Poli-Sci Perspective is a weekly Wonkblog feature in which Georgetown University's Dan Hopkins and George Washington University's Danny Hayes and John Sides offer an empirical perspective on the issues dominating Washington. In this edition, Hopkins looks at research showing which words best predict the passage -- or failure -- of legislation.

Tax rates, discretionary federal spending, immigration, the farm bill, gun control — the set of high-profile issues awaiting the 113th Congress is a long one. And in light of the failure of House Speaker John Boehner's “Plan B” on Dec. 20, observers are left to wonder whether any substantial legislation will be able to wind its way through the next Congress. So as leaders of both parties try to find compromises to address the “fiscal cliff” and the challenges beyond, they might want to know if there are any magic words that help bills become laws. 

In a 2012 conference paper, computer scientists Tae Yano and Noah A. Smith teamed up with political scientist John D. Wilkerson to look at a variant of that question. They studied the 51,762 bills introduced in the House between 1993 and 2010. Just a warning, though: their results tell us more about how we find ourselves in our current fiscal predicament than how we might escape it.   

Most of the bills died in committee, with just 13 percent having won committee approval and just 7 percent having passed the House.  Unsurprisingly, bills are more likely to survive the committee stage when they have powerful sponsors: members in the majority party, members on the relevant committee, and members who chair that committee.  Somewhat curiously, bills are less likely to make it if they're sponsored by a member from New York or introduced in December.  Bills sponsored by Ron Paul (R-Tex.) or Bob Filner (D-Calif.) are also less likely to survive.  (Incidentally, or perhaps not so incidentally, both recently announced their retirement from Congress.) 

Still, one of the paper's main contributions is to show that if we want to predict bill survival, knowing what words it uses can be very useful.  In fact, once you know the words in a bill, the added predictive power from other information (like the bill's sponsor) is limited. And what are those seemingly magic words that help bills run the legislative gauntlet? In the texts of the bills, words like “resources,” “authorization,” “authorize,” and “amounts” predict passing through committee, presumably because that is the language of ongoing appropriations.  (Also predictive of success: “administration,” “judiciary,” “march,” and “Texas.”)  At the same time, words like “percent,” “revenue,” “security,” “energy,” and “require” are among those that make bills less likely to win committee support.  The word “armed” is a hindrance, not a help.  Same with “human” and “concerned.”        

Top words predicting that a bill will survive committee...

and words predicting that it will not.

























The researchers also examined what words in a bill's title are predictive, and found that words like “internal” and “revenue” were among the most likely to predict a bill's demise. You might think that certain evocative words in their titles make bills more likely to pass — think of the USA Patriot Act.  In truth, though, it's the dry language of legislating, from “title as” to “other purposes,” that predicts committee approval. 

There are certainly institutional explanations for many of these correlations.  As the paper notes, within tax policy, many bills become bundled at the committee level, helping explain why words like the elements of “Internal Revenue Service” are predictors of bills that do not see the light of day. And if a bill's title uses a word like “designate,” it could well be legislating about a non-controversial topic like a federal post office, easing its trip to enactment. 

But reviewing those terms with predictive power, one can't help but see a trace of our nation's recurrent budget deficits. Bills that authorize spending are more likely to survive, while bills that talk about revenues or mandates are less so. 

Correlation does not equal causation. Still, as lawmakers think about legislation to address a wide range of issues in the next Congress, from the “fiscal cliff” to immigration, they might want to find a way to keep the word “revenue” out of their bill texts and titles, just in case. They should be wary of words like “require” and “make.” They should think about mentioning the state of Texas in the bill's text, and making sure that the title highlights its “other purposes.” And if the past is any guide, they might not want to introduce the legislation in the month of December.

For a swifter resolution to the year-end fiscal cliff, I'd propose to call whatever the compromise is “An Act to Extend the Relief of the Public and for Other Purposes.”  Unless, of course, I didn't like the compromise, in which case I'd call it “An Act to Prohibit the Internal Revenue Service from Raising More Revenue.”