and Persistence of Low-Income Students

Though counseling is one commonly pursued intervention to improve college enrollment and completion for disadvantaged students, there is relatively little causal evidence on its efficacy. We use a regression discontinuity design to study the impact of intensive college counseling provided by a Massachusetts program to college-seeking, low-income students that admits applicants partly on the basis of a minimum grade point average requirement. Counseling shifts enrollment toward four-year colleges that are less expensive and have higher graduation rates than alternatives students would otherwise choose. Counseling also improves persistence through at least the second year of college, suggesting a potential to increase the degree completion rates of disadvantaged students.

Intensive Math Instruction and Educational Attainment

We study an intensive math instruction policy that assigned low-skilled ninth graders to an algebra course that doubled instructional time, altered peer composition and emphasized problem solving skills. A regression discontinuity design shows substantial positive impacts of double-dose algebra on credits earned, test scores, high school graduation, and college enrollment rates. Test score effects underpredict attainment effects, highlighting the importance of long-run evaluation of such a policy. Perhaps because the intervention focused on verbal exposition of mathematical concepts, the impact was largest for students with below-average reading skills, emphasizing the need to target interventions toward appropriately skilled students.