Skip to main content









Donation Heart Ribbon
Visit the Midday Edition homepage

Forecast: If Vote Was Today, Immigration Reform Would Fail House

June 4, 2013 1:24 p.m.


Tom Wong, Assistant Professor Political Science, UC San Diego

Related Story: Forecast: If Vote Was Today, Immigration Reform Would Fail House


This is a rush transcript created by a contractor for KPBS to improve accessibility for the deaf and hard-of-hearing. Please refer to the media file as the formal record of this interview. Opinions expressed by guests during interviews reflect the guest’s individual views and do not necessarily represent those of KPBS staff, members or its sponsors.

CAVANAUGH: Leaders of both political parties in Washington have to get very good at a crucial legislative skill: Counting votes. To be effective, they have to know if they have the votes to pass a piece of legislation, and if they don't, where those extra votes might come from. But that kind of counting is still done old-school where political wheeling and dealing plays a big part. A new-school kind of counting subbing conducted here in San Diego. It's an attempt to forecast legislation from deep data mining, and the results do not measure up well for the upcoming vote on immigration reform. My get, Tom Wong is assistant professor of political science at UC study. Tom, welcome to the program.

WONG: Thank you for having me.

CAVANAUGH: We've all heard about forecasting the results of an election. Is forecasting legislation a relatively new concept?

WONG: Not really. A lot of what political scientists do is try to explain outcomes. So we use the methodological tools we have to do just that. The difference here is that I'm just building these models to predict forward. And that's something that we're actually trained against because predicting future events, well, we can be wrong. And in our craft, if we're wrong, we are very publicly wrong. But with immigration specifically, I think that's enough stickiness in how legislators vote on this issue where we can model past votes to make informed predictions about how current legislators are going to vote on the immigration reform bill.

CAVANAUGH: And your frustrate conducted in any way similar to the kinds of things that people do when they're forecasting elections?

WONG: So yes and no. For somebody like Nate Silver during the elections, there is a lot of reliance on opinion polls.

CAVANAUGH: He predicted the correct vote in each of the 50 states in our latest presidential election.

WONG: Yes, for the most part. And in terms of forecasting election outcomes, analysts and political scientists are relying on things like polling data, and trying to think about how individual voters may act, and what their preferences and incentives may be to support a certain candidate over another. What I'm doing is trying to apply that same consent, but to think about the incentives and preferences of legislators. And so that can be more tricky in some ways, but less tricky in others.

CAVANAUGH: Is this -- are you coming up with a mathematical model for this?

WONG: Yes, yes. So we're talking about taking all past votes on immigration-related bills if 2005 to present. And so 2005 is an important date because that's when we had this thing called HR4437, a restrictive immigration policy that was actually approved in the House, but subsequently led to the 5 million people out in the streets protesting in support of immigrant rights in 2006. I take that as my point of departure and collect all immigration-related votes since then. That gives us about 20,000 pieces of information in which to try to make sense of, and in making sense of that, that's how the model is constructed. What factors explain those yes votes and what factors explain the no votes. So we come up with these matrices of hundreds of thousands of pieces of information all just to sort of get a sense of what matters for immigration policy for legislators.

CAVANAUGH: And so does political party affiliation, is that part of what you take into account, the demographics of districts?

WONG: Uh-huh. So data mining in some ways is a bad word in the social sciences because that kind of applies the absence of theory, that you're just digging into the data and shooting in the dark. Before these models, we're talking about the factors that previous researchers identified as being important. And so yes, definitely a battery of political variables are included in the model, like party affiliation, the partisanship of certain states and districts. And when you think about demographic factor, those are critically important. There are sort of dozens and dozens of iterations of different models based on different assumptions until the model that returns the highest percentage of correctly classified votes is found.

CAVANAUGH: Okay, I don't want to keep the big reveal to last. So let me just tell everybody, tell us what your forecast shows.

WONG: For the Senate, the data shows 67-71 yes votes. And that's very good. We just need 60 votes. For the House, things don't look good. At best, my models predict 203 yes votes. But the story doesn't end there. Because after identifying all of the yes and the no votes, then the real work of advocacy begins. And so on both sides of this debate, there is intense lobbying that is already going on and has been going on for quite sometime now. And so for all of those house representatives who are on the fence and leaning toward no, we can use this model, identify those individuals, and then dig more deeply into what their districts look like in order to sort of figure out what may be an influential pitch for that representative to say, you know what? Supporting immigration reform is in the best interests of my constituency.

CAVANAUGH: Now, the vote in the Senate is expected next week. Senate chuck Schumer says he expects about 70 senators will vote in favor of it. That's pretty close to your prediction. Does that bode well for your prediction in the House, do you think?

WONG: Well, so the actual vote in the Senate won't be next week. That's further down the road. It will be introduced to the floor next week. And the gang of eight, the bipartisan group working on the bill, they have lofty goals. They want 70-75 yes votes. There is some disagreement even within the gang of eight as to how many votes there currently are. Schumer and McCain says 70, Marco Rubio says we don't even have 60 yet. That is somewhere around the range of what I predicted for the Senate. Maybe that's a good thing. I hope it's not a good thing. Because for the House, I would be very happy to be wrong. Because 203 does not get us to an immigration reform bill.

CAVANAUGH: And just to remind everyone, the vote level that this has to reach in the Senate is 60 and beyond.

WONG: In the Senate, it's 60 for a filibuster-proof vote. In the house, it's 218 for a majority. In the House, we can expect some representatives to abstain and not vote. So 218 is the bench mark, but somewhere around there is what's needed.

CAVANAUGH: Give us an idea of the kind of -- the way these factors combine so you can kind of figure out whether one representative is going to vote one way or another. You take their previous voting record, the demographics of their district, their party affiliation, and do you make that into a mathematical model or move it away and look at it from the theory aspect as you were talking about earlier?

WONG: The step one of the process is the mathematical model, and that's where we're combining theory with the data. The second step of the process is actually what I think is most important. We can model the outcome, but we want to get a sense of how accurate the model reflects reality. And so what I think most important in this process is step 2. I take a few key votes for the Senate and the House, and make sure that the model actually predict business those key votes correctly. At the end of the day, a model is just a prediction. I think it points us into a potential strategy in terms of getting vote, but what happens on the ground is much more important. And at the end of the day, the yes and no votes that are actually cast, that's what's most important.

CAVANAUGH: Now, you have a particular personal reason that you chose the immigration reform issue to begin this project. Can you share that with us?

WONG: Yes, yes. My personal background is being undocumented. So my parents brought me and my brother here when I was two years old. We came from Hong Kong, lived in Riverside our whole lives. And we overstayed our Visas. So we're undocumented. And I didn't learn about my status until I was 16 and just like a lot of other undocumented youth, it is shattering for one's sense of self to learn something that heavy. So that's my personal experience that motivates this interest in immigration policy. So I do have a sort of dog in this fight because at the end of the day, the models begin with sort of math and numbers. But they're a bit deeper for me than just models and numbers. I see the 11 million undocumented immigrants in my results, I see their families, I see the potential hopes and the opportunities that can come to these families if we get an immigration reform bill passed. So while the data themselves are sort of agnostic, they don't have an agenda, well, I actually want to see this passed. And I want to be able to use the tools that I know how to use to try to help make that happen.

CAVANAUGH: What does your modeling say about San Diego's congressional delegation?

WONG: We have people like Susan Davis with a very long history of supporting progressive immigration reform legislation. We have two new Democrats, Juan Vargas and Scott Peters who are predicted to lean toward support. But then we have folks like Duncan Hunter. Based on his party affiliation, his voting record, bills that he has sponsored related to immigration, as well as his father's legacy all points to Hunter being somebody who opposes the type of immigration reform bill that we currently see in the Senate. So specifically that's to say he opposes something with a path to citizenship. And in terms of the approach of the House moving forward, a lot of discussion is about a strategy of delay and dismemberment. Either delay this until 2014 midterm elections are on the minds of house reps, or dismemberment, to take this comprehensive bill apart and vote on specific measures. It's very predictable what will happen if the dismemberment strategy is actually what the House goes with. Because if you think about the key pillars of immigration reform being a path to citizenship, increased border security, and a number of initiatives, there are people like Hunter who are perfectly on board with tighter security to see that through. But in terms of a path to citizenship, the dismemberment strategy would make the most contentious part of the comprehensive immigration reform bill something that doesn't pass the House.

CAVANAUGH: If I understood you correctly, you said that part of doing this modeling, part of getting this information for lobbying purposes, lobbying in support of the immigration reform bill is to find those members of Congress who may be modeled voting against this bill, pointing out to them perhaps that the demographics of their districts have changed to such an extent that it might actually be advantageous for them to reconsider their vote. Let me take that a step further. There are some Republicans in districts where many constituents might want immigration reform but they may not vote for it to avoid a primary challenge from the right. Does any of your modeling reflect that kind of political variant?

WONG: Not so much the Grover Norquist strategy of primarying other Republicans. But there is a similar dynamic at play. In identifying all of the predicted no votes, then we look more closely into their districts and think about the young Hispanic, Latino, and Asian populations who will come of voting age in 2014 or 16, and to actually project out, thinking about 2011 election margins, how that margin may shrink in 2014 if these young people who come of age actually do register and actually do vote and remember the representatives' stance on immigration. Then we do the same thing for 2016 to sort of create that same sort of electoral concern whereas on the one hand, it's potentially a primary challenger, whereas on the other hand projecting forward in terms of demographic changes may make clear that subsequent elections are going to be very dicey for certain representatives. So somebody like Gary Miller in Southern California, Riverside, San Bernardino area, he won in 2011 by about 16,000 votes. The young Hispanic, Latino, and Asian population in his district alone who will turn 18 by next year, we're talking about over 30,000. And so with caveats that they're not all citizens, potentially, they may not all register, and they may not all vote, that number still is almost double his election victory in 2011. So that's the type of pressure -- the type of pitch that may sort of cause some electoral anxiety. And if I can just say that I do have a personal stake to see immigration reform passed. But all of the analyses are public. So for those who want to undermine immigration reform, they can use the results and do the same work we're doing.

CAVANAUGH: And finally, regardless of the outcome of this bill, you're going to be watching for key votes in both the Senate and the Congress to see whether or not your modeling actually works out correctly; is that right?

WONG: Yes. So for both houses, once the debate gets ramps up and all of the contentious amendments are voted on, I have about a dozen different analyses related to the votes that are most likely to be contentious.

CAVANAUGH: Thank you very much.

WONG: Thank you.