Policymaking seems dominated by polarized views and misinformation. But what if legislators are willing to listen to and share expertise on policy issues? Adam Zelizer provides informative briefings to state legislators on pending bills and finds that they listen, increasing their co-sponsorship of those bills and bringing along their office mates, as information spreads from one lawmaker to the next. Christian Fong finds that when legislators are assigned to new committees in the middle of a term, the legislators they usually work with start voting with them more often on bills in that committee’s issue area. Lawmakers take cues from the most informed legislators. They may be doing the best they can with limited knowledge and a lot on their plate.
Matt Grossmann: When policymakers follow informed expertise, this week on The Science of Politics. For the Niskanen Center, I’m Matt Grossmann.
We’re used to hearing that politics is dominated by polarization around existing views with everyone spreading misinformation to support or side, but what if policymakers are willing to listen to expertise and information about policy issues, even share that information with fellow lawmakers, and then even act on it in sponsoring and voting for legislation?
Perhaps policymakers are doing the best they can with limited knowledge and a lot on their plate. If so, a bit of trusted information can influence them and lead to better informed policy.
This week, I talk to Adam Zelizer of the University of Chicago about his American Political Science Review article, Is Position-Taking Contagious? He directly provides informative briefings to state legislators on pending bills and finds that they listen, increasing their cosponsorship of those bills. Not only that, the brief legislators’ office mates also change their behavior as the information spreads from one lawmaker to the next.
I also talk to Christian Fong of the University of Michigan about his Journal of Politics article, Expertise, Networks, and Interpersonal Influence in Congress. He finds that when legislators are assigned to new committees in the middle of a term, the legislators they usually cosponsor bills with start voting with them more often when it comes to bills in the committee’s issue area. Lawmakers take cues from the most informed legislators. When the most informed change, they use the new better informed signals.
Today, an optimistic account of lawmakers, starting with Zelizer field experiment.
Adam Zelizer: My research is about how legislators make decisions, why they support some bills and oppose others. And I wanted to study this through a somewhat new approach, which was to conduct experiments. So in an experiment, we randomly provide information about something, in this case, a bill, and see how that information changes a legislator’s position.
And so I worked in unnamed Southern state legislature for a couple of years and I worked on behalf of a caucus, which is a group of legislators organized around a specific interest. And I went and held meetings with about 75 legislators where I briefed them about real bills on the legislative agenda. I had briefings prepared on 16 bills and I picked four of those 16 at random for each legislator.
So we might think that whether a legislator supports a bill depends on a lot of things. It depends on what their constituents think, it depends on what party leaders or interest groups want them to do, but it also depends on their knowledge of just what the bill does.
And there are a lot of bills. They’re complicated and legislators don’t that much time. So we might think that concisely describing what these bills intend to do might be impactful. And what I found is that giving a short overview of a bill, about a one page research report and about 15 minutes of discussion for four bills made legislators reasonably more likely to come out publicly and support a bill via cosponsorship.
So talking to 20 legislators convinced one to cosponsor one of these bills. And that’s a relatively modest effect. It’s not enormous. We might be a little suspicious if the effect were too big, but it is a real effect and it is caused by this bill information that I randomized.
So I conducted that first experiment in 2015 and I went back to that same legislature in 2016 and conducted a replication, the same type of information, so bill impacts, bill research on veterans’ bills. This time there was something of a wrinkle. I wanted to know not only what was the effect of giving this information directly to legislators, what is the effect of a bill briefing, but I wanted to know if that effect traveled.
So there is a long tradition in legislative studies of studying how one legislator’s decision depends on what other legislators do. It’s called cue-taking. And it’s hard to study cue-taking because if we just ask legislators, “Do you support this bill or not and do the other legislators you’re close to support this bill or not?” we might see a correlation, but that could just be self-selection. Legislators with similar political ideologies tend to associate with each other, to be friends, and so there might not actually be an effect of one on the other. And it turns out it’s pretty hard to illustrate this effect.
And so one new way of doing so using experiments is to estimate the spillover effect of the bill briefing. So randomizing a bill briefing across legislators identifies that briefing’s effect on that legislator. It also identifies the effect of briefing that legislator on a legislator who might interact with that person.
And in the second study, I end up pooling those two together and I estimate that the cue-taking effect or the spillover effect is about 60% as large as the direct bill briefing effect. So if you have to meet with 20 legislators to convince one to cosponsor a bill through the briefing, you might have to meet with… If you meet with 33, there might be one who is convinced to cosponsor it through this cue-taking or spillover.
Matt Grossmann: Zelizer went to work. He got permission to actually intervene in a legislature.
Adam Zelizer: I was a third year PhD student coming up with my dissertation plan and I wanted to study these caucuses, these groups, but I couldn’t come up with a good research design, in part because there was a new book coming out by Jennifer Victor and co-authors on caucuses in Congress that collected all this great data. And so I thought, “They’ve already collected all the data on caucuses in Congress, what can I do? Well, maybe I can just go start one. I’ll go start a caucus somewhere.” And that was the kind of hubris I think characteristic of an early PhD student.
And so it was about to be winter break. I was going to go visit my parents who happened to live in this southern state capital. And so I reached out to someone I knew and asked for a meeting with the chief of staff to the state speaker.
And so I met with him I believe on New Year’s Eve and said, “I want to start a caucus to see how they influence policy,” which is a bit of a wild thing to suggest. I said, “I don’t want to ruffle any feathers.” This was back in 2015, well, 2014 I guess at that point, maybe even a bit earlier. And so obviously everyone had the Tea Party or the Freedom Caucus on their mind, but I wasn’t going to go into a legislature and say, “Let’s start a very ideological, very angry caucus,” so I said, “Let’s start one about veterans affairs or mental health.” These are relatively nonpartisan valence issues. He said, “Let me go talk to the Speaker.” He did. A couple days later, he said, “It’s fine with the Speaker, it’s fine with us.”
And so I started reaching out to legislators who I thought would be good to be the chair of one of these caucuses and it turned out there already was a veterans caucus. So lucky me, I guess it shows I didn’t do all that great research to find this out. It turns out caucuses are relatively hard to identify because they’re not formal groups.
And so this group already existed. I started talking to the chair. She invited me to present to the whole group about my proposed research. So I went down and was in front of a panel of legislators saying, “I want to study how you influence policy by working for you as an unpaid intern,” and they said, “Oh, sound fine to us, it’s free labor. And so, next session, come on down and you can be our intern.”
And so unfortunately state legislative sessions are short. So that was in 2014. I had to wait eight months, and by early 2015, I was back in that state capital ready to conduct research.
Matt Grossmann: He found and confirmed that information travels across shared legislator offices.
Adam Zelizer: The key to the design of the cue-taking paper is that legislators shared offices. And their assistants would sit in a shared room and then the legislator would have a private office. And what I found pretty quickly was that the whole life of the office takes place in that shared room. The assistants talk to each other. Assistants really rule the world in politics and in life. They’re the ones who know what’s going on, legislators are talking to them, all sorts of information is being shared. And I knew that because I had a tiny little desk as the intern in that shared room.
And so I spent all day with the two assistants to two legislators and we just talked all the time. And so, pretty quickly it became apparent if this legislator were to share the information I gave, it would be with the other one in the office. And so that is what motivated that interoffice design, which was really unique. Not every legislature has shared offices, but this one did.
Matt Grossmann: He gives an example of a bill that he briefed on.
Adam Zelizer: So there’s one bill that I always pick because I think it’s characteristic. It’s a bill about higher education scholarships for ROTC students. ROTC is Reserve Officers’ Training Corps. It’s for students to prepare for a career in the military. And so there are a large number of students at public and private colleges who are in this program.
In this state, there was a scholarship for all in-state residents to attend in-state colleges and universities. This scholarship I think paid for about $2,000 per semester, but it had a cap and it had a credit hour cap. So if you need a hundred credits to get a degree, it said, “We’re not going to pay out this scholarship after you get a hundred credits.” You don’t want someone in year six or seven still getting this scholarship. They wanted to target it at about four years.
The thing is, if you’re in ROTC, you take extra classes. So you take classes not because they’re targeted for your degree, but because they’re targeted for your military training. And that meant that students in this program were hitting their cap a semester or even two before they graduated. So in effect, they were getting less of the scholarship than other students.
And you can imagine that essentially taking money away from students who want a career in the military may not be the most politically popular position, so a legislator sponsored a bill that would exempt these extra classes from that cap. Okay, you would think this is a politically popular bill. It’s pro-military, it’s pro-student, it’s pro-education. Great, right?
Well, even though it might’ve appealed to legislators, it didn’t pass. This bill didn’t pass because it cost money. At $2,000 to $4,000 per student, even if it’s only 50 or 100 students, that’s a non-trivial amount of money. $500,000 may sound like a rounding error for government budgets, but a lot of state budgets are very tight. They have balance budget requirements and there’s not an extra half million dollars sitting around for programs that the back bench legislator might hope to propose. So that bill didn’t actually pass.
Matt Grossmann: Legislators have long told political scientists that they take cues from other legislators. We’ve now confirmed that.
Adam Zelizer: In John Kingdon’s classic book, Congressmen’s Voting Decisions, he asked legislators, “How often do you take cues? How often do you look to someone else to kind of decide how to vote?” And legislators say they do it overwhelmingly more than half the time.
And so we might have thought that was an overestimate. I might say I look for advice to other people, but in reality, I was probably going to take the same position anyway and I’m just reality checking. I’m seeing that what I was already going to do kind of isn’t crazy because my friend is going to do it too.
These estimates say cue-taking might happen that often. Basically for every five legislators who were convinced to support these bills because of the briefing, about three decided to support them because of cue-taking. So that’s about 60%.
Now, we don’t want to necessarily extrapolate that spillover effect or that contagion rate, if you want to call it that, to other issues, other bills. These bills are relatively nonpartisan. Veterans’ issues is not the most partisan issue. It’s relatively popular. So we might think that on some very… popular. So we might think that on some very contested or much more salient issues, the amount of spillover or cue taking would be much smaller because legislators would have much better information. But on these relatively low salience and non-controversial bills, there was about as much cue taking as legislators said there was.
Matt Grossmann: Fong also went back to the fundamental question of how legislators make voting decisions.
Christian Fong: What I was really interested in was how Congressman decided how to vote, which sounds to a lot of people like a kind of obvious question where we already know the answers, right? We think, “Okay, well, they want to get reelected.” So maybe things that their constituents want, maybe things that their donors want. These people often have very strong beliefs about public policies, so maybe there’s some ideological stuff that goes in there. But what I thought a lot of people didn’t appreciate was that, often, those are not very useful guides for the kinds of questions that members of Congress are asked to answer, right? I might know that I represent Michigan or might be an economic conservative, but all of a sudden I’m being asked to vote on this driftnet by catch reduction bill, right? And first I’ve got to go, “Okay, well, do I want driftnet by catch to be reduced? And if I do, is this a sensible way to go about doing this?”
And all of a sudden, my usual means by which I decide what to do for all apart very quickly. So I was curious, how do legislators decide how to vote in these sorts of situations, which are, in fact, most of the votes that they’re asked to take are on these sort of technical issues that don’t have clear ideological answers? Or even if they do, it’s not immediately clear to a layperson how you would go from knowing, “Okay, I’m a conservative or I’m a liberal,” to knowing how you want to vote on this obscure regulation or this government program that you’ve never heard of before. So there is actually a really old hypothesis on how legislators deal with these votes by Matthews and [Stinson 00:14:55], which says that when you don’t know what to do, you can look at somebody who does know what they’re doing, right?
And that immersive Congress tend to specialize in particular issue areas. And maybe you know that there’s somebody who you tend to agree with a lot who knows a lot more about fishing issues than you do. And you can just ask them, “Yeah. What do you think about this bill? What are the things that we should be worried about? What are the considerations that you’ve put into deciding what you wanted to do?” And you can decide how to vote that way, right? You can essentially exploit the expertise that other people have generated on this particular issue and vote accordingly. And hopefully, if you know an expert who’s a lot like you, you can approximate how you would vote if you were a real expert on the issue. So the hypothesis was really old. The tricky part was actually figuring out a credible way to test this hypothesis. We would love to be able to get into Congress and observe these conversations happening, even if we could, even then figuring out whether the conversations made a difference would be a little bit hard.
But as a general rule, we can’t even do that, right? Members of Congress tend to be a little evasive when you ask them how they decide how to vote. They like to give off the impression of being really in control, of knowing a lot of things, and of being independently minded. So if you ask them, this is the kind of thing that they’d be less inclined to tell you about. So the trick was, we needed two things. First, we needed to find an expert, right? And then we would need to find a person who would be receptive to that expert’s advice. So the way in the paper that I operationalize who’s an expert is I look at members who are assigned to committees. You learn a lot by doing in Congress. So when you’re in a committee, you’re going to hearings, you’re marking up legislation, you’re actually reading a lot of bills.
And in the process of doing that, you and your staffers are going to learn a lot about this particular set of issues that falls under the committee’s jurisdiction, so that’s good. We found our experts, so the committee members. Now you need to figure out who’s going to be listening to these committee members. And for that, I use this other relatively old idea of the co-sponsorship network. So when a legislator introduces a bill, other members of Congress have the ability to cosponsor the bill, which basically says, “I think this is a good bill. I would like to see this passed into law.” It’s not necessarily voting for it, but it’s kind of saying that you’re going to support the bill.
The idea is that if there’s some person who joins the committee and I co-sponsor a lot of their bills, that’s probably going to be somebody whose advice I’m really receptive to, right? For a few reasons. One, we know that I tend to like the sorts of bills that they write, because I go to the trouble of co-sponsoring them. And then two, we know that this other person is pretty active, pretty entrepreneurial, because they’re going around and writing bills that people like. So these co-sponsorship networks give us a way to get in there and figure out who’s going to be receptive to what we call cues. Who’s going to be receptive to this advice that’s coming from experts that are on the congressional committee?
Matt Grossmann: His project started from asking basic questions about how to make sense of bills.
Christian Fong: Not knowing anything about Congress, I asked maybe the most elementary question you could possibly ask, which is, how do members of Congress decide how to vote? And then the way I went to answer that is I went and read some bills and tried to figure out how I would vote. And then very quickly figured out that I had no way of deciding what I wanted to do on this bill, right? Even knowing my own opinions intimately and having no reelection motives, I couldn’t make heads or tails out of probably three quarters of the things that members of Congress were being asked to vote on. And then I did some reading and it turns out that there are other people who believed that members of Congress couldn’t make heads or tails of three quarters of the things that they were being asked to vote on either.
So I started digging through some kind of older literature that looked at these more foundational questions, and that’s how I found this Matthews and Stinson book on cue taking. John Kingdom also wrote a book that makes a similar point. And I thought, “This is interesting, right?” They’ve got some nice interviews from an earlier age, back in the 1970s, when members of Congress were a lot more open to talking to academics and transparent about the ways that they made their decisions, not always to their benefit. So I thought, “Well, that’s interesting. I wonder if this hypothesis they developed based on their interviews would hold up in a modern quantitative analysis.”
Matt Grossmann: He took advantage of mid session changes in committee memberships.
Christian Fong: So the big problem that you have to deal with when you want to use committee assignments as your way of measuring expertise, and you want to look at, okay, what happens when a person joins a committee? Is that almost all committee assignments are made right at the beginning of a session. And what just happened at the beginning of the session? Elections, right? So first of all, you’ve got a whole bunch of new members showing up. Members of Congress like to think that they’ve learned something from what happened in the election. So they are often changing what they think the most important issues to work on are. The scope of things that are possible to work on changes because new members are showing up and the partisan composition has changed. So you don’t want to be in a position of having to deal with all these changes happening at the same time.
Really, you just want one person acquiring expertise and then everything else in Congress remaining kind of the same. So that’s what motivated this strategy. Instead of looking at all people who are assigned to committees, looking at people who are assigned to committees right in the middle of a session, right? And there are a couple reasons that this will happen. Typically, it’s because the member of Congress who formerly held that committee seat doesn’t have it for some reason, right? They might have died, they might have resigned, or they might have transferred onto another committee themselves. And so once that committee seat opens up, the party leaders kind of go in, they see who wants it, and then will pick somebody who they think is most qualified to take that seat.
Matt Grossmann: That means we can compare the people who just missed the cutoff and then got on the committee later.
Christian Fong: What’s nice for us from the perspective of this study is that the members who are assigned to a committee mid session look a lot like the members who joined the committee at the beginning of the session. Ideologically, they look very similar. In terms of their legislative effectiveness, they look very similar. For the purposes of estimating, what kinds of queues are being sent by members of committees? Probably the best approximation we have are these people who are getting mid session assignments. You could think of as the people who just barely missed the cutoff of getting onto the committee the first time around.
Christian Fong: When you add a new member onto a committee, the people who were co-sponsoring that legislator’s bills will vote with that member 3.9 more times over the course of a session. So about four more votes. Now, that doesn’t sound like a lot when you consider that Congress takes hundreds upon hundreds of roll call votes over the course of a session. But when you consider how much members of Congress already vote together, how high that baseline rate of agreement is, right? Typically, members on this network vote together 73% of the time anyways.
Matt Grossmann: Fong found that committee memberships have a big impact on voting queues.
Christian Fong: This is resolving a pretty high percentage of the disagreement that those legislators previously had. Before, when I wasn’t paying attention to what you thought on these particular issue versus now. Another way to look at this is you consider, okay, well, that’s just when you add one member to a committee. But in fact, there are a bunch of members on committees, right? And we might want to add up all of the cues that are being sent from all of the different members who have committee assignments. So once we add up all the sources of cues and all the people who are taking cues from them, the estimate that I get is that you’re looking at something like five to 30% of all roll call votes are decided on the basis of these cues.
Matt Grossmann: Bipartisan cooperation was common, and the effects were actually stronger across parties.
Christian Fong: You might think that these co-sponsorship networks are just tracking buddies within a particular party who tend to work together. And there’s definitely some of that. When I was in the Senate as a staffer, I took cues from other legislative offices, and a lot of those were from within our own party. But we also observe a lot of this co-sponsorship behavior across parties as well. It turns out that there are a lot of bills that are dealing with more technical issues or issues on which there’s broad consensus about what should be done, and just a little bit of disagreement over how important it is to work on that particular thing. And there, we see bipartisan cooperation all the time. One of the coolest findings I think of the paper is that actually, we observe a much bigger cue taking effect across parties than within parties.
Christian Fong: Now, part of that is just because the people who are in the same party tend to vote at very high levels of agreement anyways. So there’s not much disagreement left there to resolve. But in fact, cue taking is is, in significant part, a cross party phenomenon. You are looking at people from outside of your party because they tend to possess information that’s very different from the other people that you talk to. And that added perspective can be really valuable for maybe uncovering some aspect of the policy that you hadn’t thought of before that’s going to change how you want to vote.
Matt Grossmann: The papers are similar…says the information treatments have clear links.
Adam Zelizer: So I think what’s similar about those two information treatments is that in both of our cases, some legislators were getting basically a lesson in a bill. They were being told what the bill does, they had a chance to ask questions. They could poke and prod and figure out if this bill is a good thing or a bad thing. Committee service is an intensive experience. You, over time, presumably become an expert in that issue in a way that the legislators I spoke to and randomly provided information to, were not. My briefings were 15 or 20 minutes, somewhat long, but they’re brief compared to committee service, which can be hours and hours and hours every week for years. So the dosage of the treatment in a way in committee is much stronger, but in Christian’s paper, if we think about a legislator who is assigned to a committee, they are getting bill background quite a bit like the legislators I met with.
So there are good reason to think that caucuses supplement the committee system. Caucuses are groups of like-minded legislators organized around issues, so this one was a veterans caucus. It was bipartisan, it was… as well. And they get together to talk about veterans issues. I suspect part of the reason this legislature had such a caucus was they had gotten rid of their veterans subcommittee a few years earlier. So the caucus does a lot of what a committee would do. Committees produce research reports or bill reports after they consider the bill. A legislator might read that report to learn everything that was put forth a committee. That is the same kind of report that I provided to the legislators on behalf of the caucus.
Matt Grossmann: The cue taking across legislators is also similar.
Adam Zelizer: He and I are embarked one basically the same project.
Christian Fong: He and I are embarked on basically the same project. We want to try to identify the legislators who are peers, who are looking to the target legislator for guidance and influence. In my case, I used offices. It just so happened that legislators in this state got to self-select into offices, and who would you choose to share an office with? Somebody you share a political ideology with, someone you talk to a lot, someone you like. So, they were kind of declaring, hey, this is the person that I would look to for guidance. Christian looked to co-sponsorship networks. The idea being, again, that legislators, through their public supported bills are declaring how similar they are to each other. Same idea, and we look for this information spillover from the targeted legislator to their network.
Matt Grossmann: But Congress might have more information, and more staff effects.
Christian Fong: So, if you talk to anyone who works on Capitol Hill in Washington, they’ll say staff have a much bigger role. They have a pretty big role in co-sponsorship. It is often a staffer who will circulate a member’s bills, and just walk around to a bunch of different offices, and ask if anyone wants to co-sponsor it. So, if we think that there is cue taking, it might be between staffers as much or more than between legislators themselves. Members of Congress are much busier. They have a lot of layers of staff and people around them. They’re probably not interacting directly with each other, quite as much as the state legislators. Then there’s a question of whether Congress is just a higher information environment in general.
If we think that cue taking or information effects are particularly large when information is scarce, that should make these information effects a bit bigger at the state level than at Congress. Maybe congressional staffers have already done this research. Maybe the members, because they are full-time legislators who are paid relatively well, and in many cases, have served for a very long time. Maybe they have a lot of this information already. Maybe they can judge bills more quickly. So, we might expect the effects to be smaller for that reason as well. We don’t know. I would love to do an experiment in Washington or for someone to do that. I think it would be great. One huge advantage of the state context is I met with the legislators face to face. So, 75 or so, out of about 100 legislators in the Lower House, agreed to meet with me. That is a huge success rate that I don’t think I’d have in Congress
Matt Grossmann: Fong agrees that the papers show the same basic mechanisms and finding.
Christian Fong: There used to be paragraph in the conclusion of my paper, that it’s pitching how it’s really important to be really careful and thoughtful with research design in the study of Congress, because for a lot of the questions that we really care about, we can’t run field experiments, right? This isn’t psychology. This isn’t a lab science, and so we have to deal with these tricky statistical issues. Luckily, before my paper got published [inaudible 00:29:15] Adam’s paper, just very quietly removed this statement about what was impossible, because Adam had done what I thought was impossible. So, I think it’s incredible work. I view these papers as basically complimentary. Adam Zelizer’s paper is applying information directly, so their idea is that we’re going to dose these legislators with particular policy information that might be relevant for deciding how to vote. Mine being, observational study is taking a much more hands-off approach to what constitutes information expertise.
It is assuming that Members of Congress are going to acquire this expertise and this information in the process of performing their committee responsibilities. I think it’s not a terribly controversial assumption within the community of people who spend time with or think about Congress. But it’s an assumption. We basically come to the same conclusion, which is that, hey, there’s a heck of a lot of cue taking behavior. When people have information, they use that to influence the votes of others. I think that there are some kind of nice things that each study get to that the other doesn’t right. So, my study is a study on the United States Congress, a relatively professionalized legislature, one that has a lot of demands on it, but a professionalized legislature with a full-time staff, with access to a lot of outside sources of information, they still use a lot of cue taking.
Zelizer’s paper is a study of a state legislature. It’s got this nice design where you don’t have to assume that they’re acquiring expertise or information. Your giving it to them, so you know that they’ve got it, right? But he finds essentially the same result in this much less professionalized state legislature, which I think speaks again to what a fundamental issue this is, and what a fundamental strategy this is for legislatures, that we’re going to see this over and over again. I would say that based on our results jointly, I would expect that basically any legislature, where you have individual members making decisions that aren’t totally under the thumb of the party, you’re going to see some cue taking as long as there’s some constraint in how much information they have available to them, as long as they’re being asked to solve questions that are hard, in comparison to the amount of time that they have to study them.
Matt Grossmann: The scope conditions may also be similar. Fong says this is likely more common on less salient, but not necessarily less important bills.
Christian Fong: So, if by importance, you mean salience to the voters, then yes. Right? This is going to be something that happens when the voters don’t speak with a really loud and clear voice right off of the bat. But if by importance, you mean things that affect people’s lives, things that affect whether we’ve got good regulations in this country, whether we’ve got a well-functioning bureaucracy, whether we’re properly reviewing our government programs, and deciding which of these we need, and which of these we don’t. I think a lot of these things are tremendously important. There’s a reason that they demand the time of Congress to actually go out there and write laws, and bring them to the floor, and get them passed and signed by the president. Because these are things that affect all of our lives. We often just don’t notice it until they’ve already been passed.
Matt Grossmann: Zelizer agrees this might not apply to the biggest controversies, but that’s the usual policy making.
Adam Zelizer: I think the most important thing to keep in mind about generalizability, it is that most issues are relatively nonpartisan. That’s true in the States. That’s true in Congress, too. It’s true in the Supreme Court. A lot of Supreme Court decisions are still unanimous, even though we tend to focus on the most partisan and most conflictual issues. Those are the ones that we see in the media. Those are the ones that politicians choose to amplify in a lot of cases. The overwhelming majority of bills are really not that partisan. They are about good government. They are about trying to fix problems with public policy, make public policy, better, solve some problem that somebody has. I think Veterans Affairs is representative of those. Yes, Veterans Affairs is popular, but there are costs to every bill. A lot of times there are monetary costs. If not monetary costs, there are other costs.
So, even if they are popular in theory, and no legislator really tries to propose bills that are unpopular, they still might not get passed, and there might be good reasons for legislators to oppose them. So, I would say these results are generalizable to the broad class of legislation, legislators. Think about the most salient, the most conflictual, the highest information issues, I would think that these effects would be much smaller. I think it’d be much harder to move legislators on issues that they’ve thought a ton about, where they’re getting a ton of phone calls, where they’re getting a lot of party pressure. I would just think those decisions are already locked in, relative to these bills.
Matt Grossmann: They also agree that the mechanisms they identify may be broader information might be part of a lot of other processes of influence.
Adam Zelizer: So certainly interest group influence has been proposed to occur through information. So, interest groups and lobbyists are thought in many ways, broadly, to influence policy outcomes through expertise in information. There’s some work that says that party leaders use information to control their caucuses. Jim Curry has a great book, Legislating in the Dark, that says party leaders strategically keep their rank and file uninformed, so that the information from the party leaders will be even more influential. So, in a way, it could come all back to information. Who you’re getting it from, maybe with a little bit of tint or bias, because that person wants you to do a certain thing. But ultimately, a lot of the influence we see in legislatures from party leaders to interest groups, to constituents, to staffers, comes down to information.
Matt Grossmann: Fong agrees.
Christian Fong: I think committees are just one example of a place where people go to get their information. They’re really an important example. They’re nice to study, because we can observe who joins the committee, and that allows us to see expertise creation in real time. But there’s no question to me that this is happening all over the place. I think this is a really important element of how some interest groups operate. Maybe they’re interest groups like the NRA, for example, where members already have pretty fixed positions, and there’s not a lot of room to add new information there. But when you think of the range of things that Congress works on, right? Industrial loan companies, for example, right?
Not a lot of people show up to Congress with really strong opinions about whether industrial loan companies are an acceptable form of financial organization. I think there’s a ton of room for lobbyists to get in there and provide some information, if members of Congress can trust that those lobbyists are arriving at the same conclusions that the Members of Congress themselves would arrive at, if they put in the time and effort to become real experts. So, I think, yeah, you see it in lobbyists. You see it in interactions with constituents. You see it in interactions with members who are not necessarily on committees. I think this is everywhere.
Matt Grossmann: The effects Zelizer are found seem to be less than those of constituency contacts, which also might be information.
Adam Zelizer: There are a couple really nice papers by Dan Bergen at Michigan State, where he randomizes constituency outreach to state legislators. So, he has one paper in New Hampshire and one paper in Michigan, where he works with interest groups to get grassroots voters to contact their legislators to advocate for a bill. In New Hampshire, it’s a smoke-free workplace bill. In Michigan, it was anti-bullying, and he randomizes some legislators to get this contact from voters, and some don’t get this contact. He estimates that getting lobbied essentially by voters, makes legislators 10 to 15 percentage points more likely to vote for the bill. That’s a big effect. So, if, compared to that effect, the effect of what voters want and reaching out to their legislators, the effect of just telling the legislator what the bill does is about a third as large. Different outcomes, different contacts, different bills, but roughly that’s how we could benchmark it.
Matt Grossmann: Not everyone was able to successfully persuade legislators.
Adam Zelizer: I was the briefer in two of these experiments, and I was quite concerned about that. I thought, would I be particularly effective or particularly ineffective? That could have gone either way. In fact, when I first got the results in 2015, I did a quick and dirty analysis this, and I saw a big zero effect. I thought, “Eh, why would legislators listen to me? I’m just an intern.” Then I realized I had a little coding error, fixed it, all of a sudden five percentage points. So, it’s not clear whether I was particularly effective or ineffective. I was an unpaid intern at the time. There were other PhD students who were interns in that legislature at the time. So, as far as the legislator’s concerned, yes, they knew I was a PhD student, but I was working as an intern for the Veteran’s Caucus.
Not a lot of people pick up on the advocate treatment arm. So, I mentioned the 2016 experiment was a replication with a wrinkle. I wanted to know whether the source of the information mattered. I had gotten to know an advocate, who was an elderly gentleman. He was himself, a veteran. He was disabled. He had had worked in the legislature before. He was something of a gadfly. He had a Public Access TV station, where he interviewed legislators sometimes, and he led his own veterans interest group. So, I talked to him about working with me to give these briefings. So, in the 2016 experiment, we randomized the briefings across bills, but also whether he or I would administer these briefing.
He, or I would administer these briefings. We worked together. I gave him the background information that I was using. We talked about the bills, but we conducted the meetings separately. And it turned out I didn’t see any treatment effects from his meetings. So that’s why those results are in the appendix. There are a couple things we might think about those results. I think we should be very cautious about them ,because I did not want legislators to know that his work was connected to the caucus. Not because it was secret, he was a member of the caucus. He would come to our meetings. He was well known to all of us, but I just wanted the two people giving the briefings to be very distinct. I didn’t observe him actually administering the treatments or the briefings. I don’t know if he met with legislators, stuck to the information, talked about the bills he was supposed to or not. There are all sorts of reasons why that might not have worked out.
Beyond the kind of ideal world interpretation, which is that maybe advocates aren’t as effective. And there are good theoretical reasons to think that an advocate would not be as effective. Policy high demanders serve to benefit from these bills. Legislators might discount someone with an extreme policy preference or someone who’s really a high demander and thus might not learn anything.
Matt Grossmann: Overall seeing policymakers up close made Zelzer more positive about legislator’s motives.
Adam Zelizer: If you spend a lot of time in a legislature, you see that what lawmakers and their staff spend almost all their time doing is learning. They’re asking questions. They’re doing research. They’re talking to each other. They’re trying to learn about policies. They’re trying to learn about bills. They want to make good decisions, but it takes a lot of time and a lot of information to make good decisions. And at the state level, there’s often not a lot of time and not a lot of staff to help. And so giving this subsidy basically, being an unpaid employee and helping legislators quickly and at low cost learn about bills, help them make decisions. And I do think that is a saying when perspective on law making.
Lawmakers want to make good decisions. They want to incorporate information. They want to become experts, but they’re human. And they have tons of bills on which they’re asked to take a vote. So of course they’re operating with imperfect information. The fact that this information helps them take positions, I think shows that they’re trying, they want to make good policy. They want to make good choices. I think that’s a good thing.
Matt Grossmann: Fong also says, “A close up view may be more positive.”
Christian Fong: There’s a tendency to exaggerate the role that partisan conflict plays in legislative process. Is party really important? Yes, absolutely they structure all kinds of facets of Washington life. But I think that people have to remember that a lot of the challenges that we face in government are technical challenges. We tend to spend the most time thinking and talking about the things where we have fundamental disagreements on the issues. But a lot of it comes down to hammering out the most efficient way to prevent people from catching dolphins in fishing nets. Right? And I think that people might have a slightly different view of the political process if they spent a little less time focused on the things where we’re fighting a lot, because we disagree and a little more time paying attention to the parts of the policy process where actually have broad agreement on what the goals are. And we’re just trying to work together to figure out what is the most technically efficient solution to this problem.
Matt Grossmann: Congress research is generally right on he found, even when you see it up close.
Christian Fong: It’s a little scary getting involved in Congress research having never spent any time there yourself. You always worry that, I’ve spent this these four years in graduate school, reading up everything I can on Congress and what if it’s all wrong? What if I’m just this total ignorant? And I show up and I realize that everything that we ever knew was total malarkey. I was incredibly happy to see that was not the case. In fact, I think that Congress scholars basically nailed what’s going on there. Of all the major theories that we’ve got going on in Congress and all the major strands of empirical work, I don’t think I had to identified a single one that seemed to me to be fundamentally wrong.
There’s always progress to be made. There’s always new questions that we can answer, new phenomena that we can unpack. But I really don’t think that we’ve made a lot in the way of false turns. And I think part of that can be attributed to the people who were working back in the sixties and seventies. We’re spending a lot of time in the institution, trying really hard to understand how this place works at a ground level. And I think in a significant way, the more quantitative generation of which I’m a part, owes a big debt to these people for setting the agenda in the right way. Giving us this norm of lots of our leading scholars, actually spending time in the institution to double check how things work and to make sure we’re on the right track.
Just to give an example for queue taking, so I was happy to see that there were… I worked in Senator Mike Lee’s office. We happened to have these great healthcare staffers, so there were other offices that would take queues from us. And then I worked on finance issues, things like banking and trade and budget. And there was an office that I took queues from. I thought that Pat Toomey had really excellent people. And I always read the bills myself and came to my own decision to advise the Senator. But I would also call Senator Toomey’s staffers and see what they thought about things because I valued their opinion. And I thought that they might have noticed things that I didn’t notice. And sometimes they did and I was very happy that I called them.
Matt Grossmann: He says, “Not as much has changed in Congress, as we sometimes think.”
Christian Fong: It’s actually remarkable how on the money these older generation of scholars were, these people who were really embedded it in Congress were. I don’t think that there’s a lot that I find here that they wouldn’t have suspected was true back in the 1960s, where their data was coming from. I think the really remarkable thing is that Congress has changed a lot in the 50 years since they were writing. And I think if you had taken a poll of experts on Congress, they would say that the institutions changed tremendously. We shouldn’t necessarily expect things that were true 50 years ago to be true today. In fact, lots of people like to divide the textbook of Congress. What was on in the postwar period up through the 1960s, from the more contemporary Congress that we’ve seen since 1994. It’s been a bit more polarized on a partisan basis.
I think the fact that there’s this continuity, that these things that were true 50 years ago are still true today is really good news for science. That we can actually find these enduring patterns, that we can identify these basic challenges that legislators face and these fundamental tools that they used to resolve those, that those remain stable across time. That as a scientist is real exciting to me.
Matt Grossmann: What’s next? Zelzer is trying to show that information on costs matters to legislators as well.
Adam Zelizer: One project examines how fiscal information, the cost or benefits of bills impacts legislators support. Are legislators fiscally responsible? Are they responsive to fiscal information? Or do they just vote based on party line bill? Or based on ideology? And pass the costs on to somebody else? So that one project looks at their responsiveness to fiscal information. Spoiler, I do find that they seem to care. They do not want necessarily to vote for more expensive bills. Legislators are more supportive of less costly bills, which is what I think we would want. Another project looks at how legislator support depends on their peer support.
Matt Grossmann: And Fong says, “They’re good times ahead for congressional studies because of lots of better data.”
Christian Fong: I’m pretty optimistic about the creativity of social scientists, when it comes to research design. This stuff is hard. I got lucky to notice something that we haven’t noticed yet, but I tend to think that people will continue to get lucky. There’re new graduate students coming up all the time who know about research designs and methodologies that we haven’t thought about yet. I think even more critically, there’s a really great crop of assistant professors and graduate students who are uncovering tons of new sources of data for Congress. Things from where they’re traveling, to what they’re doing when they’re at home in their district visits. To unpacking the archives, to look at parts of the process that we weren’t able to look at before. That I think are going to open up tons of new research designs, right?
The conversation we’ve had, I think about my paper really underscores the relative poverty of data that people like me are using, people who aren’t going out and collecting new stuff. We’re looking at co-sponsor, we’re looking at committees, looking at roll call boats because that’s what we have. We’re going to have so much stuff 10, 20 years from now that we don’t have right now. I’m excited to see what sorts of research designs we’re going to come up with using all this new data.
Matt Grossmann: There’s a lot more to learn. The Science of Politics is available biweekly from the Niskanen Center and part of The Democracy Group Network. I’m your host, Matt Grossman. If you like this discussion, you should check out our previous related episodes. I think you’ll like, Compromise Still Works in Congress and With Voters.” “How Media Coverage of Congress Limits Policy Making.” “Does Anyone Speak For the Poor in Congress?” Are Divided Governments the Cause of Shutdowns and Delays?” And “Do Congressional Committees Still Make Policy?” Thanks to Adam Zelzer and Christian Fong for joining me. Please check out, “Is Position Taking Contagious?” And “Expertise Networks and Interpersonal Influence in Congress.” And then listen in next time.
Photo credit: iStock