relevant ideas and concepts and should also present relevant debates that pertain to your problem statement. By conducting a thorough literature review, you may gain insight into research methodologies, study designs, and variables that are directly relevant to your own study. You will also find that the literature can help to justify or legitimize the methods you will use in your own study.
Develop a hypothesis
Research is guided by a question or issue that a researcher or group of researchers thinks is worth investigating or needs to be addressed in some way. Regardless of your motivation, your research question is best guided by a hypothesis. A hypothesis is simply an idea about the research that seems reasonable and that you think explains the situation at hand, or it might be the underlying basis of debate about the topic. In other words, the hypothesis of your research is an educated guess about what you might find, but the validity of your hypothesis has yet to be proven.
When developing your hypothesis, think about the most important aspects of your proposed topic. A hypothesis is an educated guess about relationships you might find between variables. Once you can state your ideas about the research or analysis in one sentence, you will know that you have narrowed your topic enough to create a hypothesis. Your hypothesis should clearly state the main idea or the crux of your research project. In other words, a hypothesis is simply a statement that conveys what you might find or the state of some set of events. A hypothesis presents something that can be tested, explored, and further investigated.
The following hypotheses incorporate a spatial question and can be effectively analyzed in a GIS:
A final comment on the hypothesis is that it establishes what you believe to be important about your research project. As a researcher, you can think about how GIS might add to the understanding of your hypothesis by identifying key variables (figure 3.8).
Figure 3.8 A chemical factory releases smoke into the atmosphere. Nickolay Khoroshkov/Shutterstock.com.
Every hypothesis should have a single dependent and one or more independent variables. In the first hypothesis listed previously, a GIS could be used to take census data overlaid with pollution data from the Environmental Protection Agency’s Toxic Release Inventory database. Socioeconomic information would come from the census data, and you could define (according to whatever criteria you deem appropriate) what areas are polluted (figure 3.9).
Figure 3.9 Facilities that are registered on the Environmental Protection Agency’s Toxic Release Inventory are recorded as point locations in the database. Information regarding the specific pollutants released is also recorded. Here these are coded on an ordinal scale from low toxicity (white circles) to high toxicity (black circles). Assume that the polygons on the map represent the census tracts with a specified percentage of the population classified as poor (as defined from census data). We could use the GIS to determine if there is a statistical difference based solely on the point locations of facilities releasing toxic waste (do they fall in the census tract or not?) or based on other spatial concepts such as nearness or adjacency.
If you consider the spatial distribution of the pollution sources in figure 3.9, several things might be apparent. First, the majority of the mid- and high-pollution facilities are somewhat clustered to the left of center on the map in polygons 1, 2, and 7. It would be logical to assume that if the center of the map (polygon 4) were the historic city center, industrial facilities, particularly the older, higher-polluting facilities, would have grown up around the city center. Over time, newer, cleaner facilities might be expected to grow on the edges of the city. Second, you will notice that facilities are generally located on the left side of the map, with no facilities sited on the right side of the map (polygons 3, 5, and 8). Last, we might note that although the central polygon (polygon 4) has no points inside it, and polygon 7 has only one light polluter, several other sources of pollution are just across the line in the adjacent polygons (polygons 1, 2, and 6). We will further discuss this example as we go through the remaining stages of research. As an exercise, you might also consider the other three hypotheses presented earlier or others you are interested in exploring. What considerations might be important as you operationalize this for your spatial analysis of the data in each of these situations?
Develop a conceptual model
What is a conceptual model? As discussed earlier, a conceptual model or framework is a working theoretical model that explains your view of the world. In essence, it is your chance to identify key variables in your study, to explain the links between these variables, and to explain the sequence and flow of relationships by using arrows. (We introduced the concept of creating a flowchart for GIS analysis in detail in chapter 2.)
Once you complete your literature review, you should have an easy time developing this framework. A conceptual framework is different from a literature review because it guides the research process of your specific project. The conceptual model establishes factors important to your study and indicates to all who read your study the predicted relationships between key variables.
Brainstorming is a good approach to developing a conceptual model or framework. You might start by writing down the important aspects of your study on a sheet of paper. Next, draw circles around these objects and use arrows to indicate relationships between these variables. As you identify your key variables, you can decide what type of contextual, geographic, or other information can facilitate a study of the relationships among these variables. When you identify relationships between the variables in your study, consider how geographic information might enhance your understanding of these relationships.
Consider our earlier example relating poverty and pollution. Given the spatial distribution of polluting facilities on the map, we would want to conceptualize what the actual locations really mean to our hypothesis: is presence of a facility inside a polygon all that matters, or do sites affect people for some distance (perhaps in a neighboring polygon)? If there is an effect over a distance, how does the mode of transport (through air, water, or solid waste) influence the distance and direction of the effect? What about the poverty data? Are all of the people in the census block distributed equally, or does the housing cluster in certain portions of the polygon? (This relates to issues of boundaries and the modifiable area unit problem discussed more fully in chapter 5.) The list could go on, but the important point is to think beyond what is directly visible, such as points falling inside particular polygons, to consider other factors or mechanisms that might be important in assessing the validity of your hypothesis.
A