Research Profile

Justin de Benedictis-Kessner

Justin de Benedictis-Kessner is a postdoctoral researcher at the Boston Area Research Initiative, where he focuses on political behavior, public policy, urban politics, and experimental and quantitative methodology. Justin earned his PhD in Political Science from MIT, and will be joining the Political Science faculty at Boston University starting in the 2019-2020 academic year.


Boston Area Research Initiative: Can you tell me about what you’re working on right now?

Justin de Benedictis-Kessner: Sure. So, one of the projects that I’m working on, that’s partially published and partially not published, is some work with the MBTA that started when I was in grad school and basically uses their data to show that people have trouble attributing responsibility for performance. That part has been published, but there’s another part that’s still ongoing, on how recency bias can inhibit people’s ability to hold government accountable. So that part is very much a work in progress, and there’s a lot of different behavioral measures that I’m interested in bringing in.

The way that I would explain that project is essentially, we often wonder as political scientists, are voters able to hold politicians accountable for performance? We have lots of examples of voters not being able to do that or being able to do that at the national level: national politicians, having to do with national-level policy, such as how the economy is doing. We don’t have as good of an idea about how accountability functions at the local level. One of the goals of those projects starting off was, how can we figure out the conditionality of people’s ability to hold government accountable?  What does it depend upon in terms of their psychology, in terms of the institutions that surround government?

That project basically shows that, yes, people can hold the government accountable for performance at the local level, but only under certain conditions. In this project I showed that people’s ability to assign responsibility for performance of a specific public service, such as transportation, is really key for them being able to judge the responsible politicians. This arises in a lot of different political arenas where there’s a governmental body acting as a service provider, such as the MBTA. It’s not always clear whether or not the politicians who people actually elect are responsible, and which ones are responsible. The punch line coming out of this project was that setting up this kind of institution, where responsibility is not clear or direct for voters, can be really potentially harmful for accountability for performance. So, when stuff is bad, it’s not going to improve if voters aren’t able to express their different preferences for things to the politicians who can effect change.

The ongoing portion of that is more on the cognitive recency bias side. Cognitive recency bias just means people favoring recent information when they’re making judgments, choosing how to behave, and making decisions for the future. For example, a lot of research on this shows that if you have a really painful colonoscopy, you won’t go back for another one, but if the doctor makes the final part of a colonoscopy less painful, then you are more likely to go back. This is a great health outcome because you want people to be getting colonoscopies regularly. So, for this project, I’m still working on some of the behavioral outcomes that show that people’s opinions aren’t going to change as a result of performance overall, but rather as a result of recent performance. When transit is bad recently they’re going to change their opinions of government, but not when trends have just been bad for a month or so. That’s showing this extension of recency bias to performance accountability for transportation.

BARI: Interesting. It seems like the fact that elections occur in November, not right after the winter, must be a big incumbency advantage for municipal elected officials.

JDBK: One of the things that I am potentially going to try to look into is, is there any evidence on a case by case basis of this? Is there any evidence of elected officials who are up for re-election trying to improve transit, in the same way that there is commonly evidence of city governments increasing police protection on the streets right before elections to bolster their support in those elections? It could be really interesting if, say, cities and towns were doing the same with transit improvements. Of course, transit improvements are a little stickier. You can’t just immediately put resources into making buses and trains faster, but you can do things like the bus priority lane project that Watertown and Cambridge rolled out this past October. Obviously, this is not a systematic thing, but I would say that as a city or town, October is a great time to roll out this kind of transit improvement if you care about elections.

BARI: Do you have any evidence either for or against the idea that, despite the fact that things like transit affect people a great deal, and that people care about them, people are not making electoral decisions based on holding officials accountable for transit because they have trouble assigning responsibility, but just because that’s not the issue they have decided to vote on?

JDBK: This is a question that a lot of people have asked. What are people making decisions about in local politics? What are they making any kind of judgments on? That’s a very important question that I’m trying to answer in some other research. The more comprehensive research agenda for me encompasses a lot about how people make decisions in local politics, how they vote, and how institutional and psychological barriers to their decision-making process can hinder overall representation or accountability. That’s kind of a future question. I think there’s a lot that goes into it besides transportation.

BARI: Is it helpful that you’re looking at local stuff more, just because while there’s no election that doesn’t have some partisan edge to it, an election in which there is no clearly defined partisan delineation between candidates could produce cleaner data?

JDBK: I think that partisan elections are a whole other story. There’s a lot of evidence from other people’s research that partisan cues can overwhelm some of the considerations that people might use to make decisions in nonpartisan local races.

BARI: So where is this work going?

JDBK: Well, there’s a grant that [BARI Co-Director] Dan O’Brien, [BARI Associate Research Director on Social Media] Ryan Wang, and I got which enables us to extend some of this research to different outcomes. In the project we talked about earlier, I use primarily survey data combined with a bunch of transportation performance data and CharlieCard data on when people are actually using transportation. In this new project, we’re going to use data from the Transit app, that people use to find when the next train or bus is coming, and plan routes. It also has some ability to show people options other than just public transportation for getting to where they want to go. Ride hailing services like Uber and Lyft, and also things like bike share, might be non-traditional to include in a Transit app, but give people an idea of the comprehensive picture. In a city like Boston, linking multiple modes of transit together is often the fastest way to get somewhere.

It is important to understand not just how this technology effects people’s opinions, but also how it effects their behavior. For instance, if people experience really bad transit, are they opting out of using public transportation when performance is worse? What are the systemic effects of that? Is traffic going to get worse and therefore bus performance going to be worse because people are not taking the bus and going and getting in private cars? Those are some of the some of the potential questions that this project could answer. These are questions I’ve been wanting to answer for years, but that use data that I didn’t have access to until now.

BARI: I will say that as an individual user, I never saw a reason to use the Transit app because I know more or less where the subway is. I find that for a mode that is driven more by headways and not so much by a schedule, it’s a lot less useful to me. If I took the bus more, I would use it a lot.

JDBK: This is a huge issue, actually. Brushing off bus users as potentially not as interesting is something that a lot of people do, because a lot of people ride rapid transit. But bus is actually the area where performance probably could experience the most improvement or the worst performance. There is a lot more variability in performance relative to rapid transit or subway, and traffic plays a huge role. That’s the place where it’s the most interesting to see what the ramifications of poor performance are, because from a scientific perspective it’s a lot easier to see the effects of performance on other outcomes when there is variability like that. There are also distributional consequences—the people riding buses are not the same people as the ones riding the subway.

BARI: I know you haven’t dug into the data yet, this is all very new, but are there specific things you’re excited about as you get into the data? Do you have guesses or hypotheses you want to test?

JDBK: Well, performance and opting out of public transit are the first things that I’m interested in. But there’s also a set of more basic things that I’m excited to look at, such as whether people are actually opting into receiving better or worse performance via these real time apps. One thing we might think happens is when people see that a bus is two minutes away, they may run to the bus stop to catch that bus. That’s a rational thing to do, and real time apps make that possible with a host of different transportation options. And what that might lead to is a sort of bunching of people in their arrivals at stations or at bus stops. Interestingly, this is a thing that public transportation modeling doesn’t really take into account. It’s kind of just an empirical descriptive fact that we should figure out: is there bunching right before trains and buses arrive? Does that lead to non-constant arrival rates? That would be interesting. That is one of the other basic hypotheses that we can test.

BARI: Thanks for sitting down with us! We’re excited to see where this work goes. Is there a way that people can get in touch with you, or follow along with this work?

JDBK: Yes! They can check out my website,, where I’ve posted some published and unpublished versions of my research, or follow me on Twitter, where I do a lot more musing about research, politics, life in Boston, and the combination of all of those.