Data Innovation for International Development: An Overview of Natural Language Processing for Qualitative Data Analysis
with Anna Hanchar IEEE Proceedings of the 2017 International Conference on the Frontiers and Advances in Data Science (FADS), Xi'an, China, 2017. article
Availability, collection and access to quantitative data, as well as its limitations, often make qualitative data the resource upon which development programs heavily rely. Both traditional interview data and social media analysis can provide rich contextual information and are essential for research, appraisal, monitoring and evaluation. These data may be difficult to process and analyze both systematically and at scale. This, in turn, limits the ability of timely data driven decision-making which is essential in fast evolving complex social systems. In this paper, we discuss the potential of using natural language processing to systematize analysis of qualitative data, and to inform quick decision-making in the development context. We illustrate this with interview data generated in a format of micro-narratives for the UNDP Fragments of Impact project.
Individuals, Disaggregation of the State and Negotiation Tactics: Evidence from the European Union
with Nicola Chelotti
Why do states use certain negotiation tactics more than others? By arguing that the personal characteristics of individual negotiators help to explain political choices, this article contributes to the reinvigorated literature on individuals and International Relations. In particular, it takes issue with one of the most common challenges of this literature – how, given all the bureaucracies they are embedded in, individuals are able to shape political decisions. The article argues that one particular group of actors – diplomatic negotiators – have de facto acquired ultimate policy-making responsibilities, most prominently in the selection of tactics. The image of the negotiating state is thus re-conceptualized. Not only is the fiction of a hierarchical and vertically organized state rejected, evidence is presented of the opposite trend – what might be called the ‘partial individualization’ of the state. Next, by testing four individual characteristics (experience, autonomy, social and epistemic motivations) against an original database of 138 questionnaires completed by national officials in EU foreign policy, the article shows that these characteristics matter in explaining variation in the use of tactics, although this does not apply to all the negotiating activities. The article also reveals that the explanatory power of individual-level characteristics is limited under certain conditions.
Measures of public opinion on European integration are an increasingly important independent or dependent variable in many analyses of European politics, not least due to the progressing politicization of the EU. However, applied researchers typically use a single survey item as their measure of public opinion. This poses problems arising from discontinuations and interruptions of items in survey time-series as well as concerns about differential item functioning across countries, i.e. respondents from different contexts have different reference points from which they evaluate a question. We develop measures of opinion on European integration that are more comparable over time and across countries using group-level item-response theory models that recover yearly estimates of public opinion on integration in all EU member states on the following dimensions: (1) a utilitarian dimension; (2) an affective dimension; (3) an institutional dimension; (4) and an integration dimension. We select public opinion items from Eurobarometer surveys between 1974 and 2017, evaluate the face and predictive validity of our measures, discuss applications, and make the data available for academic use.
Strategic Rapporteur Selection in Informal Inter-Institutional EU-Negotiations
In informal interinstitutional decision making in the European Union, delegations from the European Parliament and the Council of the European Union meet behind closed doors to reach a compromise that is subsequently rubber stamped by the parent chambers. The literature on informal decision-making points to the risk of policy drift towards key figures in the negotiations. Both chambers and/or parties within the chambers should select the key negotiators strategically if the potential for agency drift exists. I test whether the Parliament selects its key negotiator, the rapporteur, strategically using new data on individual preferences of all members of Parliament, national and European party affiliations as well as committee memberships within parliament. To identify the effect of informal negotiations on rapporteur selection, I exploit a rule change in 1999, where the Amsterdam Treaty allowed for file conclusion using informal first reading agreements for the first time. I estimate local average treatment effects using a regression discontinuity approach.
Measuring Political Influence: Agenda-Setting Power in the European Union
This research note is on the conceptualization of influence, to improve validity of its measurement and thereby our understanding of the legislative process. I discuss a current conceptualization used in innovative research on the power of the European Commission to shape policy, showing that theoretical contradictions may arise. I design a new measure that solves these issues by making the counterfactual expectation explicit and incorporating it into the measure. I replicate the research on the Commission's power and show (1) the Commission is, indeed, successful in seeing its preferences converted into policy, (2) contrary to the original piece, the Commission seems to possess genuine agenda setting power, (3) measurement matters as it shapes outcomes. This discussion may inform research on related work, such as the influence of lobby groups, parties, members of international organizations and more.
The European Union is an exceptionally complex political system; its bicameral legislative process in particular needs to overcome a potential plethora of veto-points. At the same time, the EU proves to be a productive negotiation "machine". This paper explores the factors that "oil" this machine; more specifically, we ask how the Union can adopt legislation in the face of an unfavourable ideological (left-right) preference constellation. We argue that actors will make generous concessions on the left-right dimension of political conflict 1) to uphold the overall system of supranational cooperation, 2) to capitalise on the reciprocity that this system affords, and 3) when they act under permissive consensus. The argument is tested on a new data set of codecision files that either amend or replace existing legislation (1999-2014 ). Given ideological divergence, two-thirds of these files—labelled "gridlocked"—should not have been adopted, because one of the veto-players should have preferred the status quo. Lending support to our argument, the analysis shows that homogenous elite-level preferences on continued cooperation and the possibility for issue-linkage can explain the adoption of such gridlocked legislation; however, the potential for compromise decreases as Euroscepticism rises. These findings contribute to established debates about bargaining under complexity and the dimensionality of the EU’s political space—and raise questions about the EU’s future ability to overcome gridlock under conditions of contestation.
Multilevel regression with post-stratification (MrP) has become a standard approach for small area estimation problems. After an initially largely positive reception in the literature, there are now also more critical contributions which show that MrP does not always perform optimally. One inherent problem is how to select the appropriate response model and specifically how to include the context-level information in an optimal fashion. We formulate this as a prediction problem and explore various simple approaches from the statistical learning literature as well as model averaging approaches. We show that a disciplined approach provides an improvement. The size of this improvement will depend on how well non-systematically specified models can perform.
Power and Transparency in Political Negotiations
This article argues that transparency affects power in political negotiations. The more secretive actor has an advantage over the more transparent one. This is of concern for the design of political institutions and for assessing their balance of power. Legislative chambers in bicameral systems can often choose rules regarding their transparency themselves, making it a choice between legitimacy and influence. The European Union provides the setting for testing that link. Treaty changes allowed one legislative chamber to become less transparent in informal and early negotiations. Meanwhile, the opposing chamber remained unaffected. A new measure of relative influence is developed which captures the expectations from veto player theory more accurately. The theory is tested on 140 salient pieces of legislation, finding a strong and robust link between transparency and power.