Collecting original data is not always necessary, and it is not always possible within a student thesis timeline. Many excellent theses are built entirely on secondary data pulled from established, well documented repositories. Knowing where to look, and how to judge what you find, will save you weeks of searching.
General purpose repositories
Kaggle hosts thousands of community uploaded and organization sponsored datasets across nearly every field, along with example code notebooks that show how other researchers have already worked with the same data.
The UCI Machine Learning Repository is one of the oldest and most cited sources for clean, well documented datasets, particularly useful for coursework and thesis projects involving classification or prediction.
Google Dataset Search works like a search engine specifically for datasets, indexing repositories across universities, governments, and organizations worldwide, making it a strong first stop when you are not yet sure where your data might live.
Zenodo, maintained by CERN, hosts research outputs including datasets from a huge range of disciplines and assigns each one a permanent citable identifier, which is useful when you need to formally cite your data source.
Our World in Data combines cleaned, source-linked datasets with visualizations covering global development, health, and the environment, and is especially useful when you need long time series with clear provenance.
Government and international organization sources
World Bank Open Data offers extensive economic, social, and development indicators for countries around the world, going back decades in many cases.
Data.gov and equivalent national portals in many countries provide free access to government collected statistics across health, education, transportation, and more.
The World Health Organization Global Health Observatory is an essential source for public health research, offering standardized health data across nearly every country.
Many countries also maintain their own national bureau of statistics, which is worth checking directly if your research focuses on a specific country or region.
Discipline specific repositories
PubMed and NCBI are essential starting points for health and life science research, offering both literature and, in many cases, linked datasets.
IEEE DataPort serves engineering and computer science researchers with datasets tied directly to published technical research.
ICPSR, based at the University of Michigan, is a leading archive for social science data, including large scale survey and longitudinal datasets used across political science, sociology, and public policy research.
How to judge whether a dataset is trustworthy
Before you build your entire analysis around a dataset, check a few basic things. Confirm who collected it and when, since outdated data can quietly undermine an otherwise strong argument. Look for documentation explaining how variables were defined and measured. Check whether the dataset has already been used in published research, which is often a good sign of its reliability. Finally, always cite your data source properly, the same way you would cite a journal article, since transparency about your data is part of what makes your research credible.
Once you've sourced a dataset you trust, the next step is deciding how to work with it — see our guide on methods of analyzing research data for a clear overview of quantitative and qualitative approaches.