2.1 A gap in academic literature: representatives’ political knowledge

Political knowledge refers to factual information about politics stored in long-term memory (Carpini & Keeter 1996). Most research has focused on the citizenry, examining their knowledge of political topics and differences among social groups (Galston 2001; Tolbert, McNeal & Smith 2003; Jerit, Barabas & Bolsen 2006; Fraile 2014). However, there has been little attention given to understanding the knowledge that elected politicians possess, especially regarding the various forms of institutional citizen participation.

Current knowledge about political elites and citizen participation merely relates to recent literature on attitudes and preferences towards different processes and factors determining politicians’ and public managers’ support for participatory mechanisms (Svensson 2017; Junius et al. 2020; Pálsdóttir & Óskarsdóttir 2021; Migchelbrink & Van de Walle 2021a; Migchelbrink & Van de Walle 2021b). This literature highlights diverse factors that explain preferences. Those include general attitudinal frames like party ideology and the conception of their representative role (Junius et al. 2020). Other influential factors include their position within the party and chamber and years in office (Vallbé & Iglesias 2018; Rangoni, Bedock & Talukder 2023). Personal characteristics such as gender, socioeconomic status, or age have also been shown to influence attitudes towards participatory mechanisms (Close 2020; Heinelt 2013). However, these characteristics have been less frequently studied, and while there is evidence of their impact on attitudes, we still lack knowledge about whether and how they relate to familiarity with participatory mechanisms.

Apart from these factors, we claim that knowledge asymmetries are likely to arise primarily from two sources: 1) experience and 2) favourability or sympathy towards citizens’ demands and preferences (Close 2020; Junius et al. 2020) in the context of representative democracy. This second aspect gathers different political variables that are presented from H2 to H4.

In the case of experience in office, policymakers with no prior political positions are often more open to implementing citizen participation (Rangoni, Bedock & Talukder 2023; Niessen et al. 2019). In Spain, support for civic engagement decreases with political experience (Vallbé & Iglesias 2018). Thus, differences in attitudes may indicate differences in knowledge. However, experience may also provide insight into various aspects and possibilities of citizen involvement in politics.

H1: Elected representatives with previous political office experience will be more knowledgeable about participatory mechanisms.

Additionally, this paper considers the conception of representatives’ roles (Pitkin 1967; Junius et al. 2020; de Smedt et al. 2024). Representation involves the representative acting as the principal would. However, political representation is complex, generating the mandate-independence controversy: should representatives follow constituents’ instructions, or should they act independently in pursuit of their welfare?

This aforementioned question highlights two different interpretations of the role of representatives. Politicians who favour the mandate view see representatives as delegates or spokespersons of their constituents, whereas those who advocate for independence view them as trustees, acting on their own judgement. This perspective shapes attitudes towards citizen participation, making some representatives more supportive than others and thus more inclined to gather information about it. For instance, ‘delegates’ are more likely to feel they must be closer to citizens’ needs and opinions (De Smedt et al. 2024), so they can also be more attentive to ways for extending the influence of citizens in decision-making.

H2: Elected representatives with a ‘delegate’ view of their roles will be more knowledgeable about participatory mechanisms.

Apart from how politicians conceive their representative role, some politicians may be more receptive to citizen demands, implying favourability towards participatory mechanisms. Some studies affirm that most representatives sympathise with hearing citizens’ demands, and they only reject participation when they are uncertain about their role (Radzik-Maruszak et al. 2020) or when they believe it is not possible given the circumstances (Sønderskov 2020). Despite their final considerations to implement (or not) an innovative participatory tool, we hypothesise that representatives who view citizen participation positively will be more knowledgeable about participatory tools.

H3: Elected representatives receptive to citizen demands will be more knowledgeable about participatory mechanisms.

In spite of its widespread adoption, participatory practices have been pioneered by left-leaning parties (Sintomer 2005; Font & Galais 2011; Heinelt 2013), and differences in implementation and understanding of participation persist between parties and distinct ideologies (Becerril-Viera, Ganuza & Rico Motos 2024; Alves & Allegretti 2012). Thus, ideology’s importance emerges as a hypothesis.

H4: Elected representatives inclined to the left will be more knowledgeable about participatory mechanisms.

3.1 The Spanish case

The Spanish context is not markedly different from other Western states in terms of the development of participatory mechanisms, albeit with uneven implementation between them. Top-down citizen participation in Spain originated primarily at the local level in the late 1970s, with advisory councils as a prominent institution. This initiative was predominantly promoted by leftist governments and mayors (Navarro 1999; Martínez-Palacios 2021). During this period, neighbourhood associations, unions, and other influential stakeholders advocated for a renewed approach to citizen participation, consistent with trends in the broader European-Mediterranean region. Several municipalities in Spain subsequently embraced this participatory culture (Aguiar & Navarro 2000), which also facilitated the adoption of collaborative governance, referenda, and opinion surveys to solicit citizen input and preferences.

Then, other forms of citizen participation emerged. In the 2000s, we witnessed the establishment of the first examples of participatory budgeting, primarily in Southern Spain (Ganuza Fernandez 2005; Ganuza & Francés 2014), which continue to be among the most replicated forms of public engagement. More recently, mini-publics have gained traction in Spain’s participatory landscape, although their implementation remains limited, especially at the national level. The first national citizens’ assembly convened in December 2021, with subsequent regional and local assemblies following suit.

Given this trajectory of participatory practices, it is timely to enquire why elites are adopting these mechanisms. However, it is crucial to first understand what elected representatives know about them and why all this matters for the democratic innovation field.

3.2 Sampling and data collection

We need to measure how much elected representatives know about participatory mechanisms and how this connects with their concerns regarding citizen participation through quantitative data. This study utilises a dataset derived from a web survey administered to Spanish political elites in the context of a project funded by the EU. The survey questionnaire encompassed their perspectives on current decision-making processes, their familiarity with various participatory mechanisms and assessments of potential participation challenges.

Fieldwork began in December 2021 and concluded in February 2022, with an average response time of 27 minutes. The dataset consists of 340 responses from Spanish politicians across three administrative levels—national, regional, and local. First, at the national level, we collected information from the members of the national Parliament (Congreso de los Diputados, MPs) and the Senate (MSs). At the regional level, we approached the information of members of all regional Parliaments (from 17 regions in Spain, MRPs). Finally, at the local level, we collected information on the municipal council members, including the mayor, of the 30 largest cities in different regions.

Accessing information for MPs, senators and parliamentary groups was facilitated by publicly available institutional webpages, which enabled comprehensive coverage of these representatives at both the national and regional levels. In the case of the local level, we used web scraping and direct contact via email or phone where information was inaccessible. This methodological approach is standard in studies involving elite populations due to their limited accessibility (Kertzer & Renshon 2022).

The final sample is compositionally balanced, albeit slightly under-represented by state-level representatives and those affiliated with right-wing or local parties, as it happens in similar studies (Junius et al. 2020). Table 1 shows an overall response rate of 14%, with a noticeable decrease at the national level (11%), primarily influenced by lower responses from the Spanish Congress. Examination of response rates by a political party or party group reveals lower rates for participants from right-wing political parties and local ones. Despite these variations, the final sample diversity adequately facilitates exploration of representatives’ knowledge and perceptions of citizen participation in decision-making processes.

3.3 Measuring political knowledge through survey questions

The measurement of political knowledge in surveys has been extensively debated (Carpini & Keeter 1993; Sturgis, Allum & Smith 2008; Rapeli 2022). This discussion typically revolves around two primary questions: 1) What do respondents need to know? 2) How do you best assess that knowledge? In this study, our focus on knowledge of participatory tools concerns respondents’ familiarity with various types of mechanisms. This encompasses general awareness, rather than detailed knowledge (origins, objectives, limitations, etc.).

To effectively capture this information, multiple questions and formats were employed. It is widely acknowledged that a single question inadequately measures knowledge (Carpini & Keeter 1993; 1996; Sturgis, Allum & Smith 2008). Moreover, given the reliance on web-based survey data, measures were taken to minimise response biases, including potential dishonest responses (Rapeli 2022; Janezic & Gallego 2020; López & Hillygus 2018).

The main starting point is what has been named declared knowledge, which builds on the responses to a closed-ended question on whether or not they knew about six types of participatory mechanisms: advisory councils, mini publics, participatory budgeting, citizen-initiated referenda, opinion surveys and primaries in political parties.⁴ Nonetheless, the answers to the closed-ended question have to be checked. Hence, this paper also examines the answers to an open-ended question, which was previously asked and in which elected representatives had to briefly mention the participatory mechanisms they knew about, giving a maximum of two examples (see survey questions in Appendix 2). In our database, we have 163 completed answers with one or two methods, and 177 expressed either not knowing (155) or not answering (22). In this case, the comparison between both answers allows for the construction of the dependent variable, observed knowledge.

To gain deeper insights into their knowledge and comment briefly on the participatory mechanisms most frequently mentioned by politicians, the results will be completed with a short analysis of this open-ended question using Atlas.ti software. The responses were coded as context code units (Prasad 2008: 13) following the categories presented in the closed-ended question: 1) advisory councils, 2) mini publics, 3) participatory budgeting, 4) referenda, 5) opinion surveys, and 6) others. In the responses, these mechanisms were explicitly mentioned or were explained through an example. Since the latter type of response was more common, a detailed coding scheme with descriptions and examples of indirect mentions or references for each category is provided in the appendix (see Table 1 in Appendix 3).

To explore differences in observed knowledge among political elites, a logistic regression was employed. Thus, the dependent variable is a dummy that stems from combining the closed-ended question and the open one: politicians who provide an answer to the open-ended question and declare to know at least four mechanisms of the list are considered to possess high knowledge (N = 158, 47%, value 1). The rest are coded as low knowledge (N = 180, 53%, value 0).⁵

The independent variables (IVs) include four main variables corresponding to each hypothesis, alongside several control variables capturing the politicians’ profile and political context. Notably, previous experience in office (H1) is coded as a dummy variable indicating whether respondents held office prior to the 2019 mandate (no previous experience, value 0; previous experience, value 1). Another variable aligns with the conceptualisation of their role as representatives (H2), drawing from Pitkin’s (1967) distinction between trustees and delegates. This is operationalised by agreement or disagreement with the statement ‘I know what is best for the public based on my professional expertise’ (delegate, value 1; trustee, value 2).⁶ This distinction relies on one fundamental idea of this two-role conception: whether the representative relies on his or her own judgement or the judgement of a third party to determine the substance of the good at which he/she aims (Rehfeld 2009).

Additionally, the model captures the receptiveness towards citizens’ initiatives (H3), and following the same logic of operationalisation mentioned above, we have considered the agreement or disagreement with the following statement: ‘Governments should be more receptive to the political initiatives organised by civil society and social movements’. Then, pro-citizens’ initiatives value 1, and anti-citizens’ initiatives value 2. Furthermore, the model has also incorporated ideology as an IV in order to test H4. In this case, ideology is a discrete variable from 0 (far left) to 10 (far right).

Also, we add three control variables.⁷ Gender is captured as a dummy variable (men, value 0; women, value 1). Also, age is included as a continuous variable to further explore sociodemographic factors. Lastly, given the varying development of participatory mechanisms across different levels of government, a categorical variable distinguishes between local, regional, and national levels.

Tables 2 and 3 summarise the descriptive statistics of the numeric and categorical independent variables used in the logistic regression performed.

3.4 Potential participation shortcomings

Finally, to explore the relationship between knowledge and politicians’ concerns on citizen participation (H5), several questions were used that align with concerns related to both citizen competence and institutional efficiency. Table 4 below presents the statements that respondents rated in terms of importance (with the smallest number showing the greatest concern), along with descriptive statistics for each variable.

The next section presents the main descriptive results, and the role played by explanatory factors related to knowledge are tested with a logistic regression. Finally, the relationship between knowledge and the perception of potential problems for citizen participation is addressed.

4.1 How much do elected representatives know of participatory mechanisms?

Elected representatives were surveyed about their knowledge of various participatory mechanisms. Overall, more than 90% of respondents reported familiarity with advisory councils, participatory budgeting, referenda and opinion surveys, and 80% in the case of mini-publics (see Figure 1 in the Appendix).

Figure 1: Valid answers to the open –ended question, by declared level of knowledge.

However, this aggregate view obscures individual-level nuances and does not capture informational asymmetries among representatives. The self-reported knowledge is notably high: 71% of surveyed politicians claimed to know all participatory mechanisms, and 23% reported familiarity with at least four mechanisms.

Despite these high percentages, comparing declared knowledge with responses to the open-ended question reveals a sharp discrepancy: nearly half of the elected representatives (45.6%) did not name any mechanisms, while 6.5% did not respond. These non-responses may contradict their self-reported knowledge in the closed-ended question. However, we must also be cautious and consider the possibility of non-response due to the time-consuming nature of open-ended questions (Neuert et al. 2021). That said, these questions were placed at the beginning of the questionnaire and structured according to existing evidence to minimise non-response and improve response quality (Schuman & Presser 1979).

Figure 1 below compares these responses, indicating that elected representatives possess less knowledge than self-reported. For instance, among those claiming to know all five participatory mechanisms (70% of respondents), only half provided one or two examples in the open-ended question. Similarly, among those claiming knowledge of four mechanisms (23%), only 45.6% responded to the open-ended question.

4.2 Participatory mechanisms known by politicians

The responses to the open-ended question were categorised following the response categories of the closed-ended question. From the 163 answers, only 114 were valid⁸ to be categorised (see descriptions and examples of each category in Table 1 in Appendix 3).

Figure 2 shows that participatory budgeting and advisory councils were the most frequently mentioned, each garnering over 30% of mentions, likely due to their prevalent implementation in Spain (see 3.1 section above). Referenda and opinion surveys were also commonly cited, while mini-publics were less frequently mentioned, with respondents typically referring in this category to deliberative citizens’ assemblies.

Figure 2: Participatory mechanisms mentioned by elected representatives in the open question.

Other participatory mechanisms mentioned emphasised proximity to citizens and community engagement. Respondents described these as ways to maintain transparency and close community ties, such as ‘a citizen participation process for drafting a city model and a general urban development plan’.

Having identified the most well-known participatory tools among political elites, the next section examines the factors influencing this knowledge.

4.3 Is knowledge different between distinct representatives’ profiles?

To further understand the knowledge of Spanish political representatives, a logistic regression analysis was conducted with observed knowledge as the dependent variable.

Table 5 below shows the degree of high knowledge for each of the values of the three categorical independent variables.

Now, the results in Table 6 indicate that knowledge is associated with certain factors related to representatives’ political stances, and some hypotheses are confirmed (H3, H4) while others are rejected (H1, H2).

The odds ratio indicates, following this formula Percentage Change = (OR-1) × 100, that receptiveness to citizens’ initiatives is the variable that best explains knowledge, confirming hypothesis 3 regarding sympathy towards citizens’ voices in decision-making. In this sense, rejecting citizens’ initiatives decreases the probabilities of having a high level of knowledge by 45%. Figure 3 illustrates the predictive margins of this variable.

Figure 3: Receptiveness towards citizens’ initiatives and knowledge (predictive margins).

Also as hypothesised (H4), the negative odds ratio of ideology indicates that a right-leaning ideological stance decreases the level of knowledge. As such, an increase of one unit in the ideological scale decreases the probability by 12% of having a high level of knowledge of participatory mechanisms.

However, contrary to expectations, hypotheses 1 and 2 do not reach statistical significance. Experience in office (H1) does not influence knowledge levels, as it seems to shape attitudes. Additionally, the elected representatives’ conception of their role (H2) is not correlated with knowledge levels. It was hypothesised that ‘trustees’ who believe in making decisions independently would know less about participatory mechanisms, but the data does not support this claim. The results suggest that sympathy for citizens’ initiatives is more determinative.

Regarding control variables, the respondents’ gender and the governmental level do not significantly influence knowledge. However, age is found to be a significant factor, although the effect is weaker: older representatives are less knowledgeable about participatory mechanisms, with the probability of high knowledge decreasing by 2% per additional year of age. As some of these participatory mechanisms are entrenched in an innovative way of understanding the relationship between citizens and politicians, or political participation in general, young politicians might be more aware of them, not because experience in office is significant, but rather because a new culture of participation is gaining momentum in democracy and democratic discourses. The political and social implications of these results are further discussed in the conclusion section.

4.4 Why knowledge matters

Does this knowledge about participatory mechanisms influence the assessment of common concerns regarding citizen participation?

The results show that most politicians are more concerned about citizens’ lack of interest in politics and whether they are poorly informed than about potential conflicts or reduced effectiveness resulting from participation (see Table 7). Interestingly, conflict is the least concerning issue for politicians. This suggests that politicians might perceive participation shortcomings primarily as a citizen competence issue rather than an institutional efficiency issue, which is commonly regarded as the main concern of public managers (Migchelbrink & Van de Walle 2021b).

As hypothesised (H5), differences emerge between respondents with high and low levels of knowledge. The importance assigned to potential problems decreases among those with higher levels of knowledge, with scores actually increasing towards 5. An ANOVA test shows that a significant relationship between knowledge and politicians’ appreciation of participation exists, except for the perception that participation takes too much time and effort.

Results suggest that knowledge about participatory mechanisms affects how individuals evaluate various aspects of citizen participation’s shortcomings. However, the magnitude of this impact is relatively small.⁹ These insights imply that while increasing knowledge about participatory mechanisms may slightly shift participation perceptions and concerns, other factors likely play a substantial role in shaping these evaluations, which should be further explored in the future.