Tools: data management

Repositories, retrieval, cleaning, and documentation

The maintained version of the “Useful Tools” list closing Chapter 4, plus the data repository catalog the chapter refers to. Last reviewed: August 2026.

Data repositories and catalogs

Repository Scope Notes
ICPSR Social science Deep archive with codebooks; institutional membership often required for some holdings
Dataverse General, federated Many university and journal instances; Harvard Dataverse is the largest
OSF General Preregistration, storage, and project pages in one place
Zenodo General, CERN-hosted Issues DOIs; good default for archiving your own data
re3data Registry of repositories Use it to find the right domain repository
Google Dataset Search Index Broad but uneven; verify provenance at the source
Data Commons Harmonized public statistics Useful for demographic and economic covariates
Harvard Dataverse General Where much replication data for political science lands
UK Data Service UK social/economic Registration required; strong survey holdings
GESIS European social science ALLBUS, Eurobarometer, ISSP

Bibliographic and scholarly data

  • OpenAlex: free, no key, ~250M works. The book’s running example; see Chapter 4 resources.
  • Crossref: DOI registration metadata, authoritative for publication records.
  • Semantic Scholar API: includes abstracts and citation contexts.
  • Zotero: reference management; the author’s zotero-files2md converts a Zotero library’s attachments into Markdown for use as an AI knowledge base (Chapter 1, Chapter 2).

Retrieval and cleaning

  • requests / httpx: HTTP clients for API work.
  • pandas: the workhorse for tabular data in Python; polars is a faster alternative for large tables.
  • OpenRefine: interactive cleaning with a replayable operation history, which is what makes it defensible rather than merely convenient.
  • pyarrow: parquet support; see Working with text data.

Documentation and structure

Suggesting a change

Found a dead link or a repository that belongs here? Corrections go through the errata page.