Validating self-reported data on student loan repayment in qualitative and quantitative studies
Submission ID: 3490

Date: Thursday, 6/11/2020 1:15 PM EDT
Session: Concurrent Session B

Primary Presenter: Ama Takyi-Laryea, The Pew Charitable Trusts

Additional Authors

Abstract

Each year 1 million Americans default on their student loan, but the experiences of defaulted borrowers are poorly understood. To research this topic, The Pew Charitable Trusts is using a mixed methods research design and an innovative process to validate participants’ self-reported data.

Borrowers’ experiences repaying their federal student loans is not only a sensitive topic but also extremely complex to decipher.  We need to rely on borrowers self-reported data as they describe their experiences through their default journey. Due to the sensitivity of the topic, the self-reported data may be prone to social desirability bias, recall bias, response bias, and sampling bias. Using administrative data to verify and provide context for our self-reported data allows us to more accurately understand the loan repayment experiences of participants in the study.

During our qualitative research we required that borrowers electronically send in their National Student Loans Data System (NSLDS) transcript (loan transcript) from the Department of Education website as part of the screening process to qualify for the study. Using programming (R®) code, we distilled relevant information from the text transcript into both graphical and tabular form to help anchor interviews and aid in participant recall. Following our results with our qualitative approach, we employed the validation process in a survey using a nationally representative sample to accurately estimate the percentage of false positives and false negatives. The results of these building blocks determined the feasibility of future research with this population.

In this presentation we will review each of our approaches to both our qualitative and qualitative research, with a focus on the logistics and resources involved in requesting and storing transcripts from a reluctant participant pool, and how it helped mitigate the common errors/issues we typically see in both qualitative and quantitative research such as recall and misreporting.