TEACH_OMB_PartB_120915

OMB Clearance Package

Supporting Statement Part B

“TEACH Grant: Study of Institutional Practices and
Grant Recipient Outcomes and Experiences.”

Information Collection

Revised December 8 2015

Prepared by:

Policy and Program Studies Service

U.S. Department of Education

400 Maryland Avenue, SW

Washington, DC 20202

Revised Supporting Statement, Part B
Paperwork Reduction Act Submission

B. Collection of Information Employing Statistical Methods

B.1. Sampling Design

Potential Respondent Universe

In the 2014-2015 award year, 788 institutions of higher education (IHE) awarded a TEACH Grant to at least one student¹. This study will include a survey of the 473 IHEs that awarded a TEACH Grant to at least ten students in the 2014-2015 award year. The sampling frame for the survey is based on data from the Federal Student Aid (FSA) National Student Loan Data System (NSLDS).

This study will also include a survey of recipients who separated from an institution and entered into the eight-year period in which four years of teaching service are required. For purposes of this study, we refer to these recipients as separated recipients. Similar to the institution survey, the sampling frame for the recipient-level survey is based on data from NSLDS. The universe of recipients considered for this survey includes recipients who separated in the 2013-14 award year or prior. In addition to the four-year teaching requirement, the TEACH Grant program also requires an annual certification. Recipients who fail to annually certify their intent to teach in a high-need field at a low-income school automatically have their TEACH Grants converted to loans. As such, the universe of recipients provides the subset of recipients who have had at least one annual certification since separating from the institution. Preliminary analysis of NSLDS data indicate that approximately 74,000 recipients separated in the 2013-14 award year or prior.

The proposed survey collection will be carried out by a contractor. The Department is currently in the process of awarding a contract for this study and expects the contract award by early 2016.

Sample Selection Process

For the IHE survey, the proposed sample size represents all of the IHEs that awarded TEACH Grants to at least 10 students in the 2014-2015 award year. These 473 institutions awarded TEACH Grants to 95 percent of the students who received TEACH Grants in that award year.

The sample of 500 separated recipients will be selected through a stratified sample. Preliminary analysis of NSLDS data indicate that of the approximately 74,000 recipients that separated in 2013-14 or prior, 33,000 have had grants converted to loans and 41,000 did not have grants converted to loans. Sample strata will be based on these two categories. The study team will randomly sample 250 recipients in this universe who have had grants converted to loans. Similarly, the study team will randomly sample 250 recipients who have not had grants converted to loans.

B.2. Procedures for Collection of Information

Initial Contact and Follow up

Survey sampling (for the separated recipient survey) and data collection will occur in early 2016 after the Department awards a contract for data collection and analysis. Pending receipt of OMB approval, letters and/or emails to the 473 IHEs and 500 separated recipients will be sent out based on contact information stored in Federal Student Aid (FSA) data systems, including physical addresses, email addresses, and phone numbers for IHE financial aid administrators and selected separated recipients. Data collection will begin immediately after initial notifications via letters and/or emails are sent. Follow-up letters and/or emails will be sent to non-respondents. In addition, prompting calls will begin for nonrespondents as a reminder to respond to the survey, as an opportunity to answer any questions about the study, and potentially as an opportunity to provide a phone interview and/or paper and pencil survey format instead of the online version².

Statistical Methodology and Estimation Procedures

The study will examine descriptive statistics for each survey question. For example, the researchers will produce tables with the frequency and percentage of IHEs answering “yes” to a question about whether they use online counseling to inform TEACH Grant recipients about program requirements. These data will address study questions regarding the frequency of specific TEACH Grant activities and supports that institutions provide students. In addition, these data will be combined with administrative data so that the study team can examine how collected data are associated with other data elements. For example, by using administrative data to calculate each institution’s grant-to-loan conversion rate, the study team may be able to associate certain types of grant administration activities with relatively higher or lower grant-to-loan conversion rates.

Degree of Accuracy Needed

The target degree of accuracy for the recipient survey is at the 90 percent confidence level. For the IHE survey, this threshold can be met with the target response rate of 80 percent³. For the recipient survey, this threshold can be met with the target response rate of 70 percent in each stratum⁴. Even with a financial incentive, the Department estimates that it is implausible to achieve an 80 percent response rate for recipients. Through the National Center for Education Statistics (NCES), the Department has administered numerous surveys to individual students or former students. These past experiences, including the recent Recent Graduates Employment and Earnings Survey for former students, suggest that a 70 percent response rate is more realistic.

Use of Periodic Data Collection

The survey will only be administered once in 2016.

B.3. Methods for Maximizing Response Rate and Dealing with Nonresponse

Response Rate

A two-month data collection period is planned to allow administrators and students to participate at a time most convenient to them. The study team has the ability to follow up with nonrespondents via mail, email, and telephone. Lastly, the study team plans to provide a $20 incentive fee for recipients who respond to the survey.

Generalizability of the Sample

The research design for the recipient survey relies on a simple random sample stratified by whether or not recipients who separated in the 2013-14 year or prior had grants converted to loans. A response rate of 70 percent will ensure enough power to detect differences that may exist between these two strata. In addition, this response rate will provide generalizability to separated recipients who had at least one annual certification period since separation.

The IHE survey will be sent to all IHEs that awarded TEACH Grants to at least 10 students in the 2014-2015 award year. A response rate of 80 percent will allow the study team to generalize responses to this universe of institutions.

B.4. Test of Procedures and Methods

No test procedures and methods are planned.

B.5. Consultations on Statistical Aspects of the Design

Agency

Brian Fu of the U.S. Department of Education is the Contracting Officer’s Representative for the study. He can be reached at 202-260-1467.

Contractors

The U.S. Department of Education will award a contract for the implementation of this study with an estimated award date in early 2016.

1 https://studentaid.ed.gov/sa/about/data-center/student/title-iv

2 Pending finalization of contract award

3 While an 80 percent response rate of the proposed sample is sufficient to reach this threshold, it does not represent the minimum response rate required to reach this threshold. Based on a universe of 473 relevant institutions, an 80 percent response rate would meet the 97 percent confidence threshold.

4 While a 70 percent response rate of the proposed sample is sufficient to reach this threshold, it does not represent the minimum response rate required to reach this threshold. Based on a universe of 33,000 separated recipients with converted grants and 41,000 separated recipients without converted grants, a 70 percent response rate in each stratum would meet the 92 percent confidence threshold.