Category: maps

As explained by Mark Monmonier in “How to Lie With Maps”, data can easily be skewed when a researcher converts it into a map. By fiddling with the settings on a map, one can portray two completely different stories with the same data. By merging Connecticut Census Data with School District Data in our class, we were able to combine the two sets of information and create maps. However, it is evident that different maps can be made in order to show differing stories.

 

For the purpose of this post, I focused on the Percent Minority Data (2009-2010) for Hartford and the surrounding towns. In the map on the left, one would assume that there is widespread racial diversity throughout the towns. In order to create this effect, I changed the map settings so that there were six shades of maroon; each shade signified a certain percentage of minorities in a specific town. Because there were so many shades, the map portrays a large number and range of minorities in Hartford and the surrounding towns. The designations for the colors and percent minorities can be found in the key to the left.

 

 

In contrast, the map on the right portrays sharp racial differences between Hartford and the surrounding towns. While Hartford and Bloomfield are shaded black, all of the other surrounding towns are shaded gray. This portrays stark differences in the racial makeup of Hartford and the surrounding areas. I made the map look this way by only using two colors: black and gray (see the key to the right). This effect was further portrayed by the fact that I designated the towns with 0-90% minorities to be colored gray, while the towns who are comprised of 90-100% minorities to be colored black. It does not look like there is a wide range of racial diversity in this map, as it does in the first map.

In Part 1 I discussed lying with charts. Now, in Part 2, I illustrate that lying with maps is also possible.

Map 1 includes a breakdown of the percent of minority students in school districts, in Hartford and surrounding towns. Map 1 was created by using Google Fusion Tables to merge two data tables together. One table consisted of the breakdown of minority students in each school district. The other table contained census data for each Connecticut town. Once the two tables were merged into one table, I altered the map style. For Map 1, I created color based categories to represent certain percentages.

Percent minority students in Hartford-area school districts, 2009-10
Click the map to view underlying data  (Sources: CT Dept of Ed, MAGIC UConn Libraries)

In the Map 1 Legend, you can see the five categories. Based on these categories, Map 1 illustrates that there was racial diversity among the school districts. When creating maps, the selection of categories can influence how data is illustrated. To really understand why this is true, let’s take a look at a second map that was created using the same two original data tables.

Percent minority students in Hartford-area school districts, 2009-10
Click the map to view underlying data  (Sources: CT Dept of Ed, MAGIC UConn Libraries)

Map 2 was also created using Google Fusion Tables and the same two original data tables that were used for Map 1. However, Map 2 illustrates sharp racial divisions among school districts in the Hartford area. But, how can this be true when the exact same data was used to create both maps? Just like with charts, the way you choose to display data in maps makes a big difference. Recall that in the Map 1 Legend I used five categories for my legend.  Now notice that in the Map 2 Legend, I used only three categories. Using less categories made the map appear to have drastic differences between more towns.

In addition, you may have noticed that I defined the categories for each of the legends using two different sets of percent breakdowns. The cut off percentage for each category can be very influential in determining which category each district falls under.  Since most of the towns shown have less than 35% minority students in their school districts, more towns appear to have little diversity. This choice of categories also depicts that there are much more minority students in the center of the mapped area.

When creating maps, and charts, it is important to consider the scales and categories that are created with them. Changing scales and categories can allow you to show one set of data in multiple ways. In other words, lying with charts and maps is most certainly possible.

 

In Part 1 I discussed lying with charts. Now, in Part 2, I illustrate that lying with maps is also possible.

Map 1 includes a breakdown of the percent of minority students in school districts, in Hartford and surrounding towns. Map 1 was created by using Google Fusion Tables to merge two data tables together. One table consisted of the breakdown of minority students in each school district. The other table contained census data for each Connecticut town. Once the two tables were merged into one table, I altered the map style. For Map 1, I created color based categories to represent certain percentages.

Percent minority students in Hartford-area school districts, 2009-10
Click the map to view underlying data  (Sources: CT Dept of Ed, MAGIC UConn Libraries)

In the Map 1 Legend, you can see the five categories. Based on these categories, Map 1 illustrates that there was racial diversity among the school districts. When creating maps, the selection of categories can influence how data is illustrated. To really understand why this is true, let’s take a look at a second map that was created using the same two original data tables.

Percent minority students in Hartford-area school districts, 2009-10
Click the map to view underlying data  (Sources: CT Dept of Ed, MAGIC UConn Libraries)

Map 2 was also created using Google Fusion Tables and the same two original data tables that were used for Map 1. However, Map 2 illustrates sharp racial divisions among school districts in the Hartford area. But, how can this be true when the exact same data was used to create both maps? Just like with charts, the way you choose to display data in maps makes a big difference. Recall that in the Map 1 Legend I used five categories for my legend.  Now notice that in the Map 2 Legend, I used only three categories. Using less categories made the map appear to have drastic differences between more towns.

In addition, you may have noticed that I defined the categories for each of the legends using two different sets of percent breakdowns. The cut off percentage for each category can be very influential in determining which category each district falls under.  Since most of the towns shown have less than 35% minority students in their school districts, more towns appear to have little diversity. This choice of categories also depicts that there are much more minority students in the center of the mapped area.

When creating maps, and charts, it is important to consider the scales and categories that are created with them. Changing scales and categories can allow you to show one set of data in multiple ways. In other words, lying with charts and maps is most certainly possible.

 

Just as graphs can be used to manipulate statistics, maps can be used to manipulate data as well. Below I created two maps that show the minority student population in the Greater Hartford area during the 2009-2010 school year as a benchmark to check the progress of the Sheff settlement.

Map 1: Sharp Racial Differences

The map below shows sharp racial differences between the different towns surrounding Hartford. As you can see, Bloomfield, Windsor, Hartford, and Manchester are the only towns that appear to have a population of minority students. While the other towns appear to be heavily populated with whites.

To make the map above to appear to have sharp racial differences between the different towns in the Hartford area, I used the bucket option to highlight the differences. With the use of the bucket feature, I set the ranges from 0-50% percent and 50-100% as seen in the screenshot below. This scale limits a large number of towns that actually have a minority population because it is very rigid. This in turn, gives the illusion that none of these towns are meeting the Sheff goal of minority students in reduced isolated settings.

 

Map 2: Widespread Diversity 

The map below shows widespread diversity between the different towns surrounding Hartford. As you can see, in comparison to the map 1 above (sharp racial differences) more towns appear to have a population of minority students and fewer towns appear to be heavily populated with whites.

To make the map above to appear to have widespread diversity between the different towns in the Hartford area, I used the bucket option to highlight the differences. With the use of the bucket feature, I set the ranges from 0-10% percent, 10-30%, and 30%-100% as seen in the screenshot below. This scale allows a large number of the towns that appeared to heavily populated with whites in map 1 to appear to have a minority population. This in turn, gives the illusion that the majority of the towns (even if the minority population is small) are meeting the Sheff goal of minority students in reduced isolated settings.

As with graphs, it is important to look at the scales that are used to represent data in maps. I have learned first hand different tricks that are used to manipulate results and I urge everyone to play close attention to details when analyzing different data in the future.

 

 

 

Although both maps are created using the same information, they appear drastically different in respect to the composition of demographics of the population. In the map I created to portray the Hartford region as being racially diverse, one can look at it and presume there appears to be a mix of races within each respective district. This is attributed to the fact that I changed the map style to have numerous “breaks” in the legend. By increasing the number of gradients within the legend, I increased the appearance that the region was increasingly more racially diverse. In contrast, for the map to portray little to no racial diversity, the map is simply divided into two colors: red (white) and blue (minority). By creating the map with only two “color categories” (divided based on anything below .50 was non-minority and anything above .50 was minority), the Hartford region appears to be completely racially segregated.

The map that depicts racial segregation shows an isolated minority population in the center—Hartford, Bloomfield, Manchester, and parts of Windsor. It appears as though essentially the majority of the suburbs in this area are predominantly white and the center is entirely non-white. In contrast, the map I created to depict a racially diverse region makes it appear as though there are higher and lower concentrations of non-whites in certain districts, but it appears as though every district is mixed and composed of both whites and non-whites. Hartford and Bloomfield are the only districts that appear to have a very concentrated population of non-whites.

In summarization, the more legends I add to the scale, the more accurate my map appears to be. I find it incredibly fascinating that I could completely “lie” about the statistics of a certain area without actually “lying” or by altering any of the data I wish to present to my audience.

#next_pages_container { width: 5px; hight: 5px; position: absolute; top: -100px; left: -100px; z-index: 2147483647 !important; }