Datasets may need to be handled differently depending on their subject and the kind of data collected.
Generally, keep these things in mind when working with other people's data:
Not all datasets collected by the government may stay accessible online.
There are some alternative ways to find federal datasets that have been removed from government websites and databases.
USA Marx Library's Government Documents section contains many print copies of historic federal datasets. There are several LibGuides for how to search our government documents by topic. You can start with the FDLP Collection: Home.
For digital datasets, there are different alternative databases and websites that you can check:
These are some data repositories. They are either general and cover a wide range of topics, or they focus on social sciences.
You can search many data repositories the same way you search databases for academic articles: you build a search string by combining search terms with Boolean Operators.
Example: I want this kind of data: college students answering questions about their mental health
I can use this search string: (College OR university) AND student AND (questionnaire OR survey OR interview OR "focus group" OR qualitative) AND ("mental health" OR depression OR anxiety OR "mental well-being" OR "mental illness")
Statistics are a processed version of datasets.
Example: A dataset would be a spreadsheet of everyone in a county and their race. A statistic would be that 45% of that county is Black.
When finding and using statistics, first check for these qualities: