Combining Incompatible Spatial Data

Abstract

Global positioning systems (GPSs) and geographical information systems (G ISs) have been widely used to collect and synthesize spatial data from a variety of sources. New advances in satellite imagery and remote sensing now permit scientists to access spatial data at several different resolutions. The Internet facilitates fast and easy data acquisition. In any one study, several different types of data may be collected at differing scales and resolutions, at different spatial locations, and in different dimensions. Many statistical issues are associated with combining such data for modeling and inference. This article gives an overview of these issues and the approaches for integrating such disparate data, drawing on work from geography, ecology, agriculture, geology, and statistics. Emphasis is on state-ofthe-art statistical solutions to this complex and important problem.

On ecological fallacy, assessment errors stemming from misguided variable selection, and the effect of aggregation on the outcome of epidemiological study

“…While the former (selection bias) refers to the way in which research subjects are selected from the study population, that is lost subjects or missing data. Concurrently, the latter type of ecological bias refers to ''information loss'' due to aggregation or measurement inaccuracy, which may distort the effect estimates (Greenland, 2001;Gotway and Young, 2002).…”

“…Recent epidemiological studies, specifically those using geographic information systems (GIS) technology (see inter alia Gotway and Young, 2002;Elliott and Wartenberg, 2004;Wakefield and Shaddick, 2005;Bell, 2006), highlighted another possible source of ecological bias, commonly known as ''error propagation'' (Heuvelink and Burrough, 1989;Goodchild et al, 1992;Burrough and McDonnell, 1998). The major cause of this sort of information bias emanates from the use of interpolation models converting data available for irregularly spaced points (e.g., air monitoring stations) into regular grids or local exposure estimates at places were study subjects reside.…”

“…The major cause of this sort of information bias emanates from the use of interpolation models converting data available for irregularly spaced points (e.g., air monitoring stations) into regular grids or local exposure estimates at places were study subjects reside. Upon such interpolation procedure, any error (e.g., attributed to faulty measurements), potentially occurring at original observation points, propagates into all output layers of data created by interpolation (Gotway and Young, 2002;Bell, 2006). Furthermore, the outcome of interpolation is sensitive to the interpolation method used, that is, nearest monitor, spline, inverse distance weighted (IDW), or kriging (McCoy and Johnston, 2001).…”

“…The choice of an interpolator and its specific quantitative parameters (such as power, search radius, etc.) may result in fairly different exposure estimates at the ''target points'' and ultimately, in erroneous associations between exposure to an environmental risk factor and its health effects (Gotway and Young, 2002;Elliott and Wartenberg, 2004;Wakefield and Shaddick, 2005).…”

“…In addition, even when individual-level health data are accessible, there is a difficulty to match them with socio-economic variables which are usually aggregated into census-designated statistical areas (i.e., census blocks and tracts), and are rarely available at the individual level (Elliott et al, 1992;Elliott and Wartenberg, 2004;Nuckols et al, 2004). Although geographic information systems (GIS) technology, which has spread widely in recent years (Gotway and Young, 2002;Brauer et al, 2003;Cockings et al, 2004;Elliott and Wartenberg, 2004;Nuckols et al, 2004;Scoggins et al, 2004), may simplify the establishment of such data linkages and thus help to verify the correspondence between results obtained from individual data and those obtained from areal aggregates, such comparative studies are yet largely forthcoming.…”