Challenges of Statistical Inference in Explaining the Use of Democratic Innovations: An inquiry Into Finnish Local Direct Democracy

What explains the use of direct democratic innovations? I will answer this question as well as discuss the methodological challenges related to statistical inference in comparing democratic innovations. Despite growing use of and case research on democratic innovations, it still seems rather unpredictable where and under what conditions they are used. What we do know is that the use and impacts are often heavily controlled by elected representatives. Systematic comparison is, however, challenging for several reasons, one being the rareness of these events (c.f. King and Zeng, 2001). Statistical analysis of determinants requires very large sample sizes in order to get some variation in the dependent variable. In an alternative approach a purposive sample can be selected ensuring the presence of both positive and negative cases. This approach, however, reduces the data into a purely cross-sectional dataset discarding the temporal dimension (c.f. Beck, 2001). I will discuss the implications of these both strategies. System-level data from municipalities in Finland are used to analyze the conditions facilitating the use of local referendums and referendum initiatives, which were introduced in Finland in 1991. Since then, more than 60 referendums and as many referendum initiatives have taken place. I argue that government-initiated referendums and non-binding referendum initiatives are a good example of democratic innovations which in general are rarely binding or required by law. I consider two types of explanatory factors: 1) Socioeconomic and demographic factors such as the population size, geographical area, economic indicators and education level, and 2) political/policy factors such as particular parties’ share of seats in the local council, effective number of parties, turnout rate in municipal elections and the prospect of municipal mergers. The results of the study may give us valuable information of favorable preconditions – including favorable representative democratic settings – for democratic innovations in general.