A3-ZevWinkelman

From CS 294-10 Visualization Sp10

Jump to: navigation, search

Assignment 3 - Web Notebook

Step1: Domain

I am interested in the activities of Congress.

Step2: Initial Question

Is government 'broken' ? Is there less progress overall in all areas of Congressional deliberation over legislation, compared to the previous year ? Compared to the previous first term of a new president ? Compared to the previous first term of the last president to initiate major health care reform ?

Step 3: Database

The website: http://www.thomas.gov has the data that I need, but there are no data dumps available for download.

Using custom software to download and parse the relevant files from the website I was able to create my own data set for this assignment.

I created two structured record sets from this data:

Bill [ BillNumber, BillCode, BillSponsor, BillTitle, SessionNumber ]

and

Bill Action [ BillNumber, BillCode, BillActionDate, BillActionIndex, BillActionText, BillRecordType, SessionNumber ]


Exploratory Analysis Process


The first challenge is how to measure progress in all areas of Congressional deliberation. There are thousands of bills and hundreds of members of Congress. Each bill goes through several stages before becoming a law. Not all bills become law, and not all bills go through the same stages. But any bill that does become law must pass both houses of congress and be signed by the President.

One way to represent a session of Congress is as a time series. It is similar to the challenge of showing flight information and train schedules, only harder, because only some of the stops are shared by different bills, and the stops can be reached in different order.


A first cut visualization might try to show the various actions that Congress takes throughout one session. Some actions are taken frequently, others much more rarely. Thousands of bills are referred to committee, only a subset of those pass either house, and a subset of those are sent to the President. Plotting the count of the cumulative times these actions have been taken over time reveals the pace and type of activity in Congress.

I tried plotting all actions taken in both houses, but many of the actions were only taken a small number of times. This presented me with a dilemma: include the clutter of many overlapping small-n time series, or filter and lose completeness. (image actually filtered - there were many more)

File:Zw_a3_screenshot1.JPG

A second attempt separated House Bills from Senate Bills. This made things much clearer, but made it harder to assess Congress as a whole. Further work would be required to recombine these subsets in a meaningful manner. I also chose to filter in this second iteration further, including only actions taken at least as many times as turning a bill into law, but the choice still bothers me:

File:Zw_a3_screenshot2.JPG

I needed a baseline from a previous session of Congress, plotted on the same chart, in order to make comparisons, so I created another data set for the previous session of Congress:

File:Zw_a3_screenshot3.JPG

As the last step, I realized that comparing across was easier than comparing vertically, so I rearranged the small multiples:

File:Zw_a3_screenshot4.JPG


The next set of questions raised, but not answered in this assignment involved more granular representation of the data:

Just because a bill does not pass, does not mean that Congress has not been doing its job. Each intermediate action taken after a bill is introduced represents some work being done by Congress. Actions involving committees are one major category of activity to analyze, as most bills are referred to and reported out of various committees throughout the legislative process. For a given committee I'd like to be able to show: how many bills were referred in, how many bills were reported out, the average, minimum and maximum turnaround times for bills referred to committees, all plotted daily for a given session of Congress.



[add comment]
Personal tools