Provost Dukes Love gave a presentation (pdf) on “Access and Affordability in Higher Education” at the Alumni Leadership weekend in May. Thanks to popular demand, we will spend this week going through some highlights. Background reading: this 2016 overview of similar material from the previous provost, Will Dudley ’89, and our 2017 series about the Equality of Opportunity project. Day 1.

Although dual y-axis charts are evil, this one provides some useful time series.

Screen Shot 2018-05-28 at 1.36.43 AM

Like all of Gaul, the last 60 years of applications to Williams can be divided into three parts. First, we have the single sex era, with 1,000 to 2,000 applications and an impressive yield of over 50%. Second, the initial 30 years of co-education, with applications mostly stable between 4,000 and 5,000. Third, the current era, with much higher raw application numbers, stable yield and falling admission rates. Comments:

1) Did Williams really yield around 65% in the mid-1960s? That seems implausible. Williams has been losing out to Harvard/Yale/Princeton for decades. I doubt that the 60s were much different. Perhaps Williams used more early decision? Perhaps there was more information sharing, allowing Williams to reject students who it knew would turn it down?

2) The shown yield rate is highly misleading. Students admitted early decision should not be included. They have no (meaningful) choice so it makes no sense to include them in a yield calculation. (Doing so also makes it harder to compare our yield with our competitors who don’t use early decision.) Yield should be measured only from the regular decision admittees. Doing this leaves our yield closer to 30%.

3) Stacked bar charts are not very helpful, especially with this color scheme. (Who can see the change in class size in the light green at the bottom?) Line graphs would be better. A better design for the same data:

  • Three separate panels, stacked on top of each other, each with the same x-axis. All feature line graphs.
  • The x-axis should be academic years, not class years. We want to know how many people applied to Williams in the fall of 2017. Telling us class year just forces us to do math in our heads.
  • The top panel should be raw applications. This is where the process starts. In addition to the raw numbers, we should have some shading or other indication of major changes like the start of co-education or the first time Common Ap usage was above 500. Even better would be to show multiple line graphs which broke total applications into a couple of different categories, like domestic versus international.
  • The second panel should be line graphs of admitted and enrolled students. Since the y-axis range would be much narrower than panel 1, we could easily see the increase in the size of the college (doubling with co-education and then increasing by another 10% since the 80s). Ideally, this panel would also include the number of students admitted early decision.
  • The third panel would combine the information from the top two to show us line graphs of percentage admitted and percentage yielded.
  • Throughout the three panels, we should see information associated with changes in the process. For example, I have a vague memory of applications rising/falling depending on how specific the supplemental essay was.
Facebooktwitter
Print  •  Email