"Attending a college with a 100 point higher average SAT is associated with 3to 7% higher earnings later in life." (e.g., Kane, 1998, cited by Dale, and Krueger 1999, 1) as higher education constitutes up 40% of total U.S. educational expenditures, understanding "the impact of selective colleges on students' labor market outcomes is central for understanding the role of human capital," Dale, and Krueger (1999, 1) stress.
Previous literature cited by Dale, and Krueger (1999, 1) includes work by "Hunt (1963), Solnion (1973), Wales (1973), Solmon and Wachtel (1975), and Wise (1975),...Brewer and Ehrenberg (1996), Behrman et al. (1996), Daniel (1997), Kane (1998), and others." c. BACKGROUND:
To remove the effect of unobserved characteristics that influence college admission, this study considers information on the set of colleges where students were accepted and rejected.
A difficulty interpreting past estimates a many estimates of the effect of college quality on students' subsequent earnings.
In their study, Dale, and Krueger (1999, 1):
match students in the newly collected College and Beyond (C&B) Data Set who were admitted to and rejected from a similar set of institutions, and estimate fixed effects models. As another approach to adjust for selection bias, we [Dale, and Krueger, (1999)] control for the average SAT score of the schools to which students applied using both the C&B and National Longitudinal Survey of the High School Class of 1972."
Dale, and Krueger (1999, 2) employ two new approaches to amend non-random selection of students on the part of elite colleges. In one approach, they only compare college quality and earnings among students accepted and rejected by a comparable set of colleges; students comparable in provisions of observable variables. In their second approach, Dale, and Krueger (1999, 2) hold the average SAT score of the schools to which each student applied constant, "as well as the average SAT score of the school the student attended, the student's SAT score, and other variables. The second approach is nested in the first estimator."
The most convincing evidence the authors present in support of their arguments includes:."..the average tuition charged by the school is significantly related to the students' subsequent earnings." Dale, and Krueger (1999, 31) state, "we find a substantial internal rate of return from attending a more costly college. Lastly, the payoff to attending an elite college appears to be greater for students from more disadvantaged family backgrounds." Their weakest arguments evolve from the fact that after Dale, and Krueger 1999 (30) correct for students' unobserved characteristics, findings cast doubt on the perception that school selectivity, as measured by the average SAT score of freshmen who attend a college, constitutes a vital determinant of students' subsequent incomes.
4. How Robust is the Evidence on the Effects of College Quality? Evidence from Matching a. In BRIEF:
Black and Smith (2003, 1) contend: "The basic finding is that college quality matters for later labor market outcomes." investigated two potential weaknesses in the most commonly used econometric approach in the literature that estimates the labor market effects of college quality 23).
Based on an individuals' ability, substantial sorting exists into colleges of differing qualities for men and women in the National Longitudinal Survey of Youth (LSY). "Higher ability students disproportionately attend higher quality colleges. (some evidence of an asymmetry [found], with more high-ability students at low-quality colleges) Sorting on ability alone, however, does not break the support condition." (Black and Smith 2003, 24) Although some concerns arise, revolving around the conventional practice of utilizing "linear selection on observables models to investigate the labor market effects of college quality,...the matching estimates support the overall finding of the regression-based literature that college quality matters for labor market outcomes."
Estimates "based only on the 'thick support' region of propensity scores around 0.5 consistently turn out larger than those constructed using the full sample." b. MOTIVATION:
To inform their analysis of the support condition, Black and Smith (2003) utilize data from the 1979 cohort of the National Longitudinal Survey of Youth (NLSY) to examine how students of varying abilities, as measured by the first principal component of the ten tests that comprise the Armed Services Vocational Aptitude Battery. This researcher basically agrees with how the authors view the context of the issue. One consideration this researcher contends may have been minimized more should have been would be "the important role of family background for both labor market outcomes and college quality choices...." (Black and Smith 2003, 22) Although the researchers included this in their analysis, the approach seemed to somewhat minimize the familial factor.
Previous research includes work by "Black, Daniel, and Smith (2003a,b), Brand (2002), Brewer, Eide, and Ehrenberg (1999), Dale and Krueger (2002), Light and Strayer (2000) and Turner (1998). The basic finding is that college quality matters for later labor market outcomes." (Black and Smith 2003, 1) This paper expands knowledge as it'd. EVIDENCE:
The authors initially interviewed respondents in 1979 and attempted to re-interview them annually (biannually since 1994) since then. "Of the five sub-samples that comprise the NLSY, we use only the representative cross-section and the minority over-samples." (Black and Smith 2003, 5) Covariates affect both college quality choice and labor market outcomes, Black and Smith 2003 (24) find. The authors present matching estimates regarding the impact on wages of attending a high-quality rather than a low-quality college. They also "present matching estimates using the Epanechnikov kernel with leave-one-out cross-validated bandwidths." (Black and Smith 2003, 18)
The authors' weakest argument evolves from their admission of they end up with large standard errors. "a problem with the data. Running linear regressions hides the problem by implicitly borrowing strength from comparison observations with lower probabilities of attending a high-quality college. (Black and Smith 2003, 23)
5.Determinants of undergraduate GPAs: SAT scores, high-school GPA and high-school rank. (2004) a. In BRIEF:
This study assesses the degree to which SAT scores, high-school GPA (HSGPA) and class rank predict success in college. The authors also explore whether race - sex differences exist regarding the potential of success in college.
The degree students with varying backgrounds are likely eligible for state-wide scholarships, as well as, student's ability to retain them after enrolment are also examined. (Cohn, Cohn, Balch and Bradley 2004, 577) Research questions include:
Who is likely to benefit most from statewide college scholarship programs in South Carolina?"
Is SAT a necessary requirement, and who is most likely to be affected by such a requirement?
What percentages of students by race and sex groups are likely to achieve a 3.0 GPA in college required to maintain their scholarships, given information about their high-school attributes?
What is the predicted college GPA for students, by race and sex, given the requirements for the various state scholarship programs? (Cohn, Cohn, Balch and Bradley 2004, 577) b. MOTIVATION: This paper tackles the question of the probability that students eligible for a scholarship will be able to maintain it in college. Students' scores on the SAT or ACT tests and/or high-school record (class rank and grade point average, GPA) routinely determine student's college admissions and eligibility for merit scholarships. This study assesses "the degree to which such factors predict success in college." (Cohn, Cohn, Balch and Bradley 2004, 578; 580) This researcher contends the view these authors' present regarding the context of the issue to be agreeable.
Previous related studies include Lenning (1975, cited by Cohn, Cohn, Balch and Bradley 2004, 577) reporting that ACT and SAT scores provide good predictive validity of freshman GPA. Noble and Sawyer (1987, cited by Cohn, Cohn, Balch and Bradley 2004, 577) relate their "predictive equations for 18 college courses with ACT data from 1980 to 1984. Noble (1991, cited by Cohn, Cohn, Balch and Bradley 2004, 577) shows that models that use either ACT or high-school grades alone do not predict as well as models that include both." (Cohn, Cohn, Balch and Bradley 2004, 577) d. EVIDENCE:
Along with student questionnaires, Cohn, Cohn, Balch and Bradley (2004, 577) utilized an informed-consent form, and course-related tests and quizzes to explore the degree to which SAT scores, high-school GPA (HSGPA) and class rank predict success in college. "Data collected from students enrolled in several sections of Principles of Economics at the University of South Carolina in 2000 and 2001 are used to study the relation between college GPA (the dependent variable) and high-school rank, HSGPA, and SAT scores (the key independent variables)." (Cohn, Cohn, Balch and Bradley 2004, 577) also assess whether race - sex differences affect the success potential of the participants in college. As the study by Cohn, Cohn, Balch and Bradley (2004, 578) was limited to 521 students attending Principles of Economics at Moore School of Business, University of South Carolina, Columbia, SC, the researchers contend it to be inappropriate to generalize their results to the entire student body at…