The process of shaping the law often resembles an Indiana Jones maze. Bills and amendments run through an elaborate system of committees, sessions and hearings filled with booby traps before finally reaching the golden idol of a final approval.

Parliamentary monitoring organizations and researchers are often interested in how various pieces of legislation survive in this environment and what are the strategies to either kill or aid them. This specifically means answering two questions: What is the probability of a bill being approved and what factors determine this probability?

The legislative process is usually hierarchical: Successful completion of a step in the process is conditioned by completion of all previous steps. Therefore, we may also want to know the probabilities of completion in each consecutive step and their determinants.

A simple way how to give a satisfying answer to these questions without wandering into the land of nonlinear logistic regressions is the Sankey diagram. It is a famous flow chart in which a process is visualized using arrows. Relative quantities of outcomes in the process are represented by arrows’ widths.

A famous example is a Sankey diagram of Napoleon’s invasion of Russia. We can clearly see how the Grand Army was gradually shrinking as French soldiers were dying or defecting. Another well-known example is the Google Analytics flow chart. It shows how many visitors enter a webpage and then either leave or continue to a different page on the same website. As the number of consecutive steps increases, the number of visitors remaining on the website decreases.

The legislative process can be visualized in the same way. Progress of bills is represented by a stream between various steps in the process and width of the stream corresponds to quantities of bills. A bill can either complete all the steps of the process, or it can “drop out” of it at some point if it gets rejected.

Let’s take a look at an example of the Czech Republic. Its Parliament is bicameral, consisting of the Chamber of Deputies and the Senate. The government and the president also enter into the legislative process at certain points. The Chamber of Deputies was dissolved very recently to hold early elections. This means that all bills that had not left the Chamber before the dissolution were automatically rejected. It is interesting to find out how many bills were rejected this way and in what step of the process.

The simplest Sankey diagram shows the process of approving the Czech state budget bills. These bills are only approved by the Chamber in three readings and then signed by the president. The Senate does not enter into the process. There were only four such bills in the last term (the budgets of 2010, 2011, 2012 and 2013). Three of them were approved and one remained unfinished due to the dissolution of the Chamber.

In the visualization (click here to see full view), steps of the legislative process are represented by white moveable rectangles. Successful completion of a step is visualized by a green stream and an unsuccessful one is visualized by a red stream. A blue stream corresponds to a successful step that takes place inside the Chamber of Deputies. Let’s now move to a more complicated example.

There are several types of bill sponsors in the Czech legislative process. More than 50 % of bills are introduced by the government. Bills can also be introduced by an individual deputy; a group of deputies; a group of senators, if approved by a majority of the Senate; and by a local government. Senators and local governments only introduce a handful of bills per term.

Government bills are expected to be more likely to successfully finish the legislative process since the government is backed up by a majority in the Chamber. Let’s find out. This is a Sankey diagram of bills sponsored by the government in the last term:

As we can see (click here to see diagram in full), there is the Senate stepping into the process. Both the Senate and the president can veto a bill. If this is the case, it returns to the Chamber where it needs to be reapproved by a majority of deputies. These vetoes are represented by yellow streams. As we can see, the Senate vetoed about a half of the bills (89 out of 160) while the president only vetoed a handful of bills (10 out of 234).

We can say that the government was indeed fairly successful. Only a very few of its bills got rejected, mostly in the Chamber. Many bills, however, have not yet finished the process. If they hadn’t left the Chamber before it was dissolved, they were automatically rejected. Bills that are currently in the Senate or on the president’s desk can still successfully finish the process but these two institutions now have a “pocket veto” because there is no Chamber of overrule them.

Let’s take a look at bills introduced by deputies, senators and local governments:

The picture is very different. There is an additional step at the beginning of the legislative process: A non-governmental bill must be first commented on in the government. As we can see, there were 10 bills at this stage of the process when the Chamber was dissolved and where thus automatically rejected.

We can see that non-governmental bills have much lower success rate and that they mostly drop out of the process in the first reading of the Chamber - the first step when a bill is voted on. Most of them were introduced by opposition deputies and thus lacked a broad support.

However, if a non-governmental bill made it out of the Chamber, it was relatively safe. This is due to the fact that the Senate is currently controlled by the parties that that were in opposition in the Chamber. Therefore, it obviously had no interest in vetoing bills sponsored by opposition deputies. Also, the president is inclined to the opposition parties.

The dissolution of the Chamber penalized the government more severely than other bill sponsors: some 25 % of governmental bills got automatically rejected due to the Chamber being dissolved. Only some 10 % of bills sponsored by other entities got rejected this way.

What type of data is necessary to create this visualization for a different Parliament? It is a dataset of bills and their transitions between various steps in the process. A transition is defined by an input step, an output step and an action that moves a bill between these two steps. Transitions are then organized into a model of the legislative process and their frequencies are used to create the visualization.

In many countries, the necessary data is published on a parliamentary website. An example from the Czech Republicis a highly controversial bill on the state support of renewable energy sources introduced by the government. As you may be able to figure out using a translator, it was vetoed by both the Senate and the president but managed to finish the legislative process nevertheless.

The Czech Parliament actually publishes this and other types of parliamentary data as a downloadable database dump. Therefore, a researcher does not have to scrape the website. Thanks to a high standard of transparency and data openness, it took me less than one afternoon to make these useful visualizations.