Document is current

Article History

Received: 26 March 2024

Accepted: 24 March 2025

First Online: 6 June 2025

Declarations

:

: The author(s) declare none.

: The steps below show how Eq. () was derived from Eq. () and how Eq. () was derived from Eq. ():Cross multiply:Remove parentheses:Simplify by subtracting from both sides of the equation:Bring all the terms with to the left-hand side of the equation by subtracting from both sides of the equation:Factorize:Divide both sides of the equation by and cancel common terms:To derive Eq. () from Eq. (), note that:Equation (4.1) can be rewritten as:We can now substitute parts of Eq. (4.2) with thereby obtaining Eq. ():

: Table Table presents a systematic overview of the process for obtaining and using the regressions models. We used the Stata program’s postestimation command, Predict, to perform the required calculations.

: Our identification was based on the assumption that municipal amalgamation only influenced voter turnout, having no other effect on the share of yes votes in the ISSA referendum. However, municipalities are not randomly selected for amalgamation. Voters in amalgamated municipalities could have characteristics influencing their voting, which differ from those in non-amalgamated municipalities and which were not captured by our control variables. Therefore, we relaxed the assumption that amalgamation had no effect on vote shares and limited the dataset to the 139 municipalities that were amalgamated during the period 2000–2022 (Table ) and the 84 municipalities that were amalgamated during the period 2010–2022 (Table ). That is, we compared voting in municipalities that were amalgamated circa 2018 (without concurrent municipal council elections) with municipalities that were amalgamated just before or after that period. As shown in Tables and , this analysis yielded regression results that closely resembled those in our main analysis (see Table , columns 3 and 4). Thus, we feel confident that our results were not driven by differences between voters in amalgamated and non-amalgamated jurisdictions.

: See (Table )

: In our main regression (Table ), we weighted the regressions results for the number of eligible voters per municipality. As an additional robustness test, we reran the regressions without using weights. As Table shows, the change in the coefficients of the relevant variables and their significance was negligible. This result implies that our results were not driven by the use of weights.