University of Amsterdam
Graduate School of Communication
- Master Thesis -
Political Communication
Campaigning electoral participation of
EU immigrants in the Netherlands:
A Theory of Planned Behavior approach
Student: Kolja Siegmund Student number: 5673992
Supervisor: Dr. Yphtach Lelkes
Abstract
Research on electoral turnout was able to illustrate structural and rational choice determents of voting, yet research on campaigning electoral participation mainly focuses on first level elections. Therefore, it is doubtful if effects are applicable to campaigning participation at local elections. This paper focuses on a social-cognitive alternative to explain vote intention, the extended Theory of Planned Behavior model. Using two studies, this research first illustrates the fit of the model to explain vote intention of EU immigrants in the Netherlands, before it tests significant predictors for causality. While the fit of the model is confirmed, no causal main effects on vote intention were found. Yet it is illustrated that interaction effects of the model’s factors can enhance but also lower the vote intention of non-voters. Finally, the results are discussed regarding their implication for campaigning electoral participation.
Introduction
In several countries, non-citizen voting rights are repeatedly discussed (Bauböck, 2005). While opponents argue that only full citizenship can entitle a person to vote, advocates argue that political participation enhances integration (e.g. Gidengil, 2013; Munro, 2008; Tam Cho, 1999). Furthermore, research shows that exclusion of large social groups from voting can influence election results, at least in large cities (Hajnal, & Tounstine, 2005). However, research directed at immigrant voting generally shows low turnout rates (e.g. Gidengil, 2013; Gordon, 1970; Just, & Anderson, 2012; Tuckel, & Maisel, 1994; Xu, 2005). The European Union (EU) is one of the largest political entities that grant reciprocal local voting rights to its
citizens. Next to a better integration of minorities, the policy was meant to enhance attitude towards the EU (Groenendijk, 2008). Yet, electoral participation is poorly promoted, and, although no official turnout rates have been assessed, it was hypothesized that the turnout rates of EU immigrants are very low (European Commission, 2012; Van der Heijden, & Van Heelsum, 2010). Problems are mostly expected in countries with prospering urban areas that are attractive to EU
immigrants, like the Netherlands (Favell, 2010).
While much is known about vote participation in general elections in the US (e.g. Pattie, & Johnston, 1992; Hanks, & Grofman, 1998), results from these studies might not be applicable to the turnout of EU immigrants at local elections in the Netherlands. First, electoral systems differ greatly between the US and the
Netherlands (Myerson, 1993), which might have an effect on relevant antecedents of voting intention. Second, general elections differ greatly from local elections, e.g. in the availability and amount of funding for campaigns. Third, target populations differ not only in nationality, but whereas only national citizens are allowed to vote in general elections, this study focuses on non-citizens. In addition, many of these studies are likely ineffective for the development of campaigns. Many turnout research focuses on structural differences between political and voting systems of states (e.g. Blais, & Dobrzynska, 1998; Franklin, & Hirczy, 1998; Blais, 2006; Knack, & White, 2000; Geys, 2006) or explain voting by rational choice approaches (e.g. Edin, Gelman, & Kaplan, 2007; Ostrom, 1997). Those approaches help in explaining behavior but showed ineffective as foundation for the development of persuasive messages regarding vote motivation (Blais, & Young, 1999).
Furthermore, prior research often focused on external societal factors
social-cognitive factors in order to cause behavioral change at an individual level (Flay, & Burton, 1990). The aim of this study is therefore to test a social cognitive theory model, the Theory of Planned Behavior (TPB), for predictive and causal effects on EU immigrant voting at local elections in the Netherlands. While prior research illustrated the applicability of the theory to vote intention in general (e.g. Hansen, & Jensen, 2007; Netemeyer, Burton, & Johnston, 1991; Peterson, 2012), these researches, again, focused on general elections or US elections, only. However, the model was chosen because it demonstrated its good power to serve as predictor of behavior, but also as foundation of effective campaigning, e.g. in health
communication (e.g. Armitage, & Conner, 2001). Furthermore, it includes the factor ‘social norm’ (or ‘civic duty to vote’), which has been proven effective (e.g.
Bengtsson, 2004; Blais, 2006; Gerber, & Rogers, 2009; Knack, 1992; Opp, 2001). Additionally, TPB explains how people form intentions to perform a behavior by a set of clearly defined variables. However, research on how these variables can be
addressed in campaigns regarding vote intention was not yet conducted. Therefore, this study contributes to previous research by testing the
universality of TPB’s applicability by applying it to the prediction of vote intention, and by creating new insights about its applicability to interventions aimed to enhance vote intention. To reach this aim, two studies will be conducted. In the first study, a survey design will be employed in order to investigate whether TPB is applicable to the prediction of vote intention. In the second study, an experimental design will be employed in order to test whether relevant predictors of vote intention can be used to create effective interventions aimed at increasing vote intention among
Theoretical background
EU immigrant voting at local elections
In 1992, EU member states agreed in the Maastricht Treaty to grant active local voting rights to EU citizens who live in another EU member state than their home country (Groenendijk, 2008) and in 2007, the Treaty of Lisbon finally granted those EU immigrants full suffrage. Yet, while some countries request EU immigrants to get enlisted on the electoral register, the Netherlands only demand registration at a Dutch municipality.
It was argued that granting local voting rights should foster positive attitudes towards the EU (Bauböck, 2005). Furthermore, the necessity to include new and fast developing societal sentiments (Favell, 2003; Favell 2010) in local decision making (Groenendijk, 2008) was acknowledged. Yet, electoral participation of EU
immigrants is poorly advertised, although the groups’ turnout rates are hypothesized to be very low (Van der Heijden, & Van Heelsum, 2010). However, the issue is rarely investigated. Raw figures of participation are only available for a few European countries that require voters to register (European Commission, 2012). The data indicate that about 50% of all EU immigrants in Spain, and about 30% in Ireland, respectively, were registered to vote in 2010, the numbers for Belgium, Cyprus, Greece, Italy, Luxembourg and Portugal range between 5% and 13%. Yet, as these numbers only apply to the registration on the electoral roll, it is reasonable to expect even lower turnout rates.
Especially for states like the Netherlands that economically benefit from immigration from other EU states (van Dalen, 2001) and expect a future growth of EU immigration (Nicolaas, 2009), new challenges regarding general and political integration are expected, foremost in metropolitan areas (Favell, 2010). While EU
immigrants only accounted for 3% of all eligible voters in the Netherlands in 2010 (CBS), numbers are much higher in urban areas. For instance, EU immigrants accounted for 15% of the electorate in Amsterdam in 2012 (O+S Amsterdam, 2013). Nonetheless, no study has yet investigated possibilities to campaign EU immigrant voting at local elections.
Research on electoral participation
Prior research on vote participation mainly focuses on elections with nation wide impact using two dominant frameworks. First, cross-sectional experiments in the context of economical and structural differences try to explain differences in voter turnout between countries based on factors such as political and electoral systems (e.g. Blais, & Aarts, 2006; Blais, & Dobrzynska, 1998; Franklin, & Hirczy, 1998), voluntary or compulsory voting (e.g. Blais, 2006; Franklin, 2002), structural obstacles (Knack, & White, 2000), or population size and concentration (Geys, 2006). These studies might help to re-structure political and election systems more favorable to voters, which might enhance turnout. However, these structural differences can hardly be changed through campaigning and individual factors are largely ignored in these studies.
The second dominant approach is based on rational choice theory, which argues that internal estimates between costs and benefits decide whether a behavior is conducted or not (Feddersen, 2004). Research in the field showed that voters’
motivation was often driven by the intention to actually influence politics (e.g.
Aldrich, 1993; Matsusaka, & Palda, 1993). Yet, as the impact of one vote is extremely low, it has been argued that costs of voting are much higher than its benefits (e.g. Bendor, Diermeyer, & Ting, 2003; Brams, 1976; Green, Shapiro, & Shapiro, 1994;
Riker, 1958; Krishna, & Morgan, 2012). The phenomenon, called the paradox of voting, was further illustrated by an experiment by Blais and Young (1999), in which information over the paradox was provided to participants of some conditions. They concluded that informing on the paradox lowered the turnout by 7%. The most consistent results from rational choice research on electoral participation were found regarding social norms or civic duty to vote. For instance, Knack (1992) illustrated on national-level data that the perception of voting as civic duty enhanced electoral participation, especially if non-voters have to expect social sanctions. This was confirmed by a survey in Quebec (N = 1,023) and British Columbia (N = 998), which revealed civic duty as the strongest predictor of electoral participation (Blais, & St-‐‑ Vincent, 2011). Hereby, both subjective as descriptive norm appear to influence behavior. Gerber and Rogers (2009) were able to demonstrate in an experiment with 1,715 participants that news messages, which forecast high turnout rates were likely to positively influence turnout rates, while messages with negative forecasts lowered turnout rates. Additionally, Karp and Brockington (2005) illustrated on voter studies data from five countries that social desirability of voter turnout is higher in countries with high turnout rates. Therefore they argued that the mere presence of a high turnout serves as a social norm.
However, it is doubtful whether these, mainly US-focused, researches are applicable to campaigning EU immigrant voting at local elections. Differences in voting systems were for example found to affect citizens’ attitude towards voting (Myerson, 1993) and generalization between voting systems regarding antecedents of vote motivation should therefore not be made without caution. In addition, many studies employ quite expensive campaign strategies, like personalized phone
King, 2006). These strategies might be relevant for campaigns in elections in which candidates dispose large monetary budgets (Abramowitz, 1991; Ansolabehere, Figueiredo, & Snyder, 2003) but might not be applicable to much less extensively funded local elections in the Netherlands. Strategies to enhance local voting participation among EU immigrants should therefore focus on standardized rather than personalized messages. Furthermore, even though it is highly likely that
normative beliefs have the potential to influence local voting, subjective norms often interact with other internal factors (e.g. Aarts, & Dijksterhuis, 2003; Bonfield, 1974; Schwartz, 1973). Hereby, especially internal factors relevant for campaigns aimed at behavioral change should be considered. Consequently, as it was argued by
Persuasion theory that attitudinal change is a pre-condition for behavioral change (McGuire, 1989), attitude should also be considered as a relevant construct for developing of effective interventions regarding the enhancement of vote intention.
The extended Theory of Planned Behavior model
A subjective-cognitive model that includes subjective norm and attitude as determents of behavior is the extended Theory of Planned Behavior (Ajzen, 1991), which has proven its effectiveness to determine factors that motivate the performance of a behavior and to address audiences with standardized messages to achieve behavioral change (e.g. Armitage, & Conner, 2001; Conner, & Armitage, McEachan, Conner, Taylor, & Lawton, 2011). The Theory of Planned Behavior (TPB) prompts that behavior is directly predicted by the intention to perform a behavior. However, intention is formed through direct effects as well as through the interaction between three main determents: Attitude towards the behavior, subjective norm, and perceived behavioral control, which are derived from individual’s salient behavioral beliefs, normative beliefs, or control beliefs, respectively (Ajzen, 1991) (see also figure 1 in
Appendix C for an overview). Furthermore, two additional determents were added: past behavior (e.g. Ajzen, 2011; Hagger, Chatzisarantis, & Biddle, 2002; Ouellette, & Wood, 1998) and (political) self-efficacy (e.g. Ajzen, 2011; Rhodes, & Courneya, 2003).
First studies, adapting TPB to turnout research, have shown a general reliability of the model as a predictor of individual vote intention. For instance, Netemeyer, Burton, and Johnston (1991) found that TPB variables predicted vote intention in a state gubernatorial election primary and Hansen and Jensen (2007) illustrated that attitude towards voting had a significant effect on (Danish) voters’ party choice, yet they failed to illustrated significant effects of subjective norm or perceived behavioral control. However, Peterson (2012) illustrated the predictive effectiveness of attitude towards the behavior, subjective norm, perceived behavioral control, and past behavior on vote intention in United States general elections. Yet, again, the transferability of the results to the relative target group of this study is not conclusive given that only voting of national citizens in elections with national-impact were considered in these previous studies. Local elections are often seen as second-order elections, and respective effects in mobilization can therefore largely differ from effects found in general elections (e.g. Rallings, & Thrasher, 2005; Reif, & Schmitt, 1997). Furthermore, it has often been illustrated that immigrant groups differ essentially from native populations in regard of political participation (e.g. Logan, Darrah, & Oh, 2012; Just, & Anderson, 2012). Finally, the studies mentioned above focus on the prediction of vote intention, but do not test TPB’s applicability to the development of campaigns regarding electoral participation. Therefore, the aim of this research is to test whether TPB predicts vote intention of EU immigrants in the
development. In the following sections, elements of the extended model of TPB are explained in relation to campaigning local voting.
Elements of the extended Theory of Planned Behavior model
Following TPB, attitude towards behavior is derived from behavioral beliefs, thus an individual’s internal estimates whether a desired outcome could be achieved through the performance of a behavior (Armitage, & Conner, 2001). It was argued that “the degree to which a person has a favorable or unfavorable evaluation or appraisal of the behavior in question” (Ajzen, 1991, p. 188) strongly affects the person’s intention to conduct the behavior. Consequently, campaigning electoral participation should assess target groups’ salient attitudes towards voting at the respective election, and focus on change in salient attitudes towards the behavior. Following Petersen (2012), an estimate of attitude towards a behavior can be achieved by measuring multiple salient beliefs about the behavior in question.
Subjective norm
Normative beliefs, thus the subjective expectations whether the performance of a behavior is honored or refused by relevant social groups, are argued to have a strong influence on a person’s behavior (Ajzen, & Madden, 1986). They were illustrated to produce social pressure to actually perform or not to perform a behavior. Therefore, they serve as subjective norms, which were shown to have a large influence on vote intentions by research on rational choice. Karp, and Brockington (2005) showed that social desirability of voter turnout is higher in countries with high turnout rates and that the mere presence of high turnout rates can serve as social pressure. These results are in agreement with studies that confirm the effect of perceived civic duty (e.g. Blais, & St-Vincent, 2011; Knack, 1992;) and illustrations that norms are formed
through perceptions of social reality, which are influenced by expected turnout rates (e.g. Bengtsson, 2004; Gerber, & Rogers, 2009). Accordingly, research on
campaigning electoral participation should include a measure of subjective norm that captures expected turnout rates. In addition, not only the predictive value of
subjective norm should be tested but also its applicability to the development of campaign.
Perceived behavioral control
Perceived behavioral control (PBC) was derived from the factor self-efficacy of social-cognitive theory (Ajzen, 1991). PBC is derived from control beliefs, which depend on a person’s estimate of his or her abilities to perform a behavior
successfully (Armitage & Conner, 2001). It has been argued that a high perceived control on a behavior enhances a person’s likelihood to perform the behavior and vice versa (Ajzen & Madden, 1986). PBC is directed only at the mere functional behavior, and not to related concepts (Ajzen, 1991). Assessing and campaigning PBC should therefore focus on behavioral aspects of voting, such as voting itself, or information gathering on the behavior or on candidates and parties, but not on mental activities like conducting a deliberated vote.
Political self-efficacy
Armitage and Conner (1999) argued that the principle definition of self-efficacy in terms of control beliefs does not allow distinguishing between internal factors (such as confidence) and external factors (such as abilities and availability). In their
experiment, the authors illustrated that self-efficacy measures, focusing on motivation and confidence, are structurally different from PBC measures, focusing on abilities, in
predicting weight loss. This finding was confirmed by another experiment, which illustrated different effects of PBC and self-efficacy on the intention to engage in physical exercises (Rhodes, & Courneya, 2003).
This understanding of self-efficacy is an agreement with the concept of political self-efficacy, which was defined in terms of a person’s perceived ability to influence politics (e.g. Madsen, 1987; Caprara, Veccione, Capanna, & Mebane, 2009). It was demonstrated that political self-efficacy results in political and electoral participation (Veccione, & Caprara, 2009). Therefore, research on campaigning local voting should include measurements of political self-efficacy. However, it is doubtful that EU immigrants in the Netherlands perceive a strong ability to influence local politics, given their minority position. Accordingly, direct measures of political self-efficacy are likely to fail. Yet, Bandura (1982) argues that issue knowledge is a driving force in acquiring self-efficacy in general, which was frequently illustrated for political self-efficacy, too (e.g. Veccione, & Caprara, 2009); Newhagen, 1994;
Holbert, Lambe, Dudo, & Carlton, 2007). Research on campaigning electoral participation of EU immigrants at local elections should thus focus on participants perceived knowledge on the host country’s politics, compared to the country’s general population, as an indirect measure the groups’ of political self-efficacy.
Past behavior
Several researches have shown that past behavior influences the effect of variables of TPB (e.g. Ajzen, 2011; Hagger, Chatzisarantis, & Biddle, 2002; Oulette, & Wood, 1998). Regarding voting, Karp and Banducci (2008) were able to demonstrate that past voting enhanced the likelihood of future voting in the same elections in a very large (N = 33,799) multi-wave comparative study of electoral systems. The results
were confirmed in a large (N = 8,466) longitudinal study on child development (Denny, & Doyle, 2009). The authors concluded that past voting enhances the likelihood of future voting at the same election by 13%. Therefore, studies on campaigning voter turnout should control for effects of past behavior, or in this case for the effect of voting at the last local elections in the Netherlands.
Hypotheses
As it was illustrated above that the TPB was successfully applied to the prediction of voting behavior on an individual level, and as causal effects of the model’s variables subjective norm and past behavior were shown to affect vote intention, the model is assumed to predict vote intention for EU immigrants in the Netherlands, too. Furthermore, as it was shown that the model is a reliable approach for social
campaigning in other areas of social science, e.g. in health communication, the model is assumed to offer a framework for campaigning electoral participation in general as well as for EU immigrants. Accordingly, it is hypothesized that (H1) all factors of the TPB model are related to vote intention (H1a) and have causal effects (H1b).
Furthermore, it was argued the combination of several TPB variables should have the most powerful effect on behavior (Ajzen, 1991). Herby, past behavior and subjective norms were found to have large effect on vote intention in previous research.
Consequently, it is hypothesized that (H2) of all significant effects, the largest
predictive (H2a) and causal effect (H2b) will be an interaction of all variables of TPB, followed by effects of past behavior and subjective norms, while attitude, political self-efficacy and perceived behavioral control are expected show the smaller effects. Furthermore, to test the model’s reliability as a campaigning approach, it is
hypothesized that (H3) addressing significant predictors of vote intention through messages will affect the targeted predictors.
Two studies are conducted to test these hypotheses. Study 1 will employ a survey design in order to determine the factors correlated to EU immigrants voting behavior at local elections in the Netherlands. Study 2 will determine whether factors that are significantly related to turnout have a causal effect and an experimental design will therefore be applied.
Study 1
A survey was conducted to investigate the predictive value of the elements of TPB for EU-immigrants vote intention. Respondents were recruited at social network groups, targeting expats in the Netherlands. A total of 559 respondents participated in the survey, with a drop-off rate of 30%. After selecting respondents from EU countries, and controlling for residence at a Dutch municipality, 253 cases were included in the analysis. 58 respondents had voted at prior local elections. A majority of 198
respondents were female, yet it is not likely that this disproportion affect the results. The age ranged between 18 and 66, with a mean of 30.9 years. Students were a minority (N = 64) compared with fully employed (N = 150) and unemployed (N = 39) respondents, nevertheless 209 respondents reported to have achieved an
undergraduate or higher degree.
Measurements
Vote intention was measured on a 7-point Likert scale (M = 0.44, SD = .35), asking for the likelihood of respondents’ participation, if local elections were held during the next week. As all following scales, vote intention was recoded into a scale between
‘0’ (low or negative) and ‘1’ (high or positive). Attitude (Cronbach’s alpha = .90; M = 0.62, SD = .21) was measured on a 4-item scale (Good; Important; Pleasant; Positive). Perceived behavioral control (Cronbach’s alpha = .89; M = 0.39, SD = .26) was measured on a 4-items scale (Voting local in the Netherlands; gathering information on local voting; gathering information on parties and candidates; influencing politics), on which respondents indicated the perception of how hard (‘0’) or easy (‘1’) the performance of one of the behaviors would be. Subjective norm (Cronbach’s alpha = .74; M = 0.24, SD = .20) was measured on a 3-item scale (Other expats; friends, family and colleagues; people like me). All items were measured as the respondents’ estimated turnout rate of respective social groups. Political self-efficacy (M = 0.21,
SD = .23) was measured on a 7-point Likert scale, on which respondents indicated
how much they thought to know about Dutch politics, compared to the average of Dutch citizens. Dutch news consumption was measured on an 8-point scale (Never; 1 day; … ; 7 days), asking respondents on how much days they follow Dutch news, during an average week. Past behavior was measured by asking if respondents had voted last local elections in the Netherlands (see Appendix D for an overview of all scales).
Results and discussion
A hierarchical multiple regression, controlling on the respondents’ gender (ns), time of residence (b = .010, S.E. .00, p = .020), and exposure to Dutch news (b = 0.38, S.E. = .06, p < .001) confirmed that the extended theory of panned behavior model serves as predictor for EU immigrants’ intention to vote local elections in the Netherlands (see also table 1 in Appendix A and figure 2 in Appendix C). Of the model’s three main variables, the respondents’ attitude (b = 0.40, S.E. = .09, p < .001) and
subjective norm (b = 0.29, S.E. = .09, p = .002) on voting local elections in the
Netherlands were identified as significant predictors for vote intention. Yet, perceived behavioral control (ns) was not. Therefore, the effect will not be controlled on
causality. However, both additional variables of the extended model, political self-efficacy (b = 0.27, S.E. = .09, p = .003) and past behavior (b = 0.24, S.E. = .05, p < .001) were identified as predictors of vote intention. The significant effect of political self-efficacy compared to the insignificant effect of PBC confirms that both measures focus on different concepts. No interaction effects were evident. The control variables showed a R2 value of .245, the model including the factors of TPB showed a R2 value of .540 and thus accounted for more than 50% of the variance. However, even though PBC was not shown to have a significant effect, H1a that postulates that all factors are related to vote intention was partly accepted, as all other factors were illustrated as determents of vote intention. However, H2a states that an interaction effect of all variables would have the largest predictive effect, followed by subjective norms and past behavior. Yet, the results do not indicate any interaction effect between the variables. Furthermore, attitude towards voting had by far the largest effect (b = 0.40), compared to subjective norm (b = 0.29), political self-efficacy (b = 0.27) and past behavior (b = 0.24). Therefore, H2a was not accepted.
Finally, a multiple regression of the items of both significant scales revealed that of all items of the attitude scale, the item ‘important’ (b = 0.50, S.E. = .09, p < .001) was the only direct predictor of vote intention (see also table 2 in Appendix A). Additionally, the item ‘people like me’ (b = 0.04, S.E. = .01, p = .001) was the only direct predictor on the subjective norm scale. The results thusly suggest that
manipulations targeting at attitude change should focus on the importance of local voting, while normative statements on expected turnout rates should focus on
statements about social groups that are comparable to receivers of the statement. However, as the duration of residence in the Netherlands and Dutch news
consumption were identified as significant predictors of vote intention, the experiment should control on both variables, too.
Study 2
Method
To reach EU immigrants in the Netherlands, social media groups that focused either on expats or single European nationalities living in the Netherlands were identified before data selection. After consulting the administrators of each group, one up to three posts were placed in each group, depending on the size of the group and the lifetime of a post. The posts explained the general motivation as ‘political integration and participation’, to give possible participants a sense of the topic, without hinting too much content of the online experiment. As an incentive, each participant had the chance on winning one of three 10€ gift-cards. In total, 574 participants participated in an experiment in a 3 (High Knowledge; High Social Norm; Control) x 3 (Low Importance; High Importance; Control) factorial design. Of all participants 128 (about 22.8%) dropped out, most frequently during the knowledge test. 57 participants were excluded for having a Dutch nationality or for not having a EU-nationality.
Furthermore, 49 participants that were not registered in the Netherlands, 9 participants that reported to understand the texts and questions of the experiment ‘not at all’ or ‘almost not at all’, and 24 participants that reported a very high or high likelihood of leaving the Netherlands within the next 12 month were excluded. Finally, responses from 295 participants, from all EU member states besides Luxembourg and the Netherlands were analyzed (see Appendix G for an overview). The sample consisted
of 213 females and 82 males, and 103 participants who had voted at the last local election. The age of participants ranged between 19 and 72 years with an average age of 32.4 years. While only 73 participants reported to be a student, compared with employed or self-employed (N = 172) and unemployed participants (N = 50), the majority of participants reported an academic background with either an
undergraduate (N = 53) or a graduate or higher degree (N = 186). Much less participants reported primary school (N = 2), secondary school (N = 34), or a professional training (N = 20) as highest level of education.
Procedure
First, participants had to consent with the conditions of participation, answered selection questions regarding their nationality (Dutch, other EU nationality, or other nationality), and confirmed that they lived in the Netherlands. Following, respondents were asked for demographical information (gender, age, country of origin,
employment, education level, duration of residence in the Netherlands, and their consumption of Dutch and non-Dutch news related to the Netherlands). The aim of the questionnaire, next to measuring the control variables duration of residence and Dutch news consumption, was to be perceived as an attempt of profiling participants, to which was referred back in the first manipulation. Afterwards participants were asked to complete a knowledge test, which consisted of nine questions with each two answers regarding Dutch politics. Both, the order of questions as well as the order of answers was randomized. The questions ranged from rather easy questions (such as the name of the Dutch prime minister) to rather specific questions (such as the
number of seats of the Dutch parliament), to make sure that participants were unlikely to know for sure whether they answered a question correct or false (see Appendix E
research material of study 2). Next, participants were randomly assigned to one of three manipulations in the knowledge-norm condition. The control group (n = 95) was asked to continue with the second part of the experiment, while both, the high
knowledge group (n = 104) and the high social norm group (n = 96). The knowledge group read the message that their answers were compared to prior studies, and that they would know more on Dutch politics than 76.8% of the Dutch population. The norm group read the message that their profile was compared to results of prior studies and that 76.8% of comparable persons would regularly vote local elections in the Netherlands.
Next, in the attitude condition, participants were asked to read a newspaper article carefully, and were randomly assigned to one of three stimuli. The articles focused the importance of local voting in the Netherlands, as the perception of importance had the largest single predictor of all items of the attitude scale used in study 1. However, the low attitude group (n = 97) read an article, which stated that local voting was not important; the high attitude group (n = 95) read an article that stated a high importance of local voting; the control group (n = 103) read an article that merely summed up facts on local elections. All articles contained 118 words and were highly similar in their structure and argumentation. Additionally, all articles contained a citation of a fictive head of political science at University of Amsterdam, to present the information as relevant and reliable en to enhance persuasive effects (Martin, Lang & Wong, 2003). To enhance experimental reality, the articles were manufactured as photographs of an actual newspaper, by using an online application1. After seeing the manipulation, participants had to fill out a questionnaire to measure the depended variables. Participants were furthermore asked about past voting,
whether they were registered in the Netherlands, their estimated likelihood of leaving the Netherlands within the next 12 month, as well as their understanding of questions and texts of the experiment. Finally, participants were debriefed on the manipulations, the aim of the research, and their right to refuse further participation in the study.
Selection variables likelihood of leaving and understanding experiment
Respondents had to report how well they understood the questions and texts of the study (M = 6.25; SD = 1.257), on a seven point Likert scale that ranged between ‘not at all’ (‘1’) and ‘very well’ (‘7’). The variable was recoded as ‘0’ for participants, which reported a bad understanding (values ‘1’ and ‘2’), and as ‘1’ for participants, which reported at least a moderate understanding of the questions (values ‘3’ up to ‘7’). Additionally, the individual’s relevance of the topic was measured, as
participants had to report their likelihood of leaving the Netherlands within the next 12 months (M = 2.67; SD = 2.09), for which a 7 point Likert scale was used ranging from ‘very unlikely’ (‘1’) up to ‘very likely’ (‘7’). Contrary to the understanding of the survey, all respondents that reported more than a moderate likelihood (values ‘6’ and ‘7’) were coded as ‘0’ for exclusion, and those who reported at most a moderate likelihood of leaving the Netherlands were coded as ‘1’ for inclusion.
Dependent variables
Attitude towards local voting was measured on a four-item scale (Good; Important; Pleasant; Positive) that showed a good reliability (Cronbach’s Alpha =0.90; M = 0.66;
SD = .22). All items were measured on 7-point Likert scales and recoded on a scale
between ‘0’ (low attitude) and ‘1’ (high attitude). As the items of the first study’s attitude scale proved to be reliable, they items were re-used for the experiment.
Comparable, self-efficacy (M = 0.28; SD = .25), which was measured self-reported as political knowledge on Dutch politics, and vote intention (M = 0.56; SD = .36), both used scales from study 1. Perceived social norm (M = 0.41; SD = 0.23) was measured as the participants’ estimated percentage of individuals ‘like me’ actually vote local elections in the Netherlands, and recoded on a scale between 0 and 1, as this measure proved effective in study 1 (see Appendix for all scales of study 2).
Measurement of control variables
Participants were asked to report the year that they moved to the Netherlands on a drop down list. Furthermore, the variables gender (‘1’ = female; ‘2’ = male) and past behavior (‘1’ voted at last local elections in the Netherlands; ‘0’ non-voter) were measured dichotomous, while the knowledge on voting rights that participants possessed in the Netherlands was measured with a multiple answers (I have the right to vote Dutch local elections; Dutch regional elections; Dutch general elections; European Parliament (EP) elections). While EU immigrants are also allowed to vote EP elections in their country of residence, only right answers regarding voting local elections were recoded as ‘1’ correct or ‘0’ incorrect. The interval variable Dutch news consumption, which showed a strong influence in study 1, was also included (M = 0.17; SD = 0.18) and was measured as the participants’ estimated number of days (Never; 1 day; … ; 7 days) that Dutch news were followed, during a typical week, and recoded into a scale between dichotomous scale, indicating a low (‘1’ = following Dutch news never, one or two days a week), moderate (‘2’ = three to five days) or high (‘3’ = 6 or 7 days) amounts of Dutch news consumption.
Using ANOVAs to control on the even distribution of the control variables past voter (Attitude condition: F (2, 295) = 2.50, p = .084, η2 = .02; Knowledge-norm
condition: F (2, 295) = 0.97, p = .382, η2 = .01), knowledge on voting rights (F (2, 295) = 2.96, p = .053, η2 = .20; F (2, 295) = 2.10, p = .125, η2 = .01), and Dutch news consumption (F (2, 295) = 0.94, p = .394, η2 = .01; F (2, 295) = 0.03, p = .968, η2 = .00) showed that they were evenly distributed between the conditions. Therefore randomization was assumed successful.
In a second selection process, past voters were selected only, as the foremost objective of this study is to explore factors facilitating behavioral change through standardized messages. Therefore, 103 cases were excluded from analysis, leaving a total of 192 responses. Due to the change of the sample, ANOVAs were used to control the even distribution of control variables. Analysis showed that knowledge on voting rights (F (2, 192) = 2.91, p = .057, η2 = .03; F (2, 192) = 2.71, p = .069, η2 = .01) was evenly distributed between the conditions, yet Dutch news consumption (F (2, 192) = 3.25, p = .041, η2 = .02; F (2, 192) = 0.37, p = .694, η2 = .00) was
unevenly distributed between conditions of the attitude manipulation. Therefore, analysis using this narrow case selection has to control on the effect of Dutch news consumption.
Results
Effects on the general sample
A MANOVA was conducted to measure effects of the stimuli on the participants’ attitude towards voting, their subjective norm, and their own estimated political self-efficacy. Box’s test revealed that equality of covariance matrices can be assumed (F (250, 8288.44) = 1.03, p = .364). However, Levene’s test revealed that homogeneity of variances can be assumed for attitude (F (66, 228) = 1.26, p = .113) but not for subjective norm (F (66, 228) = 1.85, p < .000), political self-efficacy (F (66, 228) =
1.75, p = .001), and vote intention (F (66, 228) = 1.80, p = .001). However, as three of four dependent variables violate the basic assumption that the error variance of the dependent variable is equal across groups, Pillai’s trace test was chosen to identify significance of effects, for its robustness and adequate power to correct for the violation (Field, 2005).
Pillai’s trace showed no significant main effect of attitude manipulation (F (8, 452) = 1.47, p = .187, η2 = .04), yet it revealed a significant effect of the manipulation of knowledge and norm (F (8, 452) = 2.54, p = .010, η2 = .04). More precisely, a significant effect was found for the manipulation effect on participants’ political self-efficacy (F (2, 295) = 4.87, p = .008, η2 = .04), and a marginally significant effect was found for the its effect on subjective norm (F (2, 295) = 2.86, p = .059, η2 = .03), yet no effect was found on attitude (F (2, 295) = 0.70, p = .499, η2 = .01) and vote intention (F (2, 295) = 0.10, p = .902, η2 = .01). Participants who were in the knowledge group (M = 0.43, S.E. = .31), showed a significant higher political self-efficacy compared those in the norm group (M = 0.25, S.E. = .03) and those in the control group (M = 0.28, S.E. = .03). (See also graph 1 in Appendix B)
Furthermore, the marginally significant effect of the condition on participants’ subjective norm was evident, as those participants in the norm group showed higher estimates of turnout rates for a relevant social group (M = 0.46, S.E. = .03), than those in the control (M = 0.36, S.E. = .03) or knowledge group (M = 0.38, S.E. = .03). Therefore, H1b that expects a causal relation between TPB variables and vote
intention was not accepted for the general sample. However, H3 states that messages targeting individuals’ attitude, subjective norm, and self-efficacy have a causal effect on persons’ actual attitude, subjective norm, and self-efficacy towards voting local in
the Netherlands. As a significant effect was found for self-efficacy, and a marginally significant effect was found for subjective norm, the hypothesis was partly accepted.
Pillai’s trace further ruled out significant effects of the interaction between both manipulations (F (16, 912) = 2.86, p = .480, η2 = .02). However, no interaction effect was evident, and no independent variable had a main effect on vote intention. Therefore and given the prior results, H2a, that expects the largest causal effect for an interaction of all variables, followed by subjective norm, was not accepted for the general sample (see table 3 and 4 in Appendix A for all means, standard deviations and results of the analysis).
Effects on non-voters
A MANOVA was conducted to analyze if expected effects are more likely to occur, when responses of non-voters are analyzed only. Box’s test showed that equality of covariance matrices can be assumed (F (130, 6274.38) = 1.04, p = .348). However, Levene’s test showed that homogeneity of variances can be assumed for attitude (F (17, 174) = 0.60, p = .888) and vote intention (F (17, 174) = 0.71, p = .791) but not for subjective norm (F (17, 174) = 1.72, p = .043) and political self-efficacy (F (17, 174) = 2.18, p = .006). Again, Pillai’s trace test was chosen to identify significance of effects. Analysis showed that no significant effects occurred based on attitude
manipulation (F (8, 344) = 0.53, p = .831, η2 = .01), Dutch news consumption (F (4, 171) = 0.53, p = .712, η2 = .01), the interactions between Dutch news consumption and attitude manipulation (F (8, 344) = 1.29, p = .246, η2 = .03) or the knowledge-norm manipulation (F (8, 344) = 1.44, p = .181, η2 = .03), and the three-way
interaction of independent all variables in the analysis (F (16, 696) = 1.25, p = .221, η2 = .03). However, a significant main effect of the knowledge-norm manipulation
was evident (F (8, 344) = 2.52, p = .011, η2 = .06). The knowledge-norm condition affected participants’ political self-efficacy (F (2, 192) = 3.96, p = .021, η2 = .04), which was significantly higher for those who received a message regarding their political knowledge (M = 0.24; S.E. = .31), than for those who received a message regarding the turnout rates of comparable people (M = 0.18; S.E. = .03), or no message (M = 0.12; S.E. = .03). No effect was found on participants’ attitude regarding voting (F (2, 192) = 2.22, p = .112, η2 = .03), their subjective norm (F (2, 192) = 2.70, p = .070, η2 = .03), and vote intention (F (2, 192) = 0.464, p = .629, η2 = .01). Nevertheless, the results showed an interaction effect between the attitude and knowledge-norm condition (F (16, 696) = 1.72, p = .038, η2 = .04), which
significantly affected participants’ vote intention (F (4, 192) = 2.89, p = .024, η2 = .06).
The vote intention of participants in the groups attitude low and knowledge (M = 0.52, S.E. = .08), attitude control and norm (M = 0.50, S.E. = .08) was significantly higher than the vote intention of participants in the groups attitude control and knowledge (M = 0.26, S.E. = .09), attitude low and norm (M = 0.25, S.E. = .10), and attitude high and control (M = 0.21, S.E. = .09). However, participants in the latter conditions also showed a significant lower intention to vote than participants from the conditions attitude high and knowledge (M = 0.41; S.E. = .07) attitude high and norm (M = 0.41, S.E. = .07), attitude control and control (M = 0.40, S.E. = .09), and attitude low and control (M = 0.39, S.E. = .06) (see table 5 and 6 in Appendix A for all means, standard deviations, and results of analysis). Yet, vote intention for these four groups were only significantly higher than vote intentions of participants in the group attitude high and control. Interestingly, the group attitude controls and control, thus
well as higher than each four other combinations (see also graphic 2 in Appendix B). Even though significant differences were only evident to the least three groups (Control and knowledge; low attitude and norm; high attitude and control), mean differences between the groups with highest scores on intention are striking, even though not significant.
Concluding, only the interaction of both conditions showed a significant effect on vote intention. Yet, H1b expects casual effects of all TPB factors and the
interaction of those factors. Therefore, H1b was only partly accepted for a population that includes non-voters only. Although it was shown that only significant interaction effects occurred, H2b, predicting causal effect for all variables and interactions of TPB, was partly accepted for the same population. Furthermore, H3 argues that addressing factors of TPB through messages should enhance the targeted factors. Yet, the effect was only evident for political self-efficacy. Therefore the hypothesis was partly accepted for the same population.
Discussion
The aim of this study was to test whether factors of the extended theory of planned behavior model are able to predict local voting of EU immigrants in the Netherlands and if and how elements of the theory can serve as a framework for campaigns meant to enhance electoral participation, or to achieve behavioral change of non-voters, respectively.
General fit of the model
Results of study 1 indicate TGB’s usefulness to predict vote intention, based on social-cognitive factors. Even though the predicted interaction effect was not found, it
was illustrated that all factors, with the exception of perceived behavioral control, were significant predictors of vote intention. Yet, different than predicted, the attitude towards voting local had a larger impact, than subjective norms, past behavior, or political self-efficacy. Moreover, the impact of the model was illustrated by the large amount of variance it accounted for, which was about double of the variance that control variables accounted for. Consequently, the general applicability of the model on EU immigrants’ vote intention in the Netherlands can be assumed. This is in line with prior research, which demonstrated the applicability of the model to vote intention of other groups and in other contexts (e.g. Hansen & Jensen, 2007;
Netemeyer, Burton & Johnston, 1991; Peterson, 2012). Yet, while at least Netemeyer, Burton and Johnston (1991) and Peterson (2012) were able to show an effect of perceived behavioral control, this study is does not confirm the findings. However, it is questionable whether the factor is affected by the evidently high standard of education within the sample of study 1, which might render PBC less important to respondents. However, as EU immigrants are argued to differ from other social- and immigrant-groups (Favell, 2010), it is possible that this characteristic is not universal.
Causality of effects and differences between past voters and non-voters
Focusing on the general population of this research, study 2 acknowledged causal effects only for the manipulation of subjective norms and participants’ knowledge, yet no causal effects on vote intention were evident. Even though this finding is
inconsistent with prior experiments, which illustrated causal effects of subjective norms (or perceived civic duty) on vote intentions (e.g. Gerber & Rogers, 2009), it confirms the assumption that messages can manipulate the perception of political self-efficacy and subjective norms. However, based on the results of the study, it cannot
be concluded that those factors are feasible for campaigning higher vote intentions in the general population of this study.
Nevertheless, it is arguable that interventions based on the model, which mainly focuses at behavioral change through campaigning, are more effective when non-voters are targeted only. Analysis confirms that - next to a main effect of positive knowledge messages on participants’ political self-efficacy - an interaction between knowledge-norm manipulations and attitude manipulations significantly affected participants’ vote intention. The effect illustrates that combinations of messages promoting positive attitudes and high political self-efficacy, or positive attitude and high norms, or negative attitudes only, do not differ from no treatment at all (group control and control) in their effects on vote intention. However, messages that promote combinations of a low attitude combined with a high political self-efficacy, or high norms only (such as high turnout forecasts for relevant social groups) enhance the vote intention of past non-voters. While the first effect might be explained with resistance effects of persons with a high political self-efficacy (Knowles & Linn, 2004), the effect of normative messages confirms prior findings on the effect of social norms or civic duties on vote intentions (e.g. Bengtsson, 2004; Blais, 2006; Gerber & Rogers, 2009; Knack, 1992; Opp, 2001) However, the results clearly indicate that messages, which enhance readers political self-efficacy, are most feasible in
campaigning electoral participation if they try not to enhance attitude towards voting, but to create effects of resistance. Yet, it also illustrated that pure normative messages might be sufficient to enhance electoral participation of non-voters as well.
Furthermore, the results clearly indicate that messages promoting non-voters’ political self-efficacy or positive attitudes towards voting, only, or a combination of high subjective norms with negative attitudes towards voting, lower the vote intention of
non-voters significantly, compared to persons that received no treatment. While campaigns to enhance electoral participation of EU immigrants should avoid these combinations, the results confirm interaction effects of perceived norms with other factors, which can enhance but also reverse the effect. Therefore it illustrates the need of a social-cognitive framework, like TPB, to address voting behavior through
campaigning.
Limitations and future research
Regarding the results of the studies, limitations of the study should be acknowledged. First, even though students accounted only for a minority of the sample, the general level of education was very high. This might be a characteristic of the sample, as EU immigrants are argued to be highly educated in most cases (Favell, 2010), it is also possible that more less educated immigrants could have been reached, if a Dutch version of the research was conducted, also. Second, the experiment was conducted online and manipulations, at least of the attitude condition, were designed to create strong differences. Even though all manipulations were designed carefully, to create a strong experimental reality, some participants could have seen through the
manipulation. Yet, neither the results, nor comments of participants, indicate such an effect. Furthermore, the design of study 2 did not allow analyzing interaction effects of norms and political self-efficacy. Finally, the studies only focus on vote intention and past behavior. Therefore, it is only possible to analyze and interpret results based on vote intention, yet not on actual behavioral change. However, as several studies have shown that the performance of behavior is mainly predicted by the intention to perform the behavior (e.g. Armitage & Conner, 2001), is it likely that the results also account for behavioral change, even though effects might be weaker.
Despite the limitations, this research delivers important insights. First it was illustrated that campaigning electoral participation should use social-cognitive
frameworks to avoid reverse effects of single motivational factors. Second, it tests and illustrates the fit of TPB in predicting vote intention, in yet another context and for another social group than prior research. Therefore, the consistency of results is a further hint directing at the universality of the model to explain vote intention. Third, this research explores possibilities to create campaigning messages that might help to enhance electoral participation, and through that integration, of new fast growing social group in Europe. Therefore, it actively addresses future societal challenges, and as such serves as a starting point future research – and campaigning.
References
Aarts, H., & Dijksterhuis, A. (2003). The silence of the library: environment, situational norm, and social behavior. Journal of personality and social
psychology, 84(1), 18.
Abramowitz, A. I. (1991). Incumbency, campaign spending, and the decline of competition in US House elections. The Journal of Politics, 53(01), 34-56. Ajzen, I. (1991). The theory of planned behavior. Organizational behavior and
human decision processes, 50(2), 179-211.
Ajzen, I. (2010). Behavioral interventions based on the theory of planned behavior. Retrieved March 25, 2010.
Ajzen, I. (2011). Constructing a theory of planned behavior questionnaire.
Unpublished manuscript. Retrieved on October 6th, 2014 from
Ajzen, I. (2011). The theory of planned behaviour: reactions and reflections.
Psychology & Health, 26(9), 1113-1127.
Ajzen, I., & Madden, T. J. (1986). Prediction of goal-directed behavior: Attitudes, intentions, and perceived behavioral control. Journal of experimental social
psychology, 22(5), 453-474.
Aldrich, J. H. (1993). Rational choice and turnout. American Journal of Political
Science, 246-278.
Ansolabehere, S., De Figueiredo, J. M., & Snyder, J. M. (2003). Why is there so little
money in politics? (No. w9409). National bureau of economic research.
Armitage, C. J., & Conner, M. (1999). Distinguishing Perceptions of Control From Self-‐‑Efficacy: Predicting Consumption of a Low-‐‑Fat Diet Using the Theory of
Planned Behavior1. Journal of applied social psychology, 29(1), 72-90. Armitage, C. J., & Conner, M. (2001). Efficacy of the theory of planned behaviour: A
meta-‐‑analytic review. British journal of social psychology, 40(4), 471-499. Bandura, A. (1977). Self-efficacy: toward a unifying theory of behavioral change.
Psychological review, 84(2), 191.
Bandura, A. (1982). Self-efficacy mechanism in human agency. American
psychologist, 37(2), 122.
Bandura, A. (2001). Social cognitive theory of mass communication. Media
psychology, 3(3), 265-299.
Bauböck, R. (2005). Expansive citizenship—voting beyond territory and membership.
Political science and politics, 38(04), 683-687.
Bendor, J., Diermeier, D., & Ting, M. (2003). A behavioral model of turnout.
Bengtsson, Å. (2004). Economic voting: The effect of political context, volatility and turnout on voters’ assignment of responsibility. European Journal of Political
Research, 43(5), 749-767.
Blais, A. (2006). What affects voter turnout?. Annu. Rev. Polit. Sci., 9, 111-125. Blais, A., & Aarts, K. (2006). Electoral systems and turnout. Acta Politica, 41(2),
180-196.
Blais, A., & Dobrzynska, A. (1998). Turnout in electoral democracies. European
Journal of Political Research, 33(2), 239-261.
Blais, A., & St-‐‑Vincent, S. L. (2011). Personality traits, political attitudes and the
propensity to vote. European Journal of Political Research, 50(3), 395-417. Blais, A., & Young, R. (1999). Why do people vote? An experiment in rationality.
Public Choice, 99(1-2), 39-55.
Bonfield, E. H. (1974). Attitude, social influence, personal norm, and intention interactions as related to brand purchase behavior. Journal of Marketing
Research, 379-389.
Brams, S. J. (1976). Paradoxes in politics. New York, 15.
Caprara, G. V., Vecchione, M., Capanna, C., & Mebane, M. (2009). Perceived political self-‐‑efficacy: Theory, assessment, and applications. European
Journal of Social Psychology, 39(6), 1002-1020.
CBS (2014). Bevolking; kerncijfers. Retrieved on January 10th, 2015 from http://www.cbs.nl/nl-NL/menu/themas/bevolking/cijfers/default.htm
Conner, M., & Armitage, C. J. (1998). Extending the theory of planned behavior: A review and avenues for further research. Journal of applied social psychology,
Denny, K., & Doyle, O. (2009). Does voting history matter? Analysing persistence in turnout. American Journal of Political Science, 53(1), 17-35.
European Commission (2012). Report from the Commission to the European Parliament and the Council on the application of Directive 94/80/EC on the right to vote and to stand as a candidate in municipal elections by citizens of the Union residing in a Member State of which they are not nationals. Retrieved on October 15th, 2014 from
http://ec.europa.eu/justice/citizen/files/com_2012_99_municipal_elections_en. pdf
Favell, A. (2003). Eurostars and Eurocities: towards a sociology of free moving professionals in Western Europe.
Favell, A. (2010). European identity and european citizenship in three “eurocities”: a sociological approach to the European Union. Politique européenne, (1), 187-224.
Field, A. (2009). Discovering statistics using SPSS. Sage publications.
Flay, B. R., & Burton, D. (1990). Effective mass communication strategies for health campaigns.
Franklin, M. N. (2002). The dynamics of electoral participation. Comparing
democracies, 2, 148-168.
Franklin, M. N., & Hirczy, W. P. (1998). Seperated Powers, Divided Government, and Turnout in US Presidential Elections. American Journal of Political
Science, 316-326.
Gerber, A. S., & Rogers, T. (2009). Descriptive social norms and motivation to vote: everybody's voting and so should you. The Journal of Politics, 71(01), 178-191.
Geys, B. (2006). Explaining voter turnout: A review of aggregate-level research.
Electoral Studies, 25(4), 637-663.
Gidengil, E. (2013). Immigrants, voting, and electoral participation. The Encyclopedia
of Global Human Migration.
Gordon, D. N. (1970). Immigrants and municipal voting turnout: Implications for the changing ethnic impact on urban politics. American Sociological Review, 665-681.
Green, D. P., Shapiro, I., & Shapiro, I. (1994). Pathologies of rational choice theory:
A critique of applications in political science (p. 3). New Haven: Yale
University Press.
Groenendijk, K. (2008). Local voting rights for non-nationals in Europe: What we know and what we need to learn. Washington DC: Migration Policy Institute. Hagger, M. S., Chatzisarantis, N. L., & Biddle, S. J. (2002). A meta-analytic review
of the theories of reasoned action and planned behavior in physical activity: Predictive validity and the contribution of additional variables. Journal of
sport & exercise psychology.
Hajnal, Z., & Trounstine, J. (2005). Where turnout matters: The consequences of uneven turnout in city politics. Journal of Politics, 67(2), 515-535.
Hanks, C., & Grofman, B. (1998). Turnout in gubernatorial and senatorial primary and general elections in the South, 1922–90: A rational choice model of the effects of short-run and long-run electoral competition on relative turnout.
Public Choice, 94(3-4), 407-421.
Hansen, T., & Jensen, J. M. (2007). Understanding Voters’ Decisions: A Theory of Planned Behaviour Approach. Innovative Marketing, 3(4), 86-93.
Holbert, R. L., Lambe, J. L., Dudo, A. D., & Carlton, K. A. (2007). Primacy effects of The Daily Show and national TV news viewing: Young viewers, political gratifications, and internal political self-efficacy. Journal of Broadcasting &
Electronic Media, 51(1), 20-38.
Just, A., & J Anderson, C. (2012). Immigrants, citizenship and political action in Europe. British Journal of Political Science, 42(03), 481-509.
Karp, J. A., & Banducci, S. A. (2008). Political efficacy and participation in twenty-seven democracies: How electoral systems shape political behaviour. British
Journal of Political Science, 38(02), 311-334.
Karp, J. A., & Brockington, D. (2005). Social desirability and response validity: A comparative analysis of overreporting voter turnout in five countries. Journal
of Politics, 67(3), 825-840.
Knack, S. (1992). Civic norms, social sanctions, and voter turnout. Rationality and
Society, 4(2), 133-156.
Knack, S., & White, J. (2000). Election-day registration and turnout inequality.
Political Behavior, 22(1), 29-44.
Knowles, E. S., & Linn, J. A. (2004). Approach-avoidance model of persuasion: Alpha and omega strategies for change. Resistance and persuasion, 117-148. Krishna, V., & Morgan, J. (2012). Voluntary voting: Costs and benefits. Journal of
Economic Theory, 147(6), 2083-2123.
Logan, J. R., Darrah, J., & Oh, S. (2012). The impact of race and ethnicity, immigration and political context on participation in American electoral politics. Social forces, 90(3), 993-1022.
Madsen, D. (1987). Political self-efficacy tested. The American Political Science
Martin, B. A., Lang, B., Wong, S., & Martin, B. A. (2003). Conclusion explicitness in advertising: The moderating role of need for cognition (NFC) and argument quality (AQ) on persuasion. Journal of Advertising, 32(4), 57-66.
Matsusaka, J. G., & Palda, F. (1993). The Downsian voter meets the ecological fallacy. Public Choice, 77(4), 855-878.
McEachan, R. R. C., Conner, M., Taylor, N. J., & Lawton, R. J. (2011). Prospective prediction of health-related behaviours with the theory of planned behaviour: A meta-analysis. Health Psychology Review, 5(2), 97-144.
McGuire, W.J. (1989) Theoretical foundations of campaigns. In R.E. Rice & C.K. Atkin (Eds.), Public communication campaigns (2nd ed., pp. 43-54). Newbury Park, CA: Sage.
Munro, D. (2008). Integration through participation: non-citizen resident voting rights in an era of globalization. Journal of International Migration and
Integration/Revue de l'integration et de la migration internationale, 9(1),
63-80.
Myerson, R. B. (1993). Effectiveness of electoral systems for reducing government corruption: a game-theoretic analysis. Games and Economic Behavior, 5(1), 118-132.
Netemeyer, R. G., Burton, S., & Johnston, M. (1991). A comparison of two models for the prediction of volitional and goal-directed behaviors: A confirmatory analysis approach. Social Psychology Quarterly, 87-100.
Newhagen, J. E. (1994). Media use and political efficacy: The suburbanization of race and class. Journal of the American Society for Information Science, 45(6), 386-394.
Nickerson, D. W. (2007). Does email boost turnout. Quarterly Journal of Political
Science, 2(4), 369-379.
Nickerson, D. W., Friedrichs, R. D., & King, D. C. (2006). Partisan mobilization campaigns in the field: Results from a statewide turnout experiment in Michigan. Political Research Quarterly, 59(1), 85-97.
Nicolaas, H. (2009). Bevolkingsprognose 2008-2050: veronderstellingen over immigratie. Bevolkingstrends, 57(1), 23-33.
O+S Amsterdam (2013). Amsterdam in Cijfers 2012. Gemeente Amsterdam Ouellette, J. A., & Wood, W. (1998). Habit and intention in everyday life: the
multiple processes by which past behavior predicts future behavior.
Psychological bulletin, 124(1), 54.
Tam Cho, W. K. (1999). Naturalization, socialization, participation: Immigrants and (non-) voting. The Journal of Politics, 61(04), 1140-1155.
Tuckel, P., & Maisel, R. (1994). Voter turnout among European immigrants to the United States. Journal of Interdisciplinary History, 407-430.
Pattie, C., & Johnston, R. (1998). Voter turnout at the British General Election of 1992: Rational choice, social standing or political efficacy?. European Journal
of Political Research, 33(2), 263-283.
Peterson, D. K. (2012). The Colbert Bump® and The Facebook® Follow-Through for Generation Snark: A Test and Extension of The Ajzen’s Theory of Planned Behavior for 2012. Journal of Management Research, 4(3), 43-65.
Rallings, C., & Thrasher, M. (2005). Not All ‘Second-‐‑Order’Contests are the Same: Turnout and Party Choice at the Concurrent 2004 Local and European Parliament Elections in England*. The British Journal of Politics &
Reif, K., & Schmitt, H. (1997). Second-order elections. European Journal of Political
Research, 31(1-2), 109-124.
Rhodes, R. E., & Courneya, K. S. (2003). Relationships between personality, an extended theory of planned behaviour model and exercise behaviour. British
Journal of Health Psychology, 8(1), 19-36.
Riker, W. H. (1958). The paradox of voting and congressional rules for voting on amendments. American Political Science Review, 52(02), 349-366. Schwartz, S. H. (1973). Normative explanations of helping behavior: A critique,
proposal, and empirical test. Journal of Experimental Social Psychology, 9(4), 349-364.
Smets, K., & Van Ham, C. (2013). The embarrassment of riches? A meta-analysis of individual-level research on voter turnout. Electoral Studies, 32(2), 344-359. Van Dalen, H. P. (2001). Immigratie: vloek of zegen voor de Nederlandse economie?.
Bevolking en Gezin, 30(3), 3-36.
van der Heijden, T., & van Heelsum, A. (2010). Opkomst en stemgedrag van migranten tijdens de gemeenteraadsverkiezingen van 3 maart 2010. Vecchione, M., & Caprara, G. V. (2009). Personality determinants of political
participation: The contribution of traits and self-efficacy beliefs. Personality
and Individual Differences, 46(4), 487-492.
Xu, J. (2005). Why do minorities participate less? The effects of immigration, education, and electoral process on Asian American voter registration and turnout. Social science research, 34(4), 682-702.
Appendix A: Tables
Table 1. Regression coefficients and standard errors of predictors for vote intention.
Model I Model II
Gender 0.00 (.05) 0.01 (.04)
Time of residence 0.01 (.00)* -0.00 (.00)
Exposure to Dutch news 0.38 (.06)*** 0.10 (.06)
Attitude 0.41 (.09)***
Subjective norm 0.29 (.09)**
Perceived behavioral control 0.09 (.08)
Self-efficacy 0.27 (.09)**
Past behavior 0.24 (.05)***
Intercept 2.50 (.12)* -0.06 (.11)
R2 0.25 0.54
F (df) 20.14 (4, 248)*** 31.12 (5, 243)***
Notes: Controls are Gender, Time of residence, and Exposure to Dutch news. *p <= .05. **p < .01. ***p < .001.
Table 2. Regression coefficients and standard errors of single items of predictor scales for vote intention.
Model I Model II Attitude scale items
Good 0.21 (.20)
Important 0.50 (.01)***
Pleasant 0.03 (.12)***
Positive 0.07 (.20)
Norm scale items
Expats -0.01 (.01)
Friends, Family, & Colleagues 0.06 (.01)
People like me 0.04 (.01)
Intercept -0.38 (.06) 0.50 (.07)
R2 0.31 0.33
F (df) 28.86 (4, 248)*** 31.11 (4, 248)***
Table 3. Means and (standard deviation) for total sample per condition. Condition
Attitude
Condition Norm
n Attitude Norm Self-efficacy Intention Low Control 35 0.61 (.21) 0.38 (.22) 0.19 (.24) 0.53 (.36) Norm 23 0.58 (.20) 0.40 (.24) 0.18 (.23) 0.41 (.36) Knowledge 39 0.64 (.24) 0.37 (.21) 0.32 (.26) 0.56 (.34) Total 97 0.61 (.22) 0.38 (.22) 0.24 (.25) 0.52 (.35) Control Control 34 0.69 (.20) 0.40 (.21) 0.24 (.18) 0.58 (.33) Norm 38 0.70 (.19) 0.51 (.24) 0.32 (.23) 0.61 (.32) Knowledge 31 0.65 (.25) 0.30 (.21) 0.33 (.26) 0.59 (.40) Total 103 0.68 (.21) 0.41 (.24) 0.29 (.23) 0.59 (.35) High Control 26 0.74 (.25) 0.40 (.20) 0.31 (.30) 0.60 (.42) Norm 35 0.68 (.21) 0.52 (.23) 0.22 (.22) 0.55 (.35) Knowledge 34 0.67 (.19) 0.41 (.20) 0.39 (.27) 0.60 (.35) Total 95 0.69 (.28) 0.45 (.22) 0.31 (.27) 0.58 (.36) Total Control 95 0.67 (.22) 0.39 (.21) 0.24 (.24) 0.57 (.36) Norm 96 0.66 (.21) 0.49 (.24) 0.25 (.23) 0.54 (.35) Knowledge 104 0.65 (.23) 0.37 (.21) 0.35 (.26) 0.58 (.36) Total 295 0.66 (.22) 0.41 (.23) 0.28 (.25) 0.56 (.36)
Table 4. Results of univariate analyses of variance, general sample. Sum of Squares df Mean square F p η2 Attitude Corrected Model 0.53 8 0.07 1.41 .193 .04 Intercept 125.17 1 125.17 2681.36 .000 .90 Condition Attitude 0.38 2 0.19 4.09 .018 .03 Condition Norm 0.04 2 0.02 0.46 .631 .00 Condition Attitude * Condition Norm 0.16 4 0.04 0.87 .484 .01 Error 13.35 286 0.05 Total 142.94 295 Subjective Norm Corrected Model 1.26 8 0.16 3.29 .001 .08 Intercept 48.22 1 48.22 1007.36 .000 .78 Condition Attitude 0.18 2 0.09 1.87 .156 .01 Condition Norm 0.66 2 0.33 6.86 .001 .05 Condition Attitude * Condition Norm 0.30 4 0.08 1.57 .181 .02 Error 13.69 286 0.05 Total 65.19 295 Political self efficacy Corrected Model 1.30 8 0.16 2.75 .006 .07 Intercept 22.129 1 22.13 373.09 .000 .57 Condition Attitude 0.30 2 0.15 2.50 .084 .02 Condition Norm 0.71 2 0.36 5.98 .003 .04 Condition Attitude * Condition Norm 0.34 4 0.09 1.44 .221 .02 Error 16.96 286 0.06 Total 41.53 295 Vote Intention Corrected Model 0.78 8 0.10 0.76 .635 .02 Intercept 89.98 1 89.98 707.20 .000 .71 Condition Attitude 0.48 2 0.24 1.88 .154 .01 Condition Norm 0.20 2 0.10 0.79 .454 .01 Condition Attitude * Condition Norm 0.27 4 0.07 0.52 .718 .01 Error 36.39 286 0.13 Total 131.14 295 Note N = 295
