Basic Multivariate Displays (GIS and Spatial Analysis)

Mapping for Analysis, Policy, and Decision Making

The maps we have discussed in this topic so far have been basic, displaying one set of data, one variable, at one time period, in a straightforward way. However, the conceptual framework of GIS is a very powerful one, in which data can be manipulated, classified, and displayed in innovative and interesting ways. There are many instances where social science researchers will want to combine data from several different sources to create a visual presentation of multiple phenomena in the same map. Multiple variables can be displayed, relationships can be mapped, transitions over time can be displayed, and GIS can support some very powerful research questions and important public policies and decision making in a wide variety of fields such as criminology, urban planning, and disaster relief. In this section you will see a number of examples of the types of data that can be constructed and displayed, the types of maps that can be developed, and the range of decisions and questions that can be answered. Along the way you will learn a powerful set of mapping and database tools that can be applied to any area of knowledge or life, with exciting and important consequences. Having the right data and displaying it in the right form can make the difference between a puzzle and a solution, a successful policy and a failed intervention, tragedies prevented and life lost—all this with GIS and the skills you will learn in this section.


Now that you have the basic skills required to create pin maps and thematic maps, you can apply these techniques to producing more complex maps. This section will present several examples of maps that are more analytical in nature than those we looked at in the first section. To begin with, you will see the finished map product and the analytical reasoning behind it, then the steps to create the map will be presented for you to duplicate the map on your own.

The last map in Section 1, Figure 1.97, showed the number of divorced residents of each block group in the year 2000 in the city of Riverside, California. As interesting as this information is, it begs the question of the relative numbers of divorced people in these units of analysis. It may be very important to know that there are less than 10 divorced individuals in one location, for example, if you are a divorce lawyer and you want to market your services, but it would also be important in that decision to target your marketing efforts in a few places across the city rather than everywhere in the city if you knew how many adults lived in that particular area. If there were 10 divorced people, and only 11 adults living there, it would be a waste of time to market your services there as more than 90 percent of those who could be divorced are already divorced! So knowing the percentage of divorced individuals, which is just the proportion of divorced people divided by the total number of people who could possibly be divorced, namely adults, is useful in such a situation. This is a special case of a more general idea involving the concept of relative risk. The risk of something happening is the number of actual occurrences of the thing that have happened, divided by the possible number of cases in which that thing could happen; another expression for this is the rate of something happening. So we talk about the divorce rate as the number of divorces relative to the number of people who are married, since in fact only married people can become divorced. Usually these rates are expressed for a particular time period, such as in the year 2000, and for a particular place, such as the city of Riverside. Rates are very common in discussions of events in public and private, in research, newspapers, policy discussions, economics, and just about any field of activity. The birth rate (number of children born to women in the childbearing years in a country in a year), the death rate (number of deaths divided by the number of living people in a place in a time period), the crime rate (number of crimes committed divided by the number of people living in a city in a year), and so on—these are the basic building blocks of everyday knowledge and scientific research.

The rate of something, or the relative risk of something happening, may be very different from the number of events of this type that have happened, especially in their geographic distribution. For example, if you tell me that 10,000 Toyota Camry automobiles were stolen last year, I would ask, "well, what is the relative risk of a Camry being stolen?" Just knowing the number does not help me decide whether to buy a Camry based on whether or not it will be stolen. If there are 50,000 Camrys on the road, then the relative risk of a Camry being stolen is 10,000/50,000 or 0.20 (a percentage can be made by multiplying this result by 100; rates are often standardized by multiplying them by a constant, as in the crime rate per 100,000 population, or the divorce rate per 1000 married couples, and so on). On the other hand, if you also tell me that there are 100 Porche Boxters stolen in the same year, I will ask the same question—if there are only 250 Boxters on the road (at 60K per Boxter this could well be the case), you would say, "Oh, the Camry is much more likely to be stolen than the Boxter," but you would be wrong because the relative risk for Boxters is 100/250 = 0.4. So, in fact, the relative risk of a Boxter being stolen is twice that of a Camry being stolen. The use of rates is a simple but powerful tool to gain understanding about the world around us.

Next post:

Previous post: