Geoscience Reference
In-Depth Information
the metropolitan Phoenix region. Questions focused on the location, extent, and
participation rate of all programs that aimed at educating the general public about
water issues between June 2005 and June 2006 (using a method similar to Brown
2005
). Water information program information extracted from websites, annual
reports, and other organization documents to identify the location and extent of
water information programs to supplement interview information and provide some
basis for including organizations unable to participate in initial interviews.
Responses were translated into spatial extents using ArcGIS version 9.0
(ESRI
2004
). Translating interview responses about water information programs
resulted in two initial data types: points, when information was dispersed from a
central location and polygons, when information was dispersed directly to the
public (Fig.
3.3
). For example, if an interviewee representing a municipal water
supplier responded that there were brochure kiosks at all the libraries in a city, the
water information programs were initially points georeferenced to the libraries.
In a few instances, interviewees provided probabilistic information about points.
In one instance, an organization mentioned attending 5 or 6 of a possible 30
neighborhood events in a particular city. All neighborhood events were identified
and a note was made of the probability that an event occurred there. This
probability was used to give more weight to events that occurred with certainty.
In the case of information campaigns that distributed materials to the homes,
polygons were used to represent information extent. These included school-based
programs (presumed to reach adults dispersed throughout the district), mailings,
and local news sources.
The literature provides several potential remedies to reconcile differences in
the representation of point and area information. Methods vary from more com-
plicated space-time prisms, distance decay functions or path analyses used in
transportation research to more simplistic container-based count methods. Given
the relatively inconsistent reporting habits or water educators, simpler methods
were determined that might be more appropriate for this case study (as in Brody
et al.
2002
). Water information providers did not have complete records of
distances traveled by attendees. Two methods were devised that might be appro-
priate given the nature of the data: using political boundaries (census tract and/or
zip code) as a container-based method (Griffith
2005
) and creating distance
buffers to assign an area to points. This resulted in three types of representation
(illustrated in Fig.
3.4
).
Container methods simply count the number of events that occur in a pre-defined
space. It was assumed that all points falling within each polygon were designed to
serve that population offering either zip codes or census tracts as potential
boundaries. The weaknesses of these container-based methods for establishing
exposure to environmental phenomena are well known among environmental
justice scholars (Buzzelli and Jerrett
2003
; Buzzelli et al.
2003
; Mennis
2003
),
but given the relative subjectivity built in to data representation, creating multiple
representations and comparing the consistency of results across representations
