Introduction

In recent decades, democratic systems have been dealing with more and more complex and politicized issues, such as climate change, the Covid-19 pandemic, immigration and global financial crises. These issues are characterized by high levels of ambiguity and uncertainty, and their political effects generate high pressure for action for policymakers (Auld et al., 2021; Head, 2022).

Policy responses to these challenges have been often criticized by scholars for failing to achieve their declared objectives or even producing unintended effects (Castles, 2004; Czaika & De Haas, 2013; Latin, 2012; Massey et al., 2016), and for not being ‘grounded on objective evidence’ (see also Baekkeskov, 2016; Morgan & Di Giulio, 2018). Particularly in the field of international migration, scholars have often identified gaps between policy developments and objective evidence, highlighting an ‘evidence-policy gap’ (Cairney, 2016, 13; see also: Baldwin-Edwards et al., 2018; Scholten, 2020) and called upon policymakers to adopt evidence-based approaches to address policy problems (Scholten, 2020: 69) and to reduce policy failure (Castles, 2004; Czaika & De Haas, 2013; Massey et al., 2016).

The existing scholarship identifies different explanations for this ‘evidence-policy’ gap in the migration policy field. Scholars have particularly focused on knowledge availability and knowledge use by policy actors (Baldwin-Edwards et al., 2018), portraying this ‘evidence-policy gap’ as driven by constraints on information and resources (Cairney, 2016), or the irrationality of decision-making processes (Scholten, 2020), or political and strategic interests, which overrule objective evidence (Baldwin-Edwards et al., 2018, 10).

Focusing on knowledge availability and knowledge use by policy actors, the scholarship has largely neglected, though, the processes through which policymakers form their knowledge about highly contested migration policy issues. This is an important research gap, as scholars working on policy analysis have shown that policymakers’ understandings of policy problems can decisively influence policymaking processes (among others: Dupont et al., 2023; O’Connor et al., 2023; Radaelli, 1995). Ackelson (2005), for instance, has shown that information by US Border Patrol about ‘disorder’ and ‘chaos’ caused by unauthorized immigrants entering the USA has strongly contributed to the militarization of the US–Mexico border. Another, more recent, specific example are information campaigns, like the one that was set up in 2014 by the Obama Administration in response to the so-called ‘unaccompanied children crisis’ at the Southern border,Footnote 1 which was based on elite policy actors’ idea that such crisis was caused by liberal migration policies or, more precisely, the perception of policies as liberal by migrantsFootnote 2 (see Hadj Abdou, 2020: 647). Also beyond the migration policy field, the existing scholarship suggests that knowledge formation processes are often not linear, particularly in policy fields characterized by ambiguity, uncertainty and high politicization (Brunsson, 2000; Cohen et al., 1972; Kingdon, 2014).

To fill this gap in the existing literature, this article therefore asks: what is considered ‘evidence’ by elite policymakers working in the migration policy field and where does it come from? What are the sources and actors through which policymakers form their understandings of politicized and complex migration policy problems that influence their policy decisions?

To answer these questions, and with the ultimate aim to formulate potential explanations for the observed ‘evidence-policy gap’ related to knowledge formation and source effects, we focus on the ‘heuristic case’ (Eckstein, 1975) of elite policy actors responsible for asylum and ‘irregular’ migration policy in the USA under the Obama and Trump Administrations. These policy actors include government officials and civil servants from key US institutions with migration-related portfolios, such as leading representatives of the Department of Homeland Security (DHS), of the DHS’s agencies—including the US Customs and Border Protection (CBP), the US Immigration and Customs Enforcement (ICE) and the US Citizenship and Immigration Services (USCIS)—the State Department, the National Security Council (NSC) and the Domestic Policy Council (DPC).

To address our research questions, we adopt an actor- and network-centred approach, deriving theoretical insights from the existing literature on knowledge transfers within policy networks, which assumes that policy networks and interactions therein are channels through which knowledge diffuses (Füglister, 2012) and that their relational characteristics can enable or constrain knowledge diffusion processes (Bednarz & Broekel, 2019; Stone et al., 2020).

Methodologically, we combine qualitative content analysis (Zhang & Wildemuth, 2005) of 57 interviews conducted in 2015 and 2018 and social network analysis (SNA) which elaborates relational data collected through a structured survey (see Borgatti et al., 2013). Interviews were conducted with high-level representatives of the above-mentioned key institutions with migration-related portfolios within the US Federal Administration, under both the Obama and Trump Administration, and with a wide range of other actors involved in US migration policymaking. As to the US Federal Administration, we aimed to reach the top-level representative of each of the above-mentioned institutions and, when we were unable to contact the highest-level representative, we tried to reach top-level officials from each of the main directorates, agencies or components engaged in migration work within that institution: our interviewees therefore include top-level decision-makers involved in the development and enforcement of key US migration policy measures.

More specifically, our analysis is conducted in two steps. First, the SNA is developed to map the US migration policy network and identify the dominant elite actors within the US Administration dealing with migration issue (core policymakers) and the most relevant actors that can potentially broker knowledge (knowledge producers). Second, both the qualitative content analysis and the SNA are used to explore patterns of interaction between core policymakers and knowledge producers and test three guiding expectations derived from the literature on knowledge diffusion within policy networks.

Our analysis partly challenges and adds new perspectives on the three key arguments identified in the existing debate on evidence-based policymaking in the migration policy field, concerning i) the existence of an ‘evidence-policy gap’ and policymakers’ ‘bounded rationality’, ii) the presumed irrationality of decision-making processes and iii) the dominance of political interests in policymaking processes.

We show that, first, elite US migration policymakers, during the period analysed, regularly consulted external sources and interacted with a wide range of other governmental actors and international organizations but also with nongovernmental actors such as think tanks, academia and NGOs. This challenges existing explanations that see the observed ‘evidence-policy’ gap as particularly driven by knowledge unavailability due to existing constraints of time, information and resources that policymakers face when addressing complex and politicized issues such as migration.

Second, our findings challenge existing arguments about the presumed irrationality of decision-making processes, showing that US migration policymakers played an active role in collecting information, selecting their sources and attributing different weight or relevance to the ‘evidence’ acquired. We instead generate a new possible explanation for the observed evidence-policy gap, related to how dynamics of (perceived) ideational proximity guide the way in which policymakers selected their sources of information. When establishing which information qualified as ‘evidence’, US migration policymakers tended to rely primarily on internal sources of information, and, if information produced internally was not available, on ‘like-minded’ actors, i.e. actors with similar organizational mandates, or actors that shared similar political or ideological views or perspectives on the issue of migration.

Third, our findings complement existing explanations of the observed ‘evidence-policy gap’ focused on political interests, generating a new explanation of the ‘evidence-policy gap’ focused on the role of trust and credibility in knowledge acquisition processes. While we cannot and do not aim to exclude or deny that political interests might overrule objective evidence in decision-making, our findings suggest that political and ideological factors might decisively influence the processes through which policymakers gather information and determine what qualifies as 'evidence' in the first place. In the heuristic case analysed, not only actors with perceived different ideologies were less frequently consulted, but elite policymakers' assessment of the credibility of knowledge producers and their trust in the information they provide were largely rooted in considerations of the ideological profile of these knowledge producers. For instance, during the Trump administration, elite US policymakers dismissed evidence produced by NGOs, academia and progressive/nonpartisan think tanks as ‘not credible’ due to the perceived partisanship of these actors, who were seen to belong to a different camp than the Federal Administration at the time. This perceived partisanship was considered a potential source of bias in the information they provided.

Approach

This section reviews the existing literature that provides explanations for the ‘evidence-policy gap’ observed in the migration policy field, and it then illustrates the interpretative and network-centred perspective that we adopt to approach our research questions.

Existing explanations of the ‘evidence-policy gap’ in the field of migration.

Several scholarly works in the last decades have identified gaps between these policy developments and objective evidence, in different policy fields such as health (see e.g. Oliver et al., 2014; Baekkeskov, 2016; Berger et al., 2021) and climate change mitigation (among others: Auld et al., 2021; Levin et al., 2012; Morgan & Di Giulio, 2018), but also, increasingly, in the field of international migration (Scholten, 2020; Cairney, 2016; Ruhs et al., 2019). While most of these existing works on the ‘evidence-policy gap’ in the migration policy field (as in other fields) are developed to recommend solutions to close such gap (e.g. Ruhs et al., 2019), a few research works have also tried to identify explanations for this observed gap. Our review of this existing literature suggests that three main explanations have been proposed by migration policy scholars.

A first strand of research connects to the ‘bounded rationality’ argument in the policy science literature (Simon, 1982), suggesting that that the observed ‘evidence-policy gap’ is due to ‘cognitive limits of policymakers, and an unpredictable policymaking environment’ that imposes constraints on information and resources, particularly in situations of crisis and when the policy issues at stake are highly complex and ambiguous (e.g. Cairney, 2016, 13). Because of such limits, this literature suggests, policymakers tend to overlook ‘valuable insights from alternative sources’, and thus misinterpret the situation, which often led to ‘suboptimal decisions with possibly disastrous consequences’ (Berger et al., 2021, 1).

A second strand of the literature instead suggests that the evidence-policy gap is rather due to the ‘irrationality’ of policy-making processes (see Auld et al., 2021; Levin et al., 2012). For instance, Scholten (2020, 113) argues that policy processes in the migration policy field fail to capture the complexity of the migration issue and are ‘alienated’ from objective evidence. This is described by Scholten as ‘not something actors pursue on purpose’, but ‘an inclination in actor behaviour that originates in broader structural settings’ and their detachment from the complexity of real-world phenomena.

Finally, a third strand argues that the evidence-policy gap is due to policy-making being driven by ‘underlying assumptions’ and ‘vested political interests’, which overrule objective evidence (Baldwin-Edwards et al., 2018, 10). Following this third strand of research, policy-evidence gaps are due to policymakers’ deliberate choices to ignore available evidence, due to their political strategies. Similar arguments have been also raised by scholars working on other policy fields (Latin, 2012; Audl et al., 2021).

While focusing on knowledge availability and processes through which (different types of) knowledge is used (or not) by policy actors, the existing literature has largely neglected the processes through which knowledge about migration policy issues is formed. This is an important research gap, as scholars working on policy analysis have shown that policymakers’ understandings of policy problems can decisively influence policymaking processes (among others: Dupont et al., 2023; O’Connor et al., 2023; Radaelli, 1995). In the specific field of migration, policymakers face a variety of (often contrasting) signals and cues in their policy environment, which makes processes of knowledge formation even more likely to be non-linear. Boswell et al. (2011) have for instance shown that migration policy debates not only revolve around competing values, but also relate to knowledge claims held by policy actors (on the interplay between ideology and knowledge in migration policymaking see also: Collingwood et al., 2019). Several actors are currently involved in migration policymaking, and these actors hold very different interpretations of the causes and effects of migration (diagnosis) and propose very different solutions to related challenges (prognosis). Scholten (2013) similarly argued that migration is an ‘intractable policy controversy’, meaning that actors disagree on both the solutions to be adopted, and the definition of the problem.

An interpretative, network-centred approach.

To analyse processes of knowledge production in migration policymaking, this article adopts an interpretative perspective (Weick, 1995). Following such a perspective, policy responses to complex and ambiguous issues developed by elite policymakers are to be understood as the result of how policymakers construct meaning about the causes and effects of these ‘policy problems’ (and related events). Such interpretation processes are ongoing, due to the high complexity and ambiguity of the underlying phenomena, which require continuous diagnostic assessments and prognostic evaluations by policymakers (Hadj Abdou, 2020).

Building on the sense-making tradition (Weick, 1995), Geddes (2021) specifically argues that four key factor shape actors’ interpretations of the migration issue, its causes and effects: actors’ identity, their past experiences, their interactions with other actors and the ‘signals and cues’ that actors extract from their environment. These two latter points are a crucial starting point for our analysis: policy actors making sense of the ambiguity and uncertainty related to complex policy issues such as migration tend to look for available information and interpretations that are already ‘out there’ in their environment, and they develop interpretations through social interactions with other actors within such environment. In other words, this means that decision-making processes related to complex, uncertain and politicized issues, following this interpretative approach, largely depend on transfers of knowledge between actors within the policy system and that examining knowledge formation processes and processes of knowledge diffusion between actors is crucial to understand policy responses.

To operationalize our interpretative perspective, we develop a network-centred approach, in line with the works done by a number of scholars who have explored knowledge transfers involving elite policymakers (Füglister, 2012; Krenjova & Raudla, 2018; Stone et al., 2020). This approach implies conceptualizing the policy system within which knowledge diffusion processes take place (in our case, the US migration policy system) as a network of actors—including institutions and organizations—that interact with each other exchanging information (the ‘policy network’; Rhodes, 1990). Policy networks and interactions therein—we assume—are channels through which knowledge diffuse (Füglister, 2012) and the composition and relational characteristics of the policy network can enable or constraint knowledge diffusion processes (Bednarz & Broekel, 2019; Stone et al., 2020). While scholars have occasionally referred to the relevance of actors’ social interactions for knowledge production (Hadj Abdou & Pettrachin, 2023), elite processes of knowledge formation have so far not been investigated through network-centred approaches. These are arguably appropriate to explore knowledge-formation dynamics in the migration policy field and very well fit with the interpretative perspective described above. Scholars have indeed shown that, through information exchange, actors can reduce their uncertainties regarding different policy challenges and thus improve decision-making processes, particularly in complex systems with ever-changing conditions or high levels of uncertainty (Pahl-Wostl et al., 2007).

Building upon existing studies (Kammerer & Namhata, 2018; Kim et al., 2011), to study knowledge transfers within policy networks, we introduce the concepts of ‘core decision-makers’ and ‘knowledge broker’. With the term ‘core decision-makers’, we refer to the elite decision-makers that are the main object of our analysis. These—in our framework—are the potential ‘receiver’ of information within the policy network (see e.g. Graham et al., 2013). In Sect. "Conclusion", we rigorously identify the core decision-makers within the US migration policy network. Given the US governmental architecture, and existing literature on US migration policy (for an overview see Pope, 2020), we expect core decision-makers to include, first and foremost, the US Department of Homeland Security (DHS) —which is mandated to secure the border, regulate immigration and set immigration policy—and its agencies such as CBP, ICE and USCIS. We also expect core decision-makers to include other Federal Departments and agencies with migration-related mandates that remained outside the DHS structure, including the State Department.

Following network scholars (Angst & Brandenberger, 2022; Dobbins et al., 2009; Kim et al., 2011), a ‘knowledge broker’ within a policy network is a key intermediary actor which provides links, knowledge sources and in some cases knowledge itself to other actors in their network and, particularly, to core decision-makers. Douglas et al. (2015) distinguish between ‘top-down’ knowledge brokers—actors that exert top-down pressure on core decision-makers from higher-level jurisdictions—and ‘epistemic’ knowledge brokers that include actors such as advocacy groups, professional associations, trade unions or think tanks, that diffuse information, evidence about policies and knowledge, and can also contribute to construct legitimacy for certain policies and play a key role in agenda-setting (Stone, 2000) and can be critical for promoting innovative actions and policies within their networks (Dobbins et al., 2009). These actors can play the role of ‘policy entrepreneurs’ (Kingdon, 2014, 129), i.e. ‘advocates for proposals or for the prominence of ideas’, who ‘actively seek to promote policy change’ and are willing to invest their resources of expertise and persistence to produce knowledge and bring in new perspectives and information to promote such policy change (Füglister, 2012).

Three conditions for knowledge transfers

Recognizing the importance of knowledge exchange in inter-organizational policy networks, a growing body of literature has tried to identify the factors that shape, advance and impede knowledge exchange in different policy fields (Fischer, 2022; Leifeld & Schneider, 2012; Wagner & Ylä-Anttila, 2018). There is no consensus within that literature regarding the conditions or factors that are sufficient to produce knowledge transfer between knowledge producers and elite decision-makers. However, we can derive at least three conditions or factors from existing research that are considered by scholars to be necessary (though not sufficient) for knowledge transfers between knowledge producers and elite policymakers to occur. These conditions will be studied in this paper. This list does not intend to be comprehensive but includes key factors that are particularly relevant for analyses focusing on policy issues that are complex, ambiguous and contested, and, particularly on the migration policy field, considering its key features previously described in this article.

Knowledge availability

A first necessary condition for knowledge diffusion is the availability of knowledge and, more specifically, the existence of exchanges that can lead to ‘knowledge transfers’ between core decision-makers and knowledge producers (Nilsson, 2019). In other words, transfers of knowledge depend on core decision-makers’ access to the potential source of information. More specifically, this means that, for transfer of knowledge and ‘evidence’ to take place between a knowledge producer and core decision-makers, the knowledge producer needs to have regular interactions with core decision-makers (or more regularly than other actors). The existing literature suggests that the likelihood of knowledge transfers is related to the frequency of direct exchanges between core decision-makers and knowledge producers acting as knowledge brokers (see e.g. Bahar et al., 2023; Füglister, 2012; Nilsson, 2019; Yilmaz, 2023; An et al., 2023). Despite the literature suggesting that knowledge producers can also diffuse knowledge and evidence through conferences or manuals and books (Douglas et al., 2015; Karch, 2007; Krenjova & Raudla, 2018), scholars tend to agree that knowledge transfers are much more likely to happen when there are direct contacts between actors, e.g. through lobbying processes (see e.g. Stone, 2000 on the role of think tanks in policy transfer, showing that think tanks manage to diffuse ideas mainly through their networking activities).

Ideational proximity

Several scholars (Füglister, 2012; Kingdon, 2014; Stone, 2000) have pointed out that the mere availability of knowledge is not sufficient for knowledge transfer to occur. Another necessary condition for knowledge transfer is related to the ‘ideational proximity’ between core decision-makers and knowledge producers (Knoben & Oerlemans, 2006). Put differently, knowledge transfers between actors require the potential sources of information to be perceived as holding similar, or compatible, interests and perspectives by the core decision-makers that receive the information. Knowledge transfers can be indeed hampered or impeded by divergences of views and perspectives between actors (Angst & Brandenberger, 2022) which, in fact, often explain the unequal, selective and potentially too low diffusion of knowledge’ (Bednarz & Broekel, 2019, 1460). In particular, research has shown that transfers of knowledge are less likely to take place between organizations that stand on different sides of political divides and hold differing ideological or political views, or different beliefs about fundamental values and different preferences for policy solutions (Leifeld & Schneider, 2012; Wagner & Ylä-Anttila, 2018; the role of likeability for knowledge transfers is also illustrated by: Sniderman et al., 1991; Zaller, 1992). Also, they might be related to inter-institutional or inter-organizational conflicts, related to clashing mandates or competences between institutions (Bednarz & Broekel, 2019; Boschma, 2005; Buisseret et al., 1995). Acknowledging ideational proximity when analysing migration-related knowledge diffusion processes seems crucial due to the high politicization and polarizing nature of the migration issue in the USA in the past decades (Abramowitz & McCoy, 2019; Tichenor, 2021).

Trust and credibility of sources

Third, scholars have shown that exchange of complex information largely relies on the existence of trust between actors and on the perceived credibility of knowledge producers (Nilsson, 2019). Where actors have trust on the source of information, they tend to be more willing to accept a message or piece of information shared with them. In other words, a meaningful transfer of knowledge requires core decision-makers to consider the information shared by knowledge producers—but also the procedures through which this knowledge is produced—as trustworthy, valuable or credible. The importance of sources’ trustworthiness and credibility has been well documented by scholars in various fields (Olson, 2003; Popkin, 1994; see also Stone, 2000 on the role of think tanks in knowledge diffusion processes).

Building upon these theoretical insights, we formulate three guiding expectations concerning potential explanations for the observed evidence-policy gap in the migration policy field related to source effects and knowledge production processes. First, we expect that the observed evidence-policy gap might be due to knowledge unavailability, or too infrequent contacts between elite decision-makers and knowledge producers. Such explanation would be largely in line with the first strand of the existing literature identified in Sect. "Approach": if the expectation is not confirmed, this would therefore challenge arguments about the ‘evidence-policy gap’ being due to policymakers’ ‘bounded rationality’. Second, we expect that the evidence-policy gap might be due to perceived ideational differences between elite decision-makers and knowledge producers, which obstruct knowledge diffusion among them. Third, we expect that the observed evidence-policy gap might be due to lack of trust by elite decision-makers in the information produced by knowledge producers. If the second and third expectations are confirmed, they would challenge claims based on the second and third strands of the exiting literature on the evidence-policy gap identified above. In other words, these explanations, if confirmed, would challenge claims that the evidence-policy gap is due to the irrationality of decision-making processes or to policymakers’ strategic and political interests. We test these expectations for both of the analysed time periods.

Case selection

To test our guiding expectations, we focus on knowledge formation processes in the domain of ‘irregular’ and asylum-seeking migration (henceforth: migration) in the USA, which performs the function of a ‘heuristic case’, i.e. a case study that allows to ‘inductively identify new hypotheses, variables, causal mechanisms, and causal paths’ (Eckstein, 1975).

This is an interesting case to analyse processes of knowledge formation for at least three reasons. First, between 2015 and 2018, the USA experienced an increase in arrivals of unaccompanied migrant children and women from Central America, which led to several ‘border crises’.Footnote 3 Scholars have argued that situations of (perceived) crisis are particularly suited to examine knowledge formation processes (Maitlis & Christianson, 2014, 58). In such situations, knowledge is more likely to be unavailable, as actors must develop strategies and make decisions with scant information available and under high pressure for actions.

Second, between 2015 and 2018 migration was a highly salient issue in the USA (Tichenor & Miller, 2020). Under both the Obama Administration and the Trump Administrations, characterized by different ideological and discursive stances on immigration, the immigration issue has played a key role in influencing domestic US politics (Tichenor, 2021). The high salience of immigration potentially increases the instrumental and symbolic role of evidence in policy making as opposed to the substantial use of evidence and thus calls for an analysis of processes of knowledge formation.

Third, migration in the USA is very polarized. The country and its political elites have been divided over policies governing immigration and immigrants’ rights (Schreckhise & Chand, 2023). Immigration policy has produced some of the most contentious battles in American politics (Hopkins, 2010; Tichenor & Miller, 2020, 371). Most indicative for these controversies is the longstanding stalemate over immigration reform (e.g. Tichenor, 2021). Polarization has intensified in the past decades: the Democratic Party has shifted toward a more pro-immigration stance, advocating for the legalization of irregular immigrants, while the Republican Party has grown more hostile towards immigration. This divide was especially pronounced during President Trump's tenure, as he further entrenched immigration as a partisan issue (Martin & Orrenius, 2022, 89). These controversies ultimately revolve around varying interpretations of the causes and effects of immigration (Hadj Abdou, 2020), rendering this case particularly pertinent for analysing the influences of ideology and politics on decision-making processes and knowledge production.

Methodology

To address our research questions, we conducted 41 interviews in 2016, and 16 interviews in 2018, with high-level US policymakers and other actors involved in US migration policymaking (see Table 1, Appendix).

In 2015, we selected interviewees following a ‘quota sampling’ strategy: we compiled a list of relevant organizations and institutions engaged in migration policymaking by extrapolating from the existing literature and media, and we aimed to reach key interviewees occupying central positions within each of these institutions/organizations, e.g. top-level policymakers within departments of the US Administration with migration-related portfolios. If we were unable to contact the highest-level representative, our goal was to reach top-level officials from each of the main directorates, agencies or components engaged in migration work within that organization or institution. For non-governmental actors, we also accounted for variations within each actor category and conducted multiple interviews for each actor type.

Interviews in 2015 comprised both a semi-structured and a structured component. Elite policymakers were asked semi-structured questions regarding the sources they use to derive information about migration, their relevance and value for decision-making processes, and the reasons behind their choices/assessments. They were also asked to reflect upon the information and sources that played a crucial role in specific decisions or actions. Knowledge producers were asked to describe and assess their relations with elite policymakers. The structured component included a survey to collect quantitative data for a social network analysis (SNA). Interviewees were asked about the exchanges of their institution/organization with a predetermined list of actors involved in migration policymaking (extrapolated through the above-mentioned literature and media analysis, see Table 1, Appendix), and, more specifically, about (i) the frequency of these exchanges (measured on a temporal scale)Footnote 4; (ii) the perceived value of such exchanges, measured on a scale of 1 (not at all valuable) to 5 (extremely valuable); and (iii) the perceived similarity/difference between these actors’ views/perspectives on migration and the view/perspective of their organization/institution (henceforth: perceived degree of framing consonance between actors), measured on a scale of 1 (highly different) to 5 (highly similar).

In 2018, we aimed to reach policymakers appointed by the Trump Administration occupying the same positions of policymakers interviewed in 2015 (in addition to key nongovernmental actors). Interviewees were asked the same questions as in 2016. To compensate for the absence of the survey, policymakers were asked additional questions about the frequency of exchanges with key knowledge producers, their perceived value and the degree of framing consonance with these actors.

Methodologically, our data analysis was organized in three steps.

First, using the Gephi software, we elaborated quantitative data about the frequency of exchanges between organizations/institutions to map the US migration policy network in 2015 (see Fig. 1).Footnote 5 Based on the assumption that actors’ centrality is related to their power and influence (Castells, 2009), we calculated three centrality measures to identify the core decision-makers and knowledge producers within the policy network: weighted degree, i.e. the sum of the value of all the actor’s ties; betweenness centrality, i.e. the number of shortest paths from all the vertices to all the other vertices that pass through the node; and eigenvector centrality, which depends on both the number of the actors’ ties and whether actors relate to nodes that are themselves central. The SNA was not replicated in 2018 due to research constraints. However, previous research has shown that the composition of high-level policy networks tends to remain stable over time (on elite migration policy networks see: Taylor et al., 2013), while transitions between administrations can instead result in changes in the ideological orientations of bureaucratic agencies due to turnovers in agency personnel (see Lowande, 2018 regarding the Bush-Obama transition).

Fig. 1
figure 1

Network representing frequency of discussions among actors involved in US migration policymaking. Edges represent discussions among actors. The weight of edges is proportional to the frequency of interactions. Colours identify clusters of actors interacting more frequently among them. Clusters were identified by applying the ‘community detection algorithm’ of the Gephi software (Blondel et al., 2008). The partition of the network produced by this algorithm is not unique. We ran the algorithm many times to identify (few) boundary nodes that could be assigned to different clusters. The division in clusters described is the one that was more frequently generated by the software (standard resolution of 1)

Full size image

Second, we tested our three guiding expectations, combining qualitative content analysis of the 2015 semi-structured interviews with insights from the SNA analysing data about i) the frequency of interactions between core decision-makers and knowledge producers (for the first expectation about knowledge unavailability), ii) core decision-makers’ perceived degree of framing consonance with knowledge producers (for the second expectation about ideational proximity), and iii) core policy-makers’ perceived value of their exchanges with knowledge producers (for the third expectation about trust).

Third, we replicated the qualitative content analysis for the interviews conducted in 2018, with the aim to test the same guiding expectations under this second time-period.

Findings

In this section, we first analyse the compositional specifics of the US policy network and identify core decision-makers and key knowledge producers therein. Second, we analyse core decision-makers’ knowledge-formation processes testing our three guiding expectations previously identified.

The US policy network.

Figure 1 maps the frequency of exchanges between actors within the US migration policy network during the Obama Administration, identifying the actors involved in migration policymaking.

Figure 2 instead illustrates actors’ centrality values within the network. Weighted degree values and betweenness centrality values suggest that the DHS is the key dominant actors within the US migration policy system. Other core executive actors that play a central role in US migration policymaking are the DHS’s agencies (USCIS, CBP, ICE), the POTUS Executive Office (NSC and DPC) and the State Department. We also run the SNA considering the DHS and its agencies as one unique actor: in this case centrality values of the node representing the DHS become even higher. Policymakers working for these institutions and agencies can therefore be considered as the ‘core decision-makers’ within the US migration policy network. As to the key knowledge producers, the figure suggests that these include nongovernmental actors such as think tanks and NGOs, but also international organizations such as the United Nations Refugee Agency (UNHCR), trade unions and employers’ organization, which all are rather central within the policy network. The centrality scores of think tanks are particularly high, considering that they include a small number of organizations. NGOs in turn are a much bigger and heterogeneous group, meaning that their high centrality values are at least in part related to the grouping and merging strategy adopted.

Fig. 2
figure 2

Different centrality measures calculated for actors within the US migration policy system

Full size image

Knowledge availability

As explained in Sect. "Approach", knowledge availability is related to the existence of exchanges between core decision-makers and knowledge producers within the policy network (i.e. potential sources of information). Figure 3 illustrates the average frequency of such exchanges under the Obama Administration, which allows assessing which knowledge producers managed to regularly meet and exchange with the core decision-makers.

Fig. 3
figure 3

Frequency of interactions of core decision-makers with key knowledge producers within the migration policy system (scale: 1 = occasional to 4 = daily)

Full size image

The figure shows that the core decision-makers very frequently interacted with other executive actors. Exchanges with other knowledge producers are less frequent, but, overall, rather regular, at least in the cases of trade unions, employers’ organizations, NGOs and international organizations. All these actors have (at least) monthly exchanges with the core decision-makers. Exchanges with think tanks and private companies are less than regular (score < 2).

These findings are largely in line with what emerges from the interviews we conducted with elite decision-makers under the Obama Administration. When asked about their key sources of information, all the policymakers interviewed primarily referred to internal sources, within government or even within the same governmental institutions (e.g. staff on the ground, other governmental agencies, and advisory boards) as the most frequently consulted and the most valuable resources for “evidence”:

You know, I'm at the top of the Department. So there are lots of people...who are sort of operational folks who can tell U.S. that something is a bad idea and explain to U.S. why. So, I think you help avoid some of the unintended consequences by making sure that the right folks are looked into what you're thinking about doing (High-level Official, DHS, interviewed in 2015).

Aside from these internal sources, most interviewed policymakers also mentioned reports or information produced by think tanks and traditional media. Some quotes suggest that, often, under the Obama administration, elite decision-makers actively reached out to some of these organizations. As an interviewee from the US CBP reported:

The last couple of years we have been very interested in where the think tanks are, so we've done some work with Migration Policy Institute, International Defence Analysis, some private, public projects and briefings and exchange of information to try to get better on what they're seeing worldwide (High-level official, CBP, interviewed in 2015)

Other types of sources such as reports or material produced by international organizations, NGOs, academic sources and reports from other international organizations were only occasionally mentioned by the interviewed decision-makers.

The interviews we conducted under the Trump Presidency suggest an exacerbation of the above-mentioned trends. Once again, all the key decision-makers interviewed referred to internal sources as the main sources of information.

For what's going on there, it would be our own presence. ICE has attachés and liaisons in some of these locations so that’s one of the primary areas. The intelligence community, writ large, so other partners in law enforcement that contribute to the intelligence picture. The State Department is a big contributor to conditions-based and then the countries themselves, our own diplomatic, as it is, liaison information exchange. There's a fair amount of work done of people who are in custody, surveying them about why they left and what the conditions are like and so we have our own picture that we could share and use these other inputs to develop further (High-level official, ICE, interviewed in 2018)

Differently from the previous period, exchanges with NGOs (or consultation of NGO sources) are never mentioned by the elite decision-makers interviewed. However, most of the decision-makers interviewed report about interactions they have with think tanks and international organizations and/or state that they consult reports produced by these organizations. Occasionally our interviewees also report to consult other sources, such as reports from other international organizations and research institutes.

I would look principally to internal sources. For example, if I want to understand a new flow from let’s say Nicaragua for example, the most authoritative source from internal U.S. government standpoint is our Embassy in Managua and what they're seeing, and then our intelligence agencies, potentially our military but they would be reporting back through intelligence channels. As to external actors, you might have think-tanks who follow Latin America (...). We are happy to talk to those people, we appreciate what they're seeing, we appreciate their point of view but what they bring is going to have less weight than what comes in from our people in the field (High-level official, State Department, interviewed in 2018).

Overall, our first guiding expectation—according to which we expected the observed evidence-policy gap in the migration policy field to be potentially due to knowledge unavailability, or too infrequent contacts between elite decision-makers and knowledge producers—is therefore rejected by our analysis.

Ideational proximity

Figure 4 elaborates data we collected on actors’ degree of framing consonance, providing information about how the core policymakers perceive the views or perspectives of key knowledge producers with whom they interact in relation to their own views/perspectives. The figure shows that the core decision-makers perceive their views to be very similar to that of institutions/organizations with similar mandates (e.g. other governmental actors). The international organizations and various groups of non-governmental actors, on average, are perceived by elite decision-makers as holding “neither similar nor different” views or perspectives on the migration issue. The views/perspectives of NGOs and private companies are perceived to be rather different.

Fig. 4
figure 4

Perceived degree of framing consonance of federal migration policymakers with other actors within the migration policy system (scale: 1–5)

Full size image

More broadly, our social network analysis and interview material suggest that actors within the US migration policy network tend to interact more often with actors that are perceived to hold similar views or perspectives on migration, and that have similar organizational mandates. In particular, we created a network mapping the degree of framing consonance between actors; we identified clusters within such network (i.e. groups of actors sharing similar perspectives/views on migration) and observed that the composition of clusters is very similar to the one of the network mapping the frequency of exchanges (Fig. 1).

When shifting the focus to the Trump Administration, the interview material suggests quite a radical change of perspective. When describing their exchanges of information and processes of data gathering, policymakers under the Trump Administration refer much less to the role of organizational mandates, while perceptions of ideological or political differences between the potential sources of information and the Trump Administration become paramount. Particularly when referring to think tanks, NGOs and academics, most of the elite decision-makers interviewed describe them as holding very different views—generally, liberal positions in favour of open migration policies—compared to the (more restrictive) positions of the Trump administration. The interview material therefore suggests that, under the Trump Administration, knowledge transfers were decisively shaped by elite policymakers’ perceptions of political and ideological divergence between the Administration and knowledge producers. For instance, when reflecting about the sources used in its organization to inform decisions, a high-level US official suggests that information provided by some think tanks is less influential because of their different ideological stances:

The Pew Research Center…the Migration Policy Institute, I mean they're clearly on the other side of the debate (…). If they've done stuff that we haven't done, we refer to it or use it as a jumping-off point. But, if we can, we rely on other sources. For instance, there is a former immigration judge who does different kinds of research. He's got his own different network [more aligned with the governments’ positions on immigration] (High-level official, USCIS, 2018).

While previous quotes suggested that think tanks had a limited influence on policymakers’ decisions in 2015 because of their different institutional mandates, elite policymakers’ decisions under the Trump Administration are reported to be decisively shaped by conservative think tanks that are perceived to hold very similar ideological perspectives:

Trump's election was the big change and what creates opportunities and challenges for U.S. because now we have an administration that's willing to listen to us. Actually, even before listening to us, we know that they have used our research to help shape their own policy decisions (Representative of conservative think tank, 2018).

Conversely, an influential NGO representative interviewed in 2018 suggests that perceptions of ideological distance between the Trump Administration and pro-migrant NGOs completely hampered any kind of knowledge diffusion between them: in the interviewees’ words ‘anything that we say is bad they immediately embrace as good’.

Overall, our second guiding expectation—according to which we expected the observed gap between migration policy and evidence to be due to perceived ideational differences between elite decision-makers and knowledge producers, which obstruct knowledge diffusion among them—is therefore largely confirmed by our analysis. However, during the Obama administration, ideational proximity was primarily assessed based on institutional mandates, whereas under the Trump administration, it was largely associated with perceptions of political or ideological differences.

Trust and credibility

A third necessary condition for knowledge transfers to take place within policy networks is related to the perceived credibility of knowledge producers and policymakers’ trust on them. Figure 5 illustrates data we collected in 2015 about elite policymakers’ perceived value of exchanges with the knowledge producers (and of the information they provide through such exchanges). Figure 5 suggests that the elite policymakers under the Obama Administration considered exchanges with most knowledge producers as valuable or highly valuable (average values are > 3).

Fig. 5
figure 5

Value of interactions with knowledge brokers, as perceived by federal migration policymakers (scale: 1–5). Conservative think tanks are excluded due to the very infrequent contacts between them and the core decision-makers in 2015

Full size image

This finding is confirmed by our interviews conducted under the Obama Administration in 2015:

We meet with stakeholders from all sorts of different perspectives to get an outside perspective. (…) I mean you must sort of make judgements about how trustworthy the people on the outside you want to ask questions are. So sometimes we'll have conversations one-on-one with people we know we can trust (…)I think that outside perspective on that is very helpful. And there we're not telling them anything, so we don't have to be careful about what we say, they're telling U.S. things. By outside perspective I mean for instance the NGOs that work in the area (High-level official, DHS, 2015).

Such assessments of the credibility of knowledge producers seem to be largely related to perceptions that these actors belong to the same political or ideological camp. As the representative of an NGO interviewed in 2018 puts it, despite the goals of the Obama Administration being very different from those of the ‘immigrant rights community’ (and close to those of the Trump Administration, the interviewee claims), the Obama Administration trusted this community as it ‘felt beholden to it’. The same interviewee asserts that this facilitated a certain level of dialogue between the 'immigrant rights community' and the Administration.

Our interview material reveals a radical shift in policymakers’ assessment of knowledge producers’ credibility in 2018. Most policy-makers interviewed in 2018 described information provided by progressive or non-partisan think tanks, academia and NGOs as not only ‘less authoritative than internal sources’ (High-level official, DHS, 2018) but also, very often, as not reliable or not credible. This is primarily because these nongovernmental actors are perceived as 'partisan agents', and their (perceived) partisanship is assumed to bias the knowledge they produce.

The think tanks, the research institutes, they come from a different perspective. Everyone understands who they are, what their biases are, what their perspectives are…even the most reputable ones. Brookings for example is generally and probably correctly regarded as having a centre-left perspective, so that will shape how we treat that. Some of them are more open, for example The Centre for Immigration Studies, they do a lot of work, much of it quite good but the fact that they have a particular partisan position necessarily will undermine the credibility within the internal inner agency discussions about it. (...) Nobody believes that they [think-tanks] are non-partisan. (…) Some things that pass for academic study I look at and I say this is garbage. To give an example of that, I tell you about a study I read recently. The study itself was essentially a meta-survey of the literature for refugees and migrant’s health outcomes. It was particularly for refugees and migrants working in certain more dangerous or dirty fields like agriculture or construction…the main text of the thing was fine, it summarised a number of other studies but the conclusion was pretty partisan. I read the thing and I felt that the conclusion argued in favour of essentially regularisation for vast categories of people. Just logically it didn’t follow. This thing is considered an academic study, it was published in a reputable journal. I'm going to reject that (High-level official, State Department, 2018).

Judgements about the credibility of knowledge producers become strictly linked to perceptions of their political or ideological position, under the assumption that sources of information are never objective. In other words, perceived ideological differences become crucial in determining whether information is relied upon or not, and the perception of who is trustworthy greatly depends on the beholder:

If I hear oh, this is research from a think tank or an organisation I've never heard of, first I look at who is the lead author and the authors’ bio. Then I’ll look for the organisation who funds it, who is on its Board, what are their institutional affiliations. I’ll look for the lead author, what are the other things that he or she has written. Often they’ll do a research piece but then you can find opinion articles they write or an editorial for the New York Times, well I can look at the editorial and figure out what this person’s perspective is. If you look at the Migration Policy Institute for example, it’s pretty clear where they're coming from. Maybe this is an American perspective, there is no such thing as non-partisan information. I remember having this argument. The problem is not fake news, the problem is fake objectivity. There is no such thing as an objective news source, every news source has some inherent bias. (…) I would say what I've seen so far, reports from external actors, from think tanks, from academic institutions, they are typically used to justify somebody’s existing argument. (High-level official, State Department, 2018).

Academic research, in particular, is often referred to as partisan and hence not trustworthy, or irrelevant by elite policy-makers under the Trump Administration—as also by representatives of the conservative think-tanks with whom the Trump Administration frequently interacts. While discussing a (very well-known) academic work that criticized US border policies, the representative of a conservative think tank replied:

He’s showing a graph and then telling you something that doesn’t seem to follow from that. (…). It’s about how you interpret the data. I completely believe this. I think the data is right, he’s just spinning it in a way that is simply not plausible. (…) I have very little respect for this guy because, I’ve even spoken to other sociologists and he’s much more political. He wrote a whole book on [anonymised]. It’s a political book. (…) It’s an embarrassing thing. It really highlights the guy’s political agenda (Representative of conservative think tank, 2018).

While policymakers might well be accurate in their assessments, existing research has shown that individuals tend to dismiss evidence that disagrees with pre-established ideas and to avoid emotional discomfort that arises when questioning prior beliefs and conviction (Tversky & Kahneman, 1974). As an official from the State Department interviewed in 2018 puts it, ‘I'm reducing it to politics almost by saying it depends where your base of political power is and how you accept certain facts and exclude other facts’. Whether such assessments are accurate or not, this does not influence our argument.

Overall, our analysis therefore largely confirms our third guiding expectation related to the role of trust in knowledge diffusion within policy networks. Furthermore, it suggests that elite policymakers’ sources of information shift over time depending on who is holding power. For instance, in 2015, the think tank Migration Policy Institute was recognized by elite policymakers as a significant source of information, while all policymakers interviewed in 2018 expressed concerns about its reliability due to ideological differences with the new US administration.

Conclusion

This article has contributed to the ongoing debate on the role of facts and evidence in policy-making processes related to highly complex, ambiguous and politicized issues, with a specific focus on the issue of migration. The existing scholarship has often claimed that observed policy failures in democratic systems dealing with the migration policy issue are due to persistent gaps between policy and ‘evidence’ and has proposed three main explanations for such observed gap. These explanations attribute the gap to: i) cognitive limitations of policymakers, the unpredictability of their policymaking environment and the constraints policymakers face in terms of time, information and resources when dealing with highly complex, politicized and ambiguous policy issues (Cairney, 2016); ii) the irrationality of decision-making processes or their detachment from the complexity of real-world phenomena (Scholten, 2020); and iii) the prevalence of political interests over objective evidence in policy-making processes (Baldwin-Edwards et al., 2018). This article has shifted the focus of analysis to elite policymakers’ sources of information and the ‘policy environment’ in which information or ‘evidence’ about contested policy issues is collected, analysing the ‘heuristic case’ of US migration policy under the Obama and Trump Administrations. In doing so, this article challenges these existing explanations and generates additional possible explanations for the observed ‘evidence-policy gap’ that should be further tested in future research.

First, our findings question the claim that the ‘evidence-policy gap’ is due to policymakers facing limitations in terms of access to evidence about migration. Even in situations of crises such as the one analysed, federal migration policymakers, under both the Obama and Trump Administrations, regularly interacted with a wide range of knowledge producers including international organizations and several nongovernmental actors, think tanks in particular. Policymakers also proved to regularly consult reports or material produced by nongovernmental actors as well as academic publications. This means that existing constraints of time, information and resources that policymakers face cannot alone explain the observed ‘evidence-policy’ gap.

Second, our findings question existing explanations of the ‘evidence-policy gap’ in the migration policy field focused on the presumed irrationality of decision-making processes. In the heuristic case analysed, elite policymakers’ processes of gathering of information and collection of ‘evidence’ about migration are certainly not mere straightforward assessments of facts, but our analysis has shown that these processes are not ‘irrational’ or 'alienated from objective evidence’, as part of the existing scholarship suggests. Our findings indeed show that policymakers are not simply 'passive recipients of information', but rather active 'choosers, interpreters, and rationalizers' (Mutz, 2011, 12). They play an active role in collecting information, selecting their sources and in attributing relevance to the ‘evidence’ acquired. Conversely, our analysis suggests an alternative possible explanation for the evidence-policy gap, i.e. that elite policymakers’ processes of knowledge acquisition and processing of (new) information might be very much related to dynamics of trust and perceptions of like-mindedness (see also Sabatier, 1988). In the case analysed, when selecting the information upon which they need to base their decisions, policymakers tend to rely on internal sources of information. When they reach out beyond their organization, they rely on ‘like-minded’ actors, i.e. actors with similar pre-set organizational goals and mandates and/or actors that are perceived as holding similar political or ideological positions. Under the Obama Administration, external knowledge producers such as selected think tanks, NGOs and academics were perceived to belong to the same ‘political/ideological camp’ of the Administration, but their mandates and goals were perceived by policymakers to be highly different, which limited the relevance that policymakers attributed to the information they provided. Under the Trump Administration, these actors were perceived as belonging to a different political and ideological camp, further restricting the relevance that policymakers attributed to the information they provided.

This final point connects to the third finding of this article. The analysis conducted leads us to generate an additional possible explanation of the emergence of gaps between evidence and policy, which complements existing reflections on the role of politics in policymaking processes identified in the exiting scholarship. Our analysis does not dismiss the possibility that political interests outweigh 'evidence' in influencing migration-related decision-making processes, as the existing scholarship has argued. However, in the case analysed, politics and actors’ ideologies influence first and foremost the very processes through which policymakers gather information and determine what qualifies as ‘evidence’ in the first place. This article has indeed shown that, in the case analysed, knowledge producers that are perceived as holding different ideologies—and the information they produce—tend to be trusted less by elite policymakers. Considerations of knowledge producers’ ideological profile decisively shape policymakers’ assessments of the credibility of the information that these actors produce. During the Obama Administration, policymakers argued that relying on the information provided by NGOs, academia and think tanks was more risky and more time-consuming, as information produced by these actors had to be carefully evaluated and verified (because of their different organizational mandates), but they considered exchanges with knowledge producers and the information these actors provided as highly valuable, mainly because these actors were perceived to belong to the same ideological camp. During the Trump Administration, we observed a much more pronounced inclination to dismiss all kind of evidence that was not in line with the pre-established ideas and the restrictive policy approaches and political ideology of the Administration as ‘not credible’. Therefore, reports produced by think tanks, NGOs or academic works stopped representing even a potential source of information (or ‘evidence’) for policymakers. This is mainly because the perspectives of these knowledge producers were perceived to be completely different from those of the US Administration due to the opposite ideological or political positions on migration policy. Because of their perceived partisanship, elite policymakers are convinced that the processes through which knowledge is produced by these knowledge brokers are necessarily biased, and that the information provided by them cannot to be trustworthy.

These possible additional explanations for the emergence of gaps between evidence and policy in the migration policy field that this article has generated have important implications. On the one hand, pointing to the existence of circles of knowledge, they raise doubts about the potential for perspectives and voices alternative to those of the government to come through and influence migration policymaking processes. On the other hand, these findings suggest that we should be more sceptical of the claim that evidence and facts are the cure for perceived ills of migration policymaking. If policymakers are not passive recipients of information, but ‘active choosers’, this implies significant limitations of ‘facts’. Ultimately, they suggest that policy solutions depend on the ideological profile of key actors in government and on which potential sources of information are perceived by them as likeminded and trustworthy.

Future research should further explore and test the additional explanations of the ‘evidence-policy gap’ generated by this paper, to understand the extent to which these could be generalized beyond the highly politicized US migration policy environment. On the one hand, future research should replicate our analyses in other complex, ambiguous but less contested policy fields. On the other hand, migration scholars should test the claims generated by this article in different geographical contexts (particularly outside the Global North), and at other governance levels, as policymakers’ interpretations of policy problems could be influenced by the level of government at which they operate (e.g. Bazurli et al., 2022; Scholten, 2013).

At the methodological level, the paper demonstrates the relevance of interpretative approaches and network-centred approaches in the study of migration policy processes. These approaches can provide crucial insights into ideational processes, knowledge formation and transfer, without disregarding power dynamics.