Graduation Rate Discussions

September 28, 2005

A guide for looking at retention and graduation rate methods at NC State University.

Popular methodologies for reporting graduation descriptive statistics:

•  Elapsed Time - define a starting time point, typically fall semester cohort of first-time-in-college (FTIC) full-time freshmen, and look at a fixed time point later to see who is graduated. The federal standard is 150% of "normal" degree completion time.

•  Time-to-Degree - calculate length of college career based on starting and ending point for each student, not compensating for stop-outs, and average all these times for a group of interest. This would be based on a population of graduated students and is calculated after the fact. Note that the true number for a cohort changes over time as stragglers complete their degrees.

•  Seat Time - define a starting time point (FTIC freshmen) and calculate how long they have been in progress based on semesters attended, that is remove stop-out time. This definition is rarely used because: (1) it is tough to define how part-time semesters count, and (2) it changes the discussion from time-to-degree to the questions of what to do about stop-outs and drop-outs. The results of previous studies at NC State show that those who remain full-time students and do not stop out go on to graduate at a very high rate (oddly enough).

•  Forward Looking Model - follow entering students and calculate graduation rate by counting those with a degree at a later point in time (for dual degrees, the first one counts). IPEDS and NCAA use this methodology, define the start point as new freshmen and the end point is 150% of standard time for degree. This type of report would include information on those who graduate in a discipline different than starting discipline.

•  Backward Looking Model - examine all students who graduated at a point in time and look back to see where, how, and when they came into the system. The main weakness of backward looking is it does not have consistent experiences for students (the strength of the forward-looking model). UPA calls this the flow model, it is used to analyze administrative and academic policies that affect student behavior.

Definitions of cohorts and time points

See http://nces.ed.gov/ipeds/pdf/webbase2004/grs_4yr_form.pdf for 2004 forms and definitions.

•  The standard cohort is fall semester, first-time-in-college (FTIC) full-time freshmen. This definition is used by NCAA / GRS and by Student-Right-to-Know legislation. For several years before 2005 the NCAA and IPEDS definitions were merged so that only one report was made for both destinations.

•  The IPEDS cohort definition includes resident aliens (green card, etc) and excludes non-resident aliens. There is a list of visa codes that we use to identify resident non-citizens to OP, and they report the GRS cohort and subsequent data for us. Here is the UNC-OP definition for Resident Alien - Persons who are not citizens or nationals of the United States but who have been lawfully admitted for permanent residence in the United States and hold a "green card" - Form I-151. and who have been admitted as legal immigrants for the purpose of obtaining permanent resident alien status (and who hold either an alien registration card (Form I-551 or I-151), a Temporary Resident Card (Form I-688), or an Arrival-Departure Record (Form I-94) with a notation that conveys legal immigrant status such as Section 207 Refugee, Section 208 Asylee, Conditional Entrant Parolee or Cuban-Haitian) are to be reported in the appropriate racial/ethnic categories along with United States citizens. Non-resident non-citizens are reported as international ethnic status, excluding them from GRS cohort.

•  IPEDS defines full-time and part-time using the institution's policies, we report a count in each status. For NC State the undergraduates are full-time at 12 credits per semester.

•  The UNC-OP Board of Governors (BOG) definition includes the part-time freshmen and puts a time limit on how recently the student must have graduated from high school (within the last year before freshman status). The UNC system information is developed from SDF submitted to OP as reporting mandates, OP then submits the UNC system information to IPEDS.

•  Our current internal report shows 6 year graduation of 1998 cohort (n=3614) at 67%.

•  UNC-OP reporting shows a 6-year graduation rate of the 1998 cohort (n=3617) at 66.9% with another 2.2% of the cohort graduating from another UNC institution.

•  The GRS definition gives the institution the option of fall cohort only, or of using entire year cohort (fall + spring). OP uses only the fall cohort, as do all large universities. The IPEDS definition allows FTIC freshmen that start early in a summer session before their freshman year to be counted as new freshmen in the fall entering cohort.

•  The end time point is based on elapsed academic years, which would include spring and summer graduations of the 4 th and 6 th year.

•  The GRS definition has a second group in the not-graduated category that can be shown to have (1) transferred to another institution, or (2) moved to non-academic full-time service such as military, foreign aid, or religious duty, or (3) deceased. The items in (2) and (3) adjust the original cohort. The graduation rate is reported based on the adjusted cohort, and the transfer outs are identified and the reader is allowed to logically adjust attitudes or decisions.

•  The transfer out rate is problematic to calculate. OP tracks and calculates transfers within the UNC system and to the NC Community College system. There is also the National Student Loan Clearinghouse Enrollment Search service that will, for a fee, return information on students enrolled at other institutions. We send them a list of students and they return information on matches with student name, birth date, and IDs in their database, which covers 90+ % if all institutions. Both data sources have a basic weakness in that the returned transfer information is based on any enrollment mode (full-time, part-time, degree-seeking, or non-degree) not just true transfers which would be degree-seeking students enrolled full-time or part-time.

•  Our current GRS statistics show an original bachelor's cohort for 1992 of 3048, adjusted by 3 deaths to 3045. Graduation rate in 6 years is 62.8% with 6.5% transferring to another UNC institution and 4.9% transferring to a 2-year institution (NC Community College).

NC State Database

The main model we use is the elapsed time methodology required by IPEDS for all universities. UPA has a customized database built for making this reporting easier, it is based on the mandated SDF reporting file. The actual IPEDS reporting is based on the SDF submitted to UNC-OP, and reported by them to the federal government. At NC State there has been a practice, implemented by at least the early 80s, to use the fall census time point a year after fall semester entry as the successive calculation points.

The calculation of a rate is easy once an appropriate population has been defined, just count up those who graduated at a time point later on. Populations should be defined by common characteristics. Examples of a population that should not be used would be 'all students in section 001 of a certain course in their junior year' or 'all students who were full time for their first 6 semesters in a row'.

The UPA "Attrition" database is built on a logical construct of one record per student and there are arrays for data captured at each time point. The reports and analyses the we produce are based on 'walking' the arrays to see what a student's recorded status at specific time points was. There are 4 possible states for the status:

•  E = enrolled

•  S = suspended

•  G = graduated

•  W = withdrawn

Since we have positive information in the system for students who are enrolled, suspended, and graduated in each semester, the withdrawn state becomes the absence of information and is assigned to all students no in E, S, or G states. There are some students who complete a withdrawn transaction, but there is no requirement to do so.

Peer comparisons

•  Land Grant Universities compare unfavorably to the traditional institutions of higher education. The Ivy League institutions and the traditional privates and publics (Duke, UNC-CH, and UVA for example) have 6-year graduation rates that are higher by 20% or more than the average land grant institution. See table for sample of 6-year graduation rates.

•  There are two main sources of information for peer comparisons: (1) NCAA / GRS graduation rate book which has a carefully defined methodology, is submitted by practically all institutions, and publishes information by institution, and (2) Consortium for Student Retention Data Exchange (CSRDE) based in Oklahoma which includes more detailed descriptive information than the NCAA report, but does not include all institutions. NC State participates in both data exchanges.

•  Any peer comparison should address student quality and academic progression policy issues.

Internal Analysis

•  UPA maintains an attrition / retention database based on an old model, it contains detailed data on student cohorts back to 1985. The database tracks cohorts of freshmen, transfer, and graduate students. Combined with other sources of student data, there are powerful analysis capabilities. The model used is a forward looking model based on theoretical work by Alexander Astin. In general, any group specified by student ID can be studied.

•  Backward looking studies are made on an ad hoc basis from a given graduating class. There is a limit to how far back in time electronic records exist, and it is difficult to account uniformly for changes in academic policy, campus climate, and any other influences, such as budget problems or administrative changes.

Changing the institution's graduation rate

Improving the retention or graduation rate is a complex problem, and requires multiple efforts to make a real change.

•  Typically, improvements center on increasing retention rate:

•  Increasing quality of incoming students

•  Increasing amount of financial aid available

•  Identifying at-risk students accurately

•  Increasing support for at-risk students

•  Enforcing an academic progression policy

•  Increasing faculty support

•  Improving academic or cultural climate

•  Increasing general support including advising, libraries, and computing services

•  References:

•  Leaving College , Vincent Tinto

•  What Matters in College?, Alexander Astin

 

Example data from CSRDE 2005 report. Land grant institutions are above the empty row. Graduation rate is from different cohort 5 years earlier.

Institution	2003 Cohort N	2003 PT %	1-year Retention	6-year Graduation
Arizona State	6815	20%	77%	55%
Louisiana State	5361	9%	85%	56%
Colorado State	3745	9%	83%	64%
Purdue	5969	6%	86%	65%
Florida State	6078	12%	86%	65%
Iowa State	3860	6%	85%	66%
NC State	3837	18%	90%	67%
Georgia Tech	2225	8%	92%	72%
Clemson	2749	7%	89%	72%
Maryland - College Park	4050	10%	93%	73%
Michigan State	6451	11%	91%	73%
Va Tech	4860	3%	88%	74%
Texas A&M	6389	9%	90%	76%
Penn State	6039	12%	93%	84%
U Illinois	6810	3%	90%	80%
UNC-CH	3510	5%	95%	81%
UC Berkeley	3663	5%	97%	87%
U Michigan	5538	5%	96%	87%
U Va	3101	6%	96%	92%