(Photo credit: Randall Munroe at XKCD)
The recent rise of ‘data-journalism’ or ‘explanatory journalism’ is characterized by its enthusiasm for an empiricist, evidence-driven approach to understanding politics and current events in contrast with traditional political reporting and its focus on narratives and rhetoric. Explanatory journalism has charts and so on. But the problem with placing so much faith in the ability to explain politics through better use of data is illustrated by the recent arguments between Ezra Klein and Thomas Frank about the role of political science in politics and journalism in context of piece of actual political science by Jacob Hacker and Paul Pierson. In their recent article in Perspectives on Politics, Hacker and Pierson critique the focus on elections and the parties’ fight for the median voter – what they refer to as the Downsian framework – that dominates political science. They contend this represents a narrow view of politics, point us toward Schattschneider’s approach that focuses on policies over elections and the role organized interest groups in not only fighting over ‘the prize of policy’ but, even more importantly, setting the political agend. More generally, Hacker and Pierson’s argument for a ‘policy-centered political science’ reminds us particular frameworks can provide only partial explanations of politics; additional insights may require both different ways of thinking about politics, and therefore new questions and evidence.
‘All the effing geniuses’: data journalism and ‘the power of political science’
The recent spat between Ezra Klein and Thomas Frank provides insight into how contemporary data journalists understand and present political science, and their attitude about ‘power of data’ more generally. Klein recently celebrated what he argued was the newly prominent role of political science in the way journalists understand and cover politics. Thomas Frank did not like this, and responded with a critique of the ability of the kind ‘data shuffling’ approach to political science championed by Klein to explain what is important in politics while also arguing that the deference accorded to such analysis is what leads to such poor governing in Washington. Frank comes out swinging here, but unfortunately while he raises some great issues, he does not offer a truly sustained argument for them and they get muddled in his piece. But his article nonetheless raises an important counter to the recent unquestioning enthusiasm for all things data-driven. Further, the vociferous response of the data journalists to Frank’s critique actually helped underscore Frank’s criticism.
In his piece, Klein contrasts political science as a way of knowing about politics different the historically insider-focused journalistic approach. For Klein, the appeal of political science stems from two important characteristics: its focus on empirical evidence (no surprise there) and its ability to offer structural explanations of politics. The traditional focus on insider knowledge is unreliable because insiders have their own agendas. But political science, with its emphasis on structure, can tell us much more as well as debunk conventional wisdom, says Klein:
They can’t tell you what an individual senator thinks, or what message the president’s campaign will try out next. But they can tell you, in general, how polarized the Senate is by party, and whether independent voters are just partisans in disguise, and how predictable elections generally are. They can tell you when American politics is breaking its old patterns (like with the stunning rise of the filibuster) or when people are counting on patterns that never existed in the first place (like Washington’s continued faith in the power of presidential speeches).
The currency of Klein’s understanding of political science is captured in Jonathan Chait’s writing. Chait is not a data journalist himself but regularly draws on empirical analyses in his own (often great) work, where he points to ‘the power of political science’ to provide authoritative, structural explanations of contemporary political dynamics.
Frank takes issue with the data journalists understand and use political science and his concerns are both substantive and stylistic. The stylistic critique is made only implicitly, save for his title (‘all these effing geniuses…’) but I think the two go hand in hand. Frank’s concern lies in the way in which expertise is accorded in politics and political punditry. Further, Frank argues that the way in which political actors and analysts are enthralled with (a particular form of) expertise serves to narrow the debate:
The powerful in Powertown love to take refuge in bewildering professional jargon. They routinely ignore or suppress challenging ideas, just as academics often ignore ideas that come from outside their professional in-group
Not only are those with power assumed to also have expertise, (“[e]ver wonder why the economic experts never seem to change, keep coming back, despite racking up such shattering failures as the housing bubble and the financial crisis and the bank bailouts?”), but the debate among those in power is self-servingly narrow. In this regard, Frank notes Larry Summers’ advice to Elizabeth Warren that ‘insiders don’t criticize insiders.’
As a result, Frank argues data-journalists and other ‘experts’ in DC portray politics as a foregone conclusion, and this presentation is self-reinforcing. Frank is specifically arguing against political journalism (eg “Why Democrats Can’t Win the House” by Nate Cohn) that portrays Republicans and Democrats fight over the median voter in within an enduring political structure that shifts slowly as demographics change. These people vote for Democrats, and these people don’t. Democrats live here, Republicans live there.
So Frank’s substantive critique lies with the presentation of politics as a static, or at least only changing when the structural factors that influence politics (in this case demographics) change. But his stylistic critique if how data journalists present these explanations as the final answer. Frank’s general sentiment raises an important “hey, wait a minute” corrective to ‘data journalisms’ self-satisfied empiricism. This is what the data tells us. It is so.
The data journalists reinforce Frank’s stylistic critique for him by painting the notion that politics might be more malleable than the data suggest as hackery. Klein attacks Frank for privileging his preferred narrative over science and evidence. Other piled on. Frank – who actually has a PhD – is literally compared with Abe Simpson yelling at a cloud.
‘Policy-centered political science and the terms of the debate
Klein and his posse are correct that Frank’s critiques also come with an agenda. Frank not only wants to point out that the base of support for a progressive agenda is not fixed, but could expand as policy makers fight for progressive principles because this is how he wants us to understand politics generally, but also because he supports a progressive agenda in particular. But he is doing this openly, not behind the facade of science. This is where Frank’s substantive and stylistic critiques intersect. Data-journalists emphasize structure and not the role of political agency in influencing this structure. The data-driven, structural view of politics not only obscures but abets the now decades-old Republican agenda of attempting to remake-politics.
And this is where Hacker and Pierson’s review article interacts with Frank’s argument. Hacker and Pierson argue for a way of understanding politics that differs from an approach that focuses on elections and the structural factors that influence electoral outcomes. Hacker and Pierson are critical of this Downsian approach – after Anthony Downs and his median voter theory – which also serves to inform the arguments of the data journalists with whom Frank is arguing. Instead, Hacker and Pierson lay out an approach to studying politics that focuses not solely on how structure influences politics but how agents – organized interest groups – can influence the political structure in which they operate.
For Hacker and Pierson (and many others), a primary way in which political actors seek to influence political structure is through public policies. Policies serve as the terrain political struggle. Policies are the ‘prize’ political debate and struggle. Policies confer resources. They channel government authority and finances towards certain issues and constituencies over others, and are the locus around which institutions develop. And to work within and influence the landscape that policies create, individuals must act beyond their capacity as voters. They must organize. Hacker and Pierson contrast the Downsian approach with elections with what they characterize as a Schattscheideian policy-centered approach. Schattschneider not only emphasized how ‘new policies create new politics’ but focused our attention on the role that organized groups play in these processes. While the Downsian emphasis on the median voter sees politicians as pursuing policies in order to win elections, a policy-centered approach sees actors as seeking to win elections (“and do many other things”) to have the ability craft policy.
In this regard, a policy-centered approach considers how politics is a long game in which organized interests vie for power across multiple institutional environments. And while this certainly includes elections, it is not limited to them. For example, this approach considers what makes some policies have enduring influence on the polity while others can be undermined and how a lack of policy change (coupled with a shifting social environment) can also be instructive for understanding political debates. How the policy landscape is configured has an important effect on organized political advocacy, particularly who participates and the nature of this participation. Groups may benefit from the existing policy status quo and seek to defend it. And amid debates between those challenging and defending the status quo, debates and issues that do not arise are just as important as those that do, as groups that benefit from the status quo seek to keep certain issues off the agenda, and in turn stifle the voices of those raising these issues. Indeed, Schattschneider considered the ability to influence the set of choices faced by policy makers, interest groups, and of course voters as “the supreme instrument of power.” Likewise, those challenging existing arrangements may seek to influence the policy apparatus by working across institutional venues and enlisting the help of other groups in common cause, or as Schattscheider describes, challenging groups are looking to ‘expand the scope of conflict.’
We cannot use data as evidence of something that we are not looking for.
Hacker and Pierson’s call for a policy-centered approach asks that we look at politics in a different way than the Downsian focus on how parties vie for electoral advantage by courting the median voter. The data journalists that Frank is arguing with largely accept the Downsian approach and portray their work with a data-driven certainty. Much data journalism packages its articles as ‘everything you need to know’ on a given topic. They mock Frank for arguing that the Democrats might challenge demographics by pushing progressive politics and expanding the scope of conflict. Jon Chait captures these dynamics when he ends his piece attacking Frank. He closes thusly:
At the end of his rant, Frank almost seems to concede that his problem with political science is that it leads to conclusions he finds inconvenient. “The fatalism here may be science-driven,” he concedes, “but still it boggles the mind.” Let that phrase roll around in your head for a moment. Frank has just told you everything you need to know here.
The idea is that because Frank is saying that there are political possibilities outside of what the data suggests, he is inherently wrong because the numbers don’t support his argument. But we can only marshal evidence for phenomena that we are looking for. And Hacker and Pierson point out that the Downsian approach misses aspects of reality that are crucial for understanding political dynamics. While data journalists look at how demographics inform the electoral prospects of the political parties, in their work Hacker and Pierson consider more broadly how the top one percent have been able to influence the political process and reinforce both economic and political inequality. This is a much different story than one that we can tell with data about demographics and voting behavior. As Hacker and Pierson note, a policy-centered approach to understanding politics requires greater methodological eclecticism than what Frank calls the ‘data-shuffling’ of explanatory journalism, because “in the actual conduct of research, there is a real tension between gathering more and more data and actually defining and examining the most consequential features of the political environment.”
 A couple things are worth noting here. First, I am a fan of Frank’s but also most of his critics. Ezra Klein get me into reading about politics on the internet and Jon Chait is a daily read. So the point here is not to pick favorites but to explore the issues the exchange raises. Also, though I have read a lot of Frank’s work, I actually have not read What’s the Matter with Kansas? so I can’t comment on Frank’s debate with Larry Bartels about the book in the context of political science research. But I also think that is a bit outside the scope of this post since I am more interested in how data journalism talks about political science.