User:Adowell1/sandbox

Recognizing MAUP in Research

Anthony Dowell

   In “The Big Sort”, authors Bill Bishop and Robert Cushing discuss the apparently increasing tendency for people to cluster based on their views, beliefs or lifestyles. Specifically, the notion was discussed that people are most likely to migrate toward and live in close proximity to people who were most similar to them in terms of religion, political affiliation and educational levels, among other things. They labeled this phenomenon “The Big Sort” and one of the most astounding examples of this sorting was seen in the 2004 presidential election between John Kerry and George W. Bush.

On a national scale this election ended as one of the closest in our country’s history, with Bush’s margin of victory a mere 2.5 percent. On the surface these results would indicate a high level of diversity among voters. But as the vote was broken down to a county level it reveals a much different pattern, with roughly 60 percent of counties having a final margin of victory of 20 percentage points of more in favor of one party. The counties which had these types of very one sided results accounted for nearly half of all American voters.

This type of aggregation and uniformity was not limited to politics nor was it limited to the county level of geography. In fact, the research that led to the discovery of the big sort was not even politically based. Bishop and Cushing were looking to identify patterns of growth in individual cities, specifically how certain cities were able to grow at such rapid rates compared to others in terms of both population and wealth. They found that cities such as Austin, Texas and Portland, Oregon that were considered to be high-tech and innovative attracted specific types of people at very high rates. The research they did extended beyond these indicators into the realm of social structure: the high-tech city populations were more interested in cultures other than their own, more adventurous and more likely to care about politics. Comparatively, low-tech cities contained much higher percentages of people who attended church, were active in the community or volunteered, and were more family oriented. Additionally, the people living in low-tech cities were more likely to have feelings of isolation and were generally more sedentary. The conclusion was that this sorting of people based on interests and lifestyles was “a perfectly natural phenomenon: given the choice, people choose to live around others like themselves”. While they fully expected this sorting to continue they do not think that the composition of a community is necessarily representative of how it will always be made up. Those common characteristics that drove community aggregation will eventually be replaced by new ones and the types of people inhabiting them will change and evolve.

These observations and the general idea of the Big Sort relate strongly to another geographic concept, the modifiable areal unit problem (MAUP). The basis of MUAP is that the areal units that are used for statistical studies can have a major impact on results, and thus the unit of measurement can be altered to knowingly influence the results of a study. As we saw in The Big Sort, when viewed from a national level we are led to believe that our country is made up of a very even mix of people; in the 2004 national election the vote was split nearly perfectly in half between democratic and republican voters. However when these voters are broken down based on the states, counties, cities and even the individual communities in which they live we clearly see the high levels of similarities between people within close proximity to one and other. This being said, even the spatial extents of states, counties and cities may not be the most accurate way to evaluate this type of information and that is perhaps something this research needs to pay more attention to. The fact remains that MAUP will exist, to some extent, in any analytical study that pertains to data that is spatial in nature.

So what effect does this sorting have on spatial analysis in a general sense? The fact that people are clustered based on certain factors means that the results of different types analysis may be more likely to be skewed or even misleading (whether intentionally or unintentionally). Because of this, the areal boundaries and scale that are chosen for a particular analysis become even more important, and careful consideration must be taken when selecting these. Existing and commonly used areal units such as counties, census blocks or zip codes, are often drawn arbitrarily and thus may be inappropriate for the spatial analysis taking place. Several methods have been developed for creating these spatial units but the bottom line is that the size and shape of these units should be determined by the nature of the data and not the other way around. Both caution and thoughtful planning must be exercised to ensure the results of spatial analysis are an accurate reflection of the conditions that exist in the real world. The effects of MAUP and “the Big Sort” will exist when dealing with any data that is social or political in nature but the key is finding effective ways to minimize their impacts on test results.