Band 10, Heft 2

Internal working structures within parliaments are notoriously hard to capture. While analyses based on bill co-sponsorship work for the US Congress, this approach is not feasible in many parliamentary systems. Drawing on data from the European Parliament’s legislative term of 2009–2014 this article shows that parliamentary questions can be another option. Members of the European Parliament may demand information from the Council or the Commission through oral questions. We take advantage of the fact that these questions are signed by their authors and construct a social network of members of the Parliament that support each other’s oral questions. This allows investigating how members and their groups and committees cooperate to control both Council and Commission. Our approach helps to map out the internal structure of the party groups and explore which forces shape the global network. We find that cooperation is mostly driven by party group membership with ALDE, Green/EFA, and GUENGL turning out as the most cohesive groups while SD is internally rather loosely connected. The second strongest clustering characteristic is a legislators’ native country.

Big data is everywhere, both in and out of public policy. Though a rich data source, what is the impact of big data beyond the research community? We suggest such that invoking big data-related terms acts as a heuristic for assumed algorithmic quality. Such an assumption leads to greater trust in automation in public policy decision-making. We test this “big-data-effect” expectation using four tests including a conjoint experiment embedded in a recently fielded survey experiment. We find strong evidence that indeed, big data-related terms act as powerful signals of assumed quality where respondents consistently prefer algorithms with bigger data behind them, absent any mention of predictive accuracy or definitions of key terms (e.g. “training features”). As we expect this big-data-effect is likely not beholden to public policy, we encourage more research in this vein to deepen an understanding of the influence of big data on modern society.

The publication of electoral results in real time is a common practice in contemporary democracies. However, delays in the reporting of electoral outcomes often stir up skepticism and suspicion in the vote-counting process. This issue urges us to construct a systematic test to distinguish delays attributable to manipulation to those resulting from a limited administrative capacity. This paper proposes a method to assess the potential sorting of the electoral results given the moment at which polling stations publish their vote totals. To do so, we model the time span for a polling station to report its electoral results, to identify those observations whose reported times are poorly explained by the model, and to assess a potential bias in the candidates’ vote trends. We illustrate this method by analyzing the 2006 Presidential Election in Mexico, a contest that aroused suspicion from opposition parties and public opinion alike regarding how the electoral results were reported. The results suggest that polling stations’ time logs mostly respond to their specific geographic, logistic, and sociodemographic features. Moreover, those observations that took longer than expected to report their returns had no systematic effect on the electoral outcome. The proposed method can be used as an additional post-election audit to help officials and party representatives evaluate the integrity of an election.