1. Before You Start: Organize Your Files

Documenting a dataset is significantly easier and more efficient when data files, metadata, documentation, scripts, and supporting resources are organized in a clear and consistent manner from the outset. A well-structured file organization system helps ensure that:

• Files are easy to identify, locate, and understand, reducing the time spent searching for information and helping collaborators navigate project resources efficiently.
• Important files are protected from accidental overwriting, duplication, or loss, preserving the integrity of the dataset and its associated documentation throughout the project lifecycle.
• Sensitive or restricted content is kept separate from materials intended for sharing or public dissemination, supporting compliance with data protection requirements and reducing the risk of unintended disclosure.
• Documentation activities can be completed more quickly and accurately, as all relevant materials are readily available and clearly categorized.
• Collaboration is improved, enabling team members to work with a common understanding of where files are stored and which versions should be used.
• Data preservation and long-term reuse are enhanced, making it easier to archive, locate archived versions, transfer, and maintain datasets over time.

Recommended Folder Structure

The following folder structure separates original data, working files, dissemination packages, documentation, programs, analytical outputs, and archived materials. This organization helps ensure that files are easy to locate, prevents accidental overwriting, and clearly distinguishes sensitive working files from materials intended for dissemination. We recommend that, before anything else, you create the necessary directories as follows:

Folder Structure	Purpose
📁 UGA_2026_DHS	Root Directory – Create a directory for the dataset and all associated resources (the archival package). We recommend using a clear and consistent naming convention that includes the country aor geographic scope (where applicable), the reference year or period, and a short identifier for the data collection. For example, UGA_2026_DHS for a "Demographic and Health Survey" of Uganda collected in 2026 or KEN_2025_EMIS for a Kenya Education Management Information System dataset for 2025. This directory serves as the root folder for all materials related to the microdata collectio, including data files, documentation,metadata, code, and supporting resources. Avoid spaces or special characters in folder names. Use underscores (_) or hyphens (-) as separators to ensure compatibility across operating systems, software applications, and repository platforms. A consistent naming convention improves organization, discoverability, version control, and long term preservation.
├── 📁 UGA_2026_DHS_v01_M	Create sub-directories to hold different versions of the data. The original version of the data will always be designated as the Main file. In this example, the first version (v01) of the Main data would be stored in the folder named UGA_2026_DHS_v01_M. As stated above, avoid spaces or special characters in the folder name.
├──├── 📁 Data	The Data sub-directory contains datasets throughout their lifecycle, from acquisition to dissemination (if applicable). Data files are organized in marked sub-folders to minimize risk of overwriting and sharing of restricted information.
├──├──├── 📁 Original ↳ Raw, Backup	Store source data exactly as received from the producer. These files should be preserved and remain unchanged.
├──├──└── 📁 Edited	Store edited data separate from the raw data.
├──├──├── 📁 Working ↳ Data Entry, Cleaning, Harmonization, Quality Assurance, Anonymization	This folder can hold intermediate files used during processing, validation, transformation, and anonymization.
├──├──└── 📁 Dissemination ↳ v01_A_PUF, v02_A_PUF, v01_A_LUF	If data is being disseminated, create this dissemination sub-folder to separately store final approved datasets organized by release version and access type. In this example, the files prepared for dissemination are considered adaptations of the main data, so the folder names include the adaptation version information and end with _A_PUF, _A_SUF etc.
├──├── 📁 Documentation	Create documentation sub-directories to hold materials that describe, support, or govern the study and its data. As needed, materials can be further organized into separate sub-directories as shown below.
├──├──├── 📁 Administrative ↳ Acquisition, Agreements, Correspondence, Permissions	Store administrative records related to dataset acquisition, licensing, and governance.
├──├──├── 📁 Project Materials ↳ Questionnaires, Manuals	Store data collection instruments and fieldwork guidance.
├──├──├── 📁 Reports ↳ Project Reports, Methodology, Sampling Reports, Evaluations	Store reports describing the data production process, methodology, and findings.
├──├──├── 📁 Technical ↳ Codebooks, Data Dictionaries, Sample Design, Geospatial material, data processing and quality assurance documents etc.	Store documentation required to understand and use the data.
├──├──└── 📁 Photos ↳ Fieldwork, Instruments, Maps	Store photographs and visual resources associated with the study.
├──├── 📁 Programs	Create sub-directories for scripts and code used to process, validate, anonymize, tabulate, and analyze data as needed.
├──├──├── 📁 Data Management ↳ Data Entry, Cleaning, Anonymization, QA	Store programs used to prepare dissemination-ready datasets.
├──├──└── 📁 Analysis ↳ Tabulation, Indicators, Research	Store statistical and analytical programs used to produce outputs.
├──├── 📁 Outputs ↳ Tables, Figures, Maps, Publications	Store products generated from the data, including statistical outputs and publications.
└──└── 📁 Archive ↳ Previous Releases, Superseded Files, Deposits	Store historical versions and archived materials retained for preservation and audit purposes.

Example: Dissemination Folder Structure

The dissemination package is a curated subset of the archival package containing only the resources approved for distribution. While the archival package preserves all materials associated with the data lifecycle, the dissemination package includes only those resources that can be shared and are required to support data discovery, access, interpretation, and reuse.

The dissemination folder contains the dissemination package, which includes the data and resources approved for sharing, publication, or distribution to authorized users. Depending on the dissemination model and access restrictions, this package may contain data files, metadata, documentation, codebooks, analytical code, replication programs, and other materials that help users understand, interpret, analyze, and reuse the data.

Not all resources contained in the main project or archive folder should be included in the dissemination package. Certain documents may contain sensitive, confidential, proprietary, personally identifiable, or otherwise restricted information and should therefore be retained only in the archival collection. Examples include internal correspondence, administrative records, quality control materials, disclosure assessments, security-related documentation, and other materials that are not intended for public release.

Before dissemination, all files should be reviewed to ensure they comply with applicable confidentiality, privacy, licensing, legal, and organizational requirements. The dissemination package should contain only those resources necessary to support the appropriate use, interpretation, and reproducibility of the disseminated data.

📁 Dissemination
│
├── 📁 UGA_2026_DHS_v01_M_v01_A_PUF
│   ├── 📁 Data
│   ├── 📁 Documentation
│   ├── 📁 Programs
│   └── 📄 Release_Notes.txt
│
├── 📁 UGA_2026_DHS_v01_M_v02_A_PUF
│   ├── 📁 Data
│   ├── 📁 Documentation
│   ├── 📁 Programs
│   └── 📄 Release_Notes.txt
│
├── 📁 UGA_2026_DHS_v01_M_v01_A_LUF
│   ├── 📁 Data
│   ├── 📁 Documentation
│   ├── 📁 Programs
│   └── 📄 Release_Notes.txt

Dissemination Version Naming Convention

Folder Name	Description
📁 v01_A_PUF	First release of the Public Use File package
📁 v02_A_PUF	Second release of the Public Use File package
📁 v01_A_LUF	First release of the Licensed Use File package
📁 v01_A_CUF	First release of the Confidential Use File package
📄 Release_Notes.txt	Documents changes introduced in the release

Best Practice

Store each dissemination release as a complete, self-contained package containing the distributed data, documentation, programs, and release notes. This ensures that previous releases remain reproducible and that users can clearly identify which version of the data was used.

When creating structured metadata for a dataset, use the dataset folder name as the Primary ID whenever possible. This ensures a clear and traceable link between the metadata record and the corresponding data files.

For example, when documenting the main dataset, the Primary ID should be UGA_2026_DHS_v01_M. If you are documenting Version 2 of the Public Use File, the Primary ID should be UGA_2026_DHS_v01_M_v02_A_PUF.

To accommodate diverse user needs and support long-term preservation and accessibility, data should be archived and disseminated in multiple widely used formats, such as Stata, SPSS, CSV, and ASCII.

Using a consistent naming convention helps maintain traceability, supports version control, and makes it easier to manage and identify related metadata and data assets over time.

Once the folders and files have been properly organized, the next step is to assess the data to ensure it is adequately prepared for curation, archiving, and, where appropriate, dissemination. This review helps confirm that the data meet quality, documentation, and preservation requirements before the curation process begins. The key considerations for data preparation are covered in the next section.