November 2021 -- The NSF National Center for Science & Engineering Statistics (NCSES) has released the Restricted Data Analysis System (RDAS), an online tool (https://ncsesdata.nsf.gov/rdas/#/
) that expands public access to data from the Survey of Earned Doctorates (SED). The SED is an annual census of approximately 55,000 individuals receiving a new research doctorate from U.S. colleges and universities. The SED collects information on the doctorate recipient’s educational history, funding sources postgraduation plans, employment, and demographic characteristics.
The SED RDAS data is based on a sample subjected to statistical procedures to produce estimates that reflect the entire population of doctorate recipients and protect the confidentiality of individual level data. The RDAS provides a secure platform to access a comprehensive set of SED variables and empowers users to create complex data tables at the click of a button. The RDAS provides information on the doctorate recipient’s educational history, funding sources, postgraduation plans, and demographic characteristics.
Unlike other SED data products, the RDAS is based on a sample of SED data that have been subject to nonresponse weight adjustment and missing data imputation, as well as statistical disclosure avoidance methods, which modify the data to protect disclosure of confidential information. As a result, the RDAS produces estimates that may differ from published tables; however, the missing data treatment adds value to SED data by allowing users to create tables that provide estimates for the full population using responses from all survey respondents. Further, it makes the number of doctorates in different RDAS tables internally consistent rather than being limited to those who responded to the items represented in each table.
The initial SED RDAS was created using data only for 2018 and 2019. Over time, additional years will be added. SED data are also available in the NCSES Table Builder which contains SED data going back to 1958 but only for a limited number of variables (institution, field of study, race and ethnicity, sex, and citizenship), while the RDAS contains a more comprehensive set of variables (50 data items).