Support for electoral system reform among voters and politicians: Studying information effects through survey experiments

Does information about the consequences of proposals to change the Norwegian parliamentary electoral system influence voters’ and politicians’ attitudes towards the system? Is the willingness to accept change greater among voters/politicians who “lose” under the present electoral system? These questions are illuminated using empirical data from two identical survey experiments, with responses from both voters and politicians about 1) increased proportionality between parties (more seats for smaller parties) and 2) increased geographical proportionality (stronger representation for the more populous counties). The results show that being informed about the consequences of the proposals has a major effect on voters’ and politicians’ attitudes. This applies especially to the question of increased proportionality between parties, where feedback was particularly negative from respondents who were told that the proposal might weaken the larger parties’ representation and make it more difficult to establish viable governments. The responses to the question about increased proportionality between parties were also influenced by partisanship; politicians who belonged to or voters who voted for one of the smaller parties favour increased proportionality. We also find that there is limited support for the proposal to distribute parliamentary seats according to the number of inhabitants in the counties, and this support is further reduced when the respondents are informed that the measure will increase representation from the more populous parts of the country.


Introduction
The electoral system is one of the most important political institutions in a democracy. The design of the electoral system is important because it affects the structure of the party system, the ability to create viable governments, the degree of representativeness, the selection of candidates, and the extent of party discipline and voter participation (Farell, 1997;Anckar, 2002;Norris, 2004;Gallagher and Mitchell, 2005;Shugart and Taagepera, 2017). This means that changes in the electoral system can have a major impact on the dynamics of the political system.
According to Pippa Norris (1995:4), "electoral systems are rarely designed, they are born kicking and screaming into the world out of a messy, incremental compromise between contending factions battling for survival, determined by power politics". Moreover, defence of self-interests is often cloaked in the guise of democratic principles (Rasch, 2010). Consequently, it may be difficult to find a consistent and principled line of argument behind the setup of a specific electoral system. Many "engineers" have been involved (Norris, 2004) in the making of a specific electoral system. As Carey and Hix (2011) notes, trade-offs are inevitable.
In many countries, the focus of previous research into attitudes towards electoral reforms has been on politicians and political parties (Aardal, 2014;Renwick and Pilet, 2016). This is not surprising, as electoral reforms are implemented only when a sufficient number of parliamentarians (and parties) favours the changes. However, the attitudes of ordinary voters may be as important in securing the continued legitimacy of the electoral system. Extending the argument, the attitudes of both politicians and voters may be essential in providing continued support for electoral reform.
Despite the complexity of and different motivations for electoral reform, it must be ultimately accepted as legitimate among politicians and voters. According to Robert Dahl (2000:37), a political process needs to be based on an "enlightened" understanding of alternatives and consequences to be called democratic. However, the complexity and many details of the electoral system represent a challenge even for the most knowledgeable.
How do we measure the true preferences of voters and politicians when it comes to electoral system reform? Posing a straightforward survey question with no contextual information is risky. The issue of electoral system reform will be far from the "top of the head" (Zaller, 1992) for most people. We may therefore end up measuring something other than people's true preferences. The best means of measuring preferences in this respect is to provide contextual information, but the challenge then becomes that the information could bias the response in one direction or the other. In this paper, we conduct survey experiments, testing the effects of a variety of consequences of a proposed reform (Baekgaard et al., 2019). Specifically, we ask to what extent will attitudes be contingent on those consequences of a specific electoral system reform that are emphasised?
Perhaps the most important issue in electoral system change is the balance between representation and accountable government. "You cannot have both, so the mantra goes" (Carey and Hix, 2011:383). A proportional electoral system that provides a good match between votes and seats in parliament may make it difficult to establish stable majority rule. With unstable or varying governing coalitions, it becomes harder for voters to hold governments accountable. The key issue in this balance, or trade-off, is the electoral system's degree of proportionality.
In this paper, we look at proportionality from a political and geographic standpoint using data from Norway. The Norwegian parliamentary electoral system (like most electoral systems) is less than perfectly proportional, both when it comes to political party representation in parliament and with respect to territorial representation. We study attitudes toward proposals aimed at increasing 1) political and 2) territorial proportionality, using experiments with varying information about the consequences of these proposals.
The article starts with a look at electoral system reform in Norway and the debate that surrounds it. We then discuss the theoretical assumptions regarding the willingness to accept change among politicians and voters. Thereafter, we explain the data, the design of the two experiments and their results, before discussing the implications of the analyses.

Electoral system reform in Norway
According to Rokkan (1970:96-101), democratic representation is both functional and territorial. Functional refers to the representation of economic and ideological interests by parties and is to a large extent contingent on the degree of political party proportionality in the system. Territorial representation, on the other hand, is contingent on the balance between densely and sparsely populated areas. Both of these aspects of democratic representation are present in all democratic societies, perhaps with the exception of the very smallest ones. While Norway is a small country in terms of population size, it is of a significant geographic size, and the territorial element of representation has been historically important and remains quite salient today. Ever since the constitutional assembly in 1814, seats have in fact been unevenly distributed on the constituencies, favouring peripheral areas (Aardal, 2002). 1 In the northernmost constituency, for instance, farthest away from the capital, the number of votes behind a seat is only half that of the most central constituency (Oslo). The deviation from the average number of votes per seat is large enough so that the Organization for Security and Co-operation in Europe (OSCE) has advised the Norwegian authorities to change the system (OSCE/ODIHR 2009). 2 Although the geographical allocation of seats is contested, it does not significantly affect the distribution of seats between the parties (Føllesdal, 2010;Rasch, 2010;Aardal 2011).
In terms of the functional element of representation, a number of smaller parties are competing for parliamentary seats, making the question of proportionality highly relevant. Although it is a PR system, the Norwegian system is less proportional than that of the neighbouring countries Sweden and Denmark. In addition, the degree of proportionality has been an oft-debated topic for many years, leading to increased proportionality in every electoral reform since the introduction of PR in 1919 (Aardal, 2002(Aardal, , 2011(Aardal, , 2014. Furthermore, like most of the old democracies of Europe, the electoral system in Norway is complex with a long history of electoral reform. When direct elections were introduced in 1905, the system was based on majority voting in single-member districts. However, over time elections became more and more disproportional. In 1919, The two parties that had benefitted from the existing system, the Conservatives and Liberals, accepted the introduction of a proportional system (Kristvik and Rokkan, 1964;Danielsen, 1984). The calculation of seats was performed using d'Hondt's highest averages method, which benefitted the largest party, but to a significantly lesser extent than under simple majority voting. To improve proportionality, the calculation method was changed in 1952 to the Sainte-Laguë method, with the first divisor at 1.4 (rather than 1.0). This secured a certain advantage in seats for the largest party, but to a lesser extent than under the old system (Aardal, 2002). The next changes with implications for party proportionality happened when compensatory seats were introduced, initially 8 from the 1989 election, then 19 from the 2005 election (Aardal, 2011). Only parties with 4 per cent or more of the national vote have access to the 19 compensatory seats. The degree of disproportionality, as measured by Gallagher's least square index, has been reduced from 16.9 in 1918 to 3.2 in 2017. At present, nine parties are represented in parliament and changes in the electoral law may have immediate effects on their chances of gaining seats.
In sum, Norway represents a relevant case for an experiment of attitudes towards electoral reforms, in particular concerning proportionality. Like in most of the democracies in Europe, the electoral system in Norway deals with both the functional and the territorial elements of representation, and both of these elements are occasionally subjects of public debate. Also like most other European democracies, the electoral system of Norway has been through several reforms, making the end result fairly complex, but also one that has been through several political debates and compromises over the years. How well our findings travel to other European democracies is an empirical question for future research, but these similarities indicate that the Norwegian case should at least be seen as relevant.
The topic has been on the agenda for a long time, and attitudes may differ concerning whether the issue is proportionality between parties or constituencies. Thus, the results of the experiments may be of interest to multiparty PR systems in general, and in particular PR systems where territorial representation is at stake.

Attitudes toward election reforms, the research question and hypotheses
How strong is the willingness to accept change among voters and politicians? Which of them are more willing to accept change? To provide answers to these questions, the existing research has taken two directions. The first studies the legitimacy of the election system by investigating voters' satisfaction with democracy, or their trust in the authorities and various political institutions (Bowler and Donovan, 2006). This research gives important information about how closely the population is attached to the existing political institutions. One of the findings is that the inhabitants are more satisfied with democracy in countries that use a proportional representation system that includes some degree of preferential voting (Farrell and McAlllister, 2006). However, the effect of these electoral system variables disappears when factors such as corruption and income inequality are taken into account (Renwick and Pilet, 2016;Donovan and Karp, 2017). The purpose of this type of research is to discover which features of the electoral system generate a high level of trust among voters, but it tells us little about whether the voters are willing to change the electoral system and therefore sheds little light on important aspects of the current democratic institutions (Biggers, 2019).
Looking at the issue of proportionality, specifically, it is not only a topic in many discussions about electoral systems and reform in general, but particularly so in Norway. Representational proportionality is often seen as "virtually synonymous with electoral justice" (Lijphart, 1984:140), which makes a strong case in favour of PR systems. Reynolds (2011:76-77) even finds a general trend toward increasing proportionality within PR systems over time. The balance between representation and accountability have been perceived as a scale from high proportionality and low accountability to low proportionality and high accountability. However, Carey and Hix (2011: 395) show that median district magnitude electoral systems (with four to eight seats) have "highly representative parliaments and a moderate number of parties in parliament and in government". Still, the above simplification has some merit, as can be seen through the evolution of electoral reforms in Norway since the introduction of PR in 1919 (Aardal, 2011(Aardal, , 2014.
The fact that proportionality can be looked at from two different perspectives, that is, proportionality between parties on one side and electoral districts on the other, raises the issue of duelling or competing identities.
On the one hand, the political parties' views on electoral reforms may depend on how they themselves emerge under various proposals for change (Banducci and Karp, 1999;Bowler et al., 2006;Bowler and Donovan, 2007;Biggers, 2019). An electoral system is usually a result of compromises between parties, who may have widely differing interests in relation to its design (Aardal, 2011). Parties that are disadvantaged from a reform are presumed to be sceptical about changes, while parties who will benefit are assumed to favour change. For example, smaller parties will support a proposal for increased proportionality between parties, while the larger parties will be interested in having a certain "mandate to govern". The question is whether this kind of self-interest ("group thinking") only manifests itself among the politicians, or if it also does so among voters (Bowler and Donovan, 2007). Might group affiliation considerations also explain the voters' attitudes towards electoral reforms?
On the other hand, we expect not only political parties, but also affiliation with electoral districts, to function as "filters" when the citizens and politicians consider specific electoral reforms. Both politicians and voters could feel a strong attachment to their place of residence, their electoral district, and thus favour reforms that benefit the district in which they live. Klar's (2013) study of competing identities, in the political realm, shows that people may not only hold various identities, but that different identities can be activated by framing an issue in a certain way. For the purpose of this study, both partisan political framings and geographic frames could be expected to have an effect.
A number of findings from political science research supports the idea that people are motivated by an attachment to a certain group, a certain identity. According to the "home team hypothesis", the citizens who have voted for parties that have won a mandate to implement policies are assumed to be more satisfied with the policies than the citizens whose sympathies lie with the opposition parties (Citrin, 1974). Satisfaction increases when the voters get what they want. There is also a persistent gap in support for the political system between election winners (the home team) and election losers (the away team) (Dahlberg and Linde, 2016). It is well established that satisfaction with democracy per se is higher among voters who see themselves as winners in national political elections (Anderson and Guillory, 1997;Anderson and LoTempio, 2002;Anderson et al., 2005;Esaiasson, 2011). In other words, the voters have a strong tendency to adopt the views of parties and groups to which they feel attachment to (Achen and Bartels, 2017).
The voters' attitudes towards democratic decision-making processes are also heavily influenced by whether they perceive the decision outcomes as going in their favour (Arnesen, 2017;Esaiasson et al., 2019). Since what constitutes a "fair" democratic decision procedure is often a matter of debate (Doherty and Wolak, 2012), citizens may motivate their reasoning in a self-serving direction when assessing the decision-making process in political matters. For instance, it has been shown that self-interested, instrumental considerations play an important role in citizens' support for using referendums (Werner, 2019) and for following the result of a referendum if the outcome is unfavourable to them (Arnesen et al., 2019).
The question is whether such instrumental considerations also apply when the subject is attitudes towards electoral reforms. There are few experimental studies that focus on whether attitudes towards electoral systems are influenced by proposals for change that go in the voters'/ politicians' favour or to their detriment (on the other hand, see Bowler and Donovan, 2007;Biggers, 2019). Do voters stick to principled opinions about how the electoral system should be arranged, regardless of their own interests, or are they influenced by whether they see themselves as part of a winning side or losing side with respect to (changes in) the electoral system? Our expectation is that group affiliation, measured here as affiliation with parties and electoral districts, influences the voters' and politicians' attitudes towards changes in the Norwegian electoral system.
Another question of interest is whether voters and politicians are aligned in terms of how they react to proposals of electoral reform. Whereas congruence between citizens and elected representatives on policy issues have received considerable attention (see e.g. Blais and Bodet, 2006;Rosset et al., 2013), less is known about differences in process preferences between citizens and politicians. Far from being just a means to a policy end, democratic processes are important in their own right (Hibbing and Theiss-Morse, 2001). Understanding how citizens want democratic governments to work is important to maintaining legitimate democracies. Similarly, it is important to study to what extent citizens' preferences align with those who decide on the procedure, namely the political representatives. An experimental study on voters' and politicians' fairness assessments of democratic procedures on a contentious policy decision found that politicians were more strongly coloured by the outcome than the voters were (Esaiasson and Øhberg, 2019). The differences are attributed to the politicians' stronger engagement in the policy issue.
Thus, the research question may be formulated as follows: To what extent and how does additional information about the consequences of electoral reform influence the attitudes of voters and politicians with respect to 1) party representation in parliament, and 2) geographical representation in parliament? To study this, we will apply identical survey experiments to voters and politicians.
As far as we know, our investigation is the first study to use identical survey experiments with voters and politicians to research attitudes towards the electoral system. We examine whether the attitudes towards increased proportionality (partywise and geographic) are influenced by varying information about the consequences of the proposed reforms. We use simple descriptive facts: for instance, that increased party proportionality will lead to the better representation of smaller parties and that increased geographic proportionality will benefit larger densely populated electoral districts. We then test whether those who benefit from a proposed change are more strongly affected by these pieces of information than others.
We expect politicians to be more partisan and sensitive to the fortunes of their own party than voters. Politicians from smaller parties could therefore respond more strongly than voters to the "information treatment" that smaller parties will benefit from the proposed reform. Our point of departure is that both voters and politicians hold some preferences regarding the electoral system, but that these attitudes can be influenced by information-treatments: H 1 : the attitudes of voters and politicians toward electoral system reform are affected by information about the consequences of changes in proportionality.
We expect this effect to be stronger among politicians than among voters, since politicians on average should have a stronger allegiance to their party.
H 2 : Politicians will respond in a more partisan way than voters when faced with information that a proposed reform will hurt or benefit their own party.
In  3 We do the same for the politicians in our sample. 4 The small party voters (20.5%)/politicians (23%) are coded 1, while respondents who voted for and/or represent the larger parties are coded 0.
In the experiment on increased geographical proportionality, we expect that voters/politicians who live in counties that will lose Storting seats (the losers) will be less willing to accept change than those who live in counties that will receive increased representation (the winners). When analysing this proposal for reform, we separate those who live in counties that will be disadvantaged by any putative reform (Finnmark, Troms, Nordland and Oppland) from voters who live in other counties. 5 Losing counties are coded 1 and the other counties 0. There is no specific partisan bias to this part of the electoral system, nor to the suggested reform. Identification with one's place of residence, and one's electoral district, is presumably similar among voters and politicians. Thus, we have no specific expectation about the difference between voters and politicians in this regard.

Data and design
The investigation is based on data from the Norwegian Citizen Panel and the Norwegian Panel of Elected Representatives, which are Internetbased surveys of Norwegians' and politicians' attitudes towards various social issues (Ivarsflaten et al., 2018;Peters and Broderstad, 2019). The data are collected by social scientists at the University of Bergen and NORCE. The Citizen Panel recruits participants by drawing from the National Population Register on a random basis so that all Norwegian citizens over the age of 18 years have an equal probability greater than zero of being invited to participate on the panel. The data for the analysis that follows was collected in the spring of 2018. The Panel of Elected Representatives recruits elected politicians from all levels of political government in Norway (municipal councils, county councils, the Storting and the Sami Parliament). Recruitment to the Panel of Elected Representatives started in March 2018, and out of 11, 362 politicians, 4535 participated (4321 final responses). The data for our analysis were collected in January and February 2019, from a random subset of 1450 politicians. Of these, 1248 were municipal politicians (86.4%), 154 were county council politicians (10.7%), 35 were members of the Storting (2.4%) and 8 were members of the Sami Parliament (0.5%).
The investigation was designed in the form of two survey experiments. Such experiments combine the strengths of questionnaire-based surveys and experiments (Mutz, 2011). The respondents are divided randomly into stimuli groups and control groups, so that the only thing that distinguishes them is expected to be the type of treatment they receive during the experiments. This procedure enables us to discover causal links within the framework of a questionnaire-based survey.
In the two experiments, the respondents were randomly exposed to a variety of consequences resulting from the respective proposals. When presenting the empirical analysis, we first describe the design of the actual experiments and the distribution of responses for the three reform proposals. We then look at the results of the actual experiments, before concluding by examining whether the effects of the varying consequences are conditioned by whether the voters lose as a result of the proposed changes. We start with the results of the proposal for increased proportionality between parties.

Experiment I: proportionality between political parties
This experiment measures voters and politicians' attitudes on the central trade-off in electoral system designbetween representation of voters' preferences and governments' accountability (Carey and Hix, 2011;St-Vincent et al., 2016). To measure respondents' attitudes toward greater proportional representation between the parties, we used the following question: A fundamental principle in the Norwegian electoral system is that the voters' votes shall carry equal weight, regardless of which party the voter voted for in the election. Some people think that Norway should change its electoral system to improve the correlation between a party's share of seats in the Storting and the party's support in elections.
[TREATMENT] Would you support or oppose a proposal to change the parliamentary electoral system in this way?
The question had seven response options, where the extremes were 'very strongly support' (1) and 'very strongly oppose' (7). To facilitate interpretation of the results from the regressions, we have reversed the scale so that 7 stands for strong support for the proposals, while 1 represents strong opposition.
It is conceivable that the first sentence in the question could bias respondents toward supporting the proposal, but this effect will be identical across treatment-and control-groups.
The respondents were randomised into three groups. Group 1, the control group, did not receive information beyond the above question and was therefore not primed on self-interest. Group 2 received the following information (marked [TREATMENT]): "This may mean that the smaller parties will be more strongly represented in the Storting than at present". At a similar position in the text, group 3 was exposed to: "This may mean that the larger parties will have fewer seats in the Storting than at present and make it difficult to establish viable governments".
First, we take a look at the overall distribution of attitudes towards a more proportional electoral system. To make the presentation simpler, we have combined those who, to varying degrees, are for or against the proposals and retained respondents who did not take an active position on the proposals in either direction as a separate category (see online Appendix Table A for the full distribution of responses). It appears from the distribution of responses in Fig. 1 that the voters are divided on the question of a more proportional electoral system. This proposal is opposed by 39 per cent, while around 34 per cent answer that they support the proposal to a greater or lesser extent. More than 26 per cent of the respondents did not come to a clear conclusion about whether there should be closer correlation between the parties' share of the votes, and their share of seats in the Storting, compared with the present system. The politicians are also divided on the question, and compared with the voters, their opposition to a more proportional electoral system is somewhat stronger (47.5%). Around 32 per cent of the politicians support the proposal, while around 21 per cent did not come to a clear conclusion about whether increased proportionality is desirable.

Treatment effects
The empirical analysis was performed in two steps. First, we performed a simple bivariate regression analysis to compare the treatment groups to the control group. Second, we tested whether the treatments interact with support for either smaller or larger parties. The latter model includes the treatment groups, small party support and the interaction between the two as explanatory variables. The results are presented in Table 1. Overall, the results show that while voters are initially positively inclined to increase proportionality in the electoral system.

• Voters' attitudes on average turn in a negative direction
o if primed about the benefits the change leads to for the smaller parties, or o if primed about the detrimental effects of the change for the larger parties • Voters are less supportive of reform proposals that are unfavourable to their party. • Results are similar for voters and politicians with the exception that politicians from smaller parties are more influenced by being primed about the benefits to their party than the voters of smaller parties. Table 1 is based on a simple regression analysis with the control group as the reference category. The intercept in table (4.3 for the voters) is therefore the average attitude in the control group, while the individual coefficients represent deviations from the control group. Voters and politicians who did not receive contextual information and thus were not primed on winning/losing (the control group) are positively inclined towards the proposal, as the average for this group is 4.3 among the voters and 4.0 for politicians. The effects for the experimental dummies show that receiving information about the consequences of the proposal contributes significantly to changing the attitudes of both the voters and the politicians. The results are also strikingly similar for voters and politicians.

Model I in
Respondents who are informed that the proposal may provide enhanced representation for the smaller parties are less inclined to support the proposal (− 0.46 point on the 1-7 scale among the voters and − 0.47 among the politicians). The response is even more negative for those who were informed that the proposal may lead to reduced representation for the larger parties in the Storting, making it more difficult to establish viable governments. The latter information reduces the support for the increased proportionality proposal, from both voters and politicians, by more than 0.68 point on the scale from 1 to 7. It is therefore the arguments about impaired ability to form viable governments and fewer seats for the largest parties that contribute to reduced support for the proposal to increase proportionality between parties. This applies to both the voters and their politicians. Salience may also play a role in attitudes toward the proposal. Changes in attitudes can be caused not by new information, but the fact that the different treatments make a particular consideration salient to the respondents.
The experiment, as currently designed, does not enable us to say whether it is the information about weakening the larger parties, or the
implications for viable government, that weigh most heavily. In total, both the voters and the politicians are divided in their views about whether increased proportionality is desirable, but when presented with information about the consequences of such a proposal, their support weakens. This applies regardless of whether increased proportionality is described as an advantage for small parties, or as a disadvantage for large parties/the creation of viable governments. In the question of the balance between breadth of representation and viable governments, both the voters and the politicians incline towards prioritising the latter. The possibility that the proposal may weaken the larger parties' representation and make it difficult to form viable governments has a negative impact on voters' attitudes.
It is worth noting, however, that the question is about changes in the current electoral system, which already has a high degree of proportionality. When informed about some of the consequences of a change, the average voter and politician prefer not to go further in the direction of increased proportionality.
Another question is whether the willingness to change the electoral system is greater among voters of the smaller parties in 2017 than the larger parties. We find the answer in Model II Table I. In this part of the analysis, we lose 277 voters who either did not answer the question about voting for a party, did not vote in the election or were not entitled to vote in the 2017 parliamentary election. The result displays that the effect of being informed about the consequences of the proposal is heavily influenced by whether the respondents voted for one of the small parties in 2017. Voters who voted for one of the smaller parties in 2017 are, as expected, more positive about this change proposal. Among politicians, the effect is even stronger. The F-test of the interaction terms reported in the third to last row is clearly significant (Model II). The marginal effects of this interaction term, both for voters and politicians, are shown in Fig. 2. High scores indicate stronger support for the proposal, and the horizontal line on the y-axis in the figure represents the mean support (irrespective of treatment) among voters (3.9) and politicians (3.7).
In the control group, no reference is made to the possibility that increased proportionality may introduce a bias related to party size. In Fig. 2. Attitudes to increased proportionality between parties, conditioned by which parties the respondents voted for in 2017 and the politicians' party affiliation (with 95% confidence intervals). this group, the proposal receives slightly stronger support among respondents who voted for one of the smaller parties with support around the 4% threshold in 2017. This applies only to the politicians, but the overlapping confidence intervals indicate that the differences are indistinguishable from zero. However, the support for the proposal diminishes significantly among the voters of the larger parties who receive information about the possible consequences. Compared with the larger parties' voters in the control group, the support reduces by nearly 1 point on the seven-point scale when learning that increased proportionality may weaken their own representation or make the creation of a viable government more difficult. Corresponding numbers for the politicians from larger parties are also 1 point. The effect of receiving this piece of information is more or less similar among politicians and among voters. Fig. 2 also shows that in the group of respondents who were exposed to the possibility that the proposal may lead to stronger representation for the smaller parties, the biggest difference is between respondents who voted for small parties and those who voted for larger ones. In this group, as expected, the proposal is most strongly supported by voters who themselves voted for one of the smaller parties. The same applies to the politicians from the smaller parties.
However, being informed that increased proportionality between parties will potentially weaken the larger parties, and make it difficult to establish viable governments, also gives rise to concern among voters of the smaller parties. The support for the proposal declines by 0.4 points if we compare these parties' voters with the same parties' voters who were informed that the proposal might increase their own representation. The corresponding number for the politicians of smaller parties is 0.3 points.
In total, this part of the analysis shows that the willingness to accept change is greatest among voters and politicians who "lose" under the present electoral system (small party voters/politicians), but this only applies once the respondents are informed that the proposal is to their advantage. Being informed that the proposal may weaken the larger parties' representation and render the formation of a viable government more difficult also contributes to reducing the support for the proposal among voters who voted for one of the four small parties threatened by the 4% threshold for access to the 19 adjustment seats in 2017. Among the politicians, support for increased proportionality is stronger among small party representatives, irrespective of treatment.

Experiment II: geographic distribution of seats in the storting
To study the voters' and politicians' attitudes towards the outlying districts being over-represented under the present electoral system, we asked the following question: In the Norwegian electoral system, the outlying districts are provided for by giving them greater representation in the Storting than the populous areas in the country. The present electoral system allocates seats in the Storting to the counties on the basis of each county's population and geographical area. There is currently a discussion in Norway about allocating the number of members of the Storting to be elected from the various counties on the basis of population numbers alone. [TREATMENT]. How strongly do you support or oppose this proposal to allocate the number of members of the Storting to be elected from the various counties on the basis of population numbers alone?
This question also had seven response options, where the extremes were "very strongly for" (1) and "very strongly against" (7). Again, we have reversed the scale so that 7 stands for strong support for the proposals, while 1 represents strong opposition.
In addition to the control group, we randomly allocated the respondents to an additional experimental group. This group was informed about the following consequences of the proposal: "A change of this nature may lead to the more populous parts of the country being more strongly represented in the Storting, while the outlying districts will be more weakly represented than under the present electoral system." The proposal to allocate seats in the Storting to the counties based on population alone does not have strong support among the voters (see Fig. 3 and Table A in the online Appendix for the full distribution of responses). A majority of the voters (53 per cent) oppose this proposal to a varying extent, while 29 per cent support the proposal. This finding may be a bit surprising for those that have followed Norwegian public debate about this issue. The overrepresentation of certain parts of the country has been starkly criticised, while defenders of the current system have been more muted. It appears that the over-representation of less populous counties is supported by the population at large. There are also fewer respondents (18.2 per cent) who do not form a view on this proposal, compared with the proposal on increased proportionality between parties. The attitude of the politicians to the proposal is unambiguously negative, at 73 per cent, and very few politicians fail to form a view on this question (6.7 per cent).

Treatment effects
The results of experiment II are presented in Table 2. The analysis is based on the same estimation strategy as in our first experiment. Whereas voters were overall negative towards increased proportionality across geographical regions.
• voters became less supportive of the reform when learning/being primed about how the changes will be to the benefit of the populous regions and detriment to the outlying districts. Moreover.
• the effects on voters living in counties that lose seats are in the expected negative direction, but not significantly different from other voters. • politicians, who were more strongly opposed to reform without the information about the consequences, also react negatively, albeit less so than the voters do. • living in a county presumed to lose seats under the proposal has a strong independent effect on the support for the proposal both among voters and politicians.
One group was informed that allocating seats in the Storting purely by the number of inhabitants in the counties would 1) strengthen the representation of the populous parts of the country, and 2) simultaneously weaken representation from the outlying districts. It appears that this group supported the proposal to a lesser extent than the control group, which did not receive this information. Therefore, the voters basically oppose changing the electoral system in this way, and they become more negative when presented with contextual information about the consequences of the proposal (− 0.26 points). The politicians are in general even more negative than the voters about this reform proposal. Their support for the proposal decreases somewhat (− 0.25 points) when they are informed that it enhances Storting representation for the populous parts of the country. However, this effect is only significant at the 10 per cent level.
Is the distribution of responses affected by whether the respondents live in counties that may lose under this proposal? Model II in Table 2 provides the answer to this question. The interaction term is not significant at the 5 per cent level, neither for voters nor politicians. This suggests that being exposed to different information will not be affected by whether the respondents live in a county that loses in the event of a change of this nature. However, living in a county that is presumed to lose under the proposal has a strong independent effect on support for the proposal to allocate seats to counties purely on the basis of their populations. Among the voters, living in such a county reduces support for the proposal by 0.8 points on the seven-point scale. If we then remove the non-significant interaction term in the regression analysis, the number increases to almost 1 scale point. Among the politicians, the proposal has 1.1 scale point less support in counties that lose seats, compared with politicians in other counties (without the interaction term the effect size is − 0.86 points). In total, the proposal to allocate the members of the Storting to the counties based solely on the number of inhabitants in the counties does not receive strong support from the respondents. Opposition to the proposal is particularly strong among voters who live in counties that would lose seats if the proposal were to be implemented. Among the politicians, the support for the proposal is especially low in Northern Norway [Nord-Norge] (14 percent) and Trøndelag (21 percent) compared to the Oslo/Akershus (the capital area) in which a majority support the proposal (55 percent) (see online Appendix Figure A). We believe that these results illustrate the importance of the center-periphery dimension in Norwegian politics and a historical tradition of making allowances for the outlying districts in the allocation of seats (see Aardal, 2011). The overall results for voters and politicians are quite similar in this respect. Voters and politicians are about equally affected by the information-treatment.

Conclusion
The Norwegian electoral system is frequently debated and studied, and the debate is characterised by the parties' views on various proposals for reform. We know little about the voters' attitudes towards the parliamentary electoral system in general, or to various proposals for reform in particular. In our analysis, we have endeavoured to shed light on the voters' and politicians' attitudes towards the parliamentary election system, given the various consequences of specific proposals for reform.
By using two identical survey experiments dealing with proportionality in 1) political party representation and 2) geographic representation, we find a fair amount of support for the existing system. A majority opposes the proposal to allocate seats to the counties (electoral districts) purely based on the number of inhabitants (i.e. to increase proportionality), and the opposition is strongest among the politicians.
The voters and politicians are about evenly divided on the question of increased proportionality between parties, but a plurality opposes such a change, especially among politicians. For the electoral reform proposals in this study, there is thus a general preference for the status quo; voters and politicians alike object to changes toward a more proportional electoral system than what we have today. The status quo bias surfaces in various ways within the political domain among voters and politicians alike (Druckman and Lupia, 2000;Sheffer et al., 2018;Alesina and Passarelli, 2019), and our results display that they may play a significant role in how voters and politicians react to electoral system reform proposals as well.
The results from the survey experiments furthermore show that both the voters and the politicians are to a great extent influenced by information telling them the expected consequences if the proposals are implemented. The voters and politicians appear to a limited extent to have stable attitudes towards the design of the parliamentary electoral system. It is only when respondents are informed (or primed) about the consequences of the reform proposals that strong views are expressed.
Thus, we learn that the consequences of changes to the electoral system is something neither voters nor politicians are acutely aware of. We interpret this as a good sign, and a signal that the processes are not heavily politicised. It is also a good sign that voters and politicians are broadly aligned when it comes to how they view electoral reform. Just like well-functioning democracies depend upon politicians being representative of the views of the voters on policy issues, it is important that their process preferences reflect the preferences of the voters they Standard errors in parentheses. *p < 0.10, **p < 0.05, ***p < 0.001.
represent. Moreover, we observe that instrumental, self-interest considerations do influence preferences for electoral processes, for voters and politicians alike. This result aligns well with referendum studies and other democratic decision process studies that show how the fairness and acceptance of processes are tinted by how favourable the decision outcomes are to themselves. We contribute to that literature with these experiments by also displaying the presence of outcome favourability effects in questions of reforms on proportionality in electoral systems. This effect is particularly pronounced for the question about increased proportionality between parties. The voters support proportionality as a principle, but support for this principle declines significantly once the consequences are introduced. Moreover, the proposal to allocate Storting seats to counties according to the number of inhabitants in the counties receives less support among voters who live in counties that will be disadvantaged if population is used as the basis for allocating seats. It has even less support among the politicians, and this applies particularly to politicians from counties that would have fewer seats if the proposal were implemented.
In the experiment about proportionality between parties, the outcome favourability effects are stronger among politicians than among voters. When informed (or primed) that larger parties would find it more difficult to govern or that smaller parties would benefit from the proposal, politicians from larger parties become much more negative toward the proposal. This reflects the bigger consequences of electoral reform for a politician than for a voter. Being an active member, a politician identifies more strongly with the party than a voter who merely supports it. Indeed, electoral reform may directly affect their professional career. They have more to lose or win from an electoral reform that affects proportionality between parties than voters do. As such, politicians may be strong conservatory forces in the question of electoral system reform.
One important question is the external validity of our results. Research analyzing attitudes to electoral reforms derive their hypotheses either from self-interest or ideology. Our results confirm that voters/ politicians benefiting from a given electoral reform proposal, such as increased proportionality, support it more. Our finding also indicate that left-leaning voters/politicians favour increased proportionality in principle (Labour Party respondents in the control group), but that their attitudes may be mediated by whether they benefit from the reform or not. However, we know little about how specific institutional settings might shape attitudes to electoral reforms especially among voters. Still, we think it is likely our findings can be generalized to electoral system reform debates elsewhere. The Norwegian political discussions that come along with profound electoral reforms resemble those in other countries. This is especially the case when it comes proportionality between parties. The trade-off between representation and accountable government is hotly debated in many countries. We can only speculate on whether our results could inform debates in other electoral contexts. Certainly, other plausible institutional mechanisms could be important in understanding attitudes to electoral reform proposals. For instance, it is possible that voters in democracies with weak and unstable governments may be especially sensitive to information about how strict proportionality can make it difficult to establish viable governments. Support for electoral reforms may also reflect party system characteristics since PR system with many (and small) parties produce coalition governments drawing on support of small parties either in government or in parliament. Also, respondents in countries with a center-periphery dimension reflected in their electoral systems should be especially sensitive to reforms that intend to change the allocation of seats in parliament in favour of more populous electoral districts. Attitudes to the electoral system and reforms is not likely to depend solely on who benefits, but also different institutional features and their outcomes could impact attitudes both among politicians and voters. Surely comparing attitudes to possible electoral reforms in different countries with different institutional characteristics (and consequences) is an area where more research is justified.
The starting point for this article was an expectation that the voters'/ politicians' attitudes towards the parliamentary electoral system are dependent on the consequences that may result from changing the electoral system. To a large extent this is shown to be the case. Electoral systems are complex, and the voters cannot be expected to have welldeveloped attitudes towards them. Indeed, it may be viewed positively that the electoral system is a low salience issue among voters, suggesting that the electoral system overall is perceived to produce fair outcomes.
The experiments show that to discover the respondents' attitudes towards the parliamentary electoral system, it is crucial to give them contextual information. When this type of information is provided, the experiments show that there is a clear gap in attitude towards electoral reforms between those who will gain and those who will lose as a result of the reform. The effect is especially pronounced among politicians from parties that stand to lose from a proposed reform. It should be emphasised that other consequences of the electoral system may also be significant for the voters' and politicians' attitudes towards the parliamentary election system, and with respect to democracy more widely. Some debate about the fairness of the electoral system is common in many democracies. More experiments such as these could be useful in other countries, both to illuminate the true attitudes of voters and politicians and to discover what the motivation and reasoning behind those attitudes are.

Data availability
Data and code for all analysis in the manuscript are available upon request.