Tag Archives: econometrics

Fixing college admissions

103932551-gettyimages-94953697-530x298

Harvard is being sued by a group of Asian-American students for discriminating against them through affirmative action admission standards. The Trump Administration is joining on their side. I have been troubled by the reliance on standardized tests and inflated GPAs as appropriate metrics for college admission. I propose two solutions to both of these problems:

  1. Colleges and universities should create mission statements about what the university wants to accomplish through educating students. These statements should go well beyond just saying “a well-rounded individual who is prepared for a career.” It should state that it wants a set of alumni that positively impact their communities in a number of dimensions (some not easily quantified) in a manner that improves social well being in a manner chosen by the university’s policy makers. This will require developing a consensus among those policy makers (e.g., the Board of Regents), but it will prompt a larger debate about the purpose of a college education that is badly needed right now.
  2. The colleges and universities should then conduct at least two statistical/econometric studies to determine the mix of traits that it wants in its student body to accomplish its objectives:
    • The first one would establish what preparation will lead to the most likely academic success in that college. This would completely eliminate the need for standardized tests. The key is that most colleges have an extensive set of data on the GPAs of their students, the high schools that they went to and their GPAs in high school (along with other academic data). The study (a regression)  would regress the college GPA on the high school GPA weighted by the high school (this adjusts for differences between high schools) plus a specific set of other academic / extracurricular data. This would be socio-economic blind. But most importantly, these results would NOT be the singular admission standard. It would be just a measure of likely academic success that could help identify how to deploy resources for those students once they enroll.
    • The second would evaluate how alumni impact their communities. This would rely on data about alumni on occupations, income, community activities, rate of return to certain communities based on socio-economic status, and community well-being. Alumni characteristics would include incoming socio-economics traits, study major, college GPA, extracurricular activities, and other factors thought to be important. This study (again, likely econometric) would measure the relative impact on community well-being (or other chosen objectives) of admitting students with certain characteristics and likelihood of certain majors and academic performance. For example, it might show admitting a student from Watts with a lower predicted GPA has a bigger community impact than admitting another from Beverly Hills with a higher GPA.
    • The desirable traits found in the second study would be used with the results of the first study to set admission policies in a fairly transparent way. A third study might assess what mix of student body traits is most likely to achieve the mission statement objectives. That could further help make the admission process more transparent.
Advertisements

How to misconstrue statistics in your favor: an example arguing against SB 32

 

statebystatechangeinco2emissionrateThis blog post on Fox & Hounds is an example of how to take statistics of one cause-and-effect relationship and misapply them to another situation. In this case, this graphic above shows how GHG emissions have dropped dramatically in states that used to burn coal to generate electricity, but now rely much more on natural gas. The decline in coal emissions has occurred over the last half-decade due to the fall in gas prices and the increased stringency in air quality regulations. But more importantly, those states had higher emissions that California to start with because they have been laggards in protecting their environments. The chart shows that these states are finally starting to catch up! If anything, this supports adopting SB 32 as a follow on to AB 32!

Yet the blog post misconstrues this situation to argue that it’s the “free market” that somehow is generating these greater reductions, implying that California and Mississippi had started from the same place–which of course if far from the truth. Yes, the market push from natural gas fracking explains some of this, but California was already so far ahead due to its own efforts that it has less room to improve.

Watch for these types of misrepresentations. Understand the initial premises by the authors. Ask hard questions before you accept their conclusions.

Source: There’s a Better Way :: Fox&Hounds