NEW - USDA NASS Data Lab Handbook

OMB No. 0535-XXXX
Expires: MM/DD/YYY

United States
Department of
Agriculture

USDA NASS
Data Lab Handbook

National
Agricultural
Statistics
Service

Data Lab and Data Access Group
Methodology Division

Revised
Jan 2023

According to the Paperwork Reduction Act of 1995, an agency may not conduct or sponsor, and a person is not required to respond to a
collection of information unless it displays a valid OMB control number. The valid OMB number is 0535-xxxx. The time required to complete
this information collection document is estimated to average 60 minutes per response, including the time for reviewing instructions, searching
existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information

Table of Contents
Introduction ..................................................................................................................................... 3
1

2

Regulations and Policies ......................................................................................................... 4
1.1

Special Sworn Status ........................................................................................................ 4

1.2

Safeguarding Confidentiality ........................................................................................... 4

1.3

Research or Analysis Products ......................................................................................... 5

Data Access and Disclosure Review....................................................................................... 6
2.1

2.1.1

Standard Application Process ................................................................................... 7

2.1.2

ADM-042 Request to Access Unpublished Data...................................................... 7

2.1.3

Amending an Approved Project................................................................................ 7

2.2

Data Enclave .................................................................................................................... 7

2.2.1

Requirements ............................................................................................................ 8

2.2.2

Security Protocols ..................................................................................................... 8

2.2.3

Process and Procedures ............................................................................................. 8

2.2.4

Access Point Location............................................................................................... 9

2.3

3

Preparing and Submitting a Proposal ............................................................................... 6

Disclosure Review............................................................................................................ 9

2.3.1

Disclosure Methodology ......................................................................................... 10

2.3.2

Suppressions ........................................................................................................... 10

2.3.3

Disclosure Rules ..................................................................................................... 11

2.3.4

Submitting a Request for Disclosure Review ......................................................... 13

Support .................................................................................................................................. 14
3.1

Data Resources ............................................................................................................... 15

2

Introduction
Welcome to the National Agricultural Statistics Service (NASS) Data Lab. NASS values
agricultural research and seeks to provide accurate and useful data in service to U.S. agriculture.
The NASS Data Lab is managed by the Data Lab and Data Access Group (DLDAG) in the
Methodology Division and is governed by the Data Access and Disclosure Review Board
(DADRB).
This NASS Data Lab Handbook (PSM-CS-02-Attachment-A-Handbook) is an official attachment
of NASS Administrator Policy and Standards Memorandum PSM-CS-02 and is used to inform
Special Sworn Agents (approved data users) accessing agricultural census and survey data for
analytical research. All data users are required to read and follow all policies and regulations
described in this handbook – compliance is compulsory. It consists of three sections that
explain the policies, procedures, and support of the NASS Data Lab system and data access.
•

Section 1 Regulations and Policies describes NASS principals and protocols that govern
the use of NASS data. It explains the rules enforced to protect the confidentiality of the
data and those who have provided it. Finally, it discusses NASS policy regarding the
review of reports and articles prepared from the analysis performed.

•

Section 2 Data Access and Disclosure Review describes the procedural function of the
data access application process, clearance for Foreign Nationals, how to access the data,
enclave requirements, and disclosure review.

•

Section 3 Support describes the many data resources available through NASS.

Forward any questions to the DLDAG contact if clarification of NASS policies is needed.

3

1 Regulations and Policies
1.1 Special Sworn Status
Special Sworn Agents are authorized access to a specifically approved project with NASS dataset,
variables, and specified objectives. The data were collected directly by NASS from farmers,
ranchers, and agribusinesses and include statistics that quantify the agricultural practices of these
operations. These data were collected under the authority of the Confidential Information
Protection and Statistical Efficiency (CIPSEA) Act of 2018, Title III of Pub. L. No. 115-435,
codified in 44 U.S.C. Ch. 35, which requires that all information be used for statistical purposes
only. CIPSEA gives discretionary authority to NASS to swear in Special Sworn Agents to access
NASS data. No individual has a right to the appointment nor is NASS obligated to appoint any
individual. Data access can be provided to US-based academia from public or private universities,
other USDA agencies, state government agencies with NASS cooperative agreements, and other
Federal Statistical agencies for purposes that serve the public and contribute significantly to
understanding the agricultural sector or the statistical procedures used by NASS to collect and
summarize data. These data are not to be used for regulatory, enforcement, or investigative
purposes.
Data enclave staff are authorized by NASS as official NASS Data Lab representatives. Each is
trained to administer the regulations, policies, and operational procedures approved by NASS.
Each assures that the data are accessed only by authorized users and all material leaving the enclave
has been reviewed for disclosure risk.

1.2 Safeguarding Confidentiality
CIPSEA requires that no information which could be used to identify or closely approximate data
for an individual farm, establishment, or enterprise be released to the public. NASS Special Sworn
Agents must adhere to all policies, rules, and regulations of NASS, especially those governing the
usage of confidential data. Data access is restricted to the dataset and project objectives explicitly
defined in each approved agreement with NASS. These agreements include, but are not limited
to, the following forms (descriptions for each are found in the sections below).
➢
➢
➢
➢
➢

Standard Application Process Application (SAP)
NASS Memorandum of Understanding (MOU)
NASS ADM-042 Request to Access Unpublished Data (ADM-042)
NASS ADM-043 Certification and Restrictions on Use of Unpublished Data (ADM-043)
NASS ADM-044 User Attestation (ADM-044)

In compliance with CIPSEA violations or breach of the confidentiality of any individual’s data is
subject to fine and/or imprisonment. No information that can be used to identify operators of
4

individual farms, ranches, establishments, or enterprises may be removed from the data enclave at
any time, in any format. Datasets will reside within the data enclave only; they may not be copied
or replicated to other mediums of any kind, including portable drives, phones, tablets, or laptops.
Moreover, screen-sharing, snipping, or reproducing is strictly prohibited and subject to fines and
penalties.
Special Sworn Agents may not discuss any record level data from individual farms, establishments,
or enterprises with anyone outside of NASS unless they are approved to access the same dataset
on the same project. Research must remain within the scope of the approved project.
Any data that could be used, directly or indirectly, to approximate the level of activity of any
individual farm, ranch, establishment, or enterprise must be suppressed prior to release. NASS
must review all materials requested for export from the data enclave for the possible disclosure
risk of confidential information. Section 2.3 Disclosure Review of this document outlines output
restrictions and necessary criteria for the disclosure review process.
NASS publishes several market-sensitive data series through the Agricultural Statistics Board.
Release of these data is done in such a way to ensure everyone has access to the information at the
same time and no one gains a competitive advantage. Requests for data that could compromise
this objective will be denied.

1.3 Research or Analysis Products
Many NASS data users prepare reports of findings based on their research inside the enclave.
Prior to publication, data users must include a citation on the source of the data and independent
conclusions derived. NASS requests that researchers provide a copy of their finalized work for
documentation.
All researchers must cite the National Agricultural Statistics Service as the source of the data used
in the research. The analysis, interpretation, and conclusions are those of the author and are not
subject to NASS review. Researcher results do not represent the views or opinions of NASS. The
following is an example statement of source and disclaimer (adjust specifications, as needed) that
must be included in any publication:

Summaries were derived using data collected in the
2017 Census of Agriculture by the National Agricultural Statistics Service,
United States Department of Agriculture.
Any interpretations and conclusions derived from the data represent
author viewpoints and are not necessarily those of NASS.

5

2 Data Access and Disclosure Review
2.1 Preparing and Submitting a Proposal
Data access projects should be used primarily for statistical analysis seeking modeled results
(inference, regression coefficients, standard errors, cluster, and the like) and not to reproduce
summary tables of descriptive statistics. Summary statistics (e.g. variable means) are allowed, but
only to the extent that they support modeling output. In this case, sample table shells must be
provided in the SAP application.
NASS encourages users to limit the number of files they request for export from the enclave to the
output necessary to produce the final report or paper. Restrictions are applied on the type and size
of research results that can be released from the enclave. The output expected to be released are
regressions and specified tables as found in an article in a peer-reviewed academic journal.
The SAP application sections Methodology and Requested Output are important in assessing both
the substance of the proposal and the risk of disclosing confidential information. Applications
must clearly indicate that the project will emphasize modeling output. Researchers who desire
disclosure review of large volumes of tabular output must request a Special Tabulation rather than
requesting access to record-level CIPSEA protected data (see Section 2.3).
Access to NASS data via a data enclave requires the approval by the Chair, Agricultural Statistics
Board and the Data Access and Disclosure Review Board (DADRB). Depending on the type of
data, researchers may request access using the SAP online application process, or in some
circumstances, an ADM-042. The research team must be associated with a U.S. institution,
agency, or company and all researchers must access the data from within the United States.
Applications are evaluated on a project-by-project basis; that is, an MOU or ADM-042 is required
for each project, with specifications. Census of Agriculture variables must be submitted at the
time of the request and are instrumental in the decision-making process. Although amending the
variable list is possible, researchers are advised to submit a comprehensive list from the onset.
Access is granted based upon many factors, including the fitness-for-use of the data for the stated
objectives and minimally selected variables that explicitly target those objectives.
Applicants must include all components of the SAP and must submit a list of variables for Census
of Agriculture data via NASS-provided spreadsheet. The geographic level of aggregation to be
shown in the output must be included in the project requested output question.
A request can be made online through the SAP process and questions may be sent to
[email protected].

6

2.1.1 Standard Application Process
The Standard Application Process, or SAP, is the primary way to apply for access to protected
data from all 16 federal statistical agencies and units for evidence-building. For NASS, the SAP
application process results in an MOU agreement ready for review and approval consideration.
The MOU is an agreement between NASS, the institution/agency, and the researchers. Requests
by academic institutions must be made by a professor/advisor project lead; students, who are not
official agents of the institution, are not permitted to be project leads. NASS strongly
encourages advisors to have an enclave account if they would like to see the data or output prior
to the final disclosure review.
The SAP website includes a metadata catalog of protected data assets, or “restricted data,”
available across each of the federal statistical agencies. Through this site, researchers can
determine if restricted data are appropriate for specific statistical research objectives and apply
directly online at researchdatagov.org.

2.1.2 ADM-042 Request to Access Unpublished Data
For special circumstances, the ADM-042 is used to request access to record-level data. This form
is reserved for parties who already have an external project agreement with NASS for survey
design, testing, and/or collection.

2.1.3 Amending an Approved Project
If it is determined that an adjustment to the agreement is needed, this can be requested by
addendum on a limited basis. These adjustments may include additional variables, change in
researchers, outside data sources, data access points, and/or agreement extensions. All of which
require DADRB approval. Note that the project lead on any given approved agreement cannot be
changed; in this case, a new request is needed.

2.2 Data Enclave
The data enclave is a platform that provides secure remote data access to approved researchers. It
is owned and operated by a private vendor and is fee-based; all enclave accounts must be paid by
the data user. If the data user has established an external project agreement with NASS, the enclave
fees can be incorporated into the cost estimate. If not, they must be paid directly to the vendor.
The platform provides researchers with remote access to approved datasets for analysis within the
enclave via a secure internet connection using the researcher’s computer and web browser from
an approved U.S. site location. Researchers will have access to a suite of analytical software to use
within the data enclave. For the user, the data enclave appears as a cloud-based Microsoft
7

Windows environment.

2.2.1 Requirements
Data enclave managers will provide complete and comprehensive training, onboarding, invoicing,
and a full list of applications available from within the enclave. To access NASS data via the data
enclave, the following is required:
2.2.1.1 Authorization from NASS to access unpublished data via an approved MOU or
ADM-042.
2.2.1.2 A computer with internet access and a web browser. The computer must be limited
to a fixed site and IP address.
2.2.1.3 An approved virtual site inspection of the access point, or place from which the
data enclave will be used, by NASS personnel. Access points must be within the
United States.
2.2.1.4 Completion of vendor-provided data enclave training, onboarding, and invoicing.

2.2.2 Security Protocols
All data users with approved access to unpublished data will be required to attend a security
protocol briefing, in which they will receive confidentiality training and affirm compliance to the
stipulations outlined in the PSM-CS-02-Attachment-A-Handbook. Their signature on the NASS
ADM-043 and the NASS ADM-044 will be witnessed by a NASS employee. These documents
attest to the understanding of the laws governing confidentiality of NASS data and the penalties
of violating these laws. The ADM-043 contains pertinent excerpts of these laws in more detail.
Security protocols are required annually and by project.
Foreign national researchers are permitted to access NASS data after an additional mandatory
clearance process is conducted by the Department of Homeland Security. The process begins after
project approval and can take 2-3 months, or more. If the foreign national is the project lead, no
other members may access the data until the clearance is granted. If the foreign national is a
member of the team, the project lead and other approved researchers may begin while the clearance
process is underway. More information on this process can be obtained by emailing inquiries to
the Data Lab and Data Access Group ([email protected]).

2.2.3 Process and Procedures
After the MOU or ADM-042 has been approved and all security protocols have been completed,
a NASS representative will contact the vendor to request onboarding to the data enclave. An
8

enclave manager will arrange to provide full instructions on training, account options, billing, and
onboarding to the data enclave. Enclave accounts are individually obtained; they are separate and
nontransferable.
The data enclave contains controls that prevent users from printing and transferring files to and
from the environment via USB, e-mail, screen captures, or otherwise. The signed confidentiality
documents strictly prohibit snipping, photographing, screen-sharing, replicating, or any such
capturing or transferring of the data – even with other members of the approved project. Without
a separate data enclave account, project members are authorized to collaborate through planning,
oversight, and dialogue only. Researchers may not conduct any research or analysis with the
approved data outside of the enclave or away from the approved data enclave access point.
All files intended for removal from the data enclave must first be reviewed by NASS personnel
for disclosure concerns. Only files that have been reviewed by NASS and approved for removal
may be removed from the data enclave. Requests for export are made in the data enclave
environment; training will be provided by the data enclave manager, and additional criteria can be
found in Section 2.3 Disclosure Review.

2.2.4 Access Point Location
In general, the access point for the enclave must be from the professional offices of the institution
– university, state department, etc. with the associated IP address. The approved location is not
mobile; the site inspection, which must be cleared by the NASS security officer, is for a single
chosen location only. Under special circumstances, a home office may be considered for approval.

2.3 Disclosure Review
All research results requested to be released from the data enclave must undergo a disclosure
review conducted by NASS. This requirement is in accordance with the NASS obligation to
protect the confidentiality of an individual farmer, rancher, or producer.
When the research has concluded, the disclosure review of the research results is conducted as a
unit; that is, all files are reviewed together. Since the data are often correlated, multiple export
requests for one project present increased disclosure risk and often increases suppressions. In
addition to the current disclosure review, the analyst must review all past export requests from
other projects to ensure there is no overlap that results in disclosure concerns. Currently, the review
process is done manually; reviewers read through statistical programming code and all output files
– first individually, then against each other. For this and other reasons, large tabulations and
incremental or partial submissions are strongly discouraged.
NASS encourages users to limit the number of files they request to export to the necessary output
required to produce a final report or paper. Restrictions apply (see Section 2.1).
9

Revise and resubmit for peer-reviewed journal articles will be considered, if the previously
approved variables are still available. Potential updates to already released research results
requires an active access agreement, enclave account, and the detailed specifications including the
published material.

2.3.1 Disclosure Methodology
NASS employs disclosure limitation methodology commonly used by many Federal statistical
agencies. Disclosure analysis methods use one of two criteria to determine whether a cell presents
a disclosure risk.
The first criterion is a threshold rule, where a minimum number of operations must produce the
item before a total can be released. For example, if only two farms produce milk in a county,
releasing the total milk production allows the two farmers to deduce the production of the other.
NASS uses the same threshold rule for all disclosure reviews. Each summarized estimate must be
computed from a minimum of five (5) unweighted observations. This means, anytime there are
less than five unweighted operations, the cell value will be suppressed. It is recommended to
collapse cells that do not meet this threshold before submitting disclosure review. Note: Any
output intended for publishing must be weighted.
The second criterion is a dominance rule; NASS uses different dominance rules in different
circumstances.
The (n,k) rule invokes a suppression when the top n producers account for k percent or
more of the estimated total. For example, a (2,80) rule will suppress a cell when the top 2
producers represent 80 percent or more of that cell total.
The p-percent rule requires sufficient protection so that the largest producer value cannot
be approximated to within a range of p-percent. For example, a 20-percent rule will
suppress a cell if revealing that total allows someone to estimate the top producer value to
within plus or minus 20 percent.
Federal statistical agencies do not publicly disclose the actual values of n, k, or p, as revealing
them compromises the protection.

2.3.2 Suppressions
Cells that represent disclosure risk are defined as primary cells, which are always suppressed, and
are called primary suppressions.
In many instances, a primary suppression requires another suppression to maintain protection.
10

This is due to the additive relationships prevalent in tabular summaries; this is one reason, in part,
the release of tabulations from the enclave is restricted. For example, a suppressed value that is
one element of a total can be deduced by simple subtraction. If A + B + C= D, suppressing C
alone gives the value of C no protection, as its value can easily be obtained by solving the equation
for C.
Thus, additional suppressions are needed to ensure confidentiality. These are called
complementary suppressions. In some instances, two primary suppressions can serve to protect
one another. However, in general, selecting complementary suppressions is much more difficult
and time consuming.
If NASS finds values that require suppression, the NASS analyst may apply the suppression or
require the researcher to adjust and resubmit the results.
Examples of this action might be:
NASS may suppress individual table cells by explicitly replacing the cell value with an
indicator identifying a suppression. NASS uses a (D), in place of the actual number in the
affected cells, to indicate the value is withheld to avoid disclosure of an individual
operation.
NASS may request the researcher to aggregate the data at a higher level in order to reduce
the number of suppressed cells.
During every disclosure review NASS must review the results to verify fitness-for-use standards
are maintained. Additionally, NASS must consider the disclosure results from all other
publications and tabulations of the approved dataset(s). NASS may require additional
suppressions be applied to ensure consistency and confidentiality.

2.3.3 Disclosure Rules
Data access projects should be used primarily for statistical analysis seeking modeled results
(inference, regression, cluster, etc.) and not to reproduce summary tables or descriptive statistics.
The disclosure review process for NASS-created Special Tabulations is automated. Enclave
results, however, are reviewed manually and therefore require more stringent disclosure rules.
The following rules must be observed when submitting requests for disclosure review for all results
intended for removal from the enclave. These procedures are designed to expedite and simplify
the process. To avoid unnecessary delays, review the following information about disclosure prior
to producing summary statistics or aggregates. Questions may be forwarded to the Data Lab
manager for assistance.

11

2.3.3.1 Research output should mirror intervals used by the publication from which the
analysis is derived.
2.3.3.2 Every aggregated cell (modeled or summarized) in the output (counts or values)
must include a minimum of five (5) unweighted observations.
2.3.3.3 If output includes demographics, there must also be a minimum of 30 weighted
observations reporting that single demographic at the geographic level of
aggregation.
2.3.3.4 Duplicative output will not be reviewed – e.g. descriptive statistics in table and
report forms.
2.3.3.5 All summary (aggregate) statistics must use the record-level survey or census
weights that NASS provides with each data set. For modeling purposes, NASS
strongly encourages the use of weighted counts or values for robust outcomes. Any
output intended for publishing must be weighted.
2.3.3.6 Data points or maximum/minimum values will not be authorized for removal from
the data enclave. Visualizations and other graphs must be curvilinear and may not
represent actual data points or maximum/minimum values.
2.3.3.7 Histograms, graphs, and charts using bins must ensure they are wide enough
to provide reasonable uncertainty as to the values of those records in the bin and
must contain a minimum of five (5) unweighted observations.
2.3.3.8 Percentiles must be small enough so that the reported value of at least five (5)
unweighted observations exceeds the stated percentile value. This ensures that at
least 11 observations are required to report a median. No percentile reported can
equal the value of the largest contributor (the max value).
2.3.3.9 For model output, such as linear or nonlinear regression models, disclosure risk is
less. However, NASS requires a minimum of five (5) unweighted observations and
will review the model output to check for disclosure failures.
2.3.3.10 When the use of unpublished intervals is approved, the interval range must be
sufficiently wide. As a rule, if 120% of the midpoint of the range exceeds the upper
bound of the range than the interval is too narrow. Two examples are provided for
clarity:
Range = $500-$1000
Midpoint = $750

Threshold: Midpoint*1.20 =
$750*1.20 = $900

12

Conclusion: Since max of range
($1,000) is greater than the
threshold ($900), the range is
wide enough
Range = $100 - $110
Midpoint = $105

Threshold: Midpoint*1.20 =
$105*1.20 = $126
Conclusion: Since the max of the
range ($110) is less than the
threshold ($126), the range is not
wide enough and must be
redefined or suppressed

2.3.3.11 The analysis and summary results in an editable format:
MS Excel (.csv/.xlsx), or MS Word (.docx or other text file) generally, is
accepted. HTML and non-text Stata output are not accepted.
2.3.3.12 Since all output are conducted as a unit, if any output does not comply, the entire
submission must be rejected.

2.3.4 Submitting a Request for Disclosure Review
As described in the previous section, NASS must perform a disclosure review of all output
produced from the data. The entire set of analysis for a given MOU or ADM-042 is usually
reviewed at the same time. Incremental or partial submissions are strongly discouraged.
Submit all requests for disclosure review through the data enclave managers. The enclave manager
will conduct an initial review and will forward it to the Data Lab and Data Access Group for full
disclosure review. The review time depends on the complexity of the output, the number of
reviews currently in progress, and the completeness of the submission. We recommend allowing
at least 30 days for this process in the project timeline, which is requested in the SAP application.
The disclosure review submission must include documentation with the following information:
2.3.4.1 The contact information (email/phone) of the Sworn Special Data Agent who
prepared the results for disclosure review
2.3.4.2 A description of each output file, including modeling techniques, data used, and
explanation of indication
2.3.4.3 A complete list of data sources used in the analysis and summary
2.3.4.4 A complete list of variables used, with descriptions – including created variables
2.3.4.5 Details on any data or variable modifications made – data subset may be required
2.3.4.6 Specific geo-political areas being investigated or summarized – state, county, etc.

13

2.3.4.7 Each summarized tabulated data cell in the disclosure review submission must be
accompanied by:
The total weighted and unweighted number of contributing observations
The unweighted value of each of the top two largest contributing records
2.3.4.8 The statistical programming code used to create the analyses, with appropriate
documentation. That is, every code file should have a header describing the
contents of the file, including a summary of the data manipulation that takes place
in the file.
2.3.4.9 Visualizations must be accompanied by underlying data, if subset or summarized.
2.3.4.10 Non-summarized data visualizations may not include max/mins, such as boxwhisker plots, or any other means by which individual farms may be identified.
Failure to comply with disclosure rules or provide the information listed above will extend the
disclosure review time as re-submissions will be necessary.

3 Support
NASS complies with the timeframe for approval allowed by the SAP. It is recommended that
researchers allow several weeks from the time the request is submitted to when access to the data
is granted. This provides NASS adequate time to vet the request, present it for approval, and
prepare the data set for transfer. This allows for production demands and the enclave onboarding
requirements.
NASS highly recommends that researchers are aware of their institution/agency process for outside
invoicing. Data access in the enclave is contingent upon finalized fee payment, and this process
can increase the project timeline significantly.
Once access is approved, the data will be assigned for preparation and transfer to the enclave. This
process time can vary by data source and current production demands. Amendments to the
agreement for data, researchers, or extension can be request via addendum. A request to change
the project lead necessitates a new project agreement. The data will not contain personal identifiers
such as name, address, zip codes, latitude, or longitude. The default format for delivered datasets
is comma-separated values (.csv), which can be easily imported into statistical packages. Inquire
if an accommodation is needed.
If the approved project cited additional non-NASS data sources required to accomplish the
objectives, consult with the Data Lab manager for upload. Additional support documentation,
14

including survey or census questionnaires or variable description lists, are available through the
SAP.
Each research team will have a Data Lab manager, who is aware of NASS data products and
specializes in the operation of the data enclave. The role of these individuals is to help you get
started with your research by assisting through the approval process, conducting all security
protocols, transmitting the dataset, and arranging for disclosure review.

3.1 Data Resources
Agricultural Resource Management Survey (ARMS) data collection are conducted in
collaboration with USDA Economic Research Service (ERS). All requests for ARMS data are
made through the SAP and any questions may be directed to the ERS ARMS Team at
[email protected]
A few of the many online resources for NASS data are available at the following sites:
ARMS Questionnaires and Manuals
Census of Agriculture
Guide to Products and Services
Quick Stats (data downloads)
ResearchDataGov (SAP metadata and application)
Surveys and Programs

15