Mock elections, electoral participation and political engagement amongst young people in Iceland

In this paper, we analyse whether mock elections in Iceland are fulfilling their intended purpose of increasing young voters’ participation in politics and other types of political engagement. Mock elections have been carried out in Icelan-dic upper secondary schools prior to parliamentary elections since 2016 (and prior to the municipal elections in 2018), but there is no peer-reviewed research about whether these elections are followed by an increase in political participation and engagement amongst young people. Socialisation is a key factor in encouraging political participation amongst young voters as the formative years between childhood and adulthood are a key period in forming behaviour and political attitudes. In this paper, we argue that mock elections have a socialisation impact in terms of increasing young people’s political awareness and engagement, including electoral participation. Based on the results from a survey sent to four upper secondary schools in Iceland, two where mock elections were held and two where they were not held, we found that mock elections have a negligible link to participation in real elections, whether voting made a difference or whether it was considered a civic duty to vote. Other main results are that in schools with mock elections, students were more politically engaged


Introduction
This paper examines mock elections in Icelandic upper secondary schools and their possible socialisation impact on students' political engagement, both in terms of electoral participation and other types of political engagement which are the frequency with which these students discuss politics, their interest in politics, whether they consider it a civic duty to vote or whether they think that voting matters. Mock elections are simulated elections that are usually carried out in the runup to general elections. One of the aims of mock elections is to give young voters, and those coming of age, an opportunity to practice voting and encourage them to participate in real elections. Mock elections are also aimed at mobilising and encouraging young people to think about politics and make informed decisions about their civic participation (Öhrvall & Oskarsson 2018). Mock elections have been carried out in Icelandic upper secondary schools since 2016, but there has been limited research about their success in fulfilling their intended purpose. Given that turnout, specifically amongst young voters, has declined in Iceland (Önnudóttir et al. 2021) and that mock elections have been implemented as one response aimed at increasing electoral and other types of political engagement amongst young voters, it is important to analyse whether these elections are fulfilling their intended purpose.
Elections and electoral participation are amongst the cornerstones of representative democracy in modern democratic states (Dinas 2012). Electoral participation is imperative to ensuring that electoral results reflect diverse interests regarding political decision-making and the actions of political authorities. As van Deth (2001) has argued, research on political participation over the last 50 years has showcased the importance of the possibility for citizens to participate or influence actors charged with the responsibility of democratic decision-making. This is due to the expanding scope and activities of both government and the political domain that citizens might want to participate in and impact.
There are many definitions and means of political participation, but at its core, it can be described as a range of actions that citizens can apply towards influencing the actions and decisions of government and legislators or selecting citizen representatives (Verba & Nie 1987). Electoral participation has been the most pervasive way in which citizens participate (e.g. Coffé 2013; Theocharis & van Deth 2018). In the last few decades, there has been a general decline in electoral turnout in most modern established democracies (Delwit 2013;Franklin 2004), including Iceland (Önnudóttir et al. 2021). This decline is worrisome, especially given that elections and electoral participation are amongst the cornerstones of modern liberal representative democracy.

STJÓRNMÁL & STJÓRNSÝSLA
According to Franklin et al. (2004), declining electoral participation in modern democracies has been driven by decreased participation amongst young people and that a generational change has occurred, where younger cohorts today vote to a lesser extent than prior younger cohorts. This change has also taken place in Iceland (Halldórsson & Önnudóttir 2019;Önnudóttir et al. 2021). For a better understanding of why electoral participation has taken this downturn, which seems largely driven by generational change, it is important to understand the determinants affecting young people's electoral participation. Moreover, it is important to analyse the actions that have been taken to increase young people's electoral participation. Here, we focus on the role of mock elections in upper secondary schools in Iceland. In our theoretical argument, we focus on the role of socialisation in voters' formative years when it comes to forming the habit of voting and other expressive participation. We argue that mock elections can constitute part of the socialisation process with the aim of inducing electoral and other types of political engagement amongst young voters and first-time voters.
For the purpose of this paper, we used data from a survey sent to students in four upper secondary schools in November 2021, shortly after the parliamentary election in Iceland at the end of September that year. Mock elections were held in two of these schools, which gave us an opportunity to analyse whether political engagement differed amongst the students based on whether a mock election had been held in their school. Another way would be to focus on the links between individuals' participation in mock elections and political engagement. However, those who participate in a mock election might very likely also be those who would have voted anyways in a real election, regardless of whether a mock election was held in their school or not. Thus, we consider it to be more meaningful to contrast political engagement between schools depending on whether a mock election was held or not.
The paper proceeds as follows: we start by discussing the principal factors affecting people's political participation by focusing on young people's electoral participation and research on mock elections. After presenting our hypotheses, we briefly discuss turnout and mock elections in Iceland, followed by a description and the results from our analysis. The paper concludes with a summary and a discussion.

Drivers of electoral participation
In a meta-analysis, Smets and van Ham (2013, 3) categorise explanatory factors of voter turnout into six broad theoretical approaches, which are not necessarily mutually exclusive: the resource, rational choice, mobilisation, institutional, psychological and socialisation models. Furthermore, some of the explanatory factors can arguably belong to more than one model. In this paper, we focus on the role of mock elections as part of the socialisation model and whether and how they impact or relate to electoral participation and other types of political engagement, here political interest, frequency of political discussion, whether voting matters and whether it is considered a civic duty to vote. Those factors are arguably interrelated to some extent and have been found to affect electoral participation (Blais & Young 1999;Franklin 2004;Schmitt et al. 2021). Follow-

STJÓRNMÁL & STJÓRNSÝSLA
Mock elections, electoral participation and political engagement amongst young people in Iceland ing Smets and van Ham's (2013) approach, political interest and whether voting matters fall under the psychological model, civic duty can be seen as both part of the rational choice model and the psychological one, and political discussion is part of the socialisation model. Here, we briefly discuss those factors and each theoretical approach before going into greater detail about the importance of socialisation. The basic idea behind the resource model is that electoral participation is driven by the resources available to individuals for participation. They include factors such as education, (Fieldhouse et al. 2007;Rosenstone & Hansen 1993), age 1 (Fieldhouse & Cutts 2012), income (Blais et al. 2004;Verba & Nie 1987), marital status and participation in social networks (Franklin 2004;Schmitt et al. 2021). People who have attained higher levels of education, older people, those with higher incomes, those who are married and those who actively participate in social networks, such as churches, labour unions and other social movements -are more likely to vote compared to people who are less educated, younger, have less income, are unmarried or do not actively participate in social networks.
The rational choice model is based on a cost-benefit calculus of voting, meaning that the benefits should outweigh the costs of voting (Rider & Ordeshook 1968), whether these are personal benefits or social benefits. Some examples of factors that Smets and van Ham (2013) include under this model are whether people care who wins an election and evaluations of the economy and of one's own economic situation. They include a sense of civic duty to vote as an 'extended' rational choice, based on the argument that even if the chances that a single voter will impact the outcome of the election are slim, a sense of civic duty will still encourage people to vote. While this directly contradicts the premises of rational choice that people's actions are driven by their own narrow selfinterest (e.g. Grofman 1993), it has been argued that the rational choice model can be extended to include a sense of civic duty to vote as a constraint or an encouragement to vote (Blais & Young 1999;Goldfarb & Sigelman 2010).
The mobilisation model focuses on the role of various agents in getting out the vote. Associations such as political parties, social movements, the media and interest organisations can be considered as such motivators (Rosenstone & Hansen 1993). In elections, these organisations either campaign or try to highlight certain issues with the purpose of convincing voters to cast votes in their favour or show support for their issues. The more the campaign is covered by the media, the greater the mobilisation factor in getting out the vote (Geys 2006), and media use and exposure to political news have been found to positively impact turnout (Franklin 2004;Smets & van Ham 2013).
The institutional model concerns factors pertaining to the settings of the political system, the electoral system or other systematic factors that concern the elections themselves. Institutional factors include how easy it is to vote (e.g. aspects concerning the voter registry, access to polling stations, etc.) and whether voting is compulsory (Geys 2006). Other types of institutional settings that Smets and van Ham place under this model are for example the effective number of political parties and voter facilitation rules. The arguments are that the more parties voters can choose from, and the easier STJÓRNMÁL & STJÓRNSÝSLA it is made for voters to vote (e.g. with advance voting and voting over weekends), the higher the voter turnout.
According to Smets and van Ham (2013), explanatory factors regarding electoral participation under a psychological approach range from cognitive characteristics, such as political interest and political knowledge, political efficacy, to factors that relate more to partisanship and voters' bond with political parties. Those who are more interested in politics, more knowledgeable about them and have strong ties with political parties are more likely to vote. Concerning political efficacy, it is expected that those that have higher confidence in the act of voting, whether it is own voting as internal efficacy (Morrell 2003), or whether it makes a difference whether people in general vote in terms of external efficacy (Kittilson & Anderson 2010), are more likely to vote. Smets and van Ham include civic duty to vote under the psychological model, but as already discussed they also include it as an extended rational choice, which is an example of that different factors can be argued to belong to two or more of the models discussed in this paper.
Forming a habit of voting has been shown to predict future voting (Gerber et al. 2003). While it is not clear-cut whether habituation falls under a resource approach for electoral participation, the rational choice model or an alternative approach, the process of forming the habit clearly falls under the socialisation approach. In the formative years between childhood and adulthood, socialisation is usually considered key to forming the political attitudes and political behaviour that people carry with them for the rest of their lives (Plutzer 2002). Social exchange in the form of talking to friends, family or others can be seen as part of the socialisation effect when it comes to political participation, given that political discussion can convince people to participate (Smets & van Ham 2013). Furthermore, taking part in discussions about politics can increase people's ability to discuss their political views in a more consistent and sophisticated manner than if they never think about politics (Kim et al. 1999).
It has been found that when other people in close proximity to young voters exercise their vote (e.g. peers or parents), there is an increased probability that young voters will turn out to vote. For example, Fieldhouse and Cutts (2012) found that young first-time voters are over five times more likely to vote if they live in a household where another adult votes compared to first-time voters who live in a household where nobody else votes. This effect is not found amongst older voters, which could be, as argued by Fieldhouse and Cutts, because younger voters have not yet formed a habit of voting and are, thus, more sensitive to the attitudes and behaviours of others around them.
As pointed out by Plutzer (2002) the cost of voting for first-time voters is magnified as they have never voted before, they may not have developed an understanding of party differences and key issues and are usually surrounded by peers who are in the same position. Their 'start-up' cost tends to vary as some of them have more resources, whether from their own achievement or interest or those of their parents or other agents. One of the key factors in this argument is the importance of encouragement and socialisation in helping all young voters to overcome the cost of voting. Learning to vote and establishing the habit of voting increases the likelihood that people will continue to vote.

STJÓRNMÁL & STJÓRNSÝSLA
Mock elections, electoral participation and political engagement amongst young people in Iceland This political learning curve is usually mediated through agents or institutions such as the family, peers, school, the mass media (Smets & van Ham 2013) and, in recent years, social media (Loader et al. 2014). Through socialisation, people's behaviour can be impacted, especially when they are faced with something new such as voting (Franklin et al. 2004). As many modern democracies have been faced with a general decline in voter turnout, some actions have been implemented by the authorities, the school system or other institutions or movements, with the aim of halting and reversing this decline, especially amongst young voters. One such action has been to introduce young voters and those coming of age to a mock election a few weeks or days prior to a real election. In this sense, mock elections are clearly intended to socialise young voters into being politically engaged. While we place mock elections under the socialisation model in this paper, we do not exclude that mock elections can work in other ways to encourage voting, such as a mobilisation effect, or by highlighting to young voters the cost and benefits of voting in line with a rational choice model. However, given that the aim of mock elections is to provide young people with the opportunity to practice voting and to encourage them to become otherwise politically engaged, mock elections can clearly be seen as an intended socialisation impact, and for that reason we focus on them as such in this paper.
In the next section, we briefly discuss the literature on the possible impacts of mock elections on young people's electoral participation and other types of political engagement. Thereafter, we present our hypotheses. There is a lack of research on the impact of mock elections on participation and engagement, and there are varying results in the limited literature on the subject. Borge (2017) examined whether there was a link between voting in a mock election and the intention to vote in an upcoming parliamentary election in Norway and found a positive relationship. Two points should be noted here. First, mock elections are carried out in all upper secondary schools in Norway, and thus, one cannot compare turnout based on whether there was a mock election in the students' school. Second, there is a fundamental difference between the intention to vote and the actual act of voting; those who say that they intend to vote do not necessarily turn out to vote.

Mock elections and political engagement
As such, the results of a survey among students in upper secondary schools in Iceland showed that those who participated in the mock election in 2016 were more likely to say that they voted in the parliamentary election a few weeks later (Isebarn & Björgvinsdóttir 2017). Again, this does not necessarily mean that the mock elections influ-

STJÓRNMÁL
& STJÓRNSÝSLA enced electoral participation as it might well be that those who intended to vote were precisely those who participated in their school's mock election. For this reason, we consider it important to contrast political engagement between schools depending on whether they held a mock election or not, instead of focusing on differences in engagement depending on whether individual students participated in a mock election.
Öhrwall and Oscarsson (2018) found that when controlling for factors in students' background -such as whether their parents vote, average grades and the socio-economic status of the family -mock elections involving students attending upper secondary school in Sweden did not seem to increase turnout amongst young voters in real elections. However, the Öhrwall and Oscarsson noted that not finding a relationship with electoral participation did not mean that mock elections had no impact. Mock elections can impact other kinds of political engagement such as students' sense of political efficacy, political interest and political knowledge. This is in line with findings from Denmark, where mock elections seem to enhance young people's sense of internal political efficacy and political knowledge (Hansen 2017;Hansen et al. 2015). Another study from England showed that students' participation in school activities intended to increase their engagement in politics, such as taking part in debate teams and mock elections, seemed to be successful in both the short and long terms (Keating & Janmat 2016). Students who take part in these activities were generally more likely to vote in the future and take an active part through other means, such as contacting politicians and taking part in demonstrations.
This short overview and the limited research that is available, shows that there is no general consensus about whether and how mock elections impact political engagement. In some cases, these elections have been found to enhance electoral participation and other types of political engagements, while in other cases, they seemed to encourage political engagement but not necessarily electoral participation. Here it should be emphasised that even though a direct link between mock elections and electoral participation cannot be established, mock elections can raise young people's awareness of and interest in politics and other types of expressive participation, which in turn will mobilise them to turn out to vote.

Hypotheses
Based on our prior discussion about the possible socialisation impact of mock elections, that the aims of mock elections are to increase young people's awareness of politics and encourage them to vote, we present our hypotheses about whether the 2021 mock election in Iceland fulfilled those purposes when it comes to young people's political engagement: H1: In schools where mock elections are held, students are more likely to vote in general elections.

H2:
In schools where mock elections are held, students discuss politics more frequently.

H3:
In schools where mock elections are held, students are more interested in politics.

STJÓRNMÁL & STJÓRNSÝSLA
Mock elections, electoral participation and political engagement amongst young people in Iceland H4: In schools where mock elections are held, students are more likely to consider it their civic duty to vote.

H5:
In schools where mock elections are held, students are more likely to believe that who people vote for can make a big difference.
The first hypothesis concerns whether mock elections impact electoral participation, which is an action and taps directly into political behaviour. The other four hypotheses are about political engagement as reflected in young voters' expressive political participation, which is more about attitudes and awareness instead of an action. As discussed earlier in this paper, the frequency of political discussion can be considered as part of a socialisation model, political interest and whether voting matters as part of a psychological model, and civic duty as part of both an extended rational choice approach and a psychological model. Ultimately, those perceptions and expressions of political engagement should encourage the act of voting. Before turning to our analysis, we shall discuss in the next section turnout amongst young voters in Iceland and mock elections in upper secondary schools in Iceland. Figure 1 shows that for most of the 20th century, turnout in Iceland was rather stable at around 90% in parliamentary elections and 80% in municipal elections (Statistics Iceland 2020, 2022b). A slow decline in turnout seems to have begun at the end of the 20th century, with a steeper decline in municipal elections than in parliamentary elections.

10
Mock elections, electoral participation and political engagement of young people in Iceland

STJÓRNMÁL & STJÓRNSÝSLA
Statistics Iceland has registered the age of those who turn out to vote since the 2014 municipal election. In Figure 2, we see that younger people vote to a lesser extent than older voters in both parliamentary and municipal elections (Statistics Iceland 2022a). In the 2014 municipal election, the electoral participation of 18-and 19-year-old voters was 51.9%, while it was less than 50% for voters in their 20s. The uptick in the turnout amongst first-time voters aged 18 and 19, compared to first-time voters in their early 20s, is an interesting trend which is not exclusive to Iceland (e.g. Meredith 2009). This uptick could be due to a recency effect, where first-time voters who get the opportunity to vote shortly after becoming eligible to do so are more likely to turn out compared to first-time voters who must wait a bit longer before they get the opportunity to exercise their right to vote.

11
Sara Þöll Finnbogadóttir Eva Heiða Önnudóttir n höfundar[, nafn höfundar] after becoming eligible to do so are more likely to turn out compared to first-time voters who must wait a bit longer before they get the opportunity to exercise their right to vote. Week, including political debates and panels with representatives from political parties. Mock elections should be as 'real' as possible, with voting laws, ballots, polling booths and electoral committees as part of an attempt to both train and habituate young people to take part in elections (ÉgKýs 2022). The first mock election in Iceland was held shortly before the parliamentary election at the end of September in 2016. Since then, mock elections have been held prior to all parliamentary elections (2017 and 2021) and prior to the 2018 municipal elections. Schools' participation in mock elections in Iceland is not mandatory, but participation has been high throughout exceeding 70% (73% in 2016, 70% in 2017, 73% in 2018 and 87% in 2021) (ÉgKýs 2016(ÉgKýs , 2017(ÉgKýs , 2018(ÉgKýs , 2021. There is no research, at least not to our knowledge, other than the aforementioned survey (Isebarn & Björgvinsdóttir 2017)where those who participated in mock elections were found to be more likely to say that they had voted in the real election -about the impacts of mock elections on young people's political engagements in Iceland. Through this paper, we hope to contribute towards narrowing this research gap as one steppingstone closer to establishing whether mock elections serve their intended purpose.

Data and research design
Following the parliamentary election at the end of September 2021, a survey was sent to students in four upper secondary schools in Iceland. One of the main aims of the survey was to examine whether electoral participation and political engagement amongst students differed depending on whether a mock election was held in their school. Four schools were selected, two of which held a mock election. Another criterion for selecting the schools was the number of students. 2 Schools with a relatively high number of students were selected with the aim of guaranteeing that the number of respondents would be sufficient in order to run statistical analyses. 3 For the purpose of this research, the schools in our sample are kept anonymous so as to avoid the stigmatisation of both non-participation of schools in mock elections and the students in those schools where mock elections were not held. Furthermore, the focus was on the possible impact of mock elections on political engagement, not on whether particular schools participated in the mock elections.
The mock elections in the two schools in our sample were held on 9 September 2021, just over two weeks prior to the parliamentary election on 25 September. According to information from the schools' offices and websites, the mock elections in both schools were part of a larger programme about the election. The programme included panel discussions about politics, debates and presentations with candidates from the political parties running in the election. Students had the opportunity to come together for these events, learn about the parties' policies and pose questions to the candidates. The mock elections then concluded the programme.

STJÓRNMÁL & STJÓRNSÝSLA
Following the parliamentary election, the survey was sent out on 9 November via the schools' offices, which then distributed it to students 18 years and older -the minimum voting age in Iceland. A reminder was sent on 18 November, and the survey ended on 1 December. The questionnaire included 17 questions where students were asked, among other things, about their electoral participation, interest in politics, whether voting mattered, whether it was a civic duty to vote, how frequently they engaged in discussion about electoral campaigns and their awareness of electoral campaigns. Table 1 presents an overview of the schools in the sample. It shows the total number of students in each school, the number of students surveyed, survey response rates and whether a mock election was held at the school. As depicted in Table 1, the response rate was quite low, differing between the schools. In the schools where a mock election was not held, the response rates were around 15% in each school, while the response rate was higher in schools with mock elections: 37.7% and 43.4%. The differences in the response rates between schools with or without a mock election can be considered a concern. If the mock elections made students more likely to respond to the survey, it might mean that a higher ratio of non-voters did not respond, which could have resulted in an underestimation of the effect of mock elections on political engagement. There is a related but different concern, which is the fact that the schools were not randomly selected to participate in the mock elections to begin with. Thus, the relations we find between mock elections and political engagement might be due to a pre-treatment effect, i.e. the possibility that in secondary schools where students were generally more politically engaged to start with are also the schools that are likely to have participated in the mock election program. If this is the case, the impact of mock elections on political engagement could be spurious and/or overestimated. Both of those concerns underline the importance of that even if we can establish links between mock elections and political engagement, more research is needed to present more confirm arguments about possible causal mechanisms.
Given the varying student numbers in the schools and the differences in response rates in schools with or without a mock election, we weighted out the data based on the total number of students in each school, and the weights are included in the last column in Table 1. This was done to give each school an equal weight in the data analysis, regardless of its number of students or the response rates for each school in the survey. Another way to weight the data is to use the number of students surveyed. Running the data analysis with either of these two weights, or unweighted, rendered similar results, and in this paper, we opted to present the data with a weight for the total number of students.
The total number of responses in our survey were 403, with one student reporting to be 17 years old, eight reporting that they were older than 25, and 43 that they were not eligible to vote at the time of the parliamentary election. As we were primarily interested in young voters who were coming of age, we excluded respondents who said that they were over 25 years old and those who reported that they had not yet reached the eligible voting age at the time of the election, when testing our hypotheses.

STJÓRNMÁL & STJÓRNSÝSLA
Mock elections, electoral participation and political engagement amongst young people in Iceland

Data analysis
In three of the four schools, more than 90% of the respondents said that they voted in the parliamentary election. In the fourth school, 80.9% said that they had voted ( Table  2). Given that turnout in the 2021 election was 80.1% overall, the electoral participation in our sample was clearly overestimated for three of the schools. Given the small sample size there were limited options for running complex statistical analyses. For this reason, we analysed bivariate associations in our data, using crosstabs and chi-square analysis to test our hypotheses. In our results, we mark statistical significance by asterisks (*) and a plus sign (+) as follows: *** = p<0.000, ** = p<0.01, * = p<0.05 and + = p<0.1. The results of the significance tests are presented in the tables A1-A3 in Appendix I. We note here that it is useful to bear in mind that due to our small sample size, it is difficult to get statistical significance, and more conclusive results might be reached analysing a bigger sample. *The total number of responses in the school was: A=199. Due to a mistake in the data gathering, the students in school D were not asked about their participation in the mock election. **The total number of responses in each school was: A=177, B=42, C=47 and D=50.

Results
Given that campaign activities are one of the major mobilisers in getting people to vote, we considered it important to first establish whether the students in the schools with mock elections were more exposed to campaign activities that could be linked with the mock election programme. This means that it was not only the act of voting itself in a mock election that served as a socialisation effect on voting; other actions were also part of the mock election programme, such as school visits from representatives of political parties and the fact that the focus in the schools was on electoral campaigns in the few days around the mock election.
Thus, we used questions asking the respondents about their exposure to different campaign activities. These activities can generally be categorized into two themes. The first is whether the students had been exposed to campaign activities that were part of the mock election program and/or demanded some type of action from students. This included whether they were visited by representatives of the political parties and, thus, were more likely to have been exposed to, and to have read, campaign materials such as campaign booklets and the parties' policy platform. The second theme was whether they had been exposed to campaign activities where they did not have to take any active action themselves or did not initiate action themselves, such as whether they saw advertisements from political parties or were contacted by them. Figure 3 lists the differences in the students' exposure to campaign activities 4 based on whether a mock election was held in their school or not (see Table A1 in Appendix I for the results of the chi-square tests). Respondents were asked to mark all that applied to them from a list of those activities. In terms of passive campaign exposure, such as whether the students saw advertisements or other information about political parties or were contacted by the representatives from a political party, the only observed difference involved exposure to advertisements on social media, with the students in the mock election schools being more likely to say that they saw these advertisements on social media. Regarding campaign exposure demanding some action from students or were part of the mock election programme, we found that the students in the mock election schools were more likely to say that they had read a campaign booklet, read a party's policy, visited a party's website and that they had received a visit from representatives of a political party at their school. Thus, the results show that the students in the mock election schools were exposed to more campaign messages and activities than the students in schools without mock elections. This is perhaps not surprising; nevertheless, it is important in terms of validating the differences in electoral participation and other types of political engagement in the next steps of our analysis.

STJÓRNMÁL & STJÓRNSÝSLA
Mock elections, electoral participation and political engagement amongst young people in Iceland Figure 3. Exposure to campaign activities by schools with or without mock elections In our analysis, we addressed five questions: whether the respondents voted in the parliamentary election, how frequently they discussed politics during the campaign, how interested they were in politics, whether they considered it a civic duty or a free choice to vote and whether it made a big difference as to who people voted for. 5 To test our first hypothesis about whether the students in the mock election school were more likely to vote (H1), we ran a bivariate analysis with a chi-square test. Even if turnout was a bit higher in the schools with mock elections (93.6%) compared to those without mock elections (90.1%), this difference was not statistically significant (Table 3).
Here it should be noted that out of the small sample size, a very low number of students reported that they did not vote in the general election (n = 10 in schools with mock elections and n = 14 in those without), which means that it is extremely hard to get statistical significance. Thus, it can be argued that our results for H1 remain inconclusive, at least until tested on a bigger sample. Regarding other types of political engagement, we observed a relationship between mock elections and political engagement in the form of frequency of political discussion (H2) and political interest (H3). Both were measured on a five-point response scale, from none to very great for political interest, and from never to many times per day for political discussion (see endnote 5 for more detail). To ease the interpretation of the results, we grouped together those who had a very great or great interest in politics and those who discussed politics daily or more often. These results are presented in Figure 4. The full table (A2) with uncategorized results is presented in Appendix I. Figure 4 shows that the students in the mock election schools were more likely to say that they discussed politics daily or more often (57.6%) than students in schools without mock elections (44.1%). In terms of political interest, we found that the students in the mock election schools were more interested in politics, with 32.3% of them saying that they had a very great or great interest in politics, whereas only 23.1% of those in schools without mock elections reported similar interests.

STJÓRNMÁL & STJÓRNSÝSLA
Mock elections, electoral participation and political engagement amongst young people in Iceland In terms of other types of political engagement, H4 (whether it was considered a civic duty or free choice to vote) and H5 (whether students considered that who people voted for made a difference) were not supported with statistical significance. Both were measured on a five-point response scale; civic duty from 'that it is entirely own free choice to vote' to 'that it is a civic duty to vote', and whether it makes a difference who people voted for from 'won't make any difference' to 'that it can make a big difference' (see endnote 5 for more detail). Again, to facilitate interpretation, Figure 5 presents our results based only on those who either had a strong opinion that it was a civic duty to vote or that it made a big difference who people voted for (see full table, A3, in Appendix I). Eyeballing Figure 5, we see that the directions of the relations are in line with our hypotheses that students in mock election schools were more likely to consider it a civic duty to vote or that voting made a big difference. This could indicate that young voters in schools where a mock election was held, have a greater sense of civic duty to vote or that voting matters, but a larger sample size is needed to confirm or reject that suspicion. Figure 5. A civic duty to vote and whether who people vote makes a big difference based on whether mock elections were held In sum, our results showed that the impact of mock elections on political engagement was not clear-cut, at least not in the four schools surveyed. Mock elections seemed to be related to political engagement in terms of frequency of political discussion and interest in politics but not directly in terms of whether students participated in a real election, their opinion about whether voting made a difference or whether it was considered a civic duty to vote. As already noted, it is not surprising given the small sample size that some of our results were not statistically significant. The results could indicate that there are some differences between students in schools where mock elections were held and where they were not, but we would need a larger sample size to confirm this -especially in the three cases where the differences were negligible. In the next section, we discuss our results in greater detail, their implications and possible avenues for future research.

Discussion
In this paper, we have analysed the possible socialisation impact of mock elections on political engagement amongst young voters. For the purpose of our research, we hypothesised that mock elections were positively related to both electoral participation and other types of political engagement, including the frequency of discussing politics, interest in politics, whether it was considered a civic duty to vote and whether voting made a difference.
Contrasting schools in which mock elections were held with those without mock elections revealed no major difference in electoral participation amongst the students from the four schools surveyed. This arguably indicates that mock elections, at least in

STJÓRNMÁL & STJÓRNSÝSLA
Mock elections, electoral participation and political engagement amongst young people in Iceland the four schools in our sample, do not encourage young people to vote in real elections. It is noted that the number of students who said that they did not vote was very low and the sample is rather small; and as such, it can be argued that our results regarding whether there was a direct link between a school mock election and voting in general elections are inconclusive. This is a clear avenue for future research. Furthermore, concerning other types of political engagement, we found no differences in whether it was considered a civic duty or free choice to vote or whether it made a difference who people voted for. Again, a bigger sample might show different results. However, the students in the mock election schools revealed that they discussed politics more often and that they were more interested in politics compared to students in schools where mock elections were not held. Even if we cannot argue about causality, that the mock elections increased students' political engagement in terms of political discussion and political interest, this could indicate that the mock elections and the programme that accompanied them had a socialisation impact in relation to these factors.
It can tentatively be argued that mock elections have an indirect effect on electoral participation via political discussion and political interest, given that those who discuss politics more often and are interested in politics are also more likely to vote (e.g. Fieldhouse et al. 2007). It is not only the expressive participation itself which is of importance here; it is also that those who engage in frequent political discussion and are interested in politics are more likely to be well or better informed about politics and their democratic rights compared to those who are less interested and engage in less frequent discussion. The more young voters are informed about politics, the greater the likelihood of them overcoming the cost of becoming informed, which is one of the major barriers facing many first-time voters. If mock elections are successful in raising interest and discussion among young voters, they might be a steppingstone towards both helping them to be better informed about politics and increasing the possibility that they will make a habit of voting.
Whether our results would hold if the response rates were higher, if there were more schools in the sample and if panel data were collected is a clear avenue for future research. The use of panel data would present an opportunity to better establish the effect of mock elections on voting. Young voters could for example be asked in three waves about their intention to vote prior to a mock election, their participation in a mock election and their intention to vote in a real election and then whether they actually voted in the real election. This would also give a chance to further analyse whether mock elections have an impact on other forms of political engagement such as political interest, political knowledge and internal political efficacy. Sampling more schools or all upper secondary schools in Iceland for this purpose would open a range of possibilities to include other important factors that might explain differences in political engagement. These factors could include the general profile of the socio-economic status of the students' household, parental voting and the general socio-economic status of young people's environment (e.g. town or municipality). It would also open up the possibility to analyse whether the impact of mock elections is dependent on the way in which they

STJÓRNMÁL
& STJÓRNSÝSLA are carried out in each school, for example, how much focus was on the mock election within the schools, how informed students were about the mock election occurring and whether the mock election was part of a bigger programme or discussion within the schools.
In conclusion, as mock elections play a role in political engagement in terms of political discussion and political interest, then the aims of mock elections in upper secondary schools in Iceland have been partly fulfilled, at least in the four schools in our sample. Even if there is no direct link between whether mock elections are held and participation in real elections, we underline here that it does not mean that mock elections do not have merit. They can have other important effects concerning political engagement that are worth pursuing. Carrying out mock elections in upper secondary schools means that students get the chance to practice democracy in a safe environment with their peers. Mock elections can serve the purpose of increasing young people's awareness of their democratic society, whether this is via political discussion and political interest or through the practice of making an informed decision about their vote choice, which should result in a more actively engaged citizenry in our democratic society.

Endnotes
1 It should be noted that it is not necessarily the age itself that is a resource; many of the resources that people use to vote, such as education and participation in social networks, are reflected in where people are in the life cycle -with younger people often lacking in these areas due to their age. 2 The schools were also selected based on whether they were in the capital area (two schools) or not (two schools) as the authors considered it important to include both due to possible differences in students' background in terms of location. However, in our analysis, we found no differences in participation and political engagement based on the schools' location. Therefore, we focused on the impact of the mock elections themselves and not on the location of the schools. 3 It would have been ideal to sample all students in all upper secondary schools in Iceland for the purpose of this research. The reason that this was not done was that the survey was not funded by any other means than the authors' own time. 4 The respondents were presented with a list of campaign activities and asked to mark those that they had been exposed to. 5 Questions asked: Electoral participation: 'Did you vote in the Althingi election on September 25th?': Yes/No. Frequency of political discussion: 'How often, if anytime, did you discuss domestic politics with other people during the election campaign in September (e.g. with family or friends, in school, on social media)?': many times per day; daily; a few times a week; less than a few times a week; never. Political interest: 'Do you consider your interest in politics very great, great, some, little, or are you not interested in politics at all?' Civic duty to vote: 'Some people say that it is one's own free choice whether he or she votes in elections for Althingi. Others say that the public has a civic duty to vote in elections. What do you think? Please respond by using a scale from one to five, where one means it is entirely one's own free choice if he or she votes in elections, and five means that it is a civic duty to vote'. Voting makes a difference: 'Some people say that no matter who people vote for, it won't make any difference to what happens. Others say that who people vote for can make a big difference to what happens. Using the scale from one to five, where one means that voting won't make a difference to what happens and five means that voting can make a difference, where would you place yourself ?"

STJÓRNMÁL & STJÓRNSÝSLA
Mock elections, electoral participation and political engagement amongst young people in Iceland