The Connected_Politics Blog
On this blog, the Connected_Politics Lab provides data analysis and insights into political events in Ireland and abroad.
The next crowd worker? Can ChatGPT estimate approval levels in speeches in the EU Council of Ministers?
The advent of accessible artificial intelligence like ChatGPT has transformative implications for society. To explore the utility of Large Language Models (LLMs) in academic research, we compare the performance of OpenAI’s most recent Generative Pre-Trained Transformers (GPTs) to human estimations of general approval in speeches made during deliberations in the Council of the European Union. We find that LLMs have the potential to accelerate the analysis of large datasets and alleviate an enduring issue in the social sciences: the high cost, relative scarcity, and time limitations of expert text annotators. Read full article
Deterrent and Spiral Beliefs about Force and War
This research aims to develop a classifier model that identifies these spiral and deterrent beliefs from human text responses. Another objective is to uncover their relationship with other attitudes, specifically militant assertiveness and isolationism. Militant assertiveness refers to the tendency to use aggressive or confrontational measures in the international arena, whereas isolationism relates to the tendency to distance oneself from international affairs. We hypothesise that people with high militant assertiveness prefer deterrent actions, and isolationists may demonstrate more spiral ones. Read full article
How Including Food May Strengthen Peacebuilding Efforts
With rising global food crises being linked to escalating conflicts, exploring how food is addressed in peace processes could give us valuable insights into the complex relationship between food and conflict. As part of UCD’s Connected_Politics Lab, we set out to study the connection between mentions of food topics within peace agreements and the outcome of related peace processes. Read full article
Echo Chambers among Experts of Chinese Politics
This research project explores the potential for echo chambers to develop in a network of 134 self-identified experts in Chinese politics across three regions: Europe, Asia, and North America. Using a dataset of 26,688 blog publications from these experts, we used quantitative text analysis to examine whether authors write more positively or negatively about topics in Chinese politics and to what extent their sentiment is influenced by fellow experts. Read full article
Between the Lines: Analyzing the Effect of Publication Type on Sentiment in Political Scandal Coverage
The manner in which the media covers political scandals can significantly affect citizens' perceptions surrounding such controversies, helping to frame their beliefs and attitudes. As a result, numerous contemporary studies have explored how specific media characteristics impact both coverage and public digestion of such reporting. Leveraging texts consisting of newspaper articles from the United Kingdom from 2019 to 2023, we look to contribute to this scholarship, investigating whether we can find a link between publication type (broadsheet or tabloid) and article sentiment in media coverage of political scandals. Our results indicate that such a relationship can be substantiated with a high degree of confidence, as we observed that tabloid reporting on scandals was consistently more negative than similar coverage in broadsheets. Read full article
Sentiment and Emotions on Lebanese Politics
This blog post is the overview of a larger research project that seeks to determine if the Lebanese political actor Hezbollah tends to use higher levels of negative sentiment throughout public speeches than the Lebanese president. In parallel, this study touches on whether negative emotions, such as fear, appear more than positive emotions, such as trust, during and/or after pivotal events. To conduct such research, we scraped and cleaned a new Arabic data set from speeches published on official websites. We apply quantitative text analysis to perform sentiment analysis and extract the prevalence of specific emotions. The results were unexpected and countered what Western academics anticipated in their studies. Hezbollah prefers to use similar communication strategies as the President of Lebanon. Our study shows how text analysis can enhance our understanding of political conflicts. Read full article
Do Irish politicians simplify speech when given access to a wider audience?
Video recording of parliamentary proceedings has drastically changed the way in which political practice is conducted, and some have questioned whether cameras incentivize politicians to engage in performative antics rather than passing meaningful legislation. Those who oppose televising the political process believe politicians will alter their speech to appeal to a wider audience – essentially turning the legislature into a stage for electioneering. Using a dataset of transcribed Irish Dáil Éireann plenary sessions, we used quantitative text analysis to investigate whether the 2011 implementation of broadcast cameras in the Dáil led to an overall increase in the complexity of TD parliamentary speech. Our results show that the introduction of broadcasting technologies did not produce more than a marginally negative effect on TD speech complexity. Read Full Article
What do political parties talk about when it comes to climate change? (12 July 2022)
Most people care about the environment, but do the parties they vote for? Using a dataset of party manifestos from the Manifesto Project covering 24 European countries and a period of almost 40 years, we uncover the promises that parties make, and the issues they focus on in the context of the climate crisis. We find that ecological protection and infrastructure are the most common themes discussed by parties. Furthermore, we find that, in the face of climate change, traditional expectations of competitive party politics appear to break down. Government and opposition parties are equally likely to make climate-related promises, and green party representation in government does not appear to increase the likelihood of climate-related promises among other parties. Read Full Article
The Substance of the Left-right Dimension in Ireland (21 September 2021)
Empirical analyses of party competition typically take one of two broad approaches towards estimating left-right positions. Either they start from theoretical considerations regarding the content of this dimension and derive measurement rules from those – the deductive approach – or they infer such a dimension from raw data (e.g., covering the content of election manifestos) with some kind of statistical technique – the inductive approach. Advantages and disadvantages of doing it either way have been discussed for a long time, but the debate has mainly focused on questions of measurement validity. It has largely been overlooked that inductive approaches – treating left-right as a “superdimension” that best explains observed differences in textual or behavioural data – allows us to learn more about the content of the main axis of political competition. This is the core point of a (opens in a new window)new paper published in Party Politics (open access) with Kenneth Benoit. Read Full Article
Gender bias in image labeling systems (19 November 2020)
Gender bias is pervasive in our society generally, and in the tech industry and AI research community specifically. So it is no surprise that image labeling systems—tools that use AI to generate text describing pictures—produce both blatantly sexist and more subtly gender biased results. Read Full Article
New R packages to evaluate the statistical uncertainty of causal effects more informatively (14 September 2020)
You may have come across a debate on statistical significance, most recently in a piece in Nature (Amrhein, Greenland, and McShane 2019). In inferential statistics, it is conventional to say an effect is “statistically significant” if its p-value is less than 5% (or, equivalently, its 95% confidence interval does not include zero), and “not statistically significant” otherwise. While it is convenient, this dichotomous thinking has been prone to misuses and misunderstandings (Gerber and Malhotra 2008; Esarey and Wu 2016; McShane and Gal 2016, 2017; Simonsohn, Nelson, and Simmons 2014). Read Full Article
Does the programme for government in Ireland reflect the change that voters demand? (25 June 2020)
While party manifestos have been (opens in a new window)analysed extensively, only a few studies compare issue and policy salience in coalition agreements, and programmes for government (for a recent exception see (opens in a new window)Klüver and Bäck 2019). This is surprising, given that these documents often determine which policies will be enacted in government, and what issues the government will give priority to. They are also very important documents for internal debates within parties, and typically shape the decision on whether party members support entering government. Read Full Article
Newspapers focus at least twice as much on broken than on fulfilled promises (6 May 2020)
Most voters do not believe that parties deliver on their campaign promises. Previous studies, however, usually conclude that parties indeed keep a large share of their pledges. The media take a crucial role in mediating between parties, politicians, and voters. If citizens receive more information on broken promises, their assessment of government performance might be overly negative and might not correspond to the relatively high degree of campaign pledge fulfilment. Results from a (opens in a new window)new paper, published in Political Communication, show that newspapers exert a strong and consistent negativity bias in reports on pledges. Newspapers report at least twice as much on broken than on fulfilled promises. Moreover, the focus on broken promises has increased in recent years. These findings can help us to understand why voters have such a negative perception of parties’ ability to fulfil their pledges. Read Full Article
We only agree to disagree – Forming a government after the 2020 election (4 March 2020)
As also observed after recent national elections elsewhere, for example in Germany, Israel and Sweden, the recent Dáil election resulted in a very difficult government formation process. Read Full Article
Political Advertising online during the 2020 General Election Campaign (13 February 2020)
Over the course of the 2020 General Election a tidal wave of political adverts from parties and candidates have flooded Irish digital media. Over the past four weeks more than more than sixthousand adverts have been published online and as much as €32k was spent on Google and €350kon Facebook. Potentially as much as €410k has been spent by Irish political campaigners online. Read Full Article
The strategic logic of the #GE2020 electoral campaign: Fine Gael talks economy, Fianna Fáil and Sinn Féin talk housing (3 February 2020)
Almost immediately after An Taoiseach Leo Varadkar called the #GE2020 elections on 14 January, campaign posters started to appear on Ireland’s streets. We, as members of the public, knew the campaign was underway, but the issues around which the campaign would revolve were yet to come into clear view. With the manifestos yet to be published, what issues did parties and candidates rally around? And did the official start of the campaign change their issue focus compared to what they were talking about before the campaign started? Read Full Article
The Twittersphere’s reaction to the Second Leaders’ Debate of GE2020 (30 January 2020)
Sinn Feín’s Mary Lou McDonald and People Before Profit leader Richard Boyd Barrett are widely considered to have won Monday’s Claire Byrne debate on the campus of NUI Galway. In a (opens in a new window)tweet, Harry McGee (@harrymcgee), political editor of the Irish Times, put the performance of both McDonald and Boyd Barrett ahead of the rest. Read Full Article
A Two-horse race? The Twittersphere’s reaction to the First Leaders’ Debate of GE2020 (27 January 2020)
The second leaders debate of the election campaign takes place tonight on RTÉ. In contrast to the first debate, it includes all seven major party leaders. The in-studio dynamics are likely to be quite different as a result, but it remains an open question whether or not being included matters in terms of the campaign. In this first Connected_Politics Lab campaign commentary, we explore this question by looking at online engagement with the Twitter accounts of Fine Gael, Fíanna Fail, and Sinn Féin before, during, and after last Wednesday’s debate. Read Full Article