Under Review

Gender, Justice and Deliberation: Women’s Voice in Post-Conflict Reconciliation

Denisa Kostovicova, Tom Paskhalis

We study women's speaking behavior to find out why women's presence in debates about post-conflict justice does not result in gender-responsive outcomes. To explain women's presence without influence, we investigate whether processes behind those outcomes are themselves gendered. We apply multi-method text analysis to an original corpus of over half a million words in six languages from civil society debates in the Balkans. These debates preceded the adoption of the Statute of the Regional Fact-finding Commission that did not reflect women's needs and concerns. Our analysis shows that male dominance at the micro-level of turn-taking and the absence of topics addressing gender-specific experience of conflict can drive gender-insensitive outcome, rather than commonly assumed indicators of gender inequality such as women's representation, including the frequency, length and the deliberative quality of their speech. This study contributes novel insights to why gender-just peace is elusive despite women's increased participation in peace-making.

Working papers

Interest Group Access and Campaign Spending Limits: Evidence from Brexit

Tom Paskhalis

Scholars have long been focussed on studying lobbying and potential influence that such activities can have on public policy. The ability to lobby state actors, however, critically depends on having access to them in the first place. So far much of the theoretical and empirical literature on potential mechanisms of acquiring access has been limited to donations or other forms of financial transactions. In this study I argue that in pluralist states with campaign spending limits, the influence of money is more restricted and other mechanisms such as economic importance, long period of state-government interactions and ideological proximity play an important role in meeting government officials. I use government transparency reports for 2010-2017 from the ministerial departments in the UK to measure the level of access and saliency of policy issues that provide evidence of the importance of these alternative mechanisms.
The Least Unclear Language: How Avoiding Negatives Produces Positive Understanding

Tom Paskhalis, Christian Müller

Designing valid and reliable methods for coding large quantities of text is an inherently complicated process. Manual coding of political texts can be prone to a range of problems that can lead to unreliable results. In addition to complex coding schemes, unreliability can result if the coders do not fully process the sentence when deciding on a code. If this is the case, the cognitive difficulty of processing sentences will influence the reliability of assigned codes. The current literature on content analysis assumes that coding errors are uncorrelated with the quantities of interest and are, to a large extent, a function of the analytical framework. However, this can produce biased estimated if some political actors are more likely to use certain language patterns. Psycholinguistic theory suggests that sentences with negations are inherently more difficult to process. By analyzing speeches from the US Congress and the UK House of Commons as well as party manifestos, we show that the number of negated sentences varies considerably over time and in a substantively meaningful way. For instance, legislators from government parties use fewer negations than their opposition counterparts. We address the question of whether linguistic features affect human misclassification rates with a re-analysis of crowd-coded party manifesto sentences and a coding experiment where we directly randomize the presence of some linguistic features. Our results show whether coding errors can be tied to specific linguistic features of a coding unit and thus whether those features have the potential to bias human coding.

Selected work in progress

Record Linkage with Text: Merging Data Sets When Information is Limited

Tom Paskhalis

The recent years have seen the emergence of new, more scalable ways to link information about individuals across multiple data sources. However, merging data sets when the number of variables used for record linkage is restricted remains challenging. In this paper I consider the case when the information is limited to a single multi-token text string. This situation often occurs when researchers work with organisation names, user accounts or any other short labels. In this paper I run a simulation study showing the limitations of the existing approaches that under-perform due to a lack of standardisation and short length of text. Despite these caveats, some information, such as probability of encountering different words remains unexploited by the currently available methods. I propose an alternative approach, where the constituent tokens are weighted by their overall language frequencies. This method is applied to simulated data, as well as large number of government transparency reports from the UK. The proposed weighting method offers a more robust and scalable approach both to deduplicating the transparency reports and linking them to organisation-level covariates.
Emotion Shift and Transitional Justice: A Micro- and Macro-level Effects in Justice Debates in the Balkans

Denisa Kostovicova, Ivor Sokolic, Tom Paskhalis

’A Word After a Word After a Word is Power’: Automating Manipulation Checks for Experiments with Textual Responses

Krisztián Pósch, Tom Paskhalis