College Selection

In Summer 2012, we will be considering college selection and assignment:  how students are assigned to their colleges. 

  1. What are the goals of college assignment?
  2. What goals are most important?
  3. How can we best achieve those goals?

The colleges have many different attributes based on their locations, themes, dorm rooms, history, and other factors.

The 2012 student ballot included optional questions about college choice and satisfaction.  With small margins, respondants indicated the location to be most important factor in their college choice for Cowell and Stevenson; the theme at Crown, Eight, Merrill, and Porter; the rooms at Kresge, Nine, and Ten, and Diversity at Oakes.  79% of the students were satisfied with their college experience in the past year (average of per-college averages), while 8% reported no interaction with the college, and 13% were dissatisfied.  The full survey, including the free-form text responses, is attached below. 

A look at 2011-12 graduates from each of the colleges shows interesting variation among the colleges in terms of majors, ethnicities, and genders.  Crown, with more than 50% PBSci and engineering students, Porter with nearly 30% art students, both with about 1/3 social sciences students.  Over 60% of the C8 graduates being from social sciences, followed closely by Ten, Nine, and Stevenson.  More than 20% of Cowell, Krewsge, and Merrill students completing Humanities degrees.  And so forth.

There is also great variation among the colleges in terms of graduate ethnicity, ranging from about 25% non-white/unknown to 60% non-white/unknown.

The core course and composition requirements are a fundamental part of the first-year experience.  There are variations in the movement of students through ELWR satisfaction among the colleges, and from year to year.  This may primarily be the result of differing student backgrounds, but how do the differing approaches to Core, tutoring, stretch and other support effect this component of retention and graduation.  There is also variation in the 2-year frosh retention/graduation rates, which can be looked at according to incoming quintile of an academic measure (SAT R combined, in this case).  

Current Process

College assignment takes place after students SIR, quite quickly so that housing will be able to follow up with students about room selection and other items.  Michael McCawley has written a description of the process.  Michael has also provided tables for 2012 frosh interest and frosh assignment to the colleges.  The 2011 tables are included below as well.

What are the goals of college assignment?

In no particular order, here are some possibilities.  Please suggest others!  Appearance on this list is not meant to be an endorsement!  Part of our discussion will be what the real answers to this question should be. 

  1. Enhancing specific college themes
  2. Filling beds
  3. Concentrating/Dispersing/Ignoring Transfer Students within colleges - 25.4% of the incoming class are transfer students; are they well served by the current system?
  4. Concentrating/Dispersing/Ignoring Academic Areas within colleges
  5. Concentrating/Dispersing/Ignoring Genders within colleges
  6. Concentrating/Dispersing/Ignoring Ethnic Diversity within colleges
  7. Concentrating/Dispersing/Ignoring Regent's, Pister, Honors students within colleges
  8. Effectively applying resources to student success
  9. Filling specific academic-focussed residential sub-communities (e.g., honors, international, science, sustainabilitly, arts...)
  10. Filling specific activity-focussed residential communities
  11. Increasing retention and graduation
  12. Providing testbeds for program innovation and assesment
  13. Maximizing the number of top choice [or top 2, 3, 4, 5] assignments
  14. Minimizing the number of bottom or non-choice assignments
  15. Minimizing a weighting function, such as top choice counts 1, second choice 2, ....
  16. ....
  17. ....


What goals are most important?

Discussion topic.

How can we best achieve those goals?

Discussion topic.

Ranked-Choice Simulation

Based on the summer discussion, here is a spreadsheet that roughly simulates assignment by rank choice, with a few caveats.

The spreadsheet goes row by row to see if there is space in the first choice college, and then places the student there.  If not, on to the second choice, and so forth.

In this data set, that produced 99 students who were no placed in one of their 5 choices (college 11) as part of the 311 students place in college 11 (212 students indicated no preference).  This group of students would be distributed to Merrill and Oakes, the two colleges below target.  While the algorithm does not fit as many students into a listed college, it is something that reflects what students expect given that we ask for 5 ranked choices.

I did not split colleges on gender, which would produce some additional variation.  I did not randomize the list, so it is by SID order.  I suspect this is fine for the simulation; actuall practice would require randomization to avoid giving preference based on SIR date.

Up at the top of the spreadsheet you can see the resulting ethnicity distributions among the colleges in the simulation.  Because, as with the past method, this method gives the vast majority (83%) of students their first choice and 98% of students one of their choices, this distributions can be expected to continue to reflect the differing levels of interest that students from different groups have in the various colleges.  The Merrill renovation, of course, may be an interesting shakeup of this data.

This is different from what was discussed during the summer, which was that 3 unranked colleges be selected (then enabling justifiable use of any of the top 3 colleges), with picks going to the least selected college first.  Based on the number of times Merrill and Oakes appear in the data among the first 2 colleges, it appears that there are sufficient mention of those two colleges for the "3 unranked" to work quite well.

Of course, the major rennovation project at Merrill is likely to generate considerable interest in the college, either for Fall 2013 or Fall 2014, or both.

Some other thoughts include....

- How much of the 99 would appeal would depend on how invested they are in their first choice.  It seems likely that appeals are mostly based on students not getting some college that they really really really think is perfect for them, and so getting a second, third, or forth choice may be equivalent.

- This would be assignment roughly based on what we tell students.  We can do assignment in other ways, but I highly value the idea that 'whatever it is we do, we should be able to tell students at the start how we will do it.'  At UCSC, college choices are much more than just about apartments or taquerias or being yellow, but about the academic focus of the college, the available learning sub-communties, and so forth.  We need to be able to tell students about having a great experience at whichever college they wind up, and also how they wind up there.

Here is a draft of this algorithm:

  • College assignment.  Students will be randomly ordered, and then assigned their first-choice college if there is space available, followed by second choice if the first choice was full, and so own.  At each college, the ratio of Male-Identified to Female-Identified students is not allowed to diverge too far from the campus average.  ?? Students in various special groups (first-year honors, Regent's Scholars, fellowship communities, ...) are assigned before other students.
Here is a draft for what was done last year.
  • College assignment.  Each college provides an exceptional community of learning within the larger campus.  However, the number of students interested in each specific college can vary from year to year.  To ensure that each student is in one of their 5 chosen colleges, we first focus on the colleges that are least-frequently specified, and fill them with students who listed that college as one of their choices, typically as a first through third choice, but occasionally a fourth or fifth choice will need to be assigned.  We then proceed to the next college, and so forth.  The most-specified colleges will typicall be filled with students who indicated that college as a first choice.  ?? Students in various special groups (first-year honors, Regent's Scholars, fellowship communities, ...) are assigned before other students.

College Demographics of 2011-12 Graduates

College Selection Student Survey

ELWR results by college

Fall 2012 College Demand - Frosh.xls

Fall 2012 College Assignments - Frosh.xls

Fall 2011 College Demand - Frosh.xls

Fall 2011 College Assignments - Frosh.xls

Two-year 2009-10 frosh retention/graduation by SAT R combined quintile

CHES - Co-Curricular Activity Summary 2011-12

College Assignment Process Overview (from Michael McCawley)

Two-year 2009-10 frosh retention/graduation by College and SAT R combined quintile

Fall 2012 College Assignments - Frosh with Graphs.xls

Ranked-Choice Simulation for Fall 2012 (see text above, xlsx)