Postsecondary Student

The postsecondary researcher data files contain commonly requested longitudinal student-level data from the postsecondary educational environment. The data is organized into 5 datasets: base, enrollment, program, coursework, and awards. These datasets reflect information about postsecondary student demographics, enrollments, programs, courses, cumulative progress, and degrees/certifications (referred to as awards). Data from the 2010-present school years are available.

The primary data sources for the postsecondary researcher data are the Student Transcript and Academic Record Repository (STARR) and the National Student Clearinghouse (NSC). STARR is a collection of student data from all Michigan public institutions and some independent institutions. NSC is a national database of postsecondary student information from most colleges nationwide. The information collected from STARR is more detailed than that obtained from NSC, but we use the NSC data to track the progress of Michigan's students who attend college out of state, and as a backup to ensure we obtain the most complete data possible for Michigan's postsecondary students. The STARR and NSC data are merged together and standardized, and these merged datasets serve as the source for the researcher files. More specifically, CEPI submits a student list to NSC for the past 8 cohorts of Michigan high school graduates and for all Michigan postsecondary students. NSC matches the student list to their Student Tracker database and CEPI uses the returned results.

When merging these data sources, student x year x institution enrollment and awards records are first identified from the STARR and NSC. Where only a STARR record or both STARR and NSC have a matching record, data from STARR are incorporated into the panel. After having done this, any remaining unique NSC records are then added to the panel. Most commonly, these NSC records will come from out-of-state and independent/private institutions although community colleges and, to a lesser degree, four-year institutions will also show records.

Image

While this approach greatly reduces duplicate records, there are potential shortcomings. Academic Year is based on Start and End dates that are received from STARR & NSC and there are known data quality concerns around the End date where data is received differently between STARR & NSC. When those dates fall into a different Academic Year, overreporting of the outcomes from those data may occur. A second artifact is the mixing of STARR and NSC data for the same student at the same college in the same school year if, for example, a STARR enrollment record and an NSC award record are received. NSC enrollment records at that college/school year are dropped, but the NSC award record is kept, even if it was not reported in STARR.

This strategy of combining the two data sources, plus the method of data collection used, results in something of a convenience sample for the overall postsecondary dataset. Included are the universe of students enrolled at a Michigan public institution regardless of whether or not they attended Michigan public high school. Additionally, K12 Michigan public high school attendees who attended an out-of-state postsecondary institution or an in-state independent/private postsecondary institution are included, meaning a comprehensive of post-secondary outcomes for Michigan public K12 students is achieved. Beyond this, K12 students who dual-enroll will also show records.

Overall Data Table Notes

Table Hierarchy

This is the general hierarchy of the tables:

Standard Table Joins

The below grids show how the tables link together, but other joins may be appropriate depending on the requirements of the research.

Though awards do not have a direct one-to-one link to enrollment records, most student awards can be linked to a corresponding set of enrollments as shown below. 

Post-Secondary Student Base

Highlights

Record Level: Student-level 
Record Count: ~2.6 million observations 
Years Covered: 2010-Present* 
Population Coverage: Students enrolled in or receiving an award from a Michigan public institution

The base dataset contains information on student demographics: race, gender, and date of birth. The base population includes students who have enrolled in or received an award from a Michigan public institution (and some independent institutions) during the 2010 academic year or later. The data will include the entire academic history of any such student, which may include enrollments or awards prior to the 2010 academic year.

Students are identified with a research identification code (RIC), which is unique in nearly all cases in the dataset. However, due to linking and unlinking over time as data was moved from the source datasets to the final datasets, some students may have duplicated or conflicting data. In these cases, both records are retained within the final dataset, allowing researchers to decide how to eliminate duplicate RICs. In addition, in some cases there are discrepancies between STARR and NSC data.

In addition, the data file includes a stable date of birth (latest reported date of birth) for each student, and indicators for where conflicts in demographic data exist (i.e., if conflicting dates of birth, race, or gender have been reported for any given student over time).

* Note that, because the academic histories of students might extend before the 2010 academic year, the years covered for other postsecondary files may differ. For example, there are student enrollments dating back as early to 1900 in the enrollment datafile.

Post-Secondary Student Enrollment

Highlights

Record Level: Student by school year session by institution 
Record Count: 1,792,406 observations/year beginning in 2010 
Years Covered: 2010-Present* 
Population Coverage: Students with an enrollment record at a Michigan public institution

The enrollment dataset includes one record per student, for each institution attended and school year session (for example, fall 2010-2011). For each record, session beginning and end dates are provided. Additionally, the dataset includes variables that reflect academic progress, such as session GPA and credits attempted/earned, in addition to variables reflecting academic progress. The dataset also includes flags for full-time enrollment and whether any courses were taken for remedial purposes. As with the base file, students are identified using a RIC, which is unique within a student-year-IHE-session for nearly all cases.

In this dataset, Spring and Winter sessions are treated the same by CEPI. Researchers should note that in the 2016-17 collection, CEPI began renaming all Winter sessions to Spring to reflect that. Therefore, you will see no Winter sessions from the 2016-17 enrollments on. The figure below illustrates how different sessions were categorized. For example, sessions categorized as “Late Summer” have a start date between 4/15 and 7/31 and an end date between 7/16 and 8/31. Some sessions span more than one session range; a session falling entirely within an overlap range is put into the later second session of the range (i.e., the end date determines the session for these cases). Any sessions with a start date between 4/15-EndYear and 8/1-EndYear and ending on or after 7/16-EndYear are moved to the subsequent school year.

Image

In addition, researchers should note the following definitions of student transfers, which apply to those with ENROLLMENT_TYPE = “TransferIn”. CEPI has two principal definitions of a transfer:

1. If an institution tells us that a student is a transfer, that student is a transfer, full stop. This stems from an Enrollment Type of “TransferIn”. CEPI tries to find the student’s previous college or university, but if they cannot find one then their previous enrollment is labeled as ‘Unknown’.
2. CEPI looks at all enrollments within the academic year of the report and finds the most recent enrollments for the students before that point. They look for the following to be true. If these are true, then the enrollment pattern is considered a transfer by CEPI’s definition: - The previous enrollment is at a different institution - The previous enrollment is at the same Enrollment Level (Undergraduate vs Graduate) - The student has stopped enrollment at that previous institution - The student has not previously been enrolled at their current institution (that is, readmissions are not eligible to be transfers)

CEPI then takes all transfers discovered by methods 1 and/or 2, and compiles them into the data file.

View the enrollment data: record counts by institution for 2010-2019

* Years covered includes any year for which there is data available in this file. However, researchers should note that these data encompass the academic histories of any student receiving an award beginning with the 2010 school year. As such, the base file “years covered” encompasses 2010, but all other files may include earlier records as well.

Post-Secondary Student Programs

Highlights

Record Level: Student by program by school year session by institution 
Record Count: 1,994,786 observations/year beginning in 2010 
Years Covered: 2010-Present* 
Population Coverage: Students enrolled in or receiving an award from a Michigan public institution, as well as some out of state students from NSC.

The programs dataset includes one record per student, for each program the student is working toward at all institutions and school year session (for example, fall 2010-2011). For each record, session beginning and end dates are provided. Additionally, the dataset includes variables that reflect program type and description (such as “Major in Computer Science”). Program CIP codes and descriptions are also provided but due to reporting issues, many records reflect a null value.

CEPI conducts some deduplication of program records, as follows:

• Drop records with a null program name if there is a corresponding record where all other key fields match and the program name field contains a value.
• Drop records with a null program CIP code if there is a corresponding record where all other key fields match and the program CIP code field contains a value.
• If both a STARR and an NSC program record exist where all key fields are identical, keep only the STARR record.

* Years covered includes any year for which there is data available in this file. However, researchers should note that these data encompass the academic histories of any student enrolled in or receiving an award beginning with the 2010 school year. As such, the base file “years covered” encompasses 2010, but all other files may include earlier records as well.