The Political Economy of Data Production

Post developed by Catherine Allen-West, Charles Crabtree and Andrew Kerner

ICYMI (In Case You Missed It), the following work was presented at the 2017 Annual Meeting of the American Political Science Association (APSA).  The presentation, titled “The IMF and the Political Economy of GDP Data Production” was a part of the session “Economic Growth and Contraction: Causes, Consequences, and Measurement” on Sunday, September 3, 2017.

Political economists often theorize about relationships between politics and macroeconomics in the developing world; specifically, which political or social structures promote economic growth, or wealth, or economic openness, and conversely, how those economic outcomes affect politics. Answering these questions often requires some reference to macroeconomic statistics. However, recent work has questioned these data’s accuracy and objectivity. An under-explored aspect of these data’s limitations is their instability over time. Macroeconomic data is frequently revised ex post, or after the fact, and as such one could ask the same question of (ostensibly) the same data, and get different answers depending on when the question was asked.

We set out to explore the political economy of data production by examining a newly available dataset of ex post revisions to World Development Indicators (WDI) data.[1]  Ex post revisions occur when newly available information changes national statistical offices’ beliefs about the nature of the economy. Those revisions extend into the past, effectively rewriting history and, in the process, providing a reasonable proxy for the inaccuracy of the initial reports. These revisions affect a wide swath of data, but we focus on Gross Domestic Product (GDP) and GDP-derived measures, like GDP per capita and GDP growth. GDP revisions are common—most GDP data available for download at the WDI are different now than they were at the time of its initial release. Normally these changes are subtle; other times they are substantial enough to condemn prior data releases as misleading.

We use these revisions to answer two related questions. First, how sensitive are political-economy relationships to GDP revisions? Should researchers worry about revisions-driven instability in the state of political-economic knowledge? We show that they should. To illustrate, we subject a simple, bivariate statistical relationships between democracy and growth to re-estimation using alternative versions of the “same” data. The democracy-growth relationship has been a topic of sufficient interest in economics and political science that instability in this relationship should give us reason for pause. Seen in this light our estimates are worrisome. As we show in Figure 1 below, our estimates are unstable across different “observation years” and further, they are unstable in ways that suggest that initial estimates were biased. Rather than simply a diminution of standard errors as more heavily revised data are introduced (which is what we would expect to see if revisions simply reduced random “noise in the data”), the estimated coefficients for Democracy change substantially across models estimated with different revisions of the same country-year GDP growth data.

Figure 1: GDP Growth ~ Democracy

Note: Figure 1 displays the relationship between GDP Growth and Democracy using the results from 21 different regression models. Plotted points represent parameter estimates, thick bars represent 90 percent confidence intervals, and thin bars represent 95 percent confidence intervals. Each point is labeled with the revision year used. The left side of the plot contains results from models estimated using the 2000-2004 data series, while the right side of the plot contains results from models estimated using the 1995-1999 data series. See paper for more details.

This finding anticipates our second question: Given the likelihood that GDP revision are non-random, what accounts for ex post revisions? What does the “political economy” of revisions look like? We show using Kolmogorov-Smirnov tests (see Figure 2) and random forest models (see Figure 3) that the International Monetary Fund (IMF) influences the magnitude of revisions for GDP and GDP-related measures. That is not entirely surprising. Our suspicion that the IMF would have such an effect is a straightforward recognition of its well-publicized efforts to provide financial and human resources to the national statistical offices of the countries in which it works. What we have “uncovered” in this exercise is simply one consequence of the IMF doing precisely what it has publically said it is doing. But this finding’s (retrospective) obviousness does not diminish its importance. Consider the empirical challenges that this presents. Political economists often ask if the IMF affects the way economies functions, but the IMF’s independent effect on the way economies are measured substantially complicates our ability to know if it does. And it doesn’t just complicate our ability to know if the IMF’s policies affect the economy, it complicates our ability to know if anything correlated with IMF participation affects the economy. Many important things correlate with IMF participation, including, for example, democracy, a country’s relationship with the UN, and whether or not a country is an ally of the United States.

Figure 2: Distributions of GDP Growth Changes

Note: Figure 2 presents compares the distributions of GDP growth revisions for years with and without IMF programs. The y-axis indicates the height of the density function and the x-axis indicates the magnitude of GDP growth revisions in percentages points. The solid green line denotes country years with an IMF program, while the dashed black line denotes countries years without a program. See paper for more details.

Figure 3: Predictors of GDP Growth Revisions

Note: Figure 3 presents the results from a random forest model that examines the predictors of GDP growth revisions. The vertical axis ranks variables according to their importance for predicting GDP Growth Changes. The horizontal axis displays estimates of permutation accuracy for each variable, calculated as the difference in mean squared error between a model that is fitted using the observed values for a measure and a model that is fitted using random (but realistic) values for the same measure. This measure is then scaled to represent the percentage increase in mean square error caused by permuting the values of the variable. Positive values indicate that the variables increase the predictive performance of the model, while negative values indicate that the variables decrease the predictive performance of the variables. See paper for more details.

Of course, politics likely affects the way the economy is measured in a variety of ways that have nothing to do with the IMF. Our random forest analysis suggests that democracy might also have an effect, for example, as might public sector corruption, and it is not hard to tell a plausible post hoc story for why that might be. But our aim is not to provide a comprehensive picture of the political economy of data production, but simply to show that it exists, and that it exists in a manner that should alert us to its importance. Taking seriously the political provenance of ostensibly apolitical data is an important (and, we believe, interesting) step towards refining the state of political economy knowledge.


[1] The raw data used in this paper are available at http://databank.worldbank.org/data/reports.aspx?source=WDI-Archives. To facilitate researcher use of this data, we will make it available in an R package, revisions. This package will contain long- and wide-format data sets.

Andrew Kerner is an Assistant Professor in the Political Science department at the University of Michigan, and a faculty associate at the Center For Political Studies.

Charles Crabtree is a PhD student in the Department of Political Science at the University of Michigan.

The Spread of Mass Surveillance, 1995 to Present

ICYMI (In Case You Missed It), the following work was presented at the 2017 Annual Meeting of the American Political Science Association (APSA).  The presentation, titled “Big Data Innovation Transfer and Governance in Emerging High Technology States”  was a part of the session “The Role of Business in Information Technology and Politics” on Friday September 1, 2017. 

Post developed by Nadiya Kostyuk and Muzammil M. Hussain

On August 24, 2017, India’s highest court ruled that citizens have a fundamental right to privacy. Such a ruling may serve to slowdown the government’s deployment of the Aadhaar national ID program, a robust relational database connecting each of India’s 1.3+ billion citizens with their unique 12-digit identity aimed at centralize their physiological, demographic, and digital data shadows — minute pieces of data created when an individual sends an email, updates a social media profile, swipes a credit card, uses an ATM, etc. While the government has presented the Aadhaar system as an improved channel to provide social security benefits for its nationals, India’s civil society organizations have protested it as a means of furthering government surveillance. India’s trajectory in ambitiously modernizing its high-tech toolkit for governance represents a rapidly spreading trend in the contemporary world system of 190+ nations.

Take China as an other example.  China has recently mobilized its government bureaucracies to establish the worlds’ first ever, and largest, national Social Credit System covering nearly 1.4+ billion Chinese citizens. By 2020, China’s citizen management system will include each Chinese national’s financial history, online comments about government, and even traffic violations to rank their ‘trustworthiness.’ Like India’s, these unique ‘social credit’ ratings will reward and punish citizens for their behavioral allegiance with the regime’s goals by scientifically allowing the state to operationalize its vision of a “harmonious socialist society.”

Yet, the implementation of state-sponsored and ‘big data’-enabled surveillance systems to address the operational demands of governance is not limited just to the world’s largest democratic and authoritarian states. This summer, at the annual meetings of the International Communication Association (May 2017, San Diego) and the American Political Science Association (August 2017, San Francisco), the project on Big Data Innovation & Governance (BigDIG) presented findings from the first event-catalogued case-history analysis of 306 cases of mass surveillance systems that currently exist across 139 nation-states in the world system (Kostyuk, Chen, Das, Liang and Hussain, 2017). After identifying the ‘known universe’ of these population-wide data infrastructures that now shape the evolving relationships between citizens and state powers, our investigation paid particular attention to how state-sponsored mass surveillance systems have spread through the world-system, since 1995.

By closely investigating all known cases of state-backed cross-sector surveillance collaborations, our findings demonstrate that the deployment of mass surveillance systems by states has been globally increasing throughout the last twenty years (Figure 1). More importantly, from 2006-2010 to present, states have uniformly doubled their surveillance investments compared with the previous decade.

In addition to unpacking the funding and diffusion of mass surveillance systems, we are also addressing the following questions: Which stakeholders have most prominently expressed support for, benefited from, or opposed these systems, and why? What have been the comparative societal responses to the normalization of these systems for the purposes of population management in recent decades?

The observed cases in our study differ in scope and impact.

Why do stable democracies and autocracies operate similarly, while developing and emerging democracies operate differently? Access to and organization of material, financial, and technocratic resources may provide some context.

While nations worldwide have spent at least $27.1 billion USD (or $7 per individual) to surveil 4.138 billion individuals (i.e., 73 percent of the world population), stable autocracies are the highest per-capita spenders on mass surveillance. In total, authoritarian regimes have spent $10.967 billion USD to surveil 81 percent of their populations (0.1 billion individuals), even though this sub-set of states tends to have the lowest levels of high-technology capabilities. Stable autocracies have also invested 11-fold more than any other regime-type, by spending $110 USD per individual surveilled, followed second-highest by advanced democracies who have invested $8.909 billion USD in total ($11 USD per individual) covering 0.812 billion individuals (74 percent of their population). In contrast to high-spending dictatorships and democracies, developing and emerging democracies have invested $4.784 billion USD (or $1-2 per individual) for tracking 2.875 billion people (72 percent of their population).

It is possible that in a hyper-globalizing environment increasingly characterized by non-state economic (e.g., multi-national corporations) and political (e.g., transnational terror organizations) activity, nation-states have both learned from and mimicked each other’s investments in mass surveillance as an increasingly central activity in exercising power over their polities and jurisdictions. It is also likely that the technological revolution in digitally-enabled big data and cloud computing capabilities as well as the ubiquitous digital wiring of global populations (through mobile telephony and digital communication) have technically enabled states to access and organize population-wide data on their citizens in ways not possible in previous eras. Regardless of the impetuses for increases in mass surveillance efforts, our research aims to provide empirical support to advance theory and guide policy on balancing security needs and privacy concerns at a time where many governments are ambitiously upgrading their governance systems with unbridled hi-tech capabilities.

 

Inequality is Always in the Room: Language and Power in Deliberative Democracy

ICYMI (In Case You Missed It), the following work was presented at the 2017 Annual Meeting of the American Political Science Association (APSA).  The presentation, titled “Inequality is Always in the Room: Language and Power in Deliberative Democracy” was a part of the session “Is Deliberation War by Other Means?” on Thursday, August 31, 2017. 

Posted by Catherine Allen-West


In a new paper, presented at the 2017 APSA meeting, Arthur Lupia, University of Michigan, and Anne Norton, University of Pennsylvania, explore the effectiveness of deliberative democracy by examining the  foundational communicative acts that take place during deliberation.

Read the full paper here: http://www.mitpressjournals.org/doi/abs/10.1162/DAED_a_00447

Bureaucracy and Economic Markets: A Coevolutionary View of Development

by Catherine Allen-West

Political economists have long debated the causal relationship between good institutions—such as technocratic, Weberian bureaucracies—and economic development. Whereas some insist that good institutions must precede economic development, others assert that it is economic success that eventually leads to good institutions. So who’s right and who’s wrong?

Yuen Yuen Ang Faculty Associate at the University of Michigan’s Center for Political Studies answers this question using a novel, dynamic approach in a chapter entitled Do Weberian Bureacracies Lead to Markets or Vice Versa? A Coevolutionary Approach to Development. The chapter recently appeared in an edited volume, States in the Development World, that is the final product of a series of international workshops on state capacity hosted by Princeton University. The chapter is also included in the Working Paper Series of Stanford University’s Center for Democracy, Development, and the Rule of Law (CDDRL).

Development as a Three-Step, Coevolutionary Sequence

Ang injects a novel perspective into this long-standing debate. She argues that development unfolds in a three-step coevolutionary sequence:

Harness weak institutions to build markets ➡️ emerging markets stimulate strong
institutions ➡️ strong institution preserve markets.

In this chapter, Ang demonstrates this argument using a historical comparison of three local states in China. She traces the mutual adaptation of state bureaucracies and industrial markets from 1978 (beginning of market reform) through 1993 (acceleration of market reform) to the present.

The political economies of all these locales have undergone radical transformation over the past four decades. According to the conventional view of development this process must have occurred by first establishing a fleet of new bureaucracies rather than relying on any existing systems. The process is depicted here:

Ang, however, reveals a surprisingly different—non-linear—story. She finds that Chinese locales built markets by first mobilizing the preexisting bureaucracy, left over from the Maoist era, to attract capitalist investments through distinctly non-Weberian modes of operation: non-specialized and non-impartial. Normally, features that violate Weberian best practices are dismissed as “weak” or “corrupt.” Yet weak institutions, Ang argues, are precisely the raw materials for building markets from the ground up.

From there, Ang says, the process of development continued. Once local markets took off, the emergence of markets altered the preferences and resources of local state actors. These changes led local governments to replace earlier unorthodox practices with more Weberian practices that serve to preserve established markets. Ang’s three-step coevolutionary sequence is summarized in the figure below.

Market-Building ≠ Market-Preserving

Ang’s theory of development draws a sharp distinction between the institutions that bolster emerging as opposed to mature markets, which she calls “market-building” and “market-preserving” institutions respectively. Thus far, existing theories have focused exclusively on “market-preserving” institutions. Ang calls attention to the neglected variety of “market-building” institutions.

This chapter offers a preview of Ang’s broader work, contained in her book, How China Escaped the Poverty Trap. Her book further unpacks the three-step coevolutionary sequence of development, with evidence not only from China but also other national cases (Europe, the U.S., Nigeria). It also details the distinct fieldwork strategies and analytic methods that she developed to map coevolutionary sequences.

Reviews of How China Escaped the Poverty Trap recently appeared in the Straits Times, Foreign Affairs, and World Bank Development Blog. Yongmei Zhou, the Co-Director of the World Development Report 2017, writes in her review:

“The first takeaway of the book, that a poor country can harness the institutions they have and get development going is a liberating message. Nations don’t have to be stuck in the “poor economies and weak institutions” trap.  This provocative message challenges our prevailing practice of assessing a country’s institutions by their distance from the global best practice and ranking them on international league tables. Yuen Yuen’s work, in contrast, highlights the possibility of using existing institutions to generate inclusive growth and further impetus for institutional evolution.”

It is extremely promising that the development establishment questions its long-standing belief that conventionally good/strong institutions, benchmarked by Western practices, must be in place for markets to grow. Tremendous room for alternative methods of growth promotion opens up once academics and policymakers entertain the possibility that existing institutions in developing countries, even if they violate best practices, can be used to kick-start markets, as Professor Ang’s research reveals.

References

Ang, Y.Y. “Do Weberian Bureaucracies Lead to Markets or Vice Versa? A Coevolutionary Approach to DevelopmentM.” Chapter in Centeno, Kohli & Yashar (Eds.), States in the Developing World. Cambridge: Cambridge University Press, 2017.

Ang, Y.Y. How China Escaped the Poverty Trap, Cornell University Press, Cornell Studies in Political Economy, scheduled for release on September 6, 2016.

Ang, Y.Y. “Beyond Weber: Conceptualizing an Alternative Ideal-Type of Bureaucracy in Developing Contexts,” Regulation & Governance, 2016.

The Resurgence of Women’s Protest in the United States

by Megan Bayagich

Trump-WomensMarch 2017-top-1510075 (32409710246)

On January 21, 2017, nearly half a million people flocked to Washington D.C. for the Women’s March. They carried various signs about reproductive rights, anti-Trump sentiment, and intersectionality amongst feminists. The event hosted several celebrities who spoke about women’s empowerment and stressed the need for resistance against the new administration. Just one day earlier, Donald Trump had been sworn into office as the 45th President of the United States, and a different crowd of protestors organized a Counter-Inaugural protest to display their concern. Michael Heaney, a political sociologist  at the University of Michigan, collected data on the participants at both protests, shedding light on the types of people who attended and their reasons for doing so. Heaney recently presented this work at the Center for Political Studies at the University of Michigan.

The foundation of Heaney’s research centers on the theory of “mesomobilization” – an explanation for how protests form and become organized. Typically, a central group decides to take action and declares a frame for the movement. The central group then brings other people together that would be motivated by this specific frame. Therefore, the theory suggests that all protestors share certain commonalities. In this case, the protestors were all highly motivated by women’s rights. Heaney aimed to further explore participants’ identities and compare them to that of the Counter-Inaugural protest, which advantageously occurred in the same city.

Heaney hired a team to sample the two crowds. Stationed at different places throughout the protests, the team would look out over the crowd and select an individual. To reduce bias, the sampler would count five people away from their selected individual then approach this person to participate in a six-page survey. After asking three people in the surrounding area to complete their survey, the team member would begin a different round at a new location and select another random protestor. This cycle repeated until the team gathered about 180 responses at the Counter-Inaugural protest and roughly 320 at the Women’s March.

Women's March (VOA) 03

The survey produced some interesting results:

  • The two crowds showed no difference in ideology. When asked if they leaned left, right, or center, nearly every respondent answered “left.” However, protestors at the Women’s March were much more partisan. When asked about their partisan identification (i.e. independent, independent who leans Democrat/Republican, moderate Republican/Democrat, strong Republican/Democrat, or third party), they answered they leaned more towards the Democratic Party.
  • Counter-Inaugural protestors were more inclined than Women’s Marchers to believe that the current political atmosphere justified violence.
  • Demographically, the crowd at the Women’s March was significantly older than the Counter-Inaugural, but the two groups did not vary in white versus nonwhite respondents.

Next, the survey focused on how respondents framed their participation. Participants were asked why they attended their respective event. The researchers then coded these responses in terms of gender. For example, if the person said they were there to protest for reproductive rights their answer was considered gendered. If they responded that they protested on the premise of healthcare, it was coded as non-gendered. Results show that about 15 percent of the Counter-Inaugural and 35 percent of Women’s March attendees gave a gendered reason for attending their respective protest. Furthermore, people who had organizational attachments  (i.e. involvement with Planned Parenthood) were far more likely to provide a gendered response at both events.  This suggests evidence of the mesomobilization theory at each protest. Heaney asserts that people brought their cohorts to participate in politics based on the frame that a central group created.

The researchers also examined the group of people who attended both the Counter-Inaugural protest and the Women’s March. One would expect that people observed at the Counter-Inaugural, who planned on attending the Women’s March, would be more likely to provide a gendered reason for attending the Counter-Inaugural. Remarkably, this was not the case. However, people observed at the Women’s March who also attended the Counter-Inaugural very commonly provided a gendered reason. Could the event itself play a role in the protestor’s participation or even explicitly introduce a frame? Heaney plans to investigate this curious paradox.

Michael Heaney’s data from the Counter-Inaugural protest and Women’s March gives insight on the mesomobilization theory, along with demographic data on the protestors. He continues to work on connecting evidence from both movements then plans to compare it with other data from the 2016 Democratic and Republican National Conventions and a Right to Life protest.

For more, read Heaney’s working paper here: Partisanship and the Resurgence of Women’s Protest in the United States

Inside the American Electorate: The 2016 ANES Time Series Study

Post developed by Catherine Allen-West, Megan Bayagich and Ted Brader

The initial release of the 2016 American National Election Studies (ANES) Time Series dataset is approaching. Since 1948, the ANES- a collaborative project between the University of Michigan and Stanford University- has conducted benchmark election surveys on voting, public opinion, and political participation. This year’s polarizing election warranted especially interesting responses. Shanto Iyengar, one of the project’s principal investigators and Stanford professor of political science, noted, “The data will tell us the extent to which Trump and Clinton voters inhabit distinct psychological worlds.”


To learn more about the study, we asked Ted Brader (University of Michigan professor of political science and one of the project’s principal investigators) a few questions about this year’s anticipated release.

When was the data collected?

The study interviewed respondents in a pre-election survey between September 7 and November 7, 2016. Election day was November 8. The study re-interviewed as many as possible of the same respondents in a post-election survey between November 9 and January 8, 2017.

The ANES conducted face-to-face and internet interviews again for 2016. How are these samples different from 2012? What are the sample sizes and the response rates?

The study has two independently drawn probability samples that describe approximately the same population. The target population for the face-to-face mode was 222.6 million U.S. citizens age 18 or older living in the 48 contiguous states and the District of Columbia, and the target population for the Internet mode was 224.1 million U.S. citizens age 18 or older living in the 50 U.S. states or the District of Columbia. In both modes, the sampling frame was lists of residential addresses where mail is delivered, and to be eligible to participate, a respondent had to reside at the sampled address and be a U.S. citizen age 18 or older at the time of recruitment.

The response rate, using the American Association for Public Opinion Research (AAPOR) formula for the minimum response rate on the pre-election interview, was 50 percent for the face-to-face component and 44 percent for the Internet component. The response rate for the face-to-face component is weighted to account for subsampling during data collection; due to subsampling for the face-to-face mode, the unweighted response rate would not be meaningful.

Photo Credit: Mark Newman (University of Michigan)

The re-interview rate on the post-election survey was 90 percent for the face-to-face component and 84 percent for the Internet component.

Are there any other aspects of the design that you think are particularly important?

I’d emphasize the effort to collect high quality samples via both in-person and online interviews for the whole survey as obviously the most important design aspect of the 2016 study, helping us to learn more about the trade-offs between survey mode and potential benefits of mixed mode data collection.

Are there any new questions that you think users will be particularly interested in?

Along with many previous questions that allow researchers to look at short and long term trends, we have lots of new items related to trade, outsourcing, immigration, policing, political correctness, LGBT issues, gender issues, social mobility, economic inequality, campaign finance, and international affairs.

What do you think some of the biggest challenges were for the 2016 data collection?

With increasing levels of polarization and a highly negative campaign, some Americans were much more resistant to participating in the survey. Many seemed to feel alienated, distrustful, and sick of the election. Under these circumstances, we worked hard with our partners at Westat to overcome this reluctance and are pleased to have recruited such a high quality sample by Election Day.

What are you most excited about when you think of the 2016 ANES?

The 2016 contest was in many ways a particularly fascinating election, even for those of us who usually find elections interesting! The election ultimately centered on two highly polarizing candidates, and people of many different backgrounds felt a lot was at stake in the outcome. Thus, not surprisingly, there was energetic speculation throughout the year about what voters were thinking and why they supported Clinton or Trump. The 2016 ANES survey provides an incredibly rich and unparalleled set of data for examining and testing among these speculations. I expect it will take some time to arrive at definitive answers, but I’m excited to release this wealth of evidence so the search for the truth can begin in earnest.

Is there anything else you’d like to share?

I would note that future releases will include redacted open-ended comments by respondents, numerical codings of some of the open-ended answers, and administrative data (e.g., interviewer observations, timing, etc.).

For more information about ANES please visit electionstudies.org and follow ANES on Twitter @electionstudies

 

Crime in Sweden: What the Data Tell Us

by Christopher Fariss and Kristine Eck

Christopher Fariss, University of Michigan and Kristine Eck, Uppsala University

Debate persists inside and outside of Sweden regarding the relationship between immigrants and crime in Sweden. But what can the data actually tell us? Shouldn’t it be able to identify the pattern between the number of crimes committed in Sweden and the proportion of those crimes committed by immigrants? The answer is complicated by the manner in which the information about crime is collected and catalogued. This is not just an issue for Sweden but any country interested in providing security to its citizens. Ultimately though, there is no information that supports the claim that Sweden is experiencing an “epidemic.”

In a recent piece in the Washington Post, we addressed some common misconceptions about what the Swedish crime data can and cannot tell us. However, questions about the data persist. These questions are varied but are related to two core issues: (1) what kind of data policy makers need to inform their decisions and (2) what claims can be supported by the existing data.

Who Commits the Most Crime?

Policymakers need accurate data and analytical strategies for using and understanding that data. This is because these tools form the basis for decision-making about crime and security.

When considering the reports about Swedish crime, certain demographic groups are unquestionably overrepresented. In Sweden, men, for example, are four times more likely than women to commit violent crimes. This statistical pattern however has not awoken the same type of media attention or political response as other demographic groups related to ethnicity or migrant status.

Secret Police Data: Conspiracy or Fact?

In the past, the Swedish government has collected data on ethnicity in its crime reports. The most recent of these data were analyzed by the Swedish National Council for Crime Prevention’s (BRÅ) for the period 1997-2001. The Swedish police no longer collect data on the ethnicity, religion, or race of either perpetrators or victims of crime. There are accusations that these data exist but are being withheld. Such ideas are not entirely unfounded: in the past, the Swedish police have kept secret—and illegal—registers, for example about abused women or individuals with Roma background. Accusations about a police conspiracy to suppress immigrant crime numbers tend to center around the existence of a supposedly secret criminal code used to track this data. This code is not secret and, when considered, reveals no evidence for a crime epidemic.

For the period  of November 11, 2015 through January 21, 2016 the Swedish police attempted to gauge the scope of newly arrived refugees involvement in crime, as victims, perpetrators, or witnesses. It did so by introducing a new criminal code—291—into its database. Using this code, police officers could add to reports in which an asylum seeker was involved in an interaction leading to a police report. Approximately 1% of police reports filed during this period contained this code. It is important to note here that only a fraction of these police incident reports actually lead to criminal charges being filed.

The data from these reports are problematic because there are over 400 criminal codes in the police’s STORM database, which leads to miscoding or inconsistent coding. Coding errors occur because the police officers themselves are responsible for determining which codes to enter in the system. The police note that there was variation in how the instructions for using this code were interpreted. The data show that 60% of the 3,287 police reports filed took place at asylum-seeker accommodation facilities, and that the majority of the incidents contained in these reports took place between asylum seekers. Are these numbers evidence of a crime epidemic?

Is there any Evidence for Crime Epidemic in Sweden?

If asylum-seekers are particularly crime-prone, then we would expect to see crime rates in which they are overrepresented relative to how many are living in Sweden. Sweden hosted approximately 180,000 asylum-seekers during this period and the population of Sweden is approximately 10 million. Therefore, asylum-seekers make up approximately 1.8% of the people living in Sweden, while 1% of the police reports filed in STORM were attributed to asylum-seekers.

While the Code 291 data are problematic because of issues discussed above, the data actually suggests that asylum seekers appear to be committing crime in lower numbers than the general population and does not provide support for claims of excessive criminal culpability. There were four rapes registered with code 291 for the 2.5 month period, which we find difficult to interpret as indicative of a “surge” in refugee rape. We in no way want to minimize the impact that these incidents had on the individual victims, but considering wider patterns, we consider a rate of four reports of rape over 76 days for a asylum-seeking population of 180,000 as not convincing evidence of an “epidemic” perpetrated by its members.

There is no doubt that crime occurs in Sweden. This is a problem for Swedish society and an important challenge for the government to address. It is a problem shared by all other countries. There is also no doubt that refugees and immigrants have committed crimes in Sweden, just as there is no doubt that Swedish-born citizens have committed crimes in Sweden as well. But if policy initiatives are to focus on particular demographic groups who are overrepresented in crime statistics, then it is essential that the analysis of the crimes committed by members of these groups be based on careful data analysis rather than anecdotes used for supporting political causes.

The Government of Sweden’s Facts about Migration and Crime in Sweden: http://www.government.se/articles/2017/02/facts-about-migration-and-crime-in-sweden/

Christopher Fariss is an Assistant Professor of Political Science and Faculty Associate at the Center for Political Studies at the University of Michigan.  Kristine Eck is Associate Professor at the Department of Peace and Conflict Research at Uppsala University.

 

Exploring the Tone of the 2016 Campaign

By undergraduate students Megan Bayagich, Laura Cohen, Lauren Farfel, Andrew Krowitz, Emily Kuchman, Sarah Lindenberg, Natalie Sochacki, and Hannah Suh, and their professor Stuart Soroka, all from the University of Michigan.


The 2016 election campaign seems to many to have been one of the most negative campaigns in recent history. Our exploration of negativity in the campaign – focused on debate transcripts and Facebook-distributed news content – begins with the following observations.

Since the advent of the first radio-broadcasted debate in 1948, debates have become a staple in the presidential campaign process.  They are an opportunity for voters to see candidates’ debate policies and reply to attacks in real-time. Just as importantly, candidates use their time to establish a public persona to which viewers can feel attracted and connected.

Research has accordingly explored the effects of debates on voter preferences and behavior. Issue knowledge has been found to increase with debate viewership, as well as knowledge of candidates’ policy preferences. Debates also have an agenda-setting effect, as the issues discussed in debates then tend to be considered more important by viewers. Additionally, there is tentative support for debate influence on voter preferences, particularly for independent and nonpartisan viewers. While debate content might not alter the preferences of strong partisans, it may affect a significant percentage of the population who is unsure in its voting decision. (For a review of the literature on debates effects, see Benoit, Hansen, & Verser, 2003).

Of course, the impact of debates comes not just from watching them but also from the news that follows. The media’s power to determine the content that is seen and how it is presented can have significant consequences. The literatures on agenda setting, priming, and framing make clear the way in which media shape our political reality. And studies have found that media’s coverage of debates can alter the public’s perception of debate content and their attitudes toward candidates. (See, for instance, Hwang, Gotlieb, Nah & McLeod 2006, Fridkin, Kenney, Gershon & Woodall 2008.)

This is true not just for traditional media, but for social media as well. As noted by the Pew Research Center, “…44% of U.S. adults reported having learned about the 2016 presidential election in the past week from social media, outpacing both local and national print newspapers.” Social media has become a valuable tool for the public to gather news throughout election cycles, with 61% of millennials getting political news from Facebook in a given week versus 37% who receive it from local TV. The significance of news disseminated through Facebook continues to increase.

It is in this context that we explore the nature of the content and coverage of the presidential debates of 2016.  Over the course of a term-long seminar exploring media coverage surrounding the 2016 presidential election, we became interested in measuring fluctuations in negativity across the last 40 years of presidential debates, with a specific emphasis on the 2016 debates. We simultaneously were interested in the tone of media coverage over the election cycle, examined through media outlets’ Facebook posts.

To test these hypotheses, we compiled and coded debate transcripts from presidential debates between 1976 and 2016. We estimated “tone” using computer-automated analyses. Using the Lexicoder Sentiment Dictionary (LSD) we counted the number of positive and negative words across all debates. We then ran the same test over news articles posted on Facebook during the election cycle, taken news feeds of main media outlets including ABC, CBS, CNN, NBC, and FOX. (Facebook data are drawn from Martinchek 2016.)

We begin with a simple measure of the volume of tone, or “sentiment,” in debates.  Figure 1 shows the total amount of sentiment – the total number of positive and negative words combined, as a percentage of all words – in all statements made by each in candidate across all debates.  In contrast with what some may expect, the 2016 debates were not particularly emotion-laden when compared to past cycles. From 1976 through to 2016, roughy 6.9% of the words said during debates are included in our sentiment dictionary. Hillary Clinton and Donald Trump’s speeches were essentially on par with this average; neither reached the peak of 8% (like 2004) or the low of 6% (like 2012).

Figure 1: Total Sentiment in Debates, 1976-2016

Figure 2 shows the percent of all sentiment words that were negative (to be clear: negative words as a percent of all sentiment words), and here we see some interesting differences.  Negativity from Democratic candidates has not fluctuated very much over time. The average percent of negative sentiment words for Democrats is 33.6%.  Even so, Hillary Clinton’s debate speeches showed relatively high levels of negativity, at 40.2%. Indeed, Clinton was the only Democratic candidate other than Mondale to express sentiment that is more than 40% negative.

Figure 2: Negativity in Debate Speeches, By Political Party, 1976-2016

Clinton’s negativity pales in comparison with Trump’s, however.  Figure 2 makes clear the large jump in negativity for Donald Trump in comparison with past candidates. For the first time in 32 years, sentiment-laden words used by Trump are nearly 50% negative – a level similar to Reagan in 1980 and 1984. Indeed, when we look at negative words as a proportion of all words, not just words in the sentiment dictionary, it seems that nearly every one in ten words uttered by Trump during the debates was negative.

The 2016 debates thus appear to be markedly more negative than most past debates. To what extent is the tone of debate content reflected in news coverage? Does negative speech in debates produce news coverage reflecting similar degrees of negativity?  Figure 3 explores this question, illustrating negativity (again, negative words as a proportion of all sentiment words) in the text of all Facebook posts concerning either Trump or Clinton, as distributed by five major news networks.

What stands out most in Figure 3 are the differences across networks: ABC, CNN, and NBC show higher negativity for Trump-related posts, while Fox shows higher negativity for Clinton-related posts.  CBS posts reflect a more neutral position.

Figure 3: Negativity in Facebook News Postings by Major Broadcasters, By Candidate, 2016

Clearly, political news content varies greatly across news sources. Trump’s expressed negativity in debates (and perhaps in campaign communications more generally) does not necessarily translate to more negative news content, at least by these measures. For instance: even as Trump is expressing more negative sentiment than Clinton, coverage in Fox is more positive towards Trump.  Of course, news coverage isn’t (and shouldn’t be) just a reflection of what candidates say. But these make clear that the tone of coverage for candidates needn’t be in line with the sentiment expressed by those candidates.  Expressing negative sentiment can produce negative coverage, or positive coverage, or (as Figure 3 suggests), both.

This much is clear: in line with our expectations, the 2016 presidential debates were among the most negative of all US presidential debates.  The same seems true of the campaigns, or at least the candidates’ stump speeches, more generally.  Although there was a good deal of negativity during debates, however, the tone of news coverage varied across sources.  Depending on citizens’ news source, even as candidates seem to have focused on negative themes, this may or may not have been a fundamentally negative campaign cycle. For those interested in the “tone” of political debate, our results highlight the importance of considering both politicians’ rhetoric, and the mass-mediated political debate that reaches citizens.

 


This article was co-authored by U-M capstone Communication Studies 463 class of 2016, which took place during the fall election campaign. Class readings and discussion focused on the campaign, and the class found themselves asking questions about the “tone” of the 2016 debates, and the campaign more generally. Using their professor Stuart Soroka as a data manager/research assistant, students looked for answers to some of their questions about the degree of negativity in the 2016 campaign.

 


Top 10 Most Viewed CPS Blog Posts in 2016

Post written by Catherine Allen-West.

Since it’s establishment in 2013, a total of 123 posts have appeared on the Center for Political Studies (CPS) Blog. As we approach the new year, we thought to take a look back at which of these 123 posts were most viewed across 2016.

 


 

01. Tracking the Themes of the 2016 Election by Lisa Singh, Stuart Soroka, Michael Traugott and Frank Newport (from the Election Dynamics blog)

“The results highlight a central aspect of the 2016 campaign: information about Trump has varied in theme, almost weekly, over the campaign – from Russia, to taxes, to women’s issues, etc; information about Clinton has in contrast been focused almost entirely on a single theme, email.”

 


 

02. Another Reason Clinton Lost Michigan: Trump Was Listed First on the Ballot by Josh Pasek

“If Rick Snyder weren’t the Governor of Michigan, Donald Trump would probably have 16 fewer electoral votes. I say this not because I think Governor Snyder did anything improper, but because Michigan law provides a small electoral benefit to the Governor’s party in all statewide elections; candidates from that party are listed first on the ballot.”

 


 

03. Motivated Reasoning in the Perceived Credibility of Public Opinion Polls by Catherine Allen-West and Ozan Kuru

“Our results showed that people frequently discredit polls that they disagree with. Moreover, in line with motivated reasoning theories, those who are more politically sophisticated actually discredit the polls more. That is, as political knowledge increases, the credibility drops substantially for those who disagree with the poll result.”

 

 


 

04. Why do Black Americans overwhelmingly vote Democrat? by Vincent Hutchings, Hakeem Jefferson, and Katie Brown, published in 2014.

“Democratic candidates typically receive 85-95% of the Black vote in the United States. Why the near unanimity among Black voters?”

 


 

05. Measuring Political Polarization by Katie Brown and Shanto Iyengar, published in 2014.

“Both parties moving toward ideological poles has resulted in policy gridlock (see: government shutdowndebt ceiling negotiations). But does this polarization extend to the public in general?”

 


 

06. What makes a political issue a moral issue? by Katie Brown and Timothy Ryan, published in 2014.

“There are political issues and then there are moral political issues. Often cited examples of the latter include abortion and same sex marriage. But what makes a political issue moral?”

 


 

07. Moral Conviction Stymies Political Compromise, by Katie Brown and Timothy Ryan, published in 2014.

Ryan’s overarching hypothesis boils non-compromise down to morals: a moral mindset orients citizens to oppose political compromises and punish compromising politicians. There are all kinds of issues for which some citizens seem resistant to compromises: tax reform, same-sex marriage, collective bargaining, etc. But who is resistant? Ryan shows that part of the answer has to do with who sees these issues through a moral lens.

 


 

08. Exploring the Effects of Skin Tone on Policy Preferences Among African Americans by Lauren Guggenheim and Vincent Hutchings, published in 2014.

In the United States, African Americans with darker skin tones have worse health outcomes, lower income, and face higher levels of discrimination in the work place and criminal justice system than lighter skinned Blacks. Could darker and lighter skinned African Americans in turn have different policy preferences that reflect their socio economic status-based outcomes and experiences?

 


 

09. What We Know About Race and the Gender Gap in the 2016 U.S. Election by Catherine Allen-West

As of October, the latest national polls, predicted that the 2016 Election results will reflect the largest gender gap in vote choice in modern U.S. history. If these polls had proven true, the 2016 results would indicate a much larger gender gap than what was observed in 2012, where women overwhelmingly supported Barack Obama over Mitt Romney. University of Texas at Austin Professor Tasha Philpot argues that what really may be driving this gap to even greater depths, is race.

 


 

10. How do the American people feel about gun control? by Katie Brown and Darrell Donakowski, published in 2014.

As we can see, the proportion of the public supporting tougher regulation is shrinking over the time period, while satisfaction with current regulations increased. Yet, support for tougher gun laws is the most popular choice in all included years. It is important to note that these data were collected before Aurora, Newtown, and the Navy Yard shootings. The 2016 ANES study will no doubt add more insight into this contentious, important issue.

 


 

Helping the Courts Detect Partisan Gerrymanders

Post written by Lauren Guggenheim and Catherine Allen-West.

In November, a federal court ruled that the Wisconsin Legislature’s 2011 redrawing of State Assembly districts unfairly favored Republicans deeming it an unconstitutional partisan gerrymander. This ruling is the first successful constitutional challenge to partisan gerrymandering since 1986.  The case will now head to the U.S. Supreme Court—which has yet to come up with a legal standard for distinguishing between acceptable redistricting efforts and unconstitutional gerrymandering.

While there have been successful challenges to gerrymandering based on racial grounds, most recently last week in North Carolina, proving partisan gerrymandering—where the plaintiffs must show that district lines were drawn with the intent to favor one political party over another—is more difficult. One reason is that research shows that even non-partisan commissions can produce unintentional gerrymandered redistricting plans solely on the basis of the geography of a party’s supporters. Also complicating matters are legislatures’ lawful efforts to keep communities of interest together and facilitate the representation of minorities. Because traditional efforts can produce results that appear biased, showing partisan asymmetries—the main form of evidence in previous trials—is not sufficient to challenge partisan gerrymandering in the courts.

However, in recent years, scientists have devised several standards that could be used to effectively measure partisan gerrymandering. In last month’s Wisconsin ruling, the court applied one such mathematical standard called the “efficiency gap“- a method that looks at statewide election results and calculates “wasted votes.” Using this method, the court found that Republicans had manipulated districts by packing Democrats into small districts or spreading them out across many districts, which ultimately led to Republican victories across the states larger districts.

Another method to determine partisan gerrymandering, developed by political scientists Jowei Chen and Jonathan Rodden, uses a straightforward redistricting algorithm to generate a benchmark against which to contrast a plan that has been called into constitutional question, thus laying bare any partisan advantage that cannot be attributed to legitimate legislative objectives. In a paper published last year in the Election Law Journal, Chen, a Faculty Associate at the University of Michigan’s Center for Political Studies and Rodden, Professor of Political Science at Stanford University, used the controversial 2012 Florida Congressional map to show how their approach can demonstrate and unconstitutional partisan gerrymander.

First, the algorithm simulates hundreds of valid districting plans, applying criteria traditionally used in redistricting decisions—compactness, geographic contiguity, population equality, the preservation of political communities, and the protection of voting rights for minorities—while disregarding partisanship. Then, the existing plan can be compared to the partisan distribution of the simulated plans to see where in the distribution it falls. If the partisanship of the existing plan lies in the extreme tail (or outside of the distribution) that was created by the simulations, it suggests the plan is likely to have been created with partisan intent. In other words, the asymmetry is less likely to be due to natural geography or a state’s interest in protecting minorities or keeping cohesive jurisdictions together (which is accounted for by the simulations). In this way, their approach distinguishes between unintentional and intentional asymmetries in partisanship.

Using data from the Florida case, Chen and Rodden simulated the results of 24 districts in 1,000 simulated plans. They kept three African-American districts intact because of Voting Rights Act protections. They also kept 46 counties and 384 cities together, giving the benefit of the doubt to the legislature that compelling reasons exist to keep these entities within the same simulated district. The algorithm uses a nearest distance criterion to keep districts geographically contiguous and highly compact, and it iteratively reassigns precincts to different districts until equally populated districts are achieved. The figure below shows how this looks in one of the 1,000 valid plans.

screen-shot-2016-12-01-at-12-15-45-pm

Next, to measure partisanship, Chen and Rodden needed both the most recent data possible and precinct-level election results, which they found in the 2008 presidential election results. For both the existing plan and the simulated plans, they aggregated from the precinct to the district and calculated the number of districts where McCain voters outnumbered Obama voters. The figure below shows the partisan distribution of all of the plans. A majority of the plans created 14 Republican seats, and less than half of one percent of the plans produced 16 Republican seats. However, none of the simulations produced the 17 seats that were in the Florida Legislature’s plan, showing that the pro-Republican bias in the Legislature’s plan is an extreme outlier relative to the simulations.

screen-shot-2016-12-01-at-12-16-41-pm

Because the simulations they created were a conservative test of redistricting (e.g., giving the benefit of the doubt to the Legislature by protecting three African-American districts), Chen and Rodden also tried the simulations by progressively dropping some of the districts they had previously kept intact. Results suggested the Legislature’s plan was even more atypical, as they had less pro-Republican bias than the simulations with the protected districts.

Chen and Rodden note that once a plaintiff can show that the partisanship of a redistricting plan is an extreme outlier, the burden of proof should shift to the state.  Ultimately in Florida, eight districts were found invalid and, and in December 2015, new maps were approved by the court and put into use for the 2016 Election.