BAA Research Outcomes

The Broad Agency Announcement (BAA) provides research opportunities primarily to U.S. institutions of higher education and their collaborators to conduct a variety of research projects that support the strategic objectives of NCSES and partner federal statistical agencies. Project reports for past BAA awards are organized by area of research and available for download below. 

Longitudinal and Condensed Survey Design

Creating Longitudinal Panels for the National Survey of College Graduates

Awardee: NORC at the University of Chicago
Award Year: 2020
Summary: Four shortcomings of NCSES workforce surveys were investigated through this research, most of which are discussed in the 2018 Committee on National Statistics report, Measuring the 21st Century Science and Engineering Workforce Population: Evolving Needs. This research team undertook a multifaceted exploratory project focused on the National Survey of College Graduates (NSCG) to investigate potential sources for improving the sample as part of a comprehensive NCSES data collection program on the scientific and technical workforce. The resulting paper presents analytic findings and recommendations in support of NCSES's research goal of moving toward longitudinal and condensed designs for ongoing surveys.

Exploring Adaptive Design Options for the Survey of Doctorate Recipients

Awardee: University of Michigan with University of Maryland
Award Year: 2020
Summary: Adaptive survey design approaches for the Survey of Doctorate Recipients (SDR) were examined with the goal of identifying rules that formalize key decisions related to the survey, including when to switch from one mode to the next and when to stop or reallocate efforts. Defined rules were tested on existing data using predicted survey outcome variables as inputs to simulate results on existing SDR data. The findings are presented in the final report.

Modular Survey Design and Smartphone Applications: Considerations for the Survey of Doctorate Recipients

Awardee: University of Michigan with University of Maryland
Award Year: 2020
Summary: Multiple approaches were used to lay the groundwork for future use of smartphone apps and modular design approaches as a method for data collection by NCSES. After gathering information about best practices for using these methods in large-scale surveys and investigation of participants' reactions to the methods for the Survey of Doctorate Recipients (SDR), the resulting report presents findings on the evaluation of the methods for SDR data collection. 

Reducing Respondent Burden for NCSES Surveys

Awardee: University of Michigan with University of Maryland
Award Year: 2020
Summary: Data collection strategies employed by NCSES for the Survey of Doctorate Recipients (SDR) were examined with the long-term goal of modernizing these strategies. The report identifies ways in which the time to administer the SDR can be reduced and considers the quality of information used in research involving the SDR through exploration of available administrative and survey data sources. The key outcomes presented in the report evaluate the extent to which respondent burden can be reduced using linked datasets and paradata.

Split Questionnaire/Matrix Sampling

Split Questionnaire Design Literature Review

Awardee: RTI International 
Award Year: 2020
Summary: Split questionnaire design (SQD) was focused on for this project, offering an overview of the practice, a review of methods to split questionnaires and assign to respondents, and a review of the more relevant imputation methods and issues related to these methods. Relevant findings on SQD and omissions that merit further research are presented in the full literature review. This report was the first step in a simulation study on SQD to help inform its design.

Visualizing Research Networks

The Evolution of Scientific Literature as Metastable Knowledge States

Awardee: Quantitative Scientific Solutions (QS-2) with Penn State University
Award Year: 2020
Summary: This research addresses the longstanding problem of identifying common concepts in the sciences. The results presented in this paper suggest that through the combined use of natural language clustering and citation graph analysis, the evolution of ideas over time can be predicted, allowing researchers to connect a single scientific article to past and future concepts in a way that exceeds traditional citation and reference connections.