OMB memo

Memorandum United States Department of Education

Institute of Education Sciences

National Center for Education Statistics

TO: Rochelle Martinez, OMB September 26, 2007

THROUGH: Edie McArthur

Assistant to the Commissioner, NCES

FROM: James Griffith

Postsecondary Studies Division, NCES

SUBJECT: Follow-up Responses concerning Nonsubstantive Change to the 2007-08 National Postsecondary Student Aid Study (NPSAS:08), OMB Clearance for Early Contacting of Postsecondary Institutions – Additional Institutions to Derive State-Representative Data

Below is our response to questions raised by OMB when granted the “nonsubstantive change” to the sampling design of the 2007-08 National Postsecondary Student Aid Study (NPSAS:08), dated September 21, 2007.

What is the genesis of the six-state augmentation?

One of the major purposes of the National Postsecondary Student Aid Studies (NPSAS) is to collect data that will allow for policy-relevant analyses of how students finance their education; how federal, state, and institutional financial aid programs assist them; and whether the combination of federal, state, and institutional aid programs keep postsecondary education affordable and provide access to low-income students. As currently contracted, the NPSAS:08 study design would only be representative at the national level. Data would only allow for analysis of the aid programs in general and yield estimates that are nationally representative.

The federal government provides about $50 billion annually in student loans for undergraduates and about $13 billion in Pell grants. Although large, the federal programs are not sufficient to cover the financial aid needs of most students. Students also depend on state grants (about $7 billion) and about $24 billion in grants that institutions (primarily private not-for-profit) provide from their own funds. Still, about 80 percent of all undergraduates attend public institutions, and the public institution tuition levels vary substantially by state. Taking this into account, findings from other studies, and bolstered by discussions with and recommendations by members of the NPSAS:08 Technical Review Panel, we decided to examine the feasibility and potential utility of designing and implementing a state-representative “substudy” within the overall nationally representative design for NPSAS:08. There was also considerable interest in deriving estimates not only in the public 2-year, public 4-year, and private not-for-profit 4-year (as was the case in NPSAS:04 for the 12 states oversampling), but additionally, in private for-profit degree-granting institutions. While the RFP for NPSAS:08 included a design option for a 50-state-representative sample design, as well as the base nationally representative sample design, budget restrictions prohibited the government from exercising such options.

Why were the six states selected?

The NPSAS:04 study included a representative state-level sample demonstration study of undergraduates in public 2-year, public 4-year, and private not-for-profit 4-year institutions in 12 states (4 largely populated states, 4 medium populated states, and 4 states with small populations). The usefulness of the results for analysis of tuition and aid policies in the three institutional sectors within these states, and for comparisons among these states, was limited. The major limitation was that the sample sizes for each of the three institutional sectors within the states (with some exceptions) were too small to yield the number of respondents needed for the analysis and comparison of critical subsets of students. Despite these limitations, the NPSAS:04 state-level data has been used effectively in numerous policy studies. On the other hand, attempts at comparative analyses often have been precluded by sample sizes that were too small to produce statistically significant results except at very high levels of aggregation, which did not provide the level of analysis needed to address many of the basic questions.

Expansion of the NPSAS:08 samples for a purposively selected sample of states should make it possible to produce state-level analyses and comparisons of many of the most pertinent issues in postsecondary financial aid and prices. Additionally, including state-level representation for a fourth sector -- the public for-profit degree-granting institutions -- would further enrich the resulting database and its potential for policy analysis. Because budget limitations precluded exercising the 50-state-representative study option (Option 2A of the contract), several designs for the state-representative substudy were examined. The final decision was based on cost and other considerations as described below.

Given our substantive interests, desired estimate precision, and budget limitations, it was determined that the current budget could support oversampling in six states. The six states chosen for oversampling included: California, Minnesota, Illinois, Texas, Georgia, and New York. Two primary criteria served to choose states.

First, states were chosen that have varied public sector tuitions and state financial aid programs were chosen. Such variability would provide data for analyses that examine relationships of the combination of federal and state student aid programs to student outcomes (e.g., persistence, achievement, and degree obtainment). For example, in California the community colleges charge about $600, compared to about $4,000 in New York. Similarly, there are large variations in the size and type of state grant programs.  New York and Georgia have the two largest state grant programs, but the New York TAP program is entirely need-based while the Georgia Hope program is entirely merit-based. The interaction of state tuition policies (low versus high) and state grant programs (large versus small, need-based versus merit-based) with federal and institutional financial aid programs and policies can only be adequately analyzed at the state level.

Having a constrained budget, our second criterion for selecting states was maximizing “efficiencies” in obtaining the necessary sample sizes for each state and sectors within each state. To help in this decision process, we conducted a review of participation rates of states involved in the 12-states demonstration in NPSAS:04. States were recommended for inclusion in the 6-state oversampling in NPSAS:08 based their level of cooperation and participation of institutions in NPSAS:04 study.

Do we expect controversy for selecting these states?

We do not anticipate any controversy relating to the selection of the six states for oversampling.

Do we anticipate any overlap between institutions in the field-test and the full-scale data collections?

To avoid asking an institution to participate in both the field-test and full-scale studies, the field test institutional sample was selected from the complement of institutions selected for the full-scale NPSAS:08 study.  However, since the decision to augment the full-scale sample to provide state-level representation of students in selected states and sectors was made after field test data collection was completed, it is necessary to include in the full scale study a small number of institutions that participated in the field test. It is estimated that this overlap would be about 20 institutions.

In response to two other questions in the email, first, the summary notes or “minutes” for the September 2007 NPSAS:08 TRP meeting is forthcoming. Second, Section A-16 will further elaborate on how data from the states over-sampling will be analyzed.

Please let me know if you have any further questions (telephone 202-502-7387).

3