Using Personal Digital Assistants to Collect Survey Data

New technologies for computer-assisted survey information collection (CASIC) are being developed as graphical user interfaces, mouse/pen-based computing, handwriting recognition, hand-held computers, and use of the Interact become more widespread (Connett, Mockovak and Uglow, 1994; Gilluly, 1994; Nichols and Apple, 1994; Shanks, 1994; Forster and Snow, 1995; Hansen, 1996). The utility of these new technologies depends on the survey setting, including the type of survey data to be collected, the size of the survey, and conditions for collecting survey data. We consider CASIC technologies for conducting longitudinal natural resource surveys that involve collecting survey data from multiple sources, including field observations, office records, photo-interpretation and geographic information systems (GIS). Examples of such surveys include natural resource and land use surveys, soil quality surveys, soil mapping updates, forest inventories, and environmental monitoring surveys. We focus on the National Resources Inventory (NRI), conducted by the U.S. Department of Agriculture's Natural Resources Conservation Service (NRCS) in collaboration with the Iowa State University Statistical Laboratory (ISUSL). The NRI is a national longitudinal survey in which data on hundreds of variables are collected periodically from approximately 300,000 area segments (PSUs) and 800,000 points. The national survey was conducted in 1982, 1987 and 1992, and interim special studies have been conducted in 1995 and 1996 using the national sample as a frame. The survey instrument for the national survey includes sections on land use, soil characteristics, agricultural practices, wetlands, rangelands, vegetative cover, and wildlife habitat. Historical data on sample units are used to collect information and validate sample locations. NRI surveys are conducted using a mobile work force spread throughout the U.S. When specialized skills are required to obtain survey data, multiple data gatherers may collect information on the same sample unit. In recent years, the NRCS has used intemallycreated computer-assisted survey instruments (CASIs). In preparation for the 1992 NRI, NRCS developed a CASI on UNIX and DOS computers for state offices. The survey software was designed to manage the survey data base and guide data gatherers through the survey instrument with logic for navigation and branching, and rules to validate and cross-check observations. A separate program on the state office computer was used to check the data for compatibility and consistency. Once data were complete and checked, the survey information was transmitted to the ISUSL on diskettes and tapes. State data were subsequently checked by ISUSL with ancillary data bases and more extensive edit routines. Questions and problems were resolved by personal calls and exchanges of faxes. The 1992 NRI instrument was designed for direct entry of data as they were collected by photointerpreters operating in centralized remote sensing facilities. However, paper worksheets were used more frequently than expected because of limited computer resources, data gatherer preferences and abilities, and the need to gather information at locations without convenient access to computers. As a result, the CASI functioned more as a data entry system. The software was discarded after the 1992 NRI was completed.