Upload
benjamin-schroeder
View
215
Download
1
Tags:
Embed Size (px)
Citation preview
Survey of Science and Engineering Research FacilitiesFiscal Year 2005
Dual Mode Data Collection Experience
Timothy Smith, Cindy Gray, and Leslie J. Christovich
Presented at ICES-III
Montréal, Québec, Canada June 18-21, 2007
2
Discussion topics
Discussion will focus on the FY 2005 Survey of Science and Engineering Research Facilities:
Content and burden
Key characteristics of dual mode design and data collection procedures
Response rates
Differences between web and mail respondents
Lessons learned
3
1986 – Congress mandates survey
Collection of information on amount and condition of science
and engineering (S&E) research space
Academic and nonprofit biomedical institutions are included
Biennial data collection begins using paper questionnaire
1996 – Electronic version of the survey is introduced
Windows-based application on diskette
27% of respondents use the diskette version
Background
4
1998 – NSF introduces the web survey
Respondents are given option of responding by web or mail
53% of respondents submitted by web
2003 – NSF adds a second component to the survey
Focuses on computing and networking capacity
Requires involvement of different types of survey respondents
Background (continued)
5
A census of eligible academic and biomedicalinstitutions:
477 public and private colleges and
universities
191 independent hospitals and other nonprofit
biomedical organizations
Survey population
6
Institutional coordinators (ICs) were assigned by the president/director
ICs identified and collected data from institution offices: Facilities Office Office of Sponsored Research Budget and Planning Offices Medical School representatives Department deans Chief Information Officer Others
Survey respondents/contributors
7
Part 1:
Current research space in net assignable square feet (NASF) by S&E field
Condition of current NASF by field
Repair/renovation costs by field
New construction costs and NASF by field
Survey content
8
Part 1:
Planned repair/renovation (costs) and new construction (costs and NASF) by field
Deferred repair/renovation and new construction costs by field
Research animal facilities and medical school facilities
Survey content (continued)
9
Part 2: Bandwidth to Internet1 and Internet2,
current and estimated for FY 2006
Types of connections to Internet1
Bandwidth from consortia (e.g., state or regional networks)
High performance network connections (e.g., Abilene, National LambdaRail)
Survey content (continued)
10
Part 2: High performance computing systems (i.e.,
number of systems and access)
Speed of connections between desktop ports, along internal networks, and to Internet1
Wireless connectivity
Survey content (continued)
11
Academic institutions – 41 hours
Biomedical organizations – 7 hours
Survey burden
12
October 24 – Recruitment packages sent to presidents/directors
October 26 – Survey packages sent to “repeat” ICs (others sent on a flow basis)
November 3 – Start of email and telephone verification of survey receipt
December 8 – Email reminder one week prior to initial due date (December 15)
January through April – Telephone and email prompt of nonresponding institutions
April 12 – End data collection
Data collection schedule
13
Survey response rates
14
Institutional Coordinator (IC) package sent by
U.S. mail:
NSF cover letter Paper questionnaire (Part 1 and Part 2 bound
separately) with perforated pages Web survey access instructions Copy of NSF FY 2003 Survey InfoBrief
ICs had choice to respond by web or mail
Data collection procedures
15
Question wording and page formatting were
consistent across modes
Questions could be answered in any order
“Print Paper Survey” function provided on
website
Data collection procedures (continued)
16
List of Survey Questions Screen
17
Example question
18
Example question (continued)
19
Preference for paper questionnaire as data preparation tool:
“Portability” of instructions and definitions during
record searching
Distribution of questions to various offices/departments
Collection of same data elements from multiple
offices/departments
Maintain control of final data for submission
Preparation versus response
20
Preference for web survey as data submission tool:
Population uses the web extensively
Survey materials and contacts emphasize web submission
On-line edit checks (almost 100) help identify problems Documentation of completed questionnaire (i.e., question
text, instructions/definitions, responses) can be downloaded
Continued access to survey responses after submission and after edit resolution follow-up
Preparation versus response (continued)
21
Response mode by type of institution
Table 2. Survey response mode, by type of institution: FY 2005
(Number and percent of institutions)
All Web respondents Mail respondents
Type of institution respondents Number Percent Number Percent All institutions 630 580 92 50 8
All academic 452 433 96 19 4 All biomedical 178 147 83 31 17 SOURCE: National Science Foundation/Division of Science Resources Statistics, Survey of Science and Engineering Research Facilities, Fiscal Year 2005.
22
Response mode by type of academic institution
23
Response mode by type of biomedical institution
24
Median number of days for submission
All institutions:
69 days for web respondents
68 days for mail respondents
Academic institutions:
76 days for web respondents
89 days for mail respondents
Biomedical institutions:
50 days for web respondents
56 days for mail respondents
Response time
25
Pretesting, site visits, workshops, etc.
Paper questionnaire
General definitions at start of paper and web surveys
Key definitions and instructions on multiple pages
and at time needed
Gray shading of non-applicable survey questions on
web
On-line edits with respondent messages
Communication/Comprehension
26
Median number of edit problems
All institutions: 2 edit problems for web respondents
4 edit problems for mail respondents
Academic institutions: 2 edit problems for web respondents
4 edit problems for mail respondents
Biomedical institutions: 1 edit problem for web respondents
3 edit problems for mail respondents
Communication/Comprehension
27
Resolution of respondent problems:
Telephone and email help desk
Post-submission edits and follow-up with ICs and
other contributors
Edit resolution follow-up
28
Percent of institutions requiring edit resolution follow-up
All institutions: 73% of web respondents
88% of mail respondents
Academic institutions: 77% of web respondents
89% of mail respondents
Biomedical institutions: 63% of web respondents
87% of mail respondents
Edit resolution follow-up (continued)
29
Identify contributors for follow-up
Incorporate data entry features into web survey
Tailor edit messages to specific conditions of inconsistency
Lessons learned