T.1 NASS Comments

Attachment T.1 NASS Comments

OMB No. 0584-[NEW]

Best Practices in Disaster Supplemental Nutrition Assistance Program (D-SNAP) Operations and Planning

May 11, 2021

Project Officer: Eric Sean Williams

Office of Policy Support

SNAP Research and Analysis Division

Food and Nutrition Service

U.S. Department of Agriculture

1320 Braddock Place

Alexandria, VA 22314

703.305.2640

[email protected]

NASS Comments

Alison Black, Mathematical Statistician, NASS-SEIMB-SFDS

January 7, 2021

Background

The United States Department of Agriculture Food and Nutrition Service (FNS) administers the Supplemental and Assistance Nutrition Program (SNAP) and Disaster SNAP (D-SNAP). D-SNAP is a temporary assistance administered in areas that experienced disasters or emergencies. In 2017 and 2018 D-SNAP was implemented in many areas for ~6.0 million households.

Currently, FNS would like to conduct a study to improve the D-SNAP process. Specifically, FNS would like “to identify and document best practices in D-SNAP planning and operations from across the country and for a variety of disaster types.” FNS will use the findings from the study to inform and update guidance and technical assistance to State agencies. State agencies may also use the study findings to improve their D-SNAP planning and operations.

The D-SNAP phase one sampling frame (respondent universe) is comprised of state agencies that administered at least one D-SNAP from 2017 through 2020. The D-SNAP sampling frame was stratified by state-disaster type (e.g. hurricane, flood, tornado) before a non-probability sample1¹ of 5 states were selected. The selection criteria was a specific state-disaster type combination and a state agency’s wiliness to participate in the study. The five selected states are California, Louisiana, Nebraska, North Carolina, and Tennessee.

The D-SNAP phase two sample will consist of 241 staff/individuals from participating State agencies, local SNAP offices, and D-SNAP stakeholders.

At this time, FNS is requesting Office of Management and Budget (OMB) approval to collect information from a sample of state agencies, local SNAP offices, and D-SNAP stakeholders in the selected states.

Comments and Recommendations

The goals of this research are worthy of investigation. However, the sample is not a probability sample. This process may be a good way to do exploratory analysis and begin an investigation. Although, to make policy decisions, formulate plans of actions or prioritize required changes, a study based on statistical sampling methodology may be necessary. For example: states were selected based on certain disaster types and willingness to participate in the study. “Best Practices” derived from this study may not apply to states that were not willing to participate in the study.

Many survey plan items required in an OMB docket are touched upon in the various Appendices and in Supporting Statements Parts A and B. However:

1. While survey objectives were clearly stated, it was not clear how the objectives will be met.

2. The sampling frame, stratification and sample design were not clearly stated.

3. There are many open-ended survey questions that may be difficult to quantify.

4. The estimated burden time is quite long; 4 hours in some cases. A shorter survey may result in better quality survey responses.

5. The process to summarize the reported data are not clear. For example:

   1. “The study team will request and review a variety of documents from FNS and the State agencies selected for the study and systematically extract important data elements into a comprehensive data matrix for analysis.”

   2. “For States with more than one disaster, the study team will develop multiple sets of analytic measures with a set centered on each disaster.”

6. Some summary analytic measures are not clear.

   1. “The study team will assess the relationship between the disaster area, the location used for D-SNAP administration, and the relative socioeconomic conditions of the area affected by the disaster.”

1 No statistical sampling methodology will be employed, no estimation of the number of data sources or systems used will be required, and no unusual problems requiring specialized sampling procedures have been identified: OMB Part 2, Section B.2.