|
|
|
|
|
|
|
|
|||||
    
|
     |
|
Division of Surveys and Technologies (DST)The Division of Surveys and Technologies (DST) constitutes over half of the staff of SRC and provides a wide range of survey design, data collection, and data processing services. DST staff members work closely with members of the Survey Methodology Program to assure that all projects utilize cutting-edge methodology and technical systems. Below, we highlight DST’s capabilities in three areas - sampling and analysis, data collection, and data processing.
Sampling and analysis.
DST maintains a national area probability sample for SRC projects. The first-stage units for the SRC national sample are a probability selection of counties or county groups that are permanently staffed by local interviewers. The sample is concentrated in geographic areas (PSUs) which are counties or Metropolitan Statistical Areas (MSAs). SRC maintains a local interviewing staff in these areas, usually a minimum of two interviewers per area. The national sample is readily adapted to smaller or larger sample size requirements or studies of special populations defined by the presence of a rare or specific characteristic. Although DST’s primary emphasis is on national household studies, other populations sampled include: Medicare and Medicaid populations; public and private schools and universities, their administrators, teaching staff, and students; hospitals, staff, and patients; county and municipal governments; business and industrial organizations and employees; civic, social, and professional groups and their members. Since 1991, DST has been using TIGER and GIS files to produce sample segment maps, perform stratification, and conduct spatial analysis of attribute data. TIGER is an acronym for Topological Integrated Geographic Encoding and Referencing. The TIGER/Line File is a database that contains digital data for all census map features (such as roads, railroads, and rivers) and the associated census geographical areas (such as census tracts and blocks), political areas (such as cities and townships), feature names and classification codes, census geographic area codes, FIPS codes, and within metropolitan areas, address ranges and zip codes for streets. The TIGER file covers the entire United States, Puerto Rico, the Virgin Islands, and other outlying areas. A Geographic Information Systems (GIS) software package can read a database of map coordinates and build high-resolution computerized maps and graphics. DST has been using TIGER and GIS files in sample design and development. The combination has been successfully applied to the production of segment maps for the listing of housing units and second stage stratification in multi-stage designs. Analysts in the Institute’s Survey Methodology Program and DST have developed a program, IVEware, which is an imputation/variance estimation application. IVEware uses a multivariate sequential regression approach to impute item-missing data and can create multiple imputed data sets. Please see reference: Raghunathan, T. E., Lepkowski, J. M., Van Hoewyk, J., and Solenberger, P. 2001 “A Multivariate Technique for Multiply Imputing Missing Values Using a Sequence of Regression Models.” Survey Methodology, Vol. 27, No 1. This application is currently available through the ISR website: www.isr.umich.edu/src/smp/ive/. |
||
 |
|
|
|
|
|
|
|
SRC Home | Project Homepages | Division of Surveys and Technologies | ||
|
|
|
|
Much of SRC’s work uses multistage area sampling of households.
Probability methods are employed in all stages of sample selection,
including random selection of respondents from households. Many SRC
samples have also been designed to represent populations of special areas:
individual states, countries, metropolitan areas, and cities; national
samples of work, school, and social organizations, and of their members;
research institutes and universities; members of professional societies
and other groups; and a multipurpose national probability sample of
non-households. DST also uses Random Digit Dialing sampling extensively
for telephone surveys of household populations.