Find out which data a research paper used and where it came from
A paper's conclusions are only as good as its data, but the data description is often split up: the source in the methods, the preprocessing in a supplement, and access terms in a data availability statement at the end. Ask Search+ "What datasets does this paper use, and can I access them?" and the answer gathers those passages with citations, so you can see the source, the time period, the filtering and the access route in the authors' own words.
Creating an account needs no payment.
Last updated October 2026
How to trace a paper's data with Search+
- Upload the paper with its supplement
Data dictionaries, inclusion filters and variable lists are often moved to supplementary material, sometimes as an Excel file. Add the supplement alongside the article so one question can reach both.
- Ask about source, then processing, then access
Ask where the data came from, then how it was cleaned or filtered, then how others can obtain it. Each is usually described in a different place.
- Follow named datasets back to their papers
If the authors use a public dataset described elsewhere, add that dataset's own paper to the workspace and ask about collection methods across both.
Questions to ask about a paper's datasets
Which datasets or databases does the paper use, and who collected them?
What period and which regions or institutions does the data cover?
Which records were excluded, and how many remained after cleaning?
What does the data availability statement say, and is the data public, restricted or available on request?
Is the analysis code available, and where do the authors say it can be found?
Across the papers in this workspace, which ones use the same public dataset?
What to look for in a paper's data description
Some papers collect new data; many analyse existing datasets, registries or benchmarks. The difference changes what you need to check: collection procedures for the first, version and selection for the second.
Public datasets are updated over time. Look for the version, release date or access date the authors report, because results can shift between versions.
Exclusions for missing values, duplicates or outliers decide what was actually analysed. Ask how many records were removed and why.
Statements range from open repositories with identifiers, to access on reasonable request, to data that cannot be released for privacy or contractual reasons. Read the exact wording rather than assuming.
Search+ finds the relevant passages by meaning, so a question about "where the data came from" can surface a methods paragraph that never uses the word dataset, with a citation you can open.
Data wording you may see, and what to ask next
| Wording you may see | What it usually signals | A follow-up question |
|---|---|---|
| "publicly available at" with an identifier | Open data in a repository | Which version or release did the authors use? |
| "available from the corresponding author on reasonable request" | Access controlled by the authors | Does the paper explain what conditions apply? |
| "secondary analysis of" | Existing data reused for a new question | Was the original data collected for a different purpose? |
| "after excluding" | Records removed before analysis | How many were excluded at each step? |
| "proprietary" or "licensed from" | Data others may not be able to obtain | Who owns the data, and could the study be reproduced? |
| "code is available" | Analysis scripts are published | Where is the code, and does it cover all reported results? |
What is a dataset in a research paper?
A dataset is the collection of observations a study analyses, such as survey responses, measurements, records, images or text. A paper usually describes where it came from, what it covers, how it was processed and whether others can access it.
Dataset questions
Can Search+ list the datasets a paper used?
Will it find the data availability statement?
Can it tell me if the data is good enough?
Can I see which papers use the same dataset?
Can Search+ open the dataset itself?
What if details are only in a supplement?
Check the data before you trust the result
Start a workspace, upload the paper and its supplement, and ask where the data came from.
Start a workspace