This study sought to analyze advanced Geospatial Information Systems (GIS) researchers' metadata needs and information finding behavior while using the online GIS data services pages for datasets at D.H. Hill Library at North Carolina State University (NCSU). This study focused on the following aspects of use:
- How do researchers using GIS datasets want to be able to search for information?
- What metadata is important to researchers for discovery?
- What metadata is important to researchers for other purposes?
- How important is controlled vocabulary for keywords in information discovery?
Team and Partners
- Joyce Chapman, principal investigator
- Jeff Essic
Recordings were made in the NCSU Libraries Usability Research Lab using Morae Recorder usability testing software and analyzed using Morae Manager. This software records participants' screen movements and audio.
Testing took place in the Usability Lab at D.H. Hill Library at NCSU in Spring 2010. The study consisted of three parts: a demographics questionnaire, a series of three information finding tasks to be completed by each participant to be completed using the NCSU online GIS data finder, and a set of focus group questions related to what participants use and like (see "Links" section below for downloads of materials used in testing). Usability testing was conducted using Morae Usability Testing software.
The focus group was conducted individually with nine participants. The median age was 45, and ages ranged from 29-57. The median number of times participants had used NCSU's GIS site for research was 35, but ranged from 5-300 times. Two-thirds of participants were advanced GIS researchers and one-third were self-assigned beginner or intermediate GIS researchers. All but one of the participants assigned themselves an Internet proficiency level of "expert." One-third of participants were graduate students at NCSU or UNC-Chapel Hill and two-thirds were NCSU faculty, staff, or post-doctorates. One-third of participants were associated with the Forestry department; the other two-thirds were associated with an array of different departments, including Computer Science, Information Science with a GIS concentration, Atmospheric Sciences, Transportation research, Soil Science, and NCSU Libraries.
Summary of Findings
Advanced GIS researchers want to be able to search for datasets by keyword, and they want to be able to specify extent in a search. Most are not willing to browse as an alternative to searching. When searching for datasets at the Libraries GIS pages, they assume that Librarians have normalized keywords. Though many assume that metadata from providers may be incomplete for datasets, they do not understand that this might contribute to a keyword search failing to return the item. Researchers do not prefer map-based searches over text-based searches: most like both systems, though some preferred text-based systems due to the difficulty of specifying exact extent in a map-based system. Researchers feel frustration at not having trusted resources that include all available data sets for their search, or that can offer value assessments of datasets. Researchers expect search boxes to work similarly to a Google search: search systems with defaults that differ significantly from Google should advertise that prominently near the search box. Some researchers are not willing to perform very many searches: if an acceptable results does not appear in the first search set of results, some researchers will leave to search a different site; others will give up searching and simply contact data collection agencies directly. Almost all participants saw a benefit in increased instructional materials for data downloads and searching, and the results the tasks requiring participants to download data stressed the need for data download access that is obvious and intuitive. Researchers do not enjoy reading the metadata files and some often try to avoid them; on the other hand, researchers very much want for a number of metadata elements pulled from the file to be available for search and display in other locations, such as search results pages or previews of a resource. Some of the most important metadata elements for advanced researchers when determining whether or not a data set meets their needs are: date, extent, title, attributes, feature type, source, and projection.
- Demographics Questionnaire
- Tasks and Focus Group Script
- Transcripts of Sessions/Notes
- Participant Response Tallies
Reports and Presentations
Last updated: May 7, 2010