At the very first stages of your literature review, start taking notes on potential data sources. Make a habit of jotting down the data used in each study you read to make it faster when you come back later in your search for data. Also, this practice can help you see and articulate how your contribution is unique. You might want to keep these notes in a table like the following for easy reference.
|Author(s) and Year of Publication||Claim||Data||Dependent Variable/Estimation Technique||Significant Findings|
See an example of this practice in action:
Most data catalogs describe datasets at the "study level." This means that when you search by subject or keyword, your search terms are matched against words used to describe the whole study broadly, not all the specific topics addressed within it. This practice is similar to the way books are described in the library's catalog.
However, it is often the case that you would like to search at the "variable level." For example, you might like to search for all surveys that ask a question about happiness. What is a researcher to do? Because of improvements in data description, some datasets are searchable by variable. Below are some places that allow variable searching. Be aware, though, that you cannot search all data by variable.
In your search for data you may need to compare similar questions from several surveys. Below is a grid to help you keep track of this process.
The term "raw data" is being used increasingly to mean many things, but in the world of social science research data, it means something much more specific. If you are looking for microdata (data at the individual level, analyzable at the level at which it was collected, e.g., persons, households, etc) formatted for use with statistical software, then you're in the right place.
Data in ASCII format, or formatted for use with statistical packages (e.g., SPSS, Stata, R), can often be found most efficiently by searching collections and archives of data.
Here are just a few prime candidates. For more specific recommendations and ideas, contact a librarian.
Many researchers want to make their data available to others to replicate their findings, provide transparency, and to encourage secondary research using their data. Here are some places to consider when looking for replication data.
Web sites with data almost certainly have the most current numbers, but don't always include data collected in previous years. The library's paper collections afford a wealth of places to find statistics and data beyond what is available from databases online. Books often contain appendixes with tables. Government and non-government agency reports and statistical compendia are likely to contain -- or consist nearly entirely of -- statistical tables and charts.
Most statistical publications are cataloged with a subject heading that contains the word "statistics," for example: Crime - Economic aspects - United States - Statistics.
Perform an Advanced Search with one dropdown set to "Subject contains" and type the word "statistics." Example catalog search for crime statistics.
Another plausible way to find older statistics is to browse the stacks. This is a useful way of getting around the problem of needing to guess the right words for an effective catalog search. The oversize stacks on 2nd contain a rich collection of statistical compendia.
Use the books and articles you already have in hand and skim the bibliographies and "Data" sections of the text for the names of studies, datasets, or collection agencies. Search for these names in the catalog.
For example, you might see the IMF's "Direction of Trade Statistics" cited frequently in the literature. Their web site provides access to the current data. Search Catalyst and discover that the annual index is available in its entirety back to 1962 on the second floor of the library.
Some datasets are distributed on CD-rom and are housed in the Library.
Any Keyword: cd-rom
Author Contains: "bureau of the census"
You could also start by browsing the catalog for data in the following topics:
Please feel free to reuse content from this page with acknowledgment.
Questions? Contact email@example.com