Abstract
What is the role of political parties in shaping parliamentary interactions in contexts where party switching is relatively frequent, as observed in many emerging democracies? If parties have a limited coordinating function in young democracies, one would expect current party affiliation to have little influence, with party switchers maintaining their original networks of parliamentary contacts. Conversely, a shift in interaction patterns towards members of the new party would indicate the persistence of party-based organization within parliament. We examine an extreme case: in the 11th Zambian National Assembly, nearly half of the Members of Parliament had switched party affiliation at least once. Using social network analysis of data from face-to-face interviews with 103 Members of Parliament, we analyse patterns of Member of Parliament interaction in parliament. Our findings suggest that current party affiliation significantly structures these interactions. This provides strong evidence for the continuing relevance of parties in emerging democracies – encouraging news for those who view parties as fundamental building blocks of representative democracy.
Introduction
The party may change, but does the dance remain the same? The act of switching parties may seem like a strategic move, but it raises deeper questions about the stability of political structures, particularly within parliamentary systems. Nearly every political system worldwide includes a legislative body – commonly referred to as a parliament. Members of Parliament (MPs), as key actors in these institutions, engage in deliberation, negotiation and debate to shape the political trajectory on behalf of society. Beyond public debates, however, interactions unfold in more informal settings. These social exchanges among MPs form a crucial sociological dimension of the political process. Despite their importance, the factors influencing the formation and dynamics of these parliamentary networks remain largely unexplored. This paper focuses on one key disruption to these networks: the impact of party switching on informal interactions within the house.
Political parties are considered pivotal to democratic representation, serving as essential intermediaries that link society and government by aggregating and articulating societal interests (Diamond and Gunther, 2001; Lipset, 2000; Sartori, 1976). For that reason it seems likely that party affiliation determines which MPs interact (Fowler, 2006; Kopecký, 1996; Müller, 2000; Wonka and Rittberger, 2014). However, this coordinating role may be undermined in contexts characterized by weak ideological cleavages (Young, 2014), low levels of institutionalization (Basedau and Stroh, 2008; Erdmann, 2004; Kuenzi and Lambright, 2001) and frequent party switching among legislative candidates (Arriola et al., 2022; Barrow, 2007; Booysen, 2006) – dynamics often observed in younger democracies. Although research on emerging democracies has predominantly focused on party systems and the role of parties in elections (i.e. Brambor et al., 2007; Weghorst and Bernhard, 2014), there has been less exploration of how party affiliation shapes the interaction patterns of MPs in these settings. In contexts of weak institutionalization and limited ideological differentiation, one might expect parties to function primarily as electoral tools, losing their relevance once MPs begin their parliamentary work.
To better understand the role of parties in emerging democracies beyond elections, we focus on the social dynamics resulting from party switching among MPs. How does the social network of party switchers compare to that of party loyalists? If party switchers predominantly maintain their previous network of MP contacts, this would suggest a party-independent logic driving legislative interactions. Conversely, if party switchers primarily engage with MPs from their current party, it would indicate a party-driven logic of alignment.
We examine MP interactions in the 11th Zambian National Assembly (up to 2016), which serves as an extreme case due to the high prevalence of party switching. Approximately half of the 159 MPs during this legislative term switched parties at least once. For our analysis, we conducted interviews with 103 of these MPs, gathering insights on their bilateral political exchanges with colleagues. Using this unique dataset, we test whether party switchers exhibit distinct patterns of social interaction compared to party loyalists, while accounting for other factors known to influence MP networks in African legislatures (Osei and Malang, 2016). Our findings suggest that, even in highly volatile legislatures such as Zambia’s, party affiliation remains a primary explanatory factor for bilateral political exchanges. We propose that these results likely extend to less extreme cases of emerging democracies, as Zambia not only experiences frequent party switching but is also characterized by particularly weak party institutionalization (Basedau and Stroh, 2008).
The results not only enhance our understanding of legislative interactions and the consequences of party switching but also contribute to the broader research field on the role of parties in emerging democracies. Given the lower levels of institutionalization, limited ideological differentiation and frequent party switching in young democracies – particularly in sub-Saharan Africa – the relevance of parties beyond electoral functions has often been questioned. However, our results demonstrate that party affiliation continues to shape MPs’ social networks even in dynamic environments. Thus, despite the potentially temporary nature of MPs’ loyalty, parties can still rely on their members primarily interacting within their faction, thereby facilitating coordinated parliamentary action.
Our finding that parties appear to matter beyond elections is promising for both the legislative process and the overall state of democracy. Despite the fluidity of party affiliations, voters can generally trust candidates to align for their exchange with the parties they represent. This consistency makes legislators’ behaviour more predictable and enables voters to make more informed decisions at the ballot box.
Party switching in Zambia
Zambia’s unicameral parliament, the National Assembly, is elected at the same time as the president. One hundred and fifty of its seats are distributed in single-member constituencies according to the first-past-the-post system, eight members are appointed by the president, and an additional speaker is elected by the parliament majority.
Despite the introduction of a competitive multiparty system and a slight improvement of judicial independence since the one-party state during the Second Republic 1972–1991 (with the United National Independence Party (UNIP) as the only party left), Zambia still lacks a well-established pluralist political culture. For example, the V-Dem measure of ‘opposition parties autonomy’ maps Zambia in line with the general development in sub-Saharan Africa (Coppedge et al., 2021). There was some improvement in the 1990s, but Zambia never evolved into a full liberal democracy. Zambian parties lack institutionalization and are still in a state of flux rather more characteristic of political movements than of parties (Cheeseman and Hinfelaar, 2009; Erdmann, 2007; Phiri, 2006).
We collected data during the 11th Zambian National Assembly (2011–2016) that was dominated by three parties: the then ruling party Patriotic Front (PF), the Movement for Multiparty Democracy (MMD), which was the ruling party up to 2011, and the second opposition party United Party for National Development (UPND). Although PF is sometimes referred to as a socialist party 1 as opposed to the more liberal party UPND, 2 there are hardly any differences in their manifestos of 2016 (Rakner and Svåsand, 2004). 3 Hence, Zambian parties seemingly attract their electorate on the basis of valence rather than ideological policy positions (Bleck and Van de Walle, 2013). 4
The MMD had its origins in the opposition against Kenneth Kaunda’s one-party rule and state-centred economic policies fostered by economic decline and backed by trade unions (Gertzel et al., 1992; LeBas, 2011; Rakner and Svåsand, 2004; Siachiwena and Wahman, 2024). It garnered support among the rural population with its large-scale fertilizer subsidy programme (Kim, 2017). The PF successfully appealed to the urban population and mine workers on the Copperbelt (Cheeseman et al., 2014; Hinfelaar et al., 2020; Larmer and Fraser, 2007; Resnick, 2014; Sishuwa, 2021) and expanded its support base through a social cash transfer programme and massive expansion of infrastructure (Siachiwena, 2021). In light of spiralling foreign debt and PF failing to uplift local urban economies (Hinfelaar et al., 2020), the UPND, lead by the successful businessman Hakaine Hichilema, positioned itself as a party that had the competence to turn around the economic decline.
Especially striking in the Zambian case is the high volatility of political personnel and their party affiliation (Arriola et al., 2022; Ochieng’Opalo, 2019). Resnick calls this political fluidity a ‘recycling of elites’ (2022: 82). For example, in the 2006 election, although there was no change of the majority party (i.e. the party that won the seat) in 110 of 150 constituencies, only in 53 of those the same candidate won again.
We managed to interview 103 sitting Zambian MPs before the election 2016. In our dataset, the PF is represented with 58 MPs, the UPND with 21 and the MMD with 20 MPs (the other 4 MPs are without party affiliation). Table 1 presents data on party switching among MPs, including the direction of their shifts. A pronounced pattern emerges when examining the differences in switching behaviour between parties: two-thirds of all switchers are now affiliated with the ruling PF. Within our network, 32 of the 58 MPs currently representing the PF previously belonged to a different party. Of these, 27 MPs transitioned from the MMD and 5 from the UPND. In contrast, the PF experienced minimal losses, with only one MP – Sampa Miles – leaving the party to contest as an independent candidate.
Members of parliament (MPs) and information about switching in our dataset.
PF: Patriotic Front; UPND: United Party for National Development; MMD: Movement for Multiparty Democracy (MMD).
The literature highlights two factors that explain the striking pattern of party switching: (1) MPs change parties to increase their chances for re-election, and (2) party elites try to attract the most promisining candidates. We do not answer the question about the reasons for party switching here. Party switching happened before our analysis. However, some insights into the reasons for party switching substantiate our analysis.
At the level of the MP, rational office-seeking behaviour assumes that office-holders look for the best re-election chances. In low institutionalized contexts of emerging democracy and late democratizing states, parties that fail to deliver vote and office pay-offs are especially likely to witness defection (Klein, 2021; Young, 2014). MPs choose the party with the best infrastructure to gain the next election. In Zambia, each of the major parties can rely on a functioning administration at national, provincial and constituency level and own organizations for women and young people that provide office-seeking support. With the involvement of their cadres, it is possible for the candidates of parties to manage campaigns that mobilize people all over the constituencies, 5 thus serving as electoral vehicles (Arriola et al., 2022). Out of 140 independent candidates standing in the parliamentary elections 2011 only 3 won their constituency. 6
At the level of the party, the strong party elite screens and tries to attract the most prominent MPs for the upcoming elections. If MPs in Zambia intend to be re-elected on a party ticket, at first, they depend on being nominated as candidates for a specific constituency. The national party elites choose the candidates. MPs reported in interviews that only if a candidate is sufficiently backed by the central committee of the party are they approved. 7 According to the party secretaries, 8 the non-approval of candidates is a common procedure of parties to get rid of MPs who are suspected of not having shown enough loyalty in the previous term. In a positive reading, parties search for the most successful candidates.
Mechanisms: how can party-switching influence parliamentary interaction
Once candidates are elected as MPs, they begin interacting with their peers in parliament. We seek to examine the factors that explain these interactions. The publicly visible voting behaviour in legislatures is often preceded by political exchanges among MPs. Our focus is on political exchange, an important step in the process of forming political opinions. Through personal conversations, parliamentarians share information, express their views and prepare for upcoming decisions (Giesen and Malang, 2022; McClurg and Lazer, 2014; Osei and Malang, 2016). These interactions provide opportunities to discuss current issues, share insights and seek compromises (Reh et al., 2013). The individual information exchange ties of MPs collectively form a social network. We aim to understand why MPs select specific partners for these exchanges.
Party exchange mechanism: office and policy seeking
Our baseline assumption is that MPs from the same party are more likely to exchange. Parties have traditionally been identified as important ideological focal points that influence not only the formal behaviour of MPs but also their communication patterns (Fowler, 2006; Wonka and Rittberger, 2014). However, newer conceptualizations of parties as networks offer a complementary perspective, emphasizing the fluid nature of actors and connections within a party (Koger et al., 2017). Party networks consist of actors who cooperate voluntarily with a degree of autonomy and may choose to leave one party’s network to join that of an opposition party. This more fluid conceptualization of party affiliation appears particularly fitting for Zambia. Consequently, the extent to which parties influence the social behaviour of their MPs may be less clear.
What does party affiliation promise MPs even in weakly institutionalized contexts like in Zambia? First, as already explained, one main incentive to be a party member is the likelihood of winning the seat in the election, the office-seeking perspective. Second, despite lacking strong ideological underpinning, parties still are grouped together on the basis of valence (Bleck and Van de Walle, 2013). MPs work on policies with like-minded peer. There are two key reasons MPs engage in information exchange with their party peers: (a) to enhance their chances of re-election by becoming a central node in the party elite’s information network; and (b) to influence policy proposals through strategic networking.
Based on this logic, one would typically expect MPs in the current term to primarily interact with colleagues from their own party. This interaction would be aimed at securing their internal position, a crucial step for re-election, and advancing their policy proposals into law. If this assumption holds, we would observe a consistent pattern where MPs, regardless of whether they have switched parties, engage more frequently with members of their current party.
H1: MPs of the same party are more likely to interact with each other than with MPs from other parties.
Stable elites mechanism: socialization and the cost of establishing new contacts
In contrast to the party-determined view on parliamentary networking, a concurring perspective would predict a different effect of party switching on MP exchange. Weak party institutionalization could foster a non-ideological, party-independent network based on pre-existing relationships (Bussell, 2019). There are several explanations why MPs who switch parties are likely to maintain their previous contacts. These can be divided into instrumental (strategic cost-benefit) and expressive (relational and value-based) motivations for MP interactions.
Instrumental motivations are based on staying in power. A relatively closed political elite may form interpersonal ties that are independent of current party affiliation (Resnick, 2022). In Zambia, this effect is also linked to presidential elite balancing, where incumbent presidents build more inclusive coalitions, distribute clientelist resources to their political base and co-opt opposition politicians (Beardsworth and Mutuna, 2022). According to this logic, party affiliation becomes less significant after elections and does not prevent MPs from maintaining close contacts. In this scenario, we would not observe a party homophily effect in the current term. Instead, MPs should interact with one another regardless of party affiliation. Shared policy preferences are another instrumental explanation for keeping contact beyond the party lines (Grossman and Helpman, 2001). We theorize these later for special regional and ethnic interests.
Expressive motivations to keep contacts are rooted in sociological theory. Interactions may persist because of established norms of reciprocity and trust formed over time. MPs may have a sense of obligation or mutual reliance based on prior cooperation, which transcends party affiliation. This norm of reciprocity and trust is a foundation of sociological network theory (Granovetter, 1985). Additionally, forming new contacts may incur higher costs, as it requires time to build familiarity and reduce the uncertainty associated with unfamiliar interlocutors. As Crewe (2015) shows for the UK House of Commons, genuine personal friendships can develop through these repeated interactions in parliamentary settings. Such bonds may endure regardless of party changes, driving continued cooperation and social exchange.
Both perspectives emphasize the importance of initial contact and repeated interaction. If party affiliation is irrelevant, there would be no reason to sever prior ties. To test this, we hypothesize that MPs who were once in the same party will continue their interactions, even if one of them switches parties, thus maintaining their exchanges independently of current party affiliation.
H2: MPs that were members of the same party in a previous legislative term are more likely to exchange even if one MP switched the party.
Representation of inequalities mechanism: policy-seeking
Policies and funds are often targeted at addressing specific inequalities (Boone, 2024). From a policy-seeking perspective, MPs are likely to interact more frequently with colleagues who share similar attributes, regardless of party affiliation (see also Osei and Malang, 2016).
The Zambian first-past-the-post electoral system strengthens the direct accountability of MPs, who are closely tied to specific constituencies and are expected to secure tangible policy benefits for their constituents (Lizzeri and Persico, 2001). Given limited resources, MPs are more likely to succeed by collaborating on projects that benefit multiple constituencies. For large-scale infrastructure projects, it is often impractical to implement them in every constituency. Consequently, MPs tend to prefer securing such projects in nearby areas rather than missing out entirely (Fenno, 1978). This logic applies to initiatives such as well-equipped hospitals, higher education institutions, major road developments, power plants and supply systems. By collaborating with representatives from neighbouring constituencies, MPs can collectively lobby the government to allocate resources to their region. This dynamic is likely to foster a region-based homophily effect.
H3.1: MPs that come from the same region have a higher likelihood of exchange.
The role of ethnicity in political behaviour has been extensively studied (Baldwin, 2014; Basedau et al., 2011; Birnir, 2007; Conroy-Krutz, 2013; Horowitz, 1985; Koter, 2013; Posner, 2005). Contemporary research suggests that although ethnicity in Zambia may not be inherently decisive, it is closely linked to specific economic interests shaped by regional dynamics (Hallink and Siachiwena, 2023). Consequently, policy benefits are often distributed along ethnic lines.
Zambia is a multi-ethnic society comprising 73 tribes. Although ethnically based political parties do not exist and the major parties are multi-ethnic, ethnicity continues to influence voting behaviour (Erdmann, 2007). Voters tend to support candidates from their own ethnic group (Horowitz, 1985: 320; Koter, 2013: 187), as co-ethnicity fosters a sense of solidarity. This solidarity often stems from the conscious or unconscious expectation that ethnic similarity correlates with shared policy preferences, reinforcing redistributional choices (Baldwin, 2014; Birnir, 2007: 9).
H3.2: MPs with a similar ethnicity have a higher likelihood of exchange.
Gender inequality in societies is descriptively represented by female MPs. Recent research shows that women are more likely than men to collaborate on bill co-sponsorship (Wäckerle, 2023). Exchange between female MPs could be used to substantively represent women. We therefore assume that gender homophily should be found among Zambian MPs.
H3.3: Female MPs are more likely to exchange with other female MPs than with male MPs
Data and method
To get to the core of our sociological perspective on MPs, we apply a social network approach. In recent years, networks in political contexts have experienced growing attention, identifying content-based (Carpenter et al., 2004; Henry et al., 2011), social (Berardo and Scholz, 2010; Henry et al., 2011) and institutional (Malang et al., 2019, Malang, 2022) determinants for the formation of exchange relationships between different political actors. Yet this knowledge has been built on the study of western parliaments, whereas insights about non-western legislatures are very limited (Azevedo-Harman, 2012; Barkan, 2009). Exceptions are Osei (2018, 2021) and Osei and Malang (2016).
Data collection
The 11th Zambian National Assembly had 159 members. Of these, 150 were elected in a first-past-the-post electoral system, 8 were nominated by the president, and a non-partisan speaker was elected by a simple majority of the assembly.
In preparation for the data collection, we informed the administration of the parliament and obtained the authorization of the clerk of the National Assembly to carry out this project. Due to her support, it was possible to distribute a letter to all MPs providing information about the research project and asking for participation. In addition, we informed the whips and the administration of the three major parties in personal letters.
For the data collection, we recruited six Zambian research assistants. They all passed a training to ensure a professional conduct of interviews, which we organized together with an experienced programme manager of a local research institute. During their sitting period between February and May 2016 in the year of the election, we interviewed the MPs in individual face-to-face interviews. Using a standardized questionnaire, we collected attribute and relational data for 103 MPs.
Dependent variable
We are interested in the social interactions of MPs. With whom do they talk, exchange information and coordinate? We want to understand with whom MPs interact and what this could mean for the functioning of parliamentary democracy.
To get knowledge of actual exchange patterns of MPs, we asked them to report their network behaviour to us. We closely followed the approach of Osei and Malang (2016) which to our knowledge was the first standardized network study in Africa so far. The wording of the question was adapted for the case of Zambia:
‘Looking back over the last six months, who are the people in the National Assembly with whom you have discussed important political decisions?’
Osei and Malang (2016) argue that this wording is easily understandable and focuses on political exchange in the parliament in contrast to other possible environments. As with all interview data, we rely on self-reported information which might be false due to a conscious decision not to share a piece of information or due to interviewees’ incorrect memory. However, we have aimed to credibly ensure anonymity and to set a reasonable period of time that we ask about.
The exchange between MPs is understood as a directed process coded in dichotomous form – either there was no political exchange (coded 0) or there was a political discussion (coded 1).
Explanatory variables
Our network model allows for two different analytical levels. First, we simply test if a given attribute of an actor increases the likelihood of exchange. Since our data is directed (MP A says she receives information from MP B whereas MP B indicated in the survey that he does not receive information from MP A) we can test how attributes affect the likelihood of either receiving or giving information. ‘IN’ always denotes that an MP is a target (for example, female MPs receive more information) whereas ‘OUT’ indicates that an attribute leads to more information provision (for example, older MPs establish more contacts). On the second level, we test our main hypotheses with a homophily term. The ‘MATCH’ term indicates if two MPs with the same attribute have a higher likelihood of information exchange. We use this term for the effect of party affiliation and previous party affiliation, but also to detect regional and ethnic homophily.
Main predictors: party and party switching
The variable ‘MATCH PARTY’ serves to test if two MPs from the same party have a higher likelihood to exchange in parliament (H1). Our dataset encompasses 58 members of PF, 20 members of MMD, 21 from UPND, the independent MP, the single members from the Alliance for Democracy and Development (ADD) and the Democratic Front (DF), and the independent Speaker. PREVIOUS captures the old party affiliation of an MP. Sixty MPs changed their party, 43 did not change until now. If an MP never changes their affiliation, PREVIOUS has the same value than the normal party affiliation. If an MP changed their party affiliation more than once, we took the party affiliation of the last party change. The term ‘MATCH PREVIOUS’ captures whether two MPs who were previously in the same party have a higher likelihood of exchanging political information. In addition to the nominal variables PARTY and PREVIOUS, we introduce a dichotomous variable ‘SWITCH’ that captures if an MP switched party (SWITCH=1, n=60) or not (SWITCH=0, n=43).
Further independent variables
To test the other three homophily terms, we first introduce a term that investigates if MPs from the same region have a higher likelihood of political exchange (MATCH REGION). There are currently 10 provinces in Zambia. For each MP we coded the province of the constituency that they represent: Central (n=8), Copperbelt (n=14), Eastern (n=14), Luapula (n=8), Lusaka (n=9), Muchinga (n=7), Northern (n=8), North-Western (n=6), Southern (n=13) and Western (n=10).
Second, to test the ethnic homophily hypothesis, we introduce a homophily term for this variable (MATCH: ETHNICITY). It captures whether two MPs have the same ethnicity or not. In order to ease the analysis, we group the ethnic groups into major groups, namely Bemba, Nyanja Speakers, Tonga-Ila-Lenje, Lozi and Kaonde-Lunda using the classification of the Ethnic Power Relations project (Cederman et al., 2010). More information about our coding can be found in Appendix 1.
Third, we introduce a dichotomous variable for GENDER (male=1) and its homophily term to test inequality hypothesis 3.3. Out of 103 MPs, 16 are female. We additionally control if there is a gender gap in network involvement.
We control for several attributes of MPs. Experienced MPs might be better connected and maintain wide networks. This makes them more attractive discussion partners. It is very striking that there is a lot of turnover in the Zambian parliament; 58 of 103 MPs have not been members of the parliament prior to the latest elections. The variable TENURE measures the years in parliament ranging from 1 to 24 with a mean of 6.64. Since the value of experience might decrease over time, we use a log version of the variable. Successful MPs might be surrounded by other MPs who would like to learn from them. Therefore, we use the percentage of votes as a measure for SUCCESS ranging from 0% for nominated MPs to 89.75% of the most successful MP. We introduce a binary variable that captures if members of government (IN GOVERNMENT=1, n=42) receive more political exchange than normal MPs. In addition, we include a measure for AGE. The assumption is that older MPs might be more attractive discussion partners due to their experience and because they might maintain a large network even outside the parliament built over the previous life span.
Model specification
The dependent variable is the realization of a tie, being the incidence of bilateral political exchange between two MPs. We aim to test the above-stated hypotheses on factors determining the formation of ties, in the first place of party and previous party affiliation. In addition, we hope to model the exchange behaviour within the Zambian parliament as well as possible. The network that is to be modelled encompasses 103 MPs being the nodes, with directed ties or edges. 9 Hence, there are n x n – n = 10,506 possible ties constituting the units of observation. Due to the network character, it is likely that the ties are interdependent and there is matrix autocorrelation. Since an ordinary least squares (OLS) or other regression models with similar assumptions are not suitable, we make use of an exponential random graph model (ERGM) instead (Cranmer et al., 2017; Handcock et al., 2008; Hunter et al., 2008). With the ERGM we aim to model the network based on both endogenous properties and node attributes. The node attributes are the independent variables describing the characteristics of the MPs. We fit the model through a Markov Chain Monte Carlo Maximum Likelihood Estimation (MCMC MLE). The resulting coefficients for the node attributes can be interpreted as log-odds ratios of establishing a network tie conditional on the other properties of the network.
Analysis and results
Descriptives
The concrete social network of MPs is visualized in Figure 1. The 103 MPs in the network realize 158 exchanges. Whereas 35 do not exchange at all, the most central actor receives information from 11 other MPs.
Exchange network in the Zambian Parliament.
Sixty of the MPs in our network were previously members of another party. These are the nodes in the network that are visualized in darker blue. We can see that there seems no systematic difference between the switching and non-switching MPs at first glance. The ones who switched are neither solely at the periphery nor at the centre and are distributed over all parties.
Explanatory analysis
The first model in Table 2 tests the party hypothesis, followed by Model 2, which examines the effect of party switching. Model 3 incorporates both variables, and Model 4 extends the analysis by including the effects of inequality homophily. All models include the same control variables and endogenous network terms. 10
Exponential random graph models (ERGMs) of bilateral political exchange.
Note: gwideg.fxed: geometrically weighted indegree distribution.
Model fit measures: Bayesian information criterion (BIC); Akaike information criterion (AIC); Log likelihood.
p<.001, **p<.01, *p<.05
Model 1 and Model 2 differ only in how they account for party affiliation. Whereas Model 1 tests whether MPs select their exchange partners based on their current party membership (MATCH party), Model 2 examines whether previous party membership (MATCH previous) still influences peer selection after party switching. The difference in coefficients is notable. MPs are highly likely to select their bilateral contacts in parliament based on present party affiliation, as indicated by the highly significant coefficient in Model 1. In contrast, the effect of previous party affiliation on exchange networks is much weaker. However, when previous party affiliation is introduced without controlling for present party membership, we still observe a significant effect. Once we add the present party homophily term to our ERGM equation, Model 3 shows that the effect of previously shared party membership becomes insignificant.
What do these results mean in light of our theoretical expectations? First, we find that even in weakly institutionalized and volatile legislatures like Zambia’s, party membership largely determines MPs’ present interactions in the House. This provides support for Hypothesis 1. In our interpretation, political parties – whether by offering re-election benefits or facilitating the joint pursuit of policy goals – play a crucial role in shaping MPs’ social networks. Once MPs belong to a party, they tend to align their political exchanges accordingly. We also observe that these loyalties primarily shift when an MP switches parties. Hypothesis 2 does not receive confirmation. Although there is some weak evidence that old ties beyond party cleavages influence Zambian politics, our results do not support the claim that this mechanism is a primary driver of social interactions in the National Assembly. Sociological theories like behaviour on the basis of reciprocity seem to be less important than political drivers to explain parliamentary exchange. Overall, Models 1–3 suggest that party democracy is strong, even in highly dynamic legislative environments.
We now turn to the question of whether MPs who switched parties at least once behave differently from their party-loyal peers. As shown in the descriptive statistics, MPs predominantly switched to the governing party, PF. In our statistical analysis, the term ‘IN switch’ captures the probability of party switchers receiving more or fewer contacts than non-switchers. Across all four models, we find no significant difference between the two groups – MPs do not appear more or less likely to cooperate with a party switcher. The complementary question – whether party switchers are more likely to initiate bilateral political exchanges – is addressed by the term ‘OUT switch’. Here, we observe a weakly significant negative effect, suggesting that MPs who switched parties tend to initiate fewer bilateral exchanges. However, this effect is weak and inconsistent across the four models, meaning it should not be overemphasized.
To further explore the social network effects of party switching, we visualized the indegree and outdegree measures for the entire network in Figure 2. The upper panel displays the distribution of indegree (i.e. how often an MP is chosen as a target of information exchange). We observe that the MP receiving the highest number of exchange ties is a party switcher who is now a member of the UPND. The next four most central MPs, all from PF (represented as circles), include three switchers and one non-switcher. In line with the ERGM results, this visualization suggests that some highly central MPs are party switchers. We interpret this as an indication that these MPs are either strategically selecting the most promising party ahead of elections or are being recruited by party elites due to their prominence. Another important observation from the indegree distribution is that many party switchers are found at the bottom of the figure, receiving no information at all (25 switchers vs 10 non-switchers). Taken together, these two findings help to explain the non-significant effect observed in Table 2. The lower panel, which visualizes the outdegree distribution (i.e. how often an MP initiates information exchange), does not reveal a similarly clear pattern.
Visualization of indegree (upper panel) and outdegree (lower panel). Higher positions indicate more degree.
We examined whether exchange relationships are established based on regional, ethnic or gender similarity. Model 4 in Table 2 presents the results. Consistent with the findings of Osei and Malang (2016) and Boone (2024), we observe that MPs from the same region are more likely to exchange information. We interpret this as evidence that policy-seeking behaviour, particularly efforts to secure infrastructure projects for their constituencies, may be a key driver of this pattern. In contrast, shared ethnicity does not increase the likelihood of bilateral political exchange between MPs. Additionally, the gender homophily term is significant, indicating that female MPs are more likely to exchange information. This may suggest a shared interest in advancing policies related to gender equality.
Conclusion
We examined whether party affiliation influences the social networks of MPs, particularly in the context of frequent party switching, which is more common in younger democracies. This phenomenon raises important questions about the coordinating role of parties beyond elections. Using Zambia as a least-likely case for strong party influence, we analysed whether MPs who switch parties maintain their existing network of contacts or shift their interactions towards members of their new party. If party switching has little effect on MP interactions, it would suggest that parties play a limited role in structuring legislative networks. Conversely, if party switchers realign their connections to reflect their new party affiliation, this would indicate that party structures continue to shape political behaviour even in younger democracies.
To test these contrasting hypotheses, we collected data through face-to-face interviews with 103 Zambian MPs in 2016 and applied social network analysis. The results of the ERGMs indicate that political parties play a significant role in shaping MPs’ choices of exchange partners within the legislature. Regardless of their prior affiliations, MPs predominantly interact with members of their current party, underscoring the continued relevance of party structures. However, our findings also reveal that party switchers do not sever ties with their previous networks; instead, they expand their connections by incorporating new party contacts. This suggests that although Zambia’s political elite maintains stable cross-party relationships, parties continue to serve as crucial opportunity structures that influence legislative interactions. Ultimately, our study highlights the dual nature of party affiliation in a young democracy – both reinforcing partisan bonds and allowing for broader elite networking beyond party lines.
The finding that party affiliation shapes MPs’ exchanges even in a highly fluid party system underscores the enduring role of parties in emerging democracies, particularly in the broader sub-Saharan context. Despite significant differences from consolidated party systems – such as lower levels of institutionalization and weaker ideological distinctions – political parties continue to structure legislative interactions and influence political processes. This suggests that, even in settings where party affiliation is frequently in flux, parties remain important organizational forces in shaping political behaviour.
One possible explanation for this at least temporary partisan loyalty is the strategic calculus of re-election. Maintaining close ties with party colleagues helps MPs build a support network, thereby increasing their chances of renomination. For parties, this is beneficial as it encourages MPs to align their legislative exchanges along party lines, contributing to more predictable and coordinated decision-making processes. For voters, it enhances electoral transparency, as candidates’ future political alignments become more foreseeable when their interactions are primarily within the party. More broadly, this reinforces the role of parties as fundamental institutions of representative democracy, contributing to the stability and functionality of the democratic process.
Future research should explore whether party switching influences legislative voting behaviour. Given our network analysis results, we would expect MPs to demonstrate loyalty to their current party by voting in line with the party’s position. At the individual level, further studies could examine the impact of party switching on re-election prospects. Since we suggest that MPs maintain partisan loyalty due to anticipated electoral benefits, it would be valuable to investigate whether those with stronger intra-party networks have a higher likelihood of renomination. Additionally, to deepen our understanding of the role of party affiliation in shaping MPs’ social networks, future research could expand the scope to include data from multiple countries. This would enable a comparative analysis of how parliamentary networks function across different political regimes. Overall, this study underscores the resilience of party structures in shaping legislative behaviour, even in young and dynamic democracies. Although party systems may differ in their level of institutionalization, the fundamental role of parties in organizing political interactions remains evident.
Supplemental Material
sj-docx-1-ips-10.1177_01925121251343841 – Supplemental material for Late to the party: Do party switchers behave differently in parliament from party loyalists?
Supplemental material, sj-docx-1-ips-10.1177_01925121251343841 for Late to the party: Do party switchers behave differently in parliament from party loyalists? by Axel Bayer and Thomas Malang in International Political Science Review
Footnotes
Acknowledgements
We are grateful to four anonymous reviewers as well as to Anja Osei, Katharina Holzinger and Volker Schneider for helpful comments on previous versions of our paper, and to the Southern African Institute for Policy and Research, and especially Marja Hinfelaar, for providing such an excellent platform for academic exchange. Furthermore, we thank Kawide Nakasamu, Imasiku Makufele, Jimmy Walubita, Francisca Mpike, Mwewa Nkandu, Mwiche Musukuma and Josephine Chanda for their invaluable support in collecting the interview data in Zambia. We would like to thank the members of the 11th Zambian National Assembly and the parliamentary administration for their kindness, trust and time that enabled this research.
Authors’ note
Although the corresponding author is a German diplomat, this publication does not reflect the official policy or position of the German government.
Funding
The authors disclosed receipt of the following financial support for the research, authorship and/or publication of this article: This work has been supported by the German Research Foundation (grant number HO 1811/10-1). Thomas Malang received funding by the Emmy Noether programme, Grant No. 447624982.
Supplemental material
Supplemental material for this article is available online.
Notes
Author biographies
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
