You are viewing a Web site, archived on 07:55:44 Nov 18, 2004. It is now a Federal record managed by the National Archives and Records Administration.
External links, forms, and search boxes may not function within this collection.
survey methodology

Survey Methodology:
Survey of Scientific and Engineering Research Facilities

1. Overview top

a. Purpose

The Survey of Scientific and Engineering Research Facilities is a congressionally mandated survey that serves as the primary source of information on the status pf and needs for science and engineering research facilities within research-performing universities and 4-year colleges in the United States. This information is vital for planners within the Federal Government and in academia.

b. Respondents

The survey is completed by survey coordinators at the sampled institutions.

c. Key variables

2. Survey Design top

a. Target population and sample frame

The population consists of institutions that grant graduate S&E degrees and/or annually perform at least $50,000 of separately budgeted research and development (R&D). All HBCUs with R&D programs are included, regardless of the level of R&D. Academically administered federally funded research and development centers (FFRDCs) are excluded. Total population size for the 1998 survey was approximately 660.

b. Sample design

A stratified probability sample of 350 institutions was selected for the 1998 survey. The universe was divided into nine strata. The strata were defined hierarchically. Stratum 1 was the top 100 institutions in terms of 1993 R&D expenditures. Stratum 2 was the original 29 HBCUs in the sample since the 1988 survey. Stratum 3 contained the remaining 35 HBCUs not in the original survey. Stratum 4 was institutions which enrolled at least 25 percent Black students but were not HBCUs. Stratum 5 was institutions which enrolled at least 25 percent Hispanic students. The first five strata were mutually exclusive.

The remaining institutions were placed into strata based on their institution type (i.e., doctorate granting versus nondoctorate granting) and institutional control (i.e., public versus private). Within strata, institutions were sampled proportional to size. Size was defined as the square root of the 1993 expenditures in thousands.

c. Data collection techniques

The 1998 survey was conducted by Gallup, Inc., under contract to SRS. Survey instruments were distributed to institutional coordinators for completion. Respondents could provide their data using either a paper and pen method or a web-based survey. For both methods, telephone and mail followup was used.

d. Estimation techniques

A base weight was created for each institution equal to the inverse of the probability of selecting the institution into the sample. Because some institutions in the sample did not respond to the survey, the base weights were then adjusted to account for unit nonresponse. Finally, weights were adjusted again to make the number of estimated institutions equal to the known number of institutions in various categories. For the poststratification adjustment, the institutions were classified by type, control, and HBCU status.

3. Survey Quality Measures top

a. Sampling variability

Standard errors for selected 1998 estimates may be found in Scientific and Engineering Research Facilities at Colleges and Universities: 1998. The coefficient of variation for total research square footage is approximately one percent. Coefficients for other variables are generally larger.

b. Coverage

Coverage at the institutional level is believed to be quite good, given that it is relatively easy to identify schools that grant master's or doctoral degrees in S&E fields. However, it is possible that some universities with separately budgeted R&D expenditures of $50,000 or more and/or some HBCUs are inadvertently excluded.

c. Nonresponse

(1) Unit nonresponse - Of the 350 institutions in the 1998 sample, 87 percent were able to provide at least partial data. Statistical weighting techniques were used to compensate for the underrepresentation caused by the missing institutions. There may, however, be some residual nonresponse bias attributable to uncontrolled differences between responding and nonresponding institutions.

(2) Item nonresponse - Data for institutions that partially responded were imputed using information from the 1996 survey, when available. For items for which cost and net assignable square footage was requested, estimates were used when only one of the two characteristics was missing. Hot deck procedures were used in other cases. The highest nonresponse rate in 1998 was the costs for repair/renovation and new construction of research space needed but not funded. All items had response rates over 93 percent.

d. Measurement

In order to ease the institutional reporting burden, respondents were asked to report on separately budgeted R&D activities. This provides information in a form compatible with Office of Management and Budget collections. However, this definition excludes some multi-use space from consideration.

A number of the variables that are not universally collected must be estimated by universities. Other variables, such as the condition and adequacy of research facilities, are by their very nature subjective. Estimates of variables in these categories are prone to measurement error.

4. Trend Data top

This biennial survey was first conducted in 1988. Improvements in coverage may impact some of the trends. Further, using a constant cut-off of $50,000 in separately funded R&D expenditures as a minimum for inclusion for nongraduate granting programs would result in more institutions being defined as part of the sample frame if the only change over time was inflation. Both of these factors would contribute to a slight upward bias in trend estimates for the total values of the variables over time.

5. Availability of Data top

a. Publications

The data from this survey are published annually in Detailed Statistical Tables in the series Scientific and Engineering Research Facilities.

Information from this survey is also included in Science and Engineering Indicators.

b. Electronic access

Data from this survey are available in electronic format from NSF's Online Documents system.

Due to confidentiality concerns, data from this survey are not available on WebCASPAR or as microdata files.

c. Contact for more information

Additional information about this survey can be obtained by contacting:

Leslie Christovich
Director, Academic Infrastructure Project
Research and Development Statistics Program
Division of Science Resources Studies, Room 965
National Science Foundation
4201 Wilson Boulevard
Arlington, VA 22230
Phone: (703) 292-7782
Internet: lchristo@nsf.gov
Last Modified: Oct 16, 2003 Comments to srsweb@nsf.gov