Complete Visualization

Overview

 

The Proposed Legislation and Sponsors in 2017 visualization is a first attempt and test run utilizing the bulk data of Congressional activity and legislative bill statuses, a collection available from the Government Printing Office and held in a repository hosted through the Federal Digital System (FDSys) (1). Most notably, the data provides information produced in the most recent congressional meetings. Potentially, this data could be used to create a tool allowing users to view current bills and their status as they move through the legislative process. In this visualization, the data was aggregated to one row per bill and included 24 columns. Ultimate this created 5,244 of bills in a single .csv file that was uploaded in Tableau Public and used with the most recent directory of the 115th U.S. Congressional House of Representatives (2). More details on the data cleaning process can be found in the notes section online with google docs here.

A dashboard was selected to depict multiple pieces of information based on one of the filters available in the database. These filters include two wild card fields, State and House Representative, and the main visual element of the dashboard, the map. The regional boundaries seen in the map are based on the current congressional districts of each state. The geographic data is based on State and District data and by utilizing Tableau Public mapping tools, a method already containing shape data for major regions. Congressional districts, although not a widely recognizable shape, were selected as the best portrayal of the House of Representatives voting base. The top right visualization, “U.S. States by no. of proposals and dominant party,” displays the number of bills sponsored in a state and the dominant party of the state, based on the number of bills sponsored. The blocks of colors used to represent states and bill types would not have accurately represented Congressional Districts in bulk, which would contain too many results and similar names.

The red and blue show the political party affiliation of a representative based on information found in the 2017 U.S. Government Manual. The gradient depicts the number of bills a representative has sponsored in comparison to other representatives. Tableau public cannot assign more than one dimension or measure to one visualization. The “heatmap” effect seen in this visualization is achieved through the use of a calculated field and an If Then statement formula. The statement and discrepancies within the data set may have influenced the biggest challenges and mistakes discovered in the product. The final dimensions, measures, and attributes used to create this version include the following. Filters: Bill Number based on count, District & State for mapping, Full Name, the Sum of PartyCount (the measure resulting from the If Then calculated field). Labels and Attributes: Full Name, Position, State – District (appears NY – 1), Website URL, Party, State, and Bill Type.

 

Data Processing

 

The data used to create this visualization was collected and aggregated from the Government Printing Office Bill Status Bulk Data Repository, examining specifically the House bills, resolutions, joint resolutions, and concurrent resolutions proposed in the 1st session of the 115th Congress House of Representatives (2017), as well as the 2017 Government Manual House of Representatives Directory. The datasets used in this visualization require preparation and cleaning before comparison, particularly the bulk bill status data. Although the highest care was taken to provide accurate and functional data, there are many factors that can influence the final results. The Bill Status Bulk Data is thousands of XML files split between the 4 bill types, all of which lack a similar schema that would allow easy viewing or comparison. Discrepancies found in the data after cleaning include; columns with different names but like data from the same set, such as Congress and Congress26. Information was also discovered missing from some columns, such as introduced date, limiting options for filters and potential visualizations.

 

U.S. Map of Congressional Districts

 

Congressional districts were selected as the best method for depicting the House of Representatives by their voting base. However, the shapes are unrecognizable to a majority of users and do not encourage interaction with the map. On a large scale, most Congressional districts on map are essentially invisible or meaningless. This prevented success for users navigating the visualization. The map was slow to load and could not filter for state geographies alone. One participant remarked that they selected their states because they were initially visible on the map, this is only true for a handful of states represented by one At-Large Representative. Two participants reported the “U.S. States by no. of proposals and dominant party” visualization as the main element and expected this to be the method for selecting and filtering state region with the map. The filters here were specifically disabled when errors were observed in color meaning and represented party. Labels based on party name were removed for this reason as well, causing the visualization to greatly misrepresent its containing data. See color and Gradient section for more information related to color problems. See Interactions for more information related to interaction problems.

 

Usability Task Analysis Survey

 

Participants were gathered by personally approaching friends and coworkers and asking them for their support. 5 voting age adults between 26 and 65 from a pool of friends and coworkers were asked to participate in a usability study focused on evaluating the Proposed Legislation and Sponsors in 2017 visualization. Those asked are currently working for an educational institution and work with the community in teaching to some capacity. Participants were asked to answer demographic data, complete two tasks, and respond to reflective questions about their experience. A table of results can be found online with google docs here. Details as they are relevant will be discussed within different sections. All participants were asked for additional feedback and reflection in a personal interview following completion of the usability task analysis.

 

Color and Gradient

 

It was not clear if the use of color accurately portrayed the intended information. The red and blue was intended to symbolize the political party affiliation of a representative, and at first seemed to achieve this effect. The gradient was meant to represent the number of bills a representative has sponsored in comparison to other representatives. The data does not contain co-sponsor information to represent bipartisan sponsorship and should not mix the two colors (blue and red). Each color should fade to a gray, representing less bills rather than a bipartisan bill. However, when a district is selected it will often appear to contain proposals for both parties. This can be easily seen in Delaware, which contains one At-Large Congressional District and only one Representative for the entire state, Rep. Lisa Blunt Rochester. Representative Rochester has sponsored four bills this year and, according to the original data set, her party affiliation is Democratic. When Delaware is selected in the visualization the “No. of Proposals in 2017 by bill type and party” produces a Red block of color for one of Rep. Rochester’s sponsored proposal. [IMAGE] Unless a legislator has changed parties mid year, which has not occurred, we should not see this. The problems identified here are most likely a problem with formula used to create the gradient. The If Then statement used attempts to create a measure based on party affiliation, assigning all Democratic party types along a negative number scale and Republican party types on the positive number scale. This formula or field has not behaved as original perceived but the problem may originate with the datasets themselves.

Interactions

 

The Proposed Legislation and Sponsors in 2017 visualization was designed to be interactive. The map is programmed to filter the other two visualizations; however, this action does not behave the same going both directions. One respondent to the usability task analysis survey specifically noted her expectation that all parts of the visualization would be responsive. The user remarked afterward in a brief interview that she wanted to use the “U.S. States by no. of proposals and dominant party” visualization to filter the map data. They selected Arkansas via this method but was unable to report on the state’s number of districts because they weren’t able to view the state on the map. They later remarked that they had not noticed the filter tools located below the map.

 

Recommendations

 

Future iterations of this project require a different approach when processing data. The source selected contained significantly more information than was used. Bills included multiple rows for one record and additional columns that reported on recent updates, included full text descriptions, and more. Data that requires less cleaning or a visualization using the complete dataset would ideal. Overall, he current product and data sets require more work to be a meaningful and useful (or even usable) resource.

 

Bibliography

 

1. U.S. Government Printing Office. Bill State – Bulk Data. FDSys. December 12, 2017. Retrieved November 20, 2017. https://www.gpo.gov/fdsys/bulkdata/BILLSTATUS/115
2. U.S. Government Manual. The House of Representatives. Washington, DC. August 2, 2017. https://www.govinfo.gov/content/pkg/GOVMAN-2017-08-02/pdf/GOVMAN-2017-08-02.pdf

Kieran Raines