NSF BLS Fellowships

Contents:

General Information

The American Statistical Association (ASA), under a grant from the National Science Foundation (NSF), and in cooperation with the Bureau of Labor Statistics (BLS), announces the Senior Research Fellow Program. The program's objective is to bridge the gap between academic scholars and government social science research. Research Fellows are provided the unique opportunity to address some of the complex methodological problems and analytic issues relevant to BLS programs. Fellows will conduct research in residence at BLS, use BLS data and facilities, and interact with BLS staff.

Fellowship applicants should have a recognized research record and considerable expertise in their area of proposed research. A detailed research proposal must be submitted for competitive evaluation by a Program Review Board consisting of representatives of ASA, BLS, the American Economic Association, and the American Association of Public Opinion Research. The proposal will be evaluated on the applicability of the research to BLS programs, the value of the proposed research to science, and the quality of the applicant's research record. Qualified women and members of minority groups are encouraged to apply.

Current research being conducted by economists, mathematical statisticians and behavioral scientists at BLS includes:

  • Price index number theory and measurement, impacts on occupational injuries, the timing of and the returns to employer training, an examination of divergent real wage trends, analysis of contingent work, the definition of establishment births and deaths, incorporating the effect of composite estimation into microdata weights, developing improved generalized variance functions for small domain estimates, and small area labor force estimation
  • Model-based seasonal adjustment, variance estimation for seasonally adjusted time series, applications of mathematical programming to sample allocation problems, improvement of confidence intervals for small domain estimates, application of measurement error models to employment series, development of expert systems for data editing and quality control, improvement of data collection methodologies for consumer expenditures, and demand models for forecasting components of gross domestic product, technology, or composition of employment
  • Respondents' motivation and survey nonresponse, evaluation of data quality provided by proxy respondents, application of expert system technology to unstructured interview situations, pen computing for data collection, and human factors associated with touch tone data entry systems.

Research Topics

BLS is the principal fact-finding agency for the Federal Government in the broad field of labor economics and statistics. It is responsible for the development and publication of data on employment and unemployment, prices and living conditions, compensation and working conditions, productivity measurement, and economic growth and employment projections. These data are largely based on establishment surveys conducted either by BLS or under a Federal-State cooperative program, and household surveys conducted for BLS by the Bureau of the Census.

Researchers have a unique opportunity to make major advances in methodology and applications in a number of areas. Those listed in this brochure do not exhaust the full range of possibilities. Applicants are encouraged to propose studies in other areas provided they are in socioeconomic or methodological areas that are applicable to BLS data, or have the potential of encouraging further significant broad-based research.

Statistical Methodology and Computing

Proposals in this area should be for research projects generally applicable to the collection, processing, or analysis of BLS data.

Sampling Frames

A potential source of nonsampling error for sample surveys is the universe frame. Error types include both coverage (undercoverage, overcoverage, and duplicate coverage) and content (accuracy of geographic and industry coding). The Bureau's universe frame is an administrative data file of over 6.5 million establishments. Limited research is being conducted at BLS in the areas of new business formations (births), quality of industry coding, and microdata editing techniques. Areas for additional research include measurement of undercoverage, potential impact of undercoverage on economic indicators, methods for coverage improvement (including dual-frame techniques), quality control programs for large databases, longitudinal matching, and imputation techniques.

Time in Sample Effects

For many BLS establishment surveys, for example, the National Compensation Survey and the Current Employment Survey, the sample of establishments is replaced on an on-going basis. Replacement schemes vary; for instance, an establishment may be necessarily replaced by one within the same Standard Industrial Classification (SIC). It is generally recognized that some "replacement bias" occurs. It is known, for example, that for occupational wage levels in the Employment Cost Index, recently initiated establishments have lower wage levels than those in the on-going sample. Potential areas of research could estimate this effect over time, determine estimators of the bias, and ultimately develop more accurate estimators. Investigation of possible root sources of this phenomenon could be carried out.

Nonsampling Errors

Some sources of survey errors are cognitive and social in origin. In recent years, behavioral science has been increasingly applied toward understanding psychological, sociological, and anthropological factors in the survey data collection process. For example, respondent motivation, comprehension, retrieval, and response processes have all been found to affect the quality of survey data. Research at BLS might include classification systems and classification error, organizational structure and technological change, survey nonresponse, respondent-interviewer interaction, human-computer interaction in data collection and analysis, data quality and respondent data quality and respondent rules, and autobiographical memory and everyday cognition. Research could be conducted in a field setting or utilizing the BLS Behavioral Science Research Laboratory which includes a fully equipped usability lab as well as general purpose research and observation rooms with one-way mirrors, videotaping capabilities, and facilities for running computer administered experiments.

Time-Use

In January 2003, BLS launched the American Time Use Survey, a nationwide, computer-assisted telephone survey designed to collect detailed information about how respondents spend their time. This new survey provides an opportunity for methodologists to propose reliability and validity studies. The data also provide a rich source of information useful for economic, psychological, and sociological research. Studies that could be conducted might include, but are not limited to: analysis and valuation of unpaid work to provide a more comprehensive picture of aggregate economic output, consumption, growth, productivity, and technological change; comparisons of time spent in paid work activities reported through the time diary and other labor-force surveys; examinations of the amount of time spent waiting for various services and possible effects on costs; assessments of the quality of American life; re-analysis of poverty to include details about home production and leisure activities; modeling the trade-off between consumer goods and time; and analyses of time spent in caregiving of household members.

 

Computer-Assisted Interviewing

Computer-assisted telephone interviewing (CATI) is currently being developed and implemented in a number of surveys conducted by BLS. This collection methodology affords the opportunity to conduct research on nonsampling error sources, including the measurement and control of response bias and variance, interviewer variability, and the identification of problems with survey concepts (specification error). Continued research is needed to improve the use of CATI for reducing sources of nonsampling error and to evaluate the feasibility of computer-assisted personal interviewing (CAPI).

Statistical Quality Control

BLS is continuing to strengthen its quality control program. Of special interest is the development of additional techniques for assessing the quality of data obtained from establishment surveys. Some areas for research include integrated overall error models, with provisions for estimating magnitudes and interrelationships of various error sources; methods of adapting evolutionary operations techniques to BLS operations; effective patterns of quality control feedback, training, and incentives applicable to establishment surveys; redesign of management information systems to incorporate process performance and data quality monitors; and development and measurement of proxy variables for estimating components of nonsampling error.

Cost-Error Modeling

The Bureau of Labor Statistics is currently investigating the requirements for developing a complex model for operational costs of data collection and processing for use in the sample design for the next revision of the Consumer Price Index. Topics for research in this modeling effort include repaneling of the Continuing Point-of-Purchase Survey and consequent paneling effects on CPI initiation and data collection, assessment of data collection operations with respect to evaluating workload requirements, and development of integer mathematical programming strategies for determining optimal design characteristics given data collector workload and distribution constraints.

Item Imputation

Diversity in the types of BLS surveys provides a wide variety of missing-data problems. Some examples are longitudinal imputation techniques for establishment surveys where units remain in the sample for 4 to 5 years, multivariate imputation techniques for surveys like the Consumer Expenditure Survey, which collects data on all sources of income and all types of expenditures, and multiple imputation techniques for computing employment and wages on the BLS data file of establishments.

Expert Systems for Data Access and Use

For many BLS surveys, staff review data collected each month or quarter for period-to-period comparability. Development of expert systems using artificial intelligence techniques should reduce the error and improve the knowledge base associated with this review. Research is needed to assess how artificial intelligence techniques could be used and implemented at BLS.

Information Dissemination

BLS provides public access to micro-, macro-, and metadata via the Internet. Research is needed on how to incorporate expertise on the use of statistical information into an on-line public access system, on tools that facilitate data analysis, and on interface design.

Statistical Graphics and Data Visualization

BLS publishes large volumes of economic data on employment, wages, and prices. Research is needed on improved ways of graphically presenting cross-sectional and time series data that is normally published in tables. Software also needs to be developed to allow users to do interactive analyses through the BLS Internet homepage and over the BLS Intranet.

Estimation

BLS conducts several continuing surveys in which estimates of the same quantity are produced periodically. The Employment Cost Index and the Employee Benefits Survey, for example, estimate employer costs, and benefit levels annually from overlapping samples of establishments. Two potential areas of research are the application of time series methods to improve current period estimates and the estimation of variances for the resulting estimators.

Time Series Methods

Recent years have seen broad advances in the field of time series modeling and further development of these methods and their application is needed. A broad array of monthly and quarterly time series are available from the Bureau database for study and testing. Topics of particular interest include application of model-based seasonal adjustment methods, estimation of standard errors for seasonally adjusted estimates, particularly from complex surveys, treatment of outliers in time series modeling and seasonal adjustment, multivariate time series modeling, and survey estimation incorporating time series methods.

Statistical Methods for Data Analysis

Inference based upon data from complex sample surveys is a broad area of research. Examples are hypothesis testing between mean vectors of nonnormally distributed variables, analysis of truncated distributions and mixture distributions, analysis of longitudinally linked data sets from continuing surveys and graphical analysis techniques.

Statistical Disclosure Limitation Methodology

To protect the confidentiality of the statistical tables and microdata files that are released, BLS programs apply a variety of statistical disclosure limitation procedures. A recent report of the Federal Committee on Statistical Methodology, Report on Statistical Disclosure Limitation Methodology (Office of Management and Budget, Statistical Policy Office, 1994), reviewed the current disclosure limitation practices used by federal statistical agencies and made recommendations to improve these practices. Areas of research include the use of linear programming techniques to select cells for complementary suppression in statistical tables and methods that improve the confidentiality protection of sets of tables that are interrelated.

Economic Measurement and Research

Proposals under this topic would include theoretical or empirical studies designed to improve BLS statistical programs. Research could focus on defining and refining the concepts to be measured or could analyze a variety of topics relating to labor markets, prices, and productivity.

Measurement of Labor Force Characteristics

Accuracy in measuring unemployment depends upon the ability to make meaningful distinctions between the various labor force concepts particularly between unemployment and not in the labor force.. Areas where these issues might be researched include the measurement of flows between labor force categories and longitudinal uses of existing microdata sources. Research could be conducted on adjustment of flows for rotation group effects and classification error. Research is also possible on a wide variety of topics related to the comparative labor market situation of various groups of workers at a point in time or over time.

Output Definition

The need to define output arises in the development of BLS data on producer prices (the unit of output to be priced must be specified) and on productivity (labor productivity is defined as output per employee hour). Although there is some difficulty in defining output for goods-producing industries, the problems of defining output for many of the service-producing industries are particularly complex.

Incidence of Injuries and Illnesses

The BLS Annual Survey of Occupational Injuries and Illnesses measures the incidence of injuries and illnesses by an establishment survey. It has long been recognized that the survey fails to measure a large proportion of diseases which are chronic and have long latency periods or for which occupational origin is not obvious. Research in this area might explore solutions to this problem that are available through the analysis of extant data sets, including the National Health Interview Survey, the National Health and Nutrition Examination Survey, and State and local data on mortality.

Measurement and Analysis of Nonwage Benefits

A number of BLS establishment surveys collect data on wages and other employee compensation. The Employment Cost Index measures the rate of change in total compensation, which includes wages, salaries, and employers' cost of employee benefits. Other BLS surveys, such as the Occupational Compensation Survey Program, collect information on the dollar amount of wages, but only on the incidence of nonwage benefits. Others, such as the Current Employment Statistics Program, provide no information on benefits. Research is needed to determine how benefits are valued by employees and how these valuations differ from employer costs.

Measurement of Economic Growth

Every 2 years, BLS publishes long-term economic projections of employment by industry and occupation based on a series of economic models plus analysis of industrial and occupational trends. Possible areas of research include analysis of how changes at the macroeconomic level (energy prices and terms of trade) and the microeconomic level (technological change and institutional change) affect the projections, methodology for analysis of structural change at the macro and microeconomic levels, detailed studies of specific technologies focusing on the diffusion of a technology throughout the economy and its impact on employment and studies of dynamic input-output or industry-occupation models.

Productivity Research

Productivity-related research includes factors affecting productivity change such as investments in research and development, and the education and experience of the workforce, issues affecting the measurement of productivity, modeling and estimating the effects of relative price changes on the measurement of capital input, selected industry studies with emphasis on explaining productivity growth, cross-industry comparisons of productivity growth, and international comparisons of productivity growth.

Price Measurement

A number of long-standing problems in price measurement would be suitable topics for research. They include treatment of seasonal goods, new outlets and innovations in distribution, seasonal goods. treatment of style and quality changes, and measurement of prices for the consumption of services derived from consumer durables. An in-depth analysis of measurement problems and potential biases for a specific component of the index (e.g. medical services or consumer electronics) would also be a suitable topic for research.

Analysis of Labor Markets

Among the research topics related to labor markets are analysis of sources of wage differentials, issues relating to work and the family, problems of particular groups of workers such as older workers and minorities, job stability and job security, and estimating the determinants and/or impacts of changing work arrangements such as contracting-out employment, expanded use of temporary help workers, and growth of part-time employment. BLS is currently constructing a longitudinal database that will contain quarterly employment and wage data for virtually all business establishments in the U.S.; research topics that highlight the uses of this database, such as analysis of the birth, growth, and survival of establishments or how businesses adapt to structural and technical change are encouraged. Studies which are "measurement oriented" in that they use data in new ways, yield insights into what concepts to measure, and/or result in the improvement of microdata files are of particularly high priority.

Conditions of Appointments and Benefits

Research Fellows must conduct their research at the Bureau of Labor Statistics, 2 Massachusetts Ave., NE, Washington, DC. Fellows are reimbursed by the ASA and are on a guest worker arrangement with BLS.

Salaries received by Fellows will be commensurate with their qualifications and experience. Fringe benefits, travel, and relocation support are negotiable. Arrangements can usually be made to retain fringe benefits with the permanent employer.

Fellowship appointments typically range from 4 to 6 months, subject to the availability of program funds. Appointment terms are flexible and can be full time, part-time or split into two separate terms. Applicants should specify approximate dates for proposed projects.

Fellowship resources at BLS include technical and secretarial support; library facilities; and access to in-house data bases and computer facilities including a UNIX workstation, Pentium personal computers connected by a local area network, micro-to-mainframe communication links, and computer graphics facilities. Funds are available to accommodate specialized needs for microcomputer software and hardware.

Application Procedure

Applicants must provide the following:

1. Curriculum vitae

2. Names and addresses of three references

3. Three (3) copies of a detailed research proposal that includes:
(a.) A short descriptive project title (b.) An abstract one-half page or less) (c.) A proposed project term (approximate dates) (d.) Background information on research topic, references, etc. (e.) A statement of relevant work already accomplished (f.) Proposed research with sufficient detail for evaluation of expected results (g.) A statement regarding significance of expected results (h.) Statement citing the advantages of conducting the research at BLS (i.) Resource requirements (including salary, estimated relocation and travel expenses, and research support)

The application deadline is December 10. Final decisions will be made by late February. Send application materials to:

Joyce Narine
Grants Manager
American Statistical Association
732 Washington Street
Alexandria, VA 22314
phone: 703-684-1221
fax: 703-684-8069

Further information about the program or specific research topics can be obtained by sending an email message or by contacting either of the following Bureau of Labor Statistics representatives:

Dr. Jean Fox Office of Survey Methods Research 2 Massachusetts Avenue, NE, Room 4915 Washington, DC 20212 Tel: (202) 691-7370

Dr. James Spletzer Office of Employment Research and Program Development 2 Massachusetts Avenue, NE, Room 4945 Washington, DC 20212 Tel: (202) 691-7393

 

Last Modified Date: July 19, 2008