American Institutes for Research


Job ID
United States


The American Institutes for Research (AIR) is one of the leading behavioral and social science research organizations in the world. AIR is a collaborative organization that works with clients to examine a wide range of projects from early childhood development to adult education. We pride ourselves in conducting high-impact, high-stakes projects for significant federal, state, and private organizations. AIR's vision is that research-based problem solving can improve the lives of all people.

AIR’s Research and Evaluation area is seeking a Researcher to join our team in Washington, DC. The researcher would contribute to the development of statistics about education that inform the discussion, debate and planning of decision-makers at national, state, and local levels through a contract with the National Center for Education Statistics (NCES). NCES is working to analyze methodological data from the 2016 and 2017 administrations of the National Household Education Survey as well as develop adaptive design protocols in which data collection procedures (e.g., mode, incentives, questionnaire length, etc.) will be tailored based on known characteristics of sampled households. To inform this work, AIR is analyzing a wide array of experimental data from the 2016 and 2017 collections as well as developing models aimed at accurately predicting households’ probability of response under various designs, using a wide array of auxiliary data.


AIR is seeking a statistician or data scientist to conceptualize, build, and validate predictive models of survey response propensity to inform the targeting of data collection procedures (e.g., mode sequence, incentives, questionnaire length, etc.) and design and conduct statistical analyses of methodological experiments embedded in the 2016 and 2017 data collections.


• Ph.D. in statistics, economics, educational policy, or other related social sciences field;
• Expertise with survey and experimental data;
• Understanding of parametric and non-parametric approaches to predictive modeling of binary and categorical outcomes (e.g., logistic regression, classification/regression trees, random forests);
• Knowledge of and experience with model-building and validation techniques (e.g., cross-validation);
• Proficiency with statistical software packages such as R, Stata, or SAS/WPS;
• Ability to clearly communicate methods, results, and recommendations to clients and other non-technical audiences on an ongoing basis;
• Ability to work independently and as part of a team, including the ability to maintain a professional and effective working relationship with project staff and clients;
• Ability to work under conditions requiring careful attention to detail.


Sorry the Share function is not working properly at this moment. Please refresh the page and try again later.
Share on your newsfeed