Overview
Survey Design
Trend Data
Availability of data
1. Overview ![Top of Page.](/statistics/graphics/uparrow6.gif)
a. Purpose
The Survey of Academic Research Instruments and
Instrumentation Needs is a congressionally mandated survey
that serves as the primary source of information on the need,
stock, cost, and utilization of research and development
equipment within academia in the United States. It is used by
Congress and Federal agencies in planning programs for
funding academic instrumentation.
b. Respondents
The survey is completed in part by departmental
chairpersons at selected institutions of higher education and
in part by individual investigators within the
departments.
c. Key variables
Adequacy of research equipment
Age of academic research equipment
Field of science and engineering
Instrumentation needs
Maintenance/repair expenditures for academic research
equipment
Provision for maintenance of academic research
equipment
Purchase price of academic research equipment
Type of equipment
Use of academic equipment (i.e., research, instruction,
combined)
2. Survey design ![Top of Page.](/statistics/graphics/uparrow6.gif)
a. Target population and sample frame
The 1994 institutional population consisted of the 214
colleges and universities (other than medical schools,
military schools, and academic FFRDCs) with nonmedical
R&D expenditures of at least $3 million in any year plus
the 104 medical schools with NIH extramural funding of $3
million or more. The instrument population consisted of
instrument systems originally costing $20,000 or more within
one of the following fields: agricultural, biological,
computer, or environmental sciences, chemistry,
physics/astronomy, and engineering.
b. Sample design
A three-stage sample design was used. The first stage
consisted of sampling institutions, the second stage
consisted of sampling research units (departments and
research facilities) within those institutions, and the third
consisted of sampling research instruments costing $20,000 or
more within the sampled departments and facilities. The
institutional sample size in 1992-1994 consisted of 55
institutions in the nonmedical sample and 24 medical schools.
These institutions were originally selected in 1986-87 using
stratified random sampling with probability approximately
proportionate to size. The measure of size used for the
nonmedical sample was R&D expenditures in S&E in FY
1984. For the medical sample, the measure of size was FY 1982
awards from NIH.
In the 1994 survey, the sampled institutions contained a
total of 1,541 research units identified as having at least
one instrument costing $20,000 or more in 1993. A total of
988 of these units were sampled. Sampling at the departmental
level was done only in the S&E fields with large numbers
of research units (engineering and the agricultural,
biological, and environmental sciences). Eligible units in
other science fields were sampled with certainty. A total of
34,508 instruments were identified as being within scope in
the sampled units. A sample of 8,784 instruments was selected
from those in-scope instruments identified. This total
includes 66 "supersystems," i.e., large specialized
research units built around a single integrated instrument
system that cannot be disaggregated in a meaningful way (such
as an observatory or a central computer center). All
identified supersystems were included in the survey.
c. Data collection techniques
The 1992-94 survey was conducted by Quantum ResearchCorporation under contract to SRS. Several different survey
instruments were utilized. A Department/Facility
Questionnaire was used to obtain information from heads of
departments and research facilities. This questionnaire
elicited information about the instrumentation expenditures
in the unit and opinions on the adequacy of research
instrumentation in the unit as a whole. An Instrument Data
Sheet was used to obtain information about instruments
costing at least $20,000 from principal investigators
responsible for them. Two different Supersystem Data Sheets
were used for collecting information about supersystems. One
questionnaire was used for central computer facilities and
the second for other supersystems.
Data were collected in two phases. During phase one
(collected in 1992), only the Department/Facility
Questionnaire was used to collect data at the unit level. In
phase two, information was collected for both units and
instruments, using all questionnaires.
Data collection was coordinated by institutional
coordinators appointed by the presidents or chancellors of
the institutions.
Except for the 1992 report, information about current
equipment needs and priorities is obtained with reference to
the year of the survey and information about dollar amounts
is for the year preceding the survey. All information for the
1992 survey is reported with 1992 as the reference year.
d. Estimation techniques
The nonmedical institutions were originally sampled in
1986, using 1984 expenditure information. The medical sample
was selected in 1983, using 1982 expenditure information. In
order to reflect changes in expenditures by institutions in
the intervening years, the 1992-1994 institutional sample was
restratified and sampling weights were adjusted accordingly.
Departments/facilities and instruments were weighted by the
inverse of the probability of selection in the 1992-1994
samples. All three stages were adjusted for nonresponse. At
the institutional level, adjustment was done within strata.
At the department/facility level weighting was done within
cells, defined by type of institution (medical versus
nonmedical), field of science and engineering (six
categories), size of the department (large versus small) and
type of control (private versus public). When cell size was
below 25, adjacent cells were combined to reach the minimum
size of 25. A total of 17 cells were used in 1992 for
academic departments and 7 were used for other
facilities.
e. Possible sources of error
(1) Sampling -
The coefficients of variation for overall estimates (not
broken out by field of S&E) of key variables are in the 2
to 6 percent range.
(2) Coverage -
Coverage problems could arise through incorrect
classification of colleges and universities as being
out-of-scope. Institutions may also incorrectly classify
departments/facilities, instruments, or supersystems. The
degree of such coverage error is not known.
(3) Unit nonresponse -
Of the 79 sampled colleges, universities, and medical
schools in 1992, 97 percent were able to provide at least
partial data. Usable questionnaire responses were received
for 84.4 percent of sampled departments/facilities. In order
to minimize the impact of this source of error, results are
adjusted for nonresponse through the use of statistical
weighting techniques. These unit nonresponse rates are within
the normally accepted ranges.
(4) Item nonresponse -
Item response rates ranged from 95.0 percent to 100.0
percent in the 1992 survey. Data for institutions that
partially responded were imputed for most questions using hot
deck procedures.
(5) Measurement -
The subjective nature of many of the variables in this
survey (e.g., adequacy of research equipment and
instrumentation needs) indicates that measurement error is
likely to be relatively high. Information on more objective
measures (e.g., maintenance/repair costs) is probably less
subject to measurement error. However, there have not been
empirical studies of measurement error for this survey.
3. Trend data ![Top of Page.](/statistics/graphics/uparrow6.gif)
This survey was first conducted in 1983-84. Surveys were
originally conducted on a triennial basis; beginning in 1992
they were conducted biennially. Changes in coverage may
impact some of the trends. Using a constant cut-off of $3
million in R&D expenditures as a minimum for inclusion
would result in more institutions being defined as part of
the sample frame if the only change over time was inflation.
This factor would contribute to a slight upward bias in trend
estimates for the total values of the variables over time. On
the other hand, the dollar value on instruments, as a minimum
for inclusion, was increased between 1989 and 1992 (from
$10,000 to $20,000). This would lead to a decrease in the
total dollar estimates of the stock of instruments.
4. Availability of data ![Top of Page.](/statistics/graphics/uparrow6.gif)
a. Publications
The data from this survey are published in several
different series of Detailed Statistical Tables. The most
recent report is Academic Research Instruments: Expenditures
1993, Needs 1994 (NSF 96-324). Reports are also
published on academic equipment in more narrowly defined
areas, e.g., Characteristics of Science and Engineering
Instrumentation in Academic Settings: 1993 (NSF 98-311).
b. Electronic access
Due to confidentiality concerns, data from this survey
are not available on CASPAR or on public use files.
c. Contact for more information
Additional information about this survey can be obtained
by contacting:
Leslie Christovich
Director, Academic Infrastructure Program
Research and Development Statistics Program
Division of Science Resources Statistics
National Science Foundation
2415 Eisenhower Avenue, Suite W14200
Alexandria, VA 22314
Phone: (703) 292-7782
Internet: lchristo@nsf.gov