Analyzing Election Results using the

 

Updated

 

Tree Map 3.2 Application (Latest Version as of March 2002)

 

CMSC 838f Application Project

 

Brian Phillips

 

March 7, 2002

 

·       Initial Acknowledgements:

 

Special Thanks to the Human Computer Interaction Lab at UMCP for providing the test data information and code for the treemap software.  I would

 

 

·       Project Introduction and Motivation:

 

Living in the United States we base much of our democratic freedoms in the hands of free elections.  Elections run much like any other business where the cash comes in the form of votes.  Each voter is a consumer so to speak and must be marketed to.  In large elections outside of private organizations a blurring occurs in the process of possessing so many votes over a broad spectrum of people.  Campaign Managers and strategists greatly benefit from knowing about different vital statistics involved in their voters.  When devising a successful strategy such as in a presidential election it behooves them to know where to focus their effort. No one man or entire campaign for that matter can visit all locations.  Therefore they must analyze date to formulate a strategy for a focus of effort.  Thorough analyzing large data sets we can help these professionals quickly and easily figure out not only the where, but the who and evolving trends over time. 

 

The application that I chose to use is a tree map application written by University of Maryland, Human-Computer Interaction Lab (HCIL UMCP).  A downloadable version of this software is located here.

 

·       Description of Tree Maps

 

Formal: Treemaps are a space-filling visualization for hierarchical and categorical data. Treemaps enable users to manipulate the size and color of the rectangular areas that represent leaf nodes in the tree, based on the attributes of the data. Treemaps are a useful tool in understanding data and recognizing unusual trends or patterns.[i]

 

Informally:  Tree maps provide a way for users to group large amounts of data into a single screen, while quickly analyzing it.  A picture is provided in figure 1 to give you a feel before any analysis.

 

Figure 1 Notice the segmentation by region and the breakdown of the states.

 

System Features:

 

In order to represent data the data is categorize hierarchically and placed into squares. The squares can conceptually be thought of as containers that hold the data. Only within each one of these containers thousands of entries may exist.  This conceptualizes a relative size in comparison to thousands of records without ever performing an actual count of data by physical observation.  In addition the system makes effective use of colors to emphasize the density of a square.  The physical interaction with the application that facilitates use is the ease of user interaction by simply clicking with a mouse on the square to inquire further about information within that square.  Often further information drops down in a sub menu to provide users with further information. 

 

·       Using the application to perform a task:

 

o       Steps and descriptions towards performing a task with the tree mapping application

§         Proper Format

·        We first need to have the data in the correct format.  This can be accomplished with a Microsoft excel file.  Instructions for the documentation can be found here.

§         Load the file into the tree map application

·        This is accomplished through loading in the file by using the file tab and loading the file into the application.

·        Now data is ready to be analyzed using the tools.

 Description of the working environment of the system

 

Border Size

First we can control the size of the border in order to resize the viewable area and sectionalize data in a more effective manner for analysis.  We can move the values up and down by using a slider. 

 

·        Size of square 

 

The size of the square represents the amount of electoral votes that are owned by each state in figure 2.  However we can change this by votes for a candidate or by population.  This representation will then quickly give a relative size to the different squares.  This is much easier than looking at the direct data.

 

·        Color

 

We are able to set colors to different attributes.  In this example it is done by party.  I used red for Democrat and green for Republican.  Notice how thick the border is between the different regions in this example.  We can apply filters in the system to black out certain blocks we wish to ignore.  Observer Figure 3 to get a visual representation.

 

 

 

 

 

FIGURE 2 Notice the difference in colors by majority political party in the state

 

FIGURE 3 Notice how some blocks are blackened out because we ran a query of the states that possess between 11 and 55 electoral votes.  The States in color still represent these states.

 

 

 

 

 

·        Data

 

The Data fed to the system must be placed in the proper format in order to be properly displayed.  However, when this is done you are able to run queries on that data.  The proper way to format data in an excel spread sheet is provided in the system documentation.

 

Legend

 

The legend allows you to see the various data labels as you wish in order to customize the way data is viewed.  Here is a picture to illustrate this in Figure 4:

 

Figure 4 This illustrates how to change the legeng by

utilizing the drop down menus

 

·        Filters

 

The filters are what allow the user to change the way that the data is displayed to the user.  By sliding the sliders the data constraints change thus changing the colors displayed by different squares. Figure 5 displays a picture of the sliders.

 

Figure 5 Notice the numbers underneath of the sliders that provide the queries for the user

 

System Description and Walk Through

 

Once the application is started the screen appears

 

FIGURE 6

 

Initially there are no colors associated with the tree map or values.  Each square is basically about the same size.  Each square represents a separate state that is embedded into another square representing region.

 

By entering the Legend Slider area one can change the representation of the data associated with each of the blocks, the way they are labeled and the color representation.

 

·        Detail on Demand

 

Upon highlighting each block a window appears with all the pertinent system information about that block.  This is a strength because it provides information quickly to the user.  The power here lies in the fact that it requires little effort to gather large amounts of information without much effort.

 

·        Zooming Filtering and Searching

 

Zooming is only a matter of double-clicking on the square.  Each square then goes in a level where it can be analyzed in greater detail.  When analyzing the square it is also possible to look to the upper right hand corner of the screen to observe more details about the highlighted square.  (See figure 6)

 

Figure 6 Notice how the highlighted squares information is displayed in the upper right hand corner and how by changing the query in the pull downs the labels have changed

 

·        Other Options

 

An additional option that makes this system extensible to the user is the ability to change the colors associated with the different things being compared.  Observing in figure 6 we see that the republican color differs from the democrat colors.  Notice that the blackened out part that is labeled as Republican.  This is done because of a query run in the filtering region.  These states do not meet the criteria for whatever is being queried.

 

·        Areas for Improvement

 

Though the data gives a visualization of relative size and organizes it by region, the human mind is still left to visualize a state as a square instead of its normal physical representation.  This still leaves room for the human mind to process the location of the states in their mind.  I would prefer that the states maintained a better physical relationship to the actual location of the states.  This would represent the data in a much more powerful way.

  

·        Description of Data

 

The data obtained for this presentation was obtained from the HCIL at the University of Maryland at College Park.  It is the data associated with the past three presidential elections in the United States.  The data gives the name of the state along with, amount of electoral votes, votes for and against the candidates, and the geographical region. 

 

Typically this data would be viewed in a spreadsheet left for the human mind to disseminate.

 

·        My Evaluation and Comments

 

Using this application to devise a campaign strategy would not be very hard.  Tree maps provide a way to bring together large amounts of data into easily understandable hierarchal relationships that can easily be understood by users.  In this way a campaign manager could observe the data and decide where his strengths and weakness lie. Therefore, he could decide where to focus his candidate’s attention because of a need or to not focus his candidate’s attention as a result of a lack of support.  The entire strategy could be devised in under an hour, rather than taking days and paying many analysts.  By looking at the election data it becomes apparent as to where different people voted and where Bush was able to win.  It is also apparent as to where in the country the Democrats congregate and the Republicans congregate. 

 

Overall I think that the application is very powerful and can be used very effectively to  analyze data visualize it in a more meaningful way.