Is urban gardening political participation? Or is it active community participation?

Is ‘liking’ something on Facebook political participation? Or is it slacktivism?

Is crowdfunding political participation? Or is it collective engagement?

The concept of political participation has been challenged due to a large number of new participatory activities such as supporting crowdfunding campaigns or signing online petitions. The two most prevalent recent trends are digital activities of participation (Theocharis 2015) and a broadened understanding of participation as manifested in an increasing importance of new “social movements and practices of horizontal, participatory, and direct democracy” (Ratto and Boler 2014, p. 1). It is important to study political participation empirically as it is an indicator of the quality of democracy (Theocharis and van Deth 2016).

Empirical studies always have to be up to date on the behavior studied. The rather pessimistic picture some scholars draw of the state of participation (e.g., Putnam 2001) may partly be attributed to their hesitation to consider digitally networked or other more recent forms of participation. It is therefore important to periodically update the concept of participation to detect recent trends in citizens’ political activity. However, staying updated raises the question what constitutes political participation nowadays and what does not, especially when we study citizens’ participatory behavior empirically. The quest to answer this question uncovers a large number of recently developed types of participation (e.g., online, passive, or expressive). Critics claim that a development of such study-specific types of participation should follow theoretical principles more closely. Even more importantly, the plurality of participation types and their nominal labeling make comparison of participation patterns across studies increasingly difficult (Hay 2007; van Deth 2014). In other words, we need guidelines for which activities to study and how to form distinct measures of different participation types.

To address these theoretical and empirical ambiguities, van Deth (2014) has developed a conceptual map based on original theoretical considerations about the very concept of political participation (Verba and Nie 1972). The map, which provides decision rules to identify to which type of participation certain activities belong, has been extended by Theocharis (2015) for digitally networked participation. Van Deth distinguishes four types of participation, namely political participation taking place in (PP I) or being targeted at (PP II) the political sphere, being targeted at community issues (PP III), or being a non-political but politically motivated activity (PP IV). This conceptualization has the potential to restructure the understanding of participation in a timely, timeless, yet theoretically thorough manner. Such rethinking can only be achieved if empirical studies find the concept applicable. The map was constructed for “the methodical identification of any phenomenon as a specimen of political participation and for a systematic distinction between various types of participation” (Van Deth 2014, p. 360). However, an application of this methodological identification has yet to come.

In its attempt to provide a more clear-cut empirical preparation of the conceptual map, our study is an important stepping stone between the theoretical concept and its empirical application. By applying van Deth’s decision rules, we explore to which extent participatory activities frequently measured in participation studies can be assigned to the four different types of political participation. Refining the theoretical considerations of the map, our study develops specific decision criteria scholars can come back to when operationalizing political participation according to van Deth’s concept. In the end, we test how distinct and thereby unique the four types of political participation are and thereby deliver information scholars can take into account when deciding about which conceptualization to take as basis in empirical studies. Two questions from the recent debate of what comprises political participation are specifically addressed by our study: Does the map’s application indeed support the notion that Internet-based activities can occur across all types of participation (Theocharis 2015), or does the conceptualization provide further evidence for an often found distinction between online and offline participation (e.g., Gibson and Cantijoch 2013)? And how do forms recently described as political participation (e.g., crowdfunding, urban gardening, boycotting products) integrate in this new conceptual proposal?

Our study shows that van Deth’s conceptual map allows for a clear-cut structuring of participatory activities, if additional empirical separators are considered. Furthermore, we find evidence for discriminant validity of the four resulting types of participation and provide tested measures for application in future studies with interest in capturing a broad spectrum of political participation.

Conceptualizing political participation

Definitions and conceptualizations of political participation have been discussed for decades; so has the question what actions qualify as political participation (Gibson and Cantijoch 2013). Verba and Nie (1972) provided a widely adapted definition of political participation and were some of the first to acknowledge that the concept is multidimensional. Scholars adapt this notion by increasingly taking a broadened understanding of participation into account (e.g., by focusing on civic engagement; see Norris 2002). However, the multidimensionality of political participation sparked an ongoing search for underlying patterns or systematic clusters of participatory actions (Fox 2014). Subsequently, various conceptualizations (e.g., Hamlin and Jennings 2011; Teorell et al. 2007) and numerous empirically driven studies developed a large number of political participation typologies (e.g., Bakker and de Vreese 2011; Linssen et al. 2014; Xenos et al. 2014).

A main reason behind the constant proliferation of participation typologies is a need to keep the concept up to date, i.e., to connect it to recent developments in citizen participation (e.g., Gibson and Cantijoch 2013; Hirzalla and van Zoonen 2011; Ratto and Boler 2014). As a consequence, numerous labels are used for participation, such as online (e.g., Gil de Zúñiga et al. 2012), passive (Bakker and de Vreese 2011), individual (Xenos et al. 2014), symbolic (Stolle et al. 2005), conventional (Linssen et al. 2014), and representational (Teorell et al. 2007). A closer look at the definitions reveals that participation types with different labels share a similar definitional basis. For instance, Teorell et al. (2007) identify protest activity as a distinct participation type, while Hirzalla and van Zoonen (2011) in their conceptualization include activism as a distinct type. Both types build on Verba and Nie (1972) and are based on the idea of cause-oriented, extra-institutional activities. However, due to sub-distinctions made by the authors, these types are labeled and measured differently.

We argue that such attempts to update the concept of political participation usually run against the need for a shared understanding of what constitutes political participation. A shared understanding is necessary to make results comparable across studies and detect general trends of participation in and across democracies. Establishing a longer-lasting participation typology requires a more thorough and theory-driven conceptualization. Van Deth (2014) potentially provides such a conceptualization. He suggests using a conceptual map to (1) arrive at an operational definition of political participation, (2) distinguish between different types (or ‘variances’) of political participation, and (3) assign participatory activities to or exclude them from one of the variances with the help of decision rules. Instead of classifying participation activities according to time-bound phenomena such as channel (e.g., online and offline) and activity costs (e.g., high and low), van Deth’s classifications rely on three aspects of participation: the sphere, the target, and the intention of an activity.

Four distinct types of political participation?

Van Deth’s map is a set of decision rules that were systematically developed to arrive at distinct types of political participation. These decision rules clearly aim at improving the distinction of types of participation made in empirical studies by depicting the “necessary features for each mode and type of participation [to identify] all aspects to be operationalized in surveys, content analyses and other data collection strategies” (van Deth 2014, p. 362). However, such an empirical application has yet to come. The conceptual map uses ‘objective criteria’ (p. 360) to recognize activities of political participation. This is at odds with the rather subjective assumptions a scholar has to make about the extent to which one criterion applies to a participatory activity. An objective criterion (e.g., an activity being targeted at the political sphere) might be theoretically unique, but the individual researcher still has to determine whether an activity genuinely matches it. Van Deth (2014) and Theocharis (2015) use concrete actions that depict not only the activity but also a specific cause, location, and actorFootnote 1 to exemplify its assignment to a certain type. This helps understand the initial ideas in their texts but creates a challenge for empirical (quantitative) studies, which have to use less detailed measures to derive general statements. So the question remains how unambiguously the criteria of the map can be applied to existing, empirical measures. Below, we describe Van Deth’s original decision rules and suggest additional, operational separators that are necessary to assign empirical categories to one of the four types of participation.

The decision rules consist of questions identifying activities and assigning them to distinct types of participation. The first three questions must be answered affirmatively to allow for a type-specific allocation: Q1: Do we deal with behaviour? Q2: Is the activity voluntary? Q3: Is the activity done by citizens? Q4: Is the activity located in the sphere of government/state/politics? allows for a first distinction between different types of political participation. If the answer to Q4 is affirmative, the activity belongs to the first type of political participation in van Deth’s map, namely Political Participation I. Such activities take place directly within the sphere of government/state/politics. Verba and Nie (1972) describe these activities as “’within the system’-ways of influencing politics that are generally recognized as legal and legitimate” (p. 3). Hence, we deal with participation in political institutions or state-initiated (i.e., legal) political decision-making processes.Footnote 2 Activities need to take place in a prescribed framework that is stipulated by a constitution or constitutional law, such as the right to form and be a member of a party or the right to elect political officialsFootnote 3 (Downs 1957). An empirical category for PP I therefore clearly has to mention participatory activities (a) in a political institution or (b) in constitutionally stipulated decision-making processes as specific separators.

If Q4 cannot be confirmed, the next step of the map applies a target definition by asking Q5: Is the activity targeted at the sphere of government/state/politics? If so, the activity belongs to the map’s second type, Political Participation II, meaning that it does not take place within but is targeted at the political sphere. This includes targeting the function of this sphere, i.e., being responsive and addressing problems formulated by the public. Awareness about an issue in the political sphere can be raised directly or in mediated ways (Lipsky 1968). Direct awareness is raised via legally provided ways based on constitutional law that creates a state-guaranteed framework of protest and feeds into the ‘awareness responsiveness’ of a political system (Foweraker 1995; Schumaker 1975). Furthermore, direct targeting happens if activities provoke a confrontation with the system and thereby urges an immediate response by public authorities (e.g., civil disobedience; Tarrow 1988). Directly targeted actions can have mediated targeting as by-product, such as raising the mass media’s interest or increasing public awareness (Lipsky 1968). However, the inclusion of mediated targeting opens this type of political participation (PP II) up to a number of activities that address media and public awareness but hardly have the political sphere as a target. Mediated targeting is therefore not useful to consider as an empirical separator. We therefore suggest that an empirical category should mention an activity that (c) refers directly to a political institution, actor or decision-making process, (d) occurs in the legal framework of protest, or (e) urges an immediate response by authorities.

For activities that do not follow this path, a sixth question is asked: Q6: Is the activity aimed at solving collective or community problems? If so, the activity belongs to the third type, Political Participation III, provided that “clearly private or non-public activities are excluded” (Van Deth 2014, p. 357). With this third type of political participation, van Deth integrates into the concept of political participation the concept of civic engagement, understood as problem solving on a community level (e.g., Norris 2002). Such activities are characterized by their “direct hands-on work in cooperation with others” (Zukin et al. 2006, p. 51) rather than reattributing the responsibility to the political sphere. An important distinction is that activities on a local level can well be PP II if they target the political sphere directly (e.g., stopping by at a local politician’s office) but are a specimen of PP III if they are based on citizens’ involvement to address the issue themselves (e.g., by collecting money). An empirical category should therefore (f) stress a citizen’s direct initiative in the process and of course (g) explicitly refer to a local level or the sphere of community.

If an activity does not belong to any of these three types of political participation, the conceptual map defines it as a non-political activity. However, according to van Deth, non-political activities (e.g., buying sneakers) can have a political purpose (e.g., choosing fair trade brands to protest child labor). With the last question Q7: Is the activity used to express political aims and intentions of participants?, van Deth suggests considering activities “as a form of political participation if [they are] used to express political aims and intentions by the participants” (2014, p. 359). Thus, this fourth and last type of political participation, Political Participation IV, includes politically motivated but private or non-public actions. However, Hooghe has criticized citizens’ intentions as being irrelevant for a concept of political participation and difficult to study empirically (Hooghe et al. 2014).

In their study of political consumerism, Stolle et al. (2005) define activities as specimens of participation if people are motivated by political or ethical considerations or wish to cause societal change, “either with or without relying on the political system” (p. 255). Hence, politically motivated activities can address the political system in mediated ways (e.g., raising media’s or the publics’ awareness) or with own initiatives. Despite the challenge for empirical research to assess motivations exhaustively, considering the intention of activities creates a pathway to include a growing number of expressive, novel, and individually performed actions into the concept of political participation but formulates manageable restrictions. These restrictions need to be reflected in the formulation of an empirical category. We suggest two separators to help identify the political nature of formally non-political activities. Activities (h) in most instances have to be non-political, and (i) the political considerations or purpose has to be mentioned explicitly in an empirical category. Adding the adjective ‘political’ to motivations and purposes is an important detail here, as stressed by van Deth (2014). A simplified version of van Deth’s conceptual map is depicted in Fig. 1.

Fig. 1
figure 1

Simplified version of van Deth’s (2014) conceptual map of political participation

Full size image

Exploring to which extent this systematic approach can be used to structure participatory activities is the first aim of our study. The above-mentioned additional considerations about the proposed decision rules and the suggested separators aim at an unambiguous application of empirical categories to one of the four types of political participation. Existing empirical measures will be used as a test case to see if they clearly fit in or are excluded from one of the proposed constructs.

The distinctiveness (i.e., how different the dimensions are from each other) is a quality criterion of an empirical measurement and therefore a core interest for researchers. Addressing how distinct and unique the four dimensions of participation are, hence, is the second aim of the study. The so-called discriminant validity indicates that theoretically distinct constructs (e.g., dimensions of political participation) can be mapped by empirical representation (i.e., measures) of these constructs. It furthermore tests if a “broader construct has been erroneously separated into two or more factors” (Brown 2006, p. 3) and vice versa. How distinct an empirical construct is determines the application of the underlying theoretical concept in scientific studies because only distinctive constructs allow for valid statements about different construct dimensions. A number of recent studies did not test for the distinctiveness of the dimensions used (e.g., Bakker and de Vreese 2011; Gil de Zúñiga et al. 2012), while others thoroughly tested their theoretical conceptualization empirically (e.g., Gibson and Cantijoch 2013; Hirzalla and van Zoonen 2011). Discriminant validity depends on covariances and correlations of the latent constructs, in our case the different types of political participation. In general, evidence for discriminant validity is found with small factor covariances (see Hill and Hughes 2007; Kenny and Kashy 1992). However, no general rule for acceptable factor covariances exists, since they depend on the size of data. Correlation coefficients are better suited to compare the distinctiveness of factors across data sets, due to their fixed range between −1 and +1. Brown (2006) describes a value that exceeds .80 or .85 as a criterion for poor discriminant validity (p. 131) but mentions that the interpretation always depends on the specific research context. In their empirical tests of political participation concepts, Gibson and Cantijoch (2013) report covariances (but no correlations) between .46 and .82; Hirzalla and van Zoonen (2011) report correlations between .33 and .86; and Teorell et al. (2007) correlations between −.08 and .90. Hence, different types of participatory activities are rarely undertaken completely independent from each other. The above-mentioned values furthermore show rather high correlations between different types and do not all fully meet discriminant validity criteria. Our first research question (RQ 1) therefore asks if the results of empirically testing van Deth’s systematic approach reveal a satisfying distinctiveness of the four participation types; both in terms of correlations below .80 and in comparison with previously proposed concepts.

Digital and ‘individualized collective’ actions

Besides the distinctiveness of the proposed participation types, van Deth’s concept raises questions about two topics that recently received considerable attention in the discussion about what depicts political participation: digitally networked and other newly emerged participatory activities in social movements or as ‘critical, creative makings’ (Ratto and Boler 2014).

First, Hosch-Dayican criticized the original version of the conceptual map for not providing proper guidelines on how to include online participatory activities (Hooghe et al. 2014). This shortcoming was addressed by Theocharis’ (2015) examples of digitally networked acts as actions/behavior (Q1) conducted by citizens (Q2) on a voluntary basis (Q3). Digitally networked participation can thereby take place directly within the government arena (PP I), it can be targeted at political or community concerns (PP II and III), and it can, like non-digital activities, comprise non-political acts that are politically motivated (PP IV). At the same time, he demonstrates that van Deth’s conceptual map allows for the exclusion of certain forms of digitally networked participation such as a ‘like’ on Facebook. According to Theocharis, such activities are an expression of preference, but they are not an action; this leads to exclusion after the very first question (Q1) in van Deth’s map. Nevertheless, he notes that this type of activity can be an “act of political significance [that] may have serious consequences” (2015, p. 11; see pp. 7–12 for a more elaborate explanation). Theocharis addresses the common ‘slacktivism’ criticism (Christensen 2011; Morozov 2009) with the argument that “none of the traditional definitions exclude acts based on their costs, their symbolic nature or their low impact, as long as other definitional requirements are fulfilled” (2015, p. 8). These remarks have implications for the discussion about the “‘hierarchy’ of online and offline political acts” (Hooghe et al. 2014, p. 324) and the question whether they are separate constructs. Prior research has successfully used the distinction between online and offline participation activities (e.g., Bakker and de Vreese 2011; Gibson and Cantijoch 2013; Hargittai and Shaw 2013; Jensen 2013). However, Theocharis (2015) suggests that online as well as offline activities can—theoretically—be found in all four types of participation proposed by van Deth. This claim remains empirically untested. We therefore ask whether online and offline activities conform to the four types of political participation or whether they form distinct types (RQ 2).

Second, the need to broaden the concept of political participation is stressed by van Deth as well as by critics of his concepts (Hooghe et al. 2014). The map arrives at a broadened approach by first of all including politically motivated non-political activities. In a present-day understanding, it is furthermore important to include activities that are unconventional and often take place at a local level. These ‘individual collective actions,’ such as street art or urban gardening in a community, receive increasing attention nowadays (Micheletti and McFarland 2011; Ratto and Boler 2014; Rosol 2010; Visconti et al. 2010) and are mostly subsumed under an ever-broadened concept of civic engagement (see Adler and Goggin 2005; Ekman and Amnå 2012; Zukin et al. 2006). On the one hand, this is an acknowledgement of the societal and political relevance of such new and rather alternative forms of citizen participation (Anduiza et al. 2012; Ekström and Östman 2013; Ratto and Boler 2014). On the other hand, if a broadened understanding of civic engagement means that it is seen as political participation, one easily gets the impression that nowadays “political participation can be almost everything” (Van Deth 2014, p. 353).

To address this dilemma, van Deth (2014) suggests that civic engagement be re-integrated into the concept of political participation under the above-described premises and restrictions. Under van Deth’s target definition, it makes good sense to see civic engagement activities, which clearly aim at issues on the local sphere, as a part of political participation. The re-integration of civic engagement into the concept of political participation, however, is at odds with previous conceptualizations (Gibson and Cantijoch 2013; Teorell et al. 2007). Furthermore, whether traditional, community-based activities and individualized, collective actions indeed form a coherent factor as suggested by van Deth remains untested. Our study therefore asks if traditional as well as recently emerged activities directed at collective and community concerns form a distinct type of political participation (RQ 3).

Method

To answer the research questions and assess the construct validity, we follow a deductive approach (Brown 2006; Anderson and Gerbing 1988). First, we apply the decision rules suggested by van Deth to 34 participatory items asked in a national survey in Denmark (n = 9125). Since this allocation is a crucial step in the application of the conceptual map, assumptions behind every decision are indicated based on the suggested empirical separators (Table 1). We then apply EFA and CFA in conjunction to detect the latent variable structure of independent factors and test the discriminant validity of the latent variables. In other words, we map whether activities group together in the proposed structure, based on the frequency with which citizens have performed them, and test the relationship between the activities and the individual constructs as well as between the constructs.

Table 1 Item allocation, factor loadings, and empirical separators
Full size table

Sample

National online survey data were used to answer research questions. Data for three groups of Danish citizens were collected from a pollster’s databaseFootnote 4 and national register data. Online surveys are an appropriate strategy since 93% of Danish households have access to the Internet (Danmarks Statistik 2014). In addition to a general population sample, we included special samples of young and elderly citizens to secure an appropriate distribution of respondents at both ends of the age spectrum.

First, the general population sample was drawn from the pollster’s database, based on age, gender, education, occupation, and region of residency. 10,315 respondents were randomly invited to take the online survey; 45% agreed to participate (N = 4641). Second, an elderly sample (65 years and older) was drawn from the same online database, based on the same census characteristics except age. 60% of the 3059 older citizens invited agreed to participate (N = 1831). In a third step, a youth sample (17–21 years) was drawn using national registered data to obtain postal mail addresses. 13,700Footnote 5 young citizens were invited via postal invitations to take part in the same online survey, to which access was provided via links and QR codes; 20% (N = 2653) accepted the invitation. The three groups combined form the total sample of 9125 respondents; 47.2% were male, the average age was 45.3 years (SD = 21.2), and political interest 6.2 (SD = 2.6, Min = 0, Max = 10). Compared to census data, our sample comes close to the average age of 41 years and 49% of male citizens (Danmarks Statistik 2014), but cannot claim full representativity of the Danish population.

Measures

Respondents were asked how often they had performed participatory activities within the last 12 months on a five-point scale ranging from never to four times or more. Only questions about turnout in the last national and local election were assessed with dichotomous response categories. We consulted a large number of studies and national surveys to develop a pool of participatory activities (CivicWeb 2008; GLES 2013; Ekström and Östman 2013; Ekström et al. 2014; Portney and O’Leary 2007; Stolle et al. 2005; Yndigegn and Levinsen 2015) and added newly developed items. The items were selected so the measures would cover a broad range of recent participatory items. The items we use are a close match to what citizens mentioned as activities they use to express political views in a recent study by Theocharis and van Deth (2016). Given the focus of the study, we included a similar number of digital and non-digital activities undertaken on both the local and the national levels. This variety makes the set valuable in terms of applying it to van Deth’s conceptual map. Notably, items were not explicitly selected to test the map empirically and may therefore exemplify difficulties of fitting a theoretical concept to empirical data especially well.

Empirical analysis

One contribution of our study is to show how empirical categories may fit one of the four types of participation, based on van Deth’s proposed decision rules. We provide additional arguments based on theoretical considerations about how such allocation can be successful. To this end, the seven questions discussed above were asked (Q1–Q7) for every survey item, and the empirical separators (a–i) introduced above helped further specify the affiliation of an item. The wording of survey items and their allocation can be found in Table 1; allocation examples for each type of participation are given below.

Of the 34 surveyed items, 27 were identified as acts of political participation (Table 1).Footnote 6 Items that refer to engagement within political institutions or as constitutional, stipulated decision-making were identified as specimens of PP I (e.g., voting and being a party member). Activities are targeting the political sphere (PP II) if items directly refer to political institutions (e.g., visiting a politician, signing a petition) or happen in a state-guaranteed framework of protest (e.g., participating in a demonstration). Activities address issues at a community level (PP III) if an item refers to a direct action with immediate outcome on a local level (e.g., supported a community’s crowdfunding project, volunteered in a local organization). For items that are examples of non-political but politically motivated activities (PP IV), the consideration and political purpose need to be emphasized (e.g., boycotting products for political purposes; expressing your opinion on social media about a political issue).

To compare whether the theory-driven assignment of survey items fits with the latent dimensions of the empirical data, an exploratory factor analysis (EFA) was conducted. Using EFA is common practice in testing for latent dimensions in existing survey data and in examining the plausibility of the assumed distribution of survey items on different factors (Brown 2006). The number of possible factors was fixed to four, corresponding to the four types of participation. We used a varimax rotation and allowed items to load freely on four factors (KMO = .88, Bartlett’s test of sphericity: χ 2 = 65,428.43, p < .001). Of our 27 items, 23 had a loading on the expected factor above .20 but not all items loaded on one factor exclusively (see Table 1). This is initial evidence of the expected underlying latent dimensions in our data but also shows that some items deviate from the proposed theoretical structure. In other words, we gain initial support for the discriminant validity of the construct (Brown 2006).

Four items did not load on the factor expected: Party membership loaded moderately on factor PP II instead of on the assumed factor PP I. Party membership, however, is clearly a mode of participation within the arena of politics. The two items concerning contacting media institutions to express an opinion (sent letters or wrote articles to newspapers, magazines, or the like to comment on a political matter; called a radio or TV station to express your opinion on a political issue) were assumed to be non-political activities that are politically motivated, but loaded on factor PP II and not PP IV. Handing out flyers was assigned to PP IV but loaded on PP II. Our study set out to test van Deth’s proposed conceptualization. If multiple loadings are present, we therefore include items that load or partly load on a dimension according to their original allocation.

EFA and confirmatory factor analysis (CFA) are often used in conjunction (Brown 2006; Neel et al. 2015) to develop models based on empirical data and scales. CFA tests for the underlying latent constructs with more restrictions, such as fixed cross-loadings and uncorrelated errors, which are not part of an EFA. It is thereby a more reliable test of the estimate of covariances between different factors (Brown 2006). This test was needed for an empirical demonstration of the hypothesized distinctiveness of the four types of political participation. The goodness of fit of our model and the distinctiveness of the four types of political participation are thereby assessed.

We used the software R and the lavaan package (Rosseel 2012) for structural equation modeling to analyze the sample variance–covariance matrix. Because the model includes two categorical variables (voting in a national election; voting in a local election), the diagonally weighted least squares (DWLS) estimator was used (Rosseel et al. 2015). The goodness of fit was assessed using the weighted root mean square residual (WRMR), root mean square error of approximation (RMSEA), comparative fit index (CFI), and the Tucker–Lewis index (TLI). An acceptable model fit was defined as RMSEA (≤.06), CFI (≥.95), and TLI (≥.95), WRMR (≤.09) (Brown 2006; Hu and Bentler 1999).

The goodness-of-fit indices suggested an acceptable though not completely satisfactory fit (χ 2 (253) = 3022.821, p < .001, RMSEA = 0.47, CFI = .950, TLI = .943, WRMR = 3.1).Footnote 7,Footnote 8 The covariances and correlations between factors indicate that the four types of participation can generally be treated as distinct constructs; however, differences are found between factors. The factor PP I was most distinct from the others (with PP II: cov = .16, r = .06); PP III: cov = .18, r = .07; PP IV: cov = .21, r = .08). Low covariances and correlations were furthermore found for the connection of PP II and PP III (cov = .29, r = .34) as well as between PP III and PP IV (cov = .31, r = .27). The factors PP II and PP IV (cov = .58, r = .65) showed a moderate covariance and thereby indicate the least distinctiveness for this pair.Footnote 9 All correlations were significant at a p < 0.001 level. A visualization of the whole model, including covariances and estimates, is presented in Fig. 2.

Fig. 2
figure 2

Confirmatory factor model of four types of political participation. Highest estimates fixed to 1. Gray shading indicates a digitally networked activity. Model fit: (χ 2 (253) = 3022.821, p < .001, RMSEA = 0.47, CFI = .950, TLI = .943, WRMR = 3.1). Errors correlated for “Contacted a politician in person”/“Via email or social media contacted a politician to express your opinion” and “Taken part in demonstrations, strike actions or other protest events”/“Encouraged or invited people to take part in demonstrations, strike actions, or other protest events” based on modification indices

Full size image

To facilitate the application of these four participation types in future research, we assessed the internal consistency of the measures that consist of more than two items. The measures for Political Participation II (M = .40, SD = .59, α = .74), Political Participation III (M = .58, SD = .67, α = .73), and Political Participation IV (M = .63, SD = .81, α = .73) all showed satisfactory internal consistency.

Answering research questions

To answer the first research question (RQ 1), the study tested whether the four types of participation can be modeled as distinctive empirical constructs. We find evidence that three of the four proposed types are highly distinct (covariances between .16 and .31) from each other, while moderate covariances between PP II and PP IV (.58) indicate that both modes may be undertaken coherently. The correlations between the constructs reveal further evidence for a satisfying discriminant validity, ranging all well below the critical value of .80 formulated by Brown (2006). The proposed conceptualization furthermore seems to have advantages regarding the distinctiveness of participation types vis à vis previous studies. Compared to Gibson and Cantijoch (2013) and Hirzalla and van Zoonen (2011), we find weaker relationships between different participation types. However, in contrary to Teorell et al. (2007), we do not find negative relationships between different modes, which indicates that people participating in one type do not completely disregard the other three types of political action.

This concentration of items around the different dimensions shows a variety of distinct ways citizens can become politically active. Taking responsibility within the political sphere (e.g., by voting) differs from actions that urge a political or societal change by reattributing responsibility to the political sphere (e.g., demonstrations or petitions). This resembles recent findings by Theocharis and van Deth (2016), who also found voting to be distinct mode of participation. Local-level participation that builds on immediate and self-administered outcomes for one’s community (e.g., by supporting local crowdfunding projects) is distinct from activities that aim at changing a democratic polity in a more global and indirect way (e.g., raising public awareness or following individual lifestyles).

Our second research question (RQ 2) asks if activities relying on online services can be found across the four different types or conform into distinct types. Our results indicate that the proposed concept integrates online and offline activities into the same types. Three of the four types comprise online as well as offline activities, which supports Theocharis (2015) as well as Banaji and Buckingham (2013), who describe diminishing boundaries between online and offline behavior. However, it does not corroborate previous studies that treated online activities as a separate type of participation (e.g., Bakker and de Vreese 2011; Gil de Zúñiga et al. 2014). We note that online and offline activities do not occur with the same frequency across participation types. We did not test whether this is a general pattern that might change in the future or whether it is simply determined by the items we included.

Lastly, we asked whether traditional as well as newer activities on a local level conform into one type of participation (RQ 3). This question can be clearly answered affirmatively. We have strong evidence that PP III can accommodate traditional and unconventional activities targeted at community concerns. This shows that the conceptual map is able to account for and include new, unconventional forms of participation and hence that civic engagement activities can be re-integrated into the concept of political participation.

Discussion

The study aims to apply van Deth’s conceptual map and its theoretically derived decision rules to empirical data. The map suggests a number of different approaches to conceptualize political participation and thereby differs from previous attempts to structure this concept: First, participation types are related to achieving changes or preservations in a democratic system, which is ultimately what political participation is about. Second, its systematic decision rules strive for application across time and could help prevent the emergence of ever-new conceptualizations around upcoming phenomena. Third, waiving nominal definitions might decrease the plurality of different types of participation discussed in the literature and thereby make results more comparable across studies and countries. It is important to determine if these theoretical considerations are reflected in actual behavior among citizens and thereby valuable for future empirical studies. Such a test had yet to be conducted.

While the map relies on thorough and theory-bound considerations, the decision rules are less sound when they are applied to general participatory activities as in most empirical studies. We therefore suggested categories that facilitate the separation of activities from the different types of participation. Overall, we found a close fit between the theoretical map and our empirical model, using data from an extensive national sample in Denmark. The common theoretical and empirical basis of the four dimensions suggests that this new way of structuring the concept may be well pursued by future research. However, not all items met the theoretical principles of the conceptual map and were excluded from the final model for two reasons: they did not comply with the decision rules, and they showed a different allocation in the empirical analysis than expected.

Two kinds of activities were excluded due to their non-compliance with the proposed decision rules: digitally networked activities that are not actions and specific political information gathering. The first group of activities (such as a Facebook ‘like’ of a political article) is excluded from the model because they are not considered actions per se (Q1 in van Deth’s map). We agree with the Theocharis (2015) who argues to exclude acts such as ‘liking’ something on Facebook from the concept of political participation. However, the reasoning that a like “is an expression of preference or an attitude but not an action” (p. 8) bears difficulties. Hamlin and Jennings (2011) describe an action as “a means towards the achievement of a specific purpose” (p. 5). It is hard to deny to someone, who is liking a political post on social networks, the assessment of this being a means to an end while sharing a political message is considered as such. A more helpful approach comes from Brady (1999), who describes a political act as an independently undertaken action and excludes approvals of such political acts as well as the willingness to perform them from being political participation. Sharing a political article or posting a political opinion on social media platforms may be understood as actions citizens undertake independently, while a ‘like’ is an approval of such actions. As a metaphor, one can think about this as a pub conversation: speaking out about a political issue in this (semi)public environment is comparable to sharing a post, while the nodding of a bystander comes close to a ‘like’ on social media. Nevertheless, as recently addressed by Gil de Zúñiga et al. (2014) and Gibson and Cantijoch (2013), seeing forms of political expression such as ‘likes’ as non-political or without political impact may be problematic. The concept of political expression and its relation to political participation therefore need more attention in future research.

Furthermore, van Deth’s framework excludes acts of political information seeking and consumption, such as reading political posts on Facebook or visiting a party website, even though other studies have included such acts in their concept of participation (e.g., Bakker and de Vreese 2011). Of course, pure intake of information may not fulfill the criteria of political participation. However, even though activities through which citizens inform themselves about and process political information do not fit the concept of political participation (Gibson and Cantijoch 2013; Teorell et al. 2007), they may still be relevant for political participation. For instance, in debates on what it takes to be a good, informed and competent citizen (e.g., Bennett et al. 2012; Dalton 2008), information seeking and consumption are seen as ways in which citizens may engage with society and the political system. Therefore, the concept of political involvement (e.g., Delli Carpini 2004; Gil de Zúñiga et al. 2014) may deserve closer consideration.

Few items showed a deviant empirical manifestation compared to the theoretical expectation. The fact that party membership is not a specimen of PP I may be explained with traditionally high turnout (80–90%) but low party membership (3% of the population) in Denmark (Danmarks Statistik 2015). This indicates that country-specific validations of the conceptual map may be necessary. Furthermore, contacting the news media (i.e., a mediated way of raising awareness—PP IV—but not a direct way to target the political sphere—PP II) empirically proved to be a specimen of such direct targeting. Hamlin and Jennings (2011) describe that, on the one hand, contacting media can be seen as an instrumental activity when a citizen wants to achieve a policy change; on the other hand, it may be an expressive activity if citizens only voice dissatisfaction. Although intentions might differ here, neither method targets the political sphere in a direct way; at most in a mediated way. Still, opinions citizens voice via mass media can resonate in the political sphere and among its leaders. Citizens also make their voices heard by commenting on online news media (Anderson et al. 2014) and participating in heated debates about political topics on social media. Such activities that reside—by definition—in the dimension of PP IV are far from ineffective. Future research should look into how a media agenda might be shaped by the voice of the public and how such a reshaped media agenda subsequently can affect policy changes.

Limitations

When we evaluate the results of our study, we have to keep a handful of limitations in mind. First, we used a sample with an over-representation of young and elderly citizens. This secures a sufficient frequency distribution of the analyzed activities, especially because younger citizens use digitally networked forms of participation more often than other citizens. However, this special sample might affect the generalizability of our results. Nevertheless, we obtain an equally high model fit when we only analyze the general sample representative for the Danish population (see footnote 6).

Second, our study is part of a larger research project, which gave us the fortunate opportunity to include a wide range of measures in our model. However, the project was not designed specifically to test the empirical validity of van Deth’s conceptual map, and therefore the formulation of the survey items was not always precise enough to allow for a clear-cut allocation of the items to only one type of participation. We therefore suggest that future studies pay special attention to how the political purpose of an action can be included in empirical measures.

Third, our decision to include a few items in the CFA that did not load exclusively on the expected factor in the EFA was an informed choice. Alternative solutions would have been to include these items into the types suggested by the EFA result or delete the respective item from the model. Both decisions would have resulted in a better model fit. However, our study set out to conduct a full test of the conceptual map with the widest variety of items available and as close to the theoretical principles possible. Although our tested model still reaches an acceptable model fit, we have to keep in mind that the EFA indicates only a weak distinction between PP II and PPIV, confirmed by the mediocre covariances between these two constructs. Formulating the items more closely to the proposed decision rules and empirical separators may increase the distinctiveness between these two factors.

Fourth, we use empirical and internal validation strategies to test the conceptual map, but do not use external validation. Theocharis and van Deth (2016) used an external validation strategy and find evidence of their suggested taxonomy, although their prosed types deviate from the ones tested in our study. In their external validation, they test how well typical antecedents predict six different types of participation. They find, for example, that higher education predicts participation in all six cases, and that older citizens are more likely to vote. In regard to the four types of political participation we tested, external validation should be pursued by future research to discover the characteristics of citizens engaged in each type.

These limitations notwithstanding, our study provides empirical evidence of four distinct types of political participation, derived from van Deth’s conceptual map (2014) and Theocharis’ (2015) refinement. We believe that this conceptualization has advantages over other concepts of political participation because it provides a clear framework, is built on classic assumptions in participation research, and avoids the use of misleading labels. By providing tested measurements for these four types of participation, we hope to make the concept applicable to future studies and to contribute to a common understanding of political participation as well as improved comparability of results.