Research Overview

The research presented in this report is based on analysis of data and information collected through several sources:

• Administrative data collected by 12 Graduate! Network communities (as of early March 2020): Albuquerque, NM; Cleveland, OH; Corpus Christi, TX; Detroit, MI; Greensboro, NC; Kansas City, MO; Louisville, KY; Philadelphia, PA; Rhode Island; San Antonio, TX; St. Louis, MO; and Spokane, WA.
• National Student Clearinghouse (NSC) enrollment and graduation records (as of early March 2019)
• A survey of comebackers conducted in December 2019 (n=325) and fielded in seven Graduate! Network communities
• Focus groups of comebackers conducted in January/February 2020 in three Graduate! Network communities: Cleveland, OH; Kansas City, MO; and Philadelphia, PA
• One-on-one phone interviews conducted in February 2020 with comebackers in five Graduate! Network communities: Albuquerque, NM (2); Cleveland, OH (4); Kansas City, MO (3); Philadelphia, PA (5); and Spokane, WA (1)

Quantitative and qualitative research methods were used for this report.

Quantitative analysis: The full study set for the quantitative analysis consisted of 7,843 adults who had some degree of contact with a Graduate! Network program and were in the “some college, no degree” category, meaning they had completed at a minimum one term of coursework but were not matriculating when they contacted the program. These student-level administrative data records were augmented by enrollment and graduation records obtained from the NSC. Unless otherwise noted, the analysis presented in this report is for this study set.

Using NSC’s methodology for determining how much college students completed before stopping out the first time around, the study set is broken down this way: single-term enrollee (959), multiple-term enrollee (3,345), and potential completer (3,539). This study set is not a random sampling of the “some college, no degree” population. Rather, it is a subset with a particular point of view—comebackers who have demonstrated some degree of motivation to return to school, as evidenced by engaging with a Graduate! Network program. This engagement ranges from completing an online intake form to extensive, ongoing contact with a Graduate! Network advisor.

Data were analyzed using standard descriptive statistics and cross tabulations. For the predictive analysis, we used the “decision tree” model.

Qualitative analysis: Three qualitative research activities were conducted for this project, in this order: (1) a survey of comebackers, (2) focus groups of comebackers, and (3) one-on-one phone interviews with comebackers. Unlike the quantitative analysis, where the study set was limited by the ability to match program data against NSC records, the survey was sent to all comebackers in seven Graduate! Network communities who meet the minimum threshold of engagement. A single link to an online survey was sent to all recipients. Upon entry into the survey, respondents were sorted into separate tracks based on their responses to two questions: first, whether they had contacted their Graduate! Network program officially in stop-out (if yes, they moved to the second sorting question; if no, they were sent to the end for a general set of questions); and second, what their actions were after engaging with the program. For this second sorting question, respondents were sent to four separate groups: (1) respondents who had not gone on to enroll, (2) respondents who went on to enroll and were still matriculating, (3) respondents who went on to enroll and were back in stop-out, and (4) respondents who went on to enroll and graduated. All respondents were asked a set of questions at the end regarding their views on the value of a college degree, the availability of employer benefits, their actions related to non-degree credentials, and basic demographic information. A total of 325 comebackers completed the survey, and responses were analyzed using basic descriptive statistics and cross tabulations.

Findings from the survey were used to craft questions and protocols for the focus groups and phone interviews, though each activity focused on a specific subset of comebackers. Those who made it to graduation were recruited for the focus groups. Those who went on to re-enroll after engagement—they might have still been matriculating, or they might have gone back into stop-out—were recruited for the phone interviews. Transcripts were generated for all the focus groups and interviews. They were reviewed and tagged for key concepts and compelling quotes.

Synthesis: The final step of the research process was to combine the findings of the quantitative and qualitative activities and to identify key concepts and trends. An advisory group of national experts from the arenas of higher education, policy, and research provided guidance and feedback throughout the entire process.

We acknowledge the following limitations of our data:

• Data collection was completed just before the onset of COVID-19 and therefore do not reflect changes in status, condition, views, attitudes, or aspirations of comebackers brought about by widespread closures, job losses, and health impacts.
• The study set was limited to adults for whom there were corresponding enrollment and graduation records from NSC, which was done through a process of matching name and date of birth from The Graduate! Network’s administrative data. Instances where there was an incorrect or missing date of birth or where there had been a name change or difference in recorded name resulted in unmatched records. Adults with unmatched records were not included in the study set.
• There was incomplete demographic information for our study set, limiting our ability to draw strong conclusions when cross tabulating against these variables, especially for family annual income. The percentages of non-blank responses for key demographics cited in this report were: gender identity, 83 percent; race/ethnicity, 74 percent; family annual income, 40 percent.
• Some data collection methods were optional and relied on self-reported perceptions and behaviors.
• Some of the data collection methods had low response rates, limiting our ability to draw strong conclusions.
• Some data collection methods were conducted in a setting where services and programs were taking place, possibly resulting in social desirability bias.