Research Interests
- Computer-assisted data collection methods
- Survey methods for the National Resources Inventory
- Accuracy assessment of spatial databases
- Social policy applications
Computer-Assisted Data Collection Methods
With the advent of emerging technologies, there is considerable interest in using geospatial data in mobile computing environments for survey data collection. Project Battuta is an NSF-sponsored project to develop general principles for using geospatial data in field data collection from both the user and the computing infrastructure perspective. This project is sponsored by NSF's Digital Government Initiative, with funding and agency collaborators from Census Bureau, Bureau of Labor Statistics (BLS), U.S. Department of Agriculture (USDA), and U.S. Geological Survey. Recent work with Census Bureau and BLS has focused on establishing principles for displaying and using map and GPS data in tablet and handheld computing environments based on spatial cognition theory. One goal of this work is to increase accuracy and efficiency of protocol implementation by users who vary in spatial abilities and skills. In collaboration with ISU computer science faculty and graduate students, we are developing design principles for computing infrastructures that accommodate multiple field settings and balance computing loads between the field and computing infrastructure environments. Collaborations with geography faculty at the University of California, Santa Barbara have focused on adaptive sampling designs for scientific field studies and extensions of interface designs for wearable computers. Future research will focus on photographic materials, particularly as they relate to reducing measurement errors in data collection.
Survey Methods for the National Resources Inventory
The Center for Survey Statistics and Methodology has held cooperative agreements with the USDA Natural Resources Conservation Service (NRCS) since 1956 to provide statistical and survey methods expertise for NRCS natural resource surveys. The National Resources Inventory (NRI), < http://www.nrcs.usda.gov/technical/NRI/> is a large-scale longitudinal survey designed to monitor status and trends in natural resource conditions on nonfederal lands in the US and its protectorates. The bulk of the data are collected by photo-interpretation; small field studies are conducted to augment the information base for variables that are not observable via remote sensing techniques.
Working with CSSM and NRCS staff, we have developed mobile handheld and distributed technologies for national data collection activities. Early work focused on using handheld computers (Apple Newtons) to serve data collection software and sample units using client-server technology. To take advantage of global positioning system (GPS) signals in the field, a simple user interface was integrated with the Newton computer-assisted data collection software to control the technical components of the GPS-Newton interaction and to reduce errors in obtaining and using location data in the field. The client-server data collection system led to a rapid expansion in the use of the web for monitoring data collection progress, serving data collection resources, and delivering remote training. This research provided the foundation for more theoretical endeavors in the use of geospatial information for mobile data collection settings (see above).
I also work with CSSM faculty and graduate students and NRCS staff on quality assessment activities for the continuous NRI. Current work involves developing sampling designs to identify sample units for data collection quality control inspection. The design uses several factors related to the area segment historical data, timing in the data collection process, and the data collector attributes in monitoring the quality of each data collector's work. In addition, we are designing a pilot study to provide initial data on the interaction between positional and land cover/use measurement errors in preparation for designing on-going measurement error evaluation studies for the NRI. Other recent work includes working with statistics graduate students to develop a self-paced web instruction module to introduce principles of sample survey methods to agency staff, and evaluation of imputation procedures for area segment data.
Accuracy Assessment of Spatial Databases
Working with Iowa Geographical Analysis Program (GAP) faculty and staff and CSSM graduate students, probability sample-based accuracy assessment methods were developed for land cover maps being produced in the Midwest. Many land cover assessment efforts use sample designs that fail to include all land areas with positive probability and that do not take advantage of sampling structures that can reduce data collection effort and improve precision of estimates. We developed a two-stage sample design for the study area that accommodated assessment objectives, operational constraints, and statistical goals. Standard survey protocols were applied in contacting landowners to generate a high rate of cooperation in gaining access to land where sample points were located. Accuracy assessment estimators were extended to account for the unequal probability design and to incorporate known information about map land cover surface areas.
Social Policy Applications
I have worked with faculty and graduate students in ISU's Human Development and Family Studies and Economics Departments to develop methodologies for investigating the impact of welfare reform on state-level policy development and affected populations. This work began with a pilot study to determine whether the Survey of Program Dynamics (SPD), a national survey established to monitor welfare reform, could be implemented at the state level to develop results comparable with national estimates. After this approach was deemed too expensive relative to information gained, a second pilot survey was implemented using a dual frame sample of welfare participants and of the general adult population in Iowa that could be linked to results from the SPD. Part of the research also involved examining methods for estimating the prevalence of food insecurity (hunger) in the US, particularly to consider shortened food security rosters for smaller surveys.
