Introduction

In recent years, the number of asylum claimants is significantly on the rise in Europe, thereby demonstrating that the world situation is certainly at a critical juncture and in a worrying state of regress with respect to fundamental human rights. Forced migration worldwide caused by wars, conflicts, and persecution has reached the highest levels recorded thus far and the numbers are rapidly increasing. By the end of 2016, 65.6 million individuals had been forcibly displaced worldwide as a result of persecution, conflict, violence or human rights violations, an increase of 300,000 people over the previous year. There were 2,826,508 asylum seekers, i.e., individuals who sought international protection and whose claims for refugee status have not yet been determined, irrespective of when they were lodged. Due to its central location in the Mediterranean Sea, Italy is the third country (after Germany and the USA) for the number of received new asylum claims in 2016 (122,972), with 247,992 people arriving by sea, often with tragic outcomes (UNHCR 2016). A major political and social media debate on the migration emergency has generated stereotyped representations of asylum seekers as bogus, as victims, or as a threat to national borders (for a review see Bottura and Mancini 2016). From 2015, Italian newspapers, for example, have significantly increased the attention paid to the issue of immigration, which has encouraged feelings of solidarity in the native population and, often, feelings of being invaded and threatened (Barretta 2015).

Strong evidence exists in psychosocial literature showing that feelings of threats are strongly linked to prejudice against migrants (e.g., Stephan et al. 2009) and threats and prejudice are associated with ideological beliefs (e.g., Duckitt and Sibley 2010). Indeed, several studies consistently showed that perceived threats and ideological beliefs are core antecedents of prejudice against out-groups (e.g., Anderson 2018; Anderson and Ferguson 2018; Duckitt and Sibley 2010; Esses et al. 2008; Stephan et al. 2009).

Nevertheless, to the best of our knowledge, two aspects are still under-investigated in psychosocial research: a) the role of threats and ideological beliefs in prejudice against a particular category of migrants, that is to say asylum seekers, and b) the relationship of threats and ideological beliefs with behavioural intention in favour of or against asylum seekers. The aim of this paper was to try to partially fill these gaps by investigating the relationship among threats (real and symbolic), ideological beliefs (RWA and SDO), prejudice, and behavioural intention toward asylum seekers in a sample of Italian people.

Threat and Prejudice towards Asylum Seekers

With the exception of some studies conducted in Australia (e.g., Canetti et al. 2016; Pedersen et al. 2005; Suhnan et al. 2012), few studies have explicitly investigated prejudice against asylum seekers. Most of the research has been focused on refugees instead, often including asylum seekers within this more general category (e.g., Schweitzer et al. 2005; Turoy-Smith et al. 2013). Compared to both economic migrants and refugees, contemporary asylum seekers are generally a more vulnerable group. The indefinite nature of their claim, their need for long-term assistance from the state, and doubts regarding the validity of their rights to seek asylum seem to bolster people’s negative attitudes towards them (e.g., Canetti et al. 2016). Accordingly, studies, albeit few in number, have consistently highlighted that threat is a core issue that leads people to increase negative attitudes towards asylum seekers. For example, Louis et al. (2007) analysed perceptions of structural and instrumental threats against asylum seekers in Australia and showed that these variables predicted unfavourable attitudes towards asylum seekers, support for exclusionary measures, and representation of asylum seekers as an over-benefited group. The topic of threat has also been highlighted in studies on the social representation of asylum seekers (e.g., Goodman and Burke 2010; Lynn and Lea 2003; Moloney 2010; Pearce and Stockdale 2009). Lynn and Lea (2003), studying the media construction of the representation of asylum seekers in the UK, underlined how the term bogus was used as a rhetorical device to reinforce a phantom threat, represented by the possibility that among asylum seekers there could be criminals or persons who had exploited resources by living off the government’s generosity. In the study by Goodman and Burke (2011), religious grounds and the associated threat of terrorism were arguments that undergraduate students participating in focus groups used to oppose asylum. Other studies (e.g., Klocker 2004; Pedersen et al. 2005; Suhnan et al. 2012) pointed out how this rhetorical mechanism was powerful in spreading false beliefs about asylum seekers, beliefs that were related to high scores of threat and prejudice. For example, Pedersen et al. (2007) found a significant and positive correlation between the perceived threat from terrorism and prejudice against asylum seekers in a sample of Australian people. In sum, the available evidence suggests that asylum seekers are often and easily perceived as a threat for the in-group.

Ideological Beliefs (SDO/RWA) and Prejudice towards Asylum Seekers

Individual and ideological-based orientations such as social dominance orientation (SDO, Pratto et al. 1994) and right-wing authoritarianism (RWA, Altemeyer 1981) have been taken into account when looking at intergroup relations in reference to asylum seekers. SDO is meant to capture people’s desire that social groups are arranged in an iniquitous social hierarchy, in which some groups dominate other groups. RWA, instead, captures people’s desire for social order and protection of the in-group’s worldview. In this way, SDO is considered as an ideology which increases people’s desire to maintain and protect the social advantage of the in-group (e.g., material resources), while RWA appears to be linked to people’s desire to defend the values, norms, and culture of the in-group (e.g., symbolic resources). Accordingly, the research confirmed that SDO and RWA, albeit related, are two independent antecedents of prejudice towards asylum seekers (Anderson 2018, Study 2; Anderson et al. 2015; Esses et al. 2008; Louis et al. 2007; Nickerson and Louis 2008). For example, Esses et al. (2008) reported a significant and positive correlation between SDO and the tendency to dehumanise refugee claimants in Canada. Nickerson and Louis (2008) showed that attitudes and feelings towards asylum seekers were more negative among Australians with high levels of RWA and SDO. Some Australian studies showed that when participants perceived asylum seekers as “cheaters” of the system, prejudicial attitudes (Anderson et al. 2015) and endorsements for restriction of asylum seekers’ access (Louis et al. 2007) predicted high scores for SDO and RWA. Conversely, when participants perceived asylum seekers as being in a disadvantaged position and therefore in need of resources, low scores for SDO and RWA predicted positive attitudes towards asylum seekers (Anderson et al. 2015). Anderson’s (2018) cross-sectional studies provided further evidence that SDO predicts negative attitudes towards asylum seekers at both an implicit and explicit level, and RWA predicts negative explicit attitudes (Study 2). Finally, Anderson and Ferguson’s (2018) recent meta-analysis showed that SDO and RWA are strongly correlated with anti-asylum seeker sentiments.

Behavioural Intentions and Prejudice towards Asylum Seekers

Behavioural intentions in favour of or against asylum seekers have also been rarely analysed in psychosocial research which, instead, has focalized mostly on prejudices and negative attitudes. This is somewhat surprising, given the relevance that behavioural intentions and actual behaviours have for the relationships between asylum claimants and native populations. As is well known, attitudes and behavioural intentions are linked to one another, even though the strength of this relationship is highly variable, from weak to moderate (Ajzen 1991; Ajzen and Fishbein 1980; Fishbein 2008). More importantly, attitudes appear to be weakly predictive of actual behaviour, while behavioural intention appears to be the most direct predictor of a range of behaviours (e.g. Albarracin et al. 2001). Meta-analytical evidence shows that behavioural intentions predict from 19 to 38% of the variance of actual behaviour (Sutton 1998), suggesting that intentions to behave in a particular way are the immediate and best predictor of behaviour (see also Hutchison and Rosenthal 2011). For these reasons, investigating the antecedents of behavioural intentions in favour of or against asylum seekers, as well as their connections with prejudice, appears to be greatly relevant in order to better understand how people would behave toward asylum claimants.

Some studies have taken into account behavioural intention and its relationship with prejudice, but they have mainly analysed behavioural intentions in terms of the local population’s support for exclusionary asylum policies (Canetti et al. 2016; Louis et al. 2007). For example, Canetti and co-workers (Canetti et al. 2016) showed that conservative political ideologies increased the threat perception held by members of both the Israeli and the Australian majority and that threat increased support for exclusionary asylum policies. Other studies considered variables related to intergroup relations, such as national identification (Nickerson and Louis 2008; Pehrson et al. 2009) or asylum seeker stereotypes (Verkuyten 2004). In a longitudinal study, Pehrson et al. (2009, study 2) confirmed that the more intense the definition of nationalism in a population, the more restrictive the behavioural intentions towards asylum seekers. Similarly, Nickerson and Louis (2008) found that Australian participants who strongly identified with asylum seekers as humans were also more positive in their behaviour towards asylum seekers than those who strongly identified with their own nationality. Using a scale of behaviours that analysed the participants’ intentions to reduce the number of asylum seekers (for example, voting in a confidential ballot/referendum, taking part in a rally, or signing a petition), Louis et al. (2007) found that Australians’ willingness to restrict the access of asylum seekers to their nation and its resources was predicted accurately by intergroup hostility and prejudice and by concerns for procedural and distributive fairness. Similarly in Europe, Verkuyten (2004) tested whether the influence of anger and sympathy on policy support, and on opportunities and rights that asylum seekers should or should not receive, changed as a function of the way asylum seekers were defined as migrants with no choice but to migrate or as cheaters of the asylum system. The author found that, for genuine asylum seekers, feelings of sympathy predicted behavioural intentions in support of their plight whereas, for bogus asylum seekers, feelings of anger affected behavioural intentions against them.

Therefore, these studies seem to confirm a negative relation between prejudice, or variables affecting prejudice, and native people’s support of political actions and policies towards asylum seekers. However, with the exception of Pehrson et al. (2009), none of these studies measured an active intention to extend aid to asylum seekers, as for example, in assisting a charity that was interested in getting young people involved in volunteer work with asylum seekers. Unlike the support for political actions or norms and policies towards immigration, “active” support assumes a more direct exposition of the host population to the needs of asylum seekers and therefore represents a more concrete and proximally related indicator of prosocial behaviour towards asylum seekers. In order to have a direct and active indicator, in this study we considered willingness to provide asylum seekers with material assistance in terms of hospitality, money or volunteer work as measure of the behavioural intentions in support of asylum seekers.

Aim of the Present Study

The aim of this work was to investigate the relationship among perceived threat, SDO, RWA, prejudice, and behavioural intention toward asylum seekers among a sample of Italian adults. In order to meet this goal, we considered the dual process model of ideology and prejudice (DPM, Duckitt 2001, 2006; Duckitt and Sibley 2010; Matthews et al. 2009) as a theoretical framework. The DPM posits that RWA and SDO are rooted in a dangerous (threatening) and competitive social worldview and that RWA and SDO would predict prejudice depending on the extent to which migrants were perceived as threatening or deviant or as competitive and economically disadvantaged (Cohrs and Stelzl 2010). In the DPM, SDO and RWA follow two parallel but distinct ways in which SDO reflects a dimension of competition for intergroup dominance, while RWA reflects a dimension of reduction of in-group threat and research of collective in-group security. Both dimensions, however, contribute to increase prejudice and negative attitudes toward the competitive and threatening out-groups.

To the best of our knowledge, several studies have tested and supported the DPM’s expectations considering prejudice against migrants as a primary outcome variable (e.g., Caricati et al. 2017; Cohrs and Stelzl 2010). However, no studies have until now verified the explicative role of the DPM on behavioral indicators, such as those related to the intention of helping someone. Nevertheless, we believe that DPM can easily be enlarged to also include expectation about behavioural intention, assuming that both prejudice and behavioural intentions are expressions of the same underlying psychosocial process. In other words, DPM leads to the expectation that both SDO- and RWA-ways decrease people’s intention to engage in prosocial behaviours toward out-groups, which is discouraged by prejudice against these out-groups. Accordingly, based on the existing literature and DPM, the first expectation of this study was that for realistic and symbolic in-group threats, SDO, and RWA was positively associated with prejudice against asylum seekers (hypothesis 1A) and negatively associated with intentions to help them (hypothesis 1B). Moreover, in line with previous studies (e.g., Canetti et al. 2016; Louis et al. 2007; Nickerson and Louis 2008; Pehrson et al. 2009; Verkuyten 2004) we expected that the more people are prejudiced against asylum seekers the less they would be oriented to behave in favour of asylum seekers (hypothesis 2). Furthermore, based on the DPM and the evidence that RWA and SDO are affected by the extent to which a group is perceived as threatening (e.g., Caricati et al. 2017; Cohrs and Stelzl 2010; Duckitt and Sibley 2010; Matthews et al. 2009), we expected that the effect of a realistic threat on both prejudice against asylum seekers and intention to help them would be mediated by SDO (hypotheses 3A and 3B). In the same way, we expected that the effect of symbolic threat on both prejudice and prosocial behavioural intentions towards them would be mediated by RWA (hypotheses 4A and 4B). Similarly, we also expected that both realistic and symbolic in-group threats would affect the prosocial behavioural intentions towards asylum seekers through SDO (or RWA) and prejudice (hypothesis 5 and 6). Figure 1 shows this model.

Fig. 1
figure 1

Results from path analysis on the tested model (model 1). Fully standardized coefficients are reported. * p < .05; ** p < .01; ns = not significant. Overall fit: χ2(3) = 37.98, p < .001, CFI = 0.912, TLI = 0.559, RMSEA = 0.241, 90% CI = .185–.302, p < .001, SRMR = 0.079

Full size image

Method

Design, Participants and Procedure

A survey questionnaire design was used. The first page of the questionnaire clearly stated that participation was voluntary and data collection was anonymous. Non-students were enrolled in the study using a snowball technique that followed a quota sampling method based on equal numbers (N = 25) of male and female participants belonging to four age groups (18–34, 35–44, 45–55, and more than 55 years old).

Sample

People living in Tuscany, Italy, were enrolled in this study through a quota sampling method stratified by gender and age groups (see Table 1). Two hundred participants, with a mean age of 43.51 (SD = 13.41, range = 18–75 years), answered a questionnaire that was distributed and presented by a researcher. Table 1 describes the distribution of participants according to their gender and age groups.

Table 1 Sample characteristics: gender and age groups
Full size table

Almost all of the participants (197) were born in Italy, and the majority (167, 83.5%) were Catholic (32, 16.5% were atheist). More than three-quarters of the sample (154, 77.0%) were in paid employment. Only 19.0% said that they knew one or more asylum seekers. Finally, 155 participants (77.4%) had attained a bachelor’s or master’s degree level.

Measures

Social dominance orientation was measured with the Italian version of the short form of the SDO scale (Pratto et al. 2013, e.g., ‘Superior groups should dominate inferior groups’), which is composed of four items rated on a 5-point Likert-type scale (1 = disagree; 5 = agree).

Right-wing authoritarianism was measured with four items (e.g., ‘Obedience and respect for authority are the most important values children should learn’) taken from the authoritarian submission and aggression dimension of the Italian version of the RWA scale (Manganelli et al. 2007). Items were rated on a 5-point Likert-type scale (1 = disagree; 5 = agree).

In-group threat was measured with a short version of the zero-sum beliefs scale (Esses et al. 2001). The scale is composed of 12 items rated on a 5-point Likert-type scale (1 = disagree, 5 = agree). Five items asked the participants to express their agreement with some statements referring to realistic threat (e.g., ‘Money spent on social services for asylum seekers means less money for services for Italians already living here’) and seven items referred to the symbolic threat from asylum seekers (i.e., ‘When asylum seekers are encouraged to maintain their cultural practices, Italian culture is weakened’). Confirmatory factor analysis with robust standard error estimation suggested that the two-factor model had an acceptable fit, χ2(51) = 88.49, p < .001, CFI = 0.924, TLI = 0.902, RMSEA = 0.061, 90%CI [0.041–0.080], p = .175, SRMR = 0.054.

Prejudice against asylum seekers was measured with 17 items from the attitudes towards asylum seekers scale (ATAS) that Pedersen, Attwell, and Heveli designed in a bottom-up fashion in an Australian context (Pedersen et al. 2005). Examples of items are: ‘Asylum seekers are manipulative in the way that they engage in self-harm protesting such as self-mutilation’, ‘I sympathise with the situation of asylum seekers’. Because the Australian context is different to the Italian one, three independent judges translated the 18 items and evaluated their application to the situation of asylum seekers in Italy. One item of the original scale (Asylum seekers who mutilate themselves would not make model citizens) was excluded because it was judged to not be applicable to Italy. Responses were rated on a 5-point Likert-type scale (1 = disagree, 5 = agree). Explorative factor analysis revealed that two items had poor loading on the latent dimension and were weakly correlated with the total score. Thus, these items were removed and the total score of the ATAS was computed on the remaining 15 items.

Proactive behavioural intention (PBI) was measured with three ad hoc items asking participants to indicate the extent to which they would engage in some behaviours directed towards asylum seekers (i.e., “host an asylum seeker for a few days”, “donate money to a charity that deals with asylum seekers”, and “perform volunteer work in an association that deals with asylum seekers”) on a 5-point Likert scale (1 = certainly not; 5 = certainly yes).

For all measures, the intended items were averaged to create a composite score in which higher scores indicated higher levels of the measured constructs.

Results

Descriptive Analysis

Table 2 shows zero-order correlations among the considered variables as well as descriptive statistics and internal reliability. As indicated, all scales had adequate internal reliability and were significantly correlated with one another. As expected, SDO, RWA, and in-group threat were positively correlated with prejudice against asylum seekers, which, in turn, was negatively correlated with PBI. Finally, SDO and RWA were positively correlated with both realistic and symbolic in-group threats. Moreover, the gender and age of the participants affected none of the considered variables (all ps > 0.30).

Table 2 Zero-order correlations, Cronbach’s alpha, mean and standard deviation of considered measures
Full size table

Testing the Models

In order to test the model, a path analysis was performed using maximum likelihood estimation with robust standard error estimation in order to take into account the possible non-normality of some variables. The R software package (R Core Team 2015) was used to perform the analyses. As indicated above, realistic and symbolic in-group threats, SDO, and RWA were the primary antecedents of both prejudice (ATAS; H1A) and pro-social behaviour intentions (PBI; H1B); prejudice was the primary antecedent of PBI (H2). Based on the DMP (Duckitt and Sibley 2010), SDO was considered as mediating the effect of realistic in-group threat on both ATAS (H3A) and PBI (H3B), and RWA as mediating the effect of symbolic in-group threat on both ATAS (H4A) and PBI (H4B). Finally, it was expected that realistic and symbolic in-group threats would indirectly affect PBI via SDO and ATAS or RWA and ATAS respectively (H5 and H6).

Figure 1 shows the results of the path analysis. As one can see in Fig. 1, in accordance with hypothesis 1A, both realistic (b = 0.11, SE = 0.05, Z = 1.96, p = .05, 95%CI [0.00, 0.21]) and symbolic threats (b = 0.32, SE = 0.06, Z = 5.03, p < .001, 95%CI [0.19, 0.44]), and both SDO (b = 0.18, SE = 0.05, Z = 3.75, p < .001, 95%CI [0.08, 0.27]) and RWA (b = 0.10, SE = 0.04, Z = 2.28, p = .023, 95%CI [0.01, 0.18]) significantly and positively predicted prejudice against asylum seekers. However, contrary to hypothesis 1B, only realistic threats significantly and negatively predicted intention to act with prosocial behaviours towards asylum seekers (b = −0.32, SE = 0.09, Z = 3.35, p = .001, 95%CI [−0.50, −0.13]). Furthermore, as expected from hypothesis 2, prejudice decreased people’s intention to act with prosocial behaviours towards asylum seekers (b = −0.61, SE = 0.11, Z = 5.48, p < .001, 95%CI [−0.82, −0.39]). Moreover, in accordance with hypotheses 3A and 4A, SDO significantly (b = 0.07, SE = 0.02, Z = 3.07, p = .002, 95%CI [0.02, 0.11]) mediated the relationship between realistic in-group threat and prejudice, while RWA significantly mediated (b = 0.06, SE = 0.03, Z = 2.12, p = .034, 95%CI [0.004, 0.11]) the relationship between symbolic in-group threat and prejudice. However, in contrast to hypothesis 3B, the relationship between realistic in-group threat and PBI was not significantly mediated by SDO (b = −0.04, SE = 0.03, Z = 1.13, p = .259, 95%CI [−0.10, 0.03]). Similarly, in contrast to hypothesis 4B, the relationship between symbolic in-group threat and PBI was not significantly mediated by RWA (b = 0.06, SE = 0.04, Z = 1.43, p = .153, 95%CI [−0.02, 0.14]). Finally, as expected from hypotheses 5 and 6, realistic in-group threat had a significant total indirect effect – through the mediation of both SDO and prejudice – on PBI (b = −0.10, SE = 0.04, Z = 2.84, p = .004, 95%CI [−0.18, −0.03]), and symbolic in-group threat had a significant total indirect effect – through the mediation of both RWA and prejudice – on PBI (b = −0.23, SE = 0.06, Z = 3.74, p < .001, 95%CI [−0.34, −0.11]). This model explained the considerable amount of variance of both PBI and prejudice (R2 = .40 and R2 = .50 respectively). On the whole, however, the model did not have an excellent fit (Hu and Bentler 1999): χ2(3) = 37.98, p < .001, CFI = 0.912, TLI = 0.559, RMSEA = 0.241, 90% CI = 0.185–0.302, p < .001, SRMR = 0.079.

Considering the high correlation between realistic and symbolic in-group threats, as well as the similar correlation between both the two types of threats and SDO and RWA (Table 2), a second model was tested, in which a single threat predictor was considered.

This second model (Fig. 2) had a good fit: χ2(1) = 0.010, p = .920, CFI = 1.00, TLI = 1.00, RMSEA = 0.000, 90% CI = .000–.065, p = 0.919, SRMR = 0.001, and accounted for 38% of the variance of the PBI. In accordance with the previous model, this second model confirmed only some of the hypotheses of the study: the in-group threat (b = 0.41, SE = 0.06, Z = 6.91, p < 0.001, 95%CI [0.29, 0.52]) and conservative beliefs (SDO, b = 0.19, SE = 0.05, Z = 3.91, p < 0.001, 95%CI [0.09, 0.28], and RWA, b = 0.09, SE = 0.04, Z = 2.10, p = .036, 95%CI [0.01, 0.17]) significantly and positively predicted prejudice against asylum seekers (hypothesis 1A); in partial contrast to hypothesis 1B, only in-group threat (b = −0.34, SE = 0.10, Z = 3.39, p = 0.001, 95%CI [−0.54, −0.14]) significantly and negatively predicted intention to act with prosocial behaviours towards asylum seekers. Confirming hypothesis 2, prejudice decreased people’s intention to act with prosocial behaviours towards asylum seekers (b = −0.58, SE = 0.11, Z = 5.06, p < 0.001, 95%CI [−0.80, −0.35]). Moreover, ingroup threat had indirect effects on prejudice through both SDO (b = 0.11, SE = 0.03, Z = 3.46, p = 0.001, 95%CI [0.05, 0.17]) and RWA (b = 0.06, SE = 0.03, Z = 2.03, p = 0.043, 95%CI [0.02, 0.12]); nevertheless, conservative beliefs did not mediate the relationship between in-group threat and PBI (indirect effect through SDO: b = −0.04, SE = 0.05, Z = 0.92, p = 0.358, 95%CI [−0.12, 0.05]; indirect effect through RWA: b = 0.05, SE = 0.05, Z = 1.11, p = 0.268, 95%CI [−0.04, 0.14]). However, the in-group threat had a significant total indirect effect – through the mediation of both SDO and prejudice – on PBI (b = −0.32, SE = 0.08, Z = 4.12, p < .001, 95%CI [−0.47, −0.17]).

Fig. 2
figure 2

Results from path analysis on the tested model (model 2). Fully standardized coefficients are reported. * p < .05; ** p < .01; ns = not significant. Overall fit: χ2(1) = 0.010, p = 0.920, CFI = 1.000, TLI = 1.000, RMSEA = 0.000, 90% CI = 0.000–0.065, p = 0.919, SRMR = 0.001

Full size image

Discussion and Conclusion

Using the dual-process model of ideology and prejudice (Duckitt 2001) as a theoretical framework, the aim of the present research was to investigate whether realistic and symbolic in-group threats and conservative ideologies (i.e., SDO and RWA), would affect both prejudice against and prosocial behavioural intentions towards asylum seekers. The present study is novel, given that psychosocial literature only rarely focuses on asylum seekers and behavioural intentions, focalizing mostly on prejudice against migrants in general or refugees. First, that of asylum seekers is a timely topic, given the impact of the surge in asylum seekers in various European countries, and especially in Italy. The capacity to meet the needs of this particularly vulnerable subgroup of migrants through adequate infrastructure, ideology, and host country attitudes is therefore an important topic of research (Mancini et al. 2018a). This study has provided some insight into how attitudinal research can help to improve the reception of asylum seekers in host countries. Second, to the best of our knowledge, the present work is one of the first studies to take into account, along with prejudice, the intention to act pro-socially towards asylum seekers and to investigate behavioural intentions in favour of asylum seekers through a measure of willingness to give active and concrete aid to asylum seekers. Specifically, this study has tried to assess the psychological processes which can affect the prosocial behavioural intentions of the local population towards asylum seekers.

Results indicated that the perception of asylum seekers as a threat for the in-group resources or values strongly and directly affected prejudice against this group. More precisely, supporting DPM and previous studies (Klocker 2004; Louis et al. 2007; Pedersen et al. 2005; Suhnan et al. 2012), higher scores of realistic and symbolic in-group threats were associated with increased prejudice. Again, supporting the DPM and previous studies (e.g., Caricati et al. 2017; Jost et al. 2003), SDO and RWA appeared to be boosted by perceived threats and to increase prejudice against asylum seekers. Moreover, SDO and RWA partially mediated the effect of threats on prejudice confirming, thus, the important role of conservative ideologies in fuelling prejudicial attitudes (Duckitt 2001, 2006). In line with studies showing the important role of in-group threat on intergroup relations (e.g., Canetti et al. 2016; Louis et al. 2007; Suhnan et al. 2012), the effects of in-group threat also extended to the conative domain so that perceived threat appeared to affect prosocial behavioural intentions directly. It is worth noting that the first model suggested that realistic more than symbolic perceived threat appears to be the most powerful motivational thrust for denying help to asylum seekers. Model two, however, indicated that a composite measure of threat is associated with pro-social behavioural intentions in both direct and indirect (i.e., via prejudice) fashions. Interestingly, prejudice was positively and significantly associated with a decrease in intention to act pro-socially (i.e., hosting an asylum seeker for a few days, donating money to a charity that deals with asylum seekers, and volunteering to work in an association that deals with asylum seekers). This finding is in line with studies showing that the more people are prejudiced against asylum seekers, the less supportive they are for inclusion policies and non-restrictive actions towards asylum seekers (Canetti et al. 2016; Louis et al. 2007; Nickerson and Louis 2008; Pehrson et al. 2009; Verkuyten 2004). This result is also consistent with the Reasoned Action Approach (Ajzen 1991; Ajzen and Fishbein 1980; Fishbein 2008) supporting the idea that attitudes and behavioural intentions are linked one to another.

Other results did not support our expectations and DPM and evidenced some new interesting findings. First, we were not able to clearly distinguish the two distinct processes which are hypothesised by the DPM. More precisely, symbolic and realistic threats have similar effects on both SDO and RWA so it is virtually impossible distinguish their effects. Thus, contrary to DPM (Duckitt 2001, 2006) and previous findings (Caricati et al. 2017; Cohrs and Stelzl 2010; Duckitt and Sibley 2010; Matthews et al. 2009), conservative beliefs increased prejudice against asylum seekers, but this did not occur through the two distinct hypothesised paths in which a) realistic in-group threat increased prejudice via SDO, and b) symbolic in-group threat increased prejudice via RWA. A possible explanation for this unexpected result might rely on both the specific target considered in this study – asylum seekers – and the particular situation in which the data have been collected. The migration emergency and the large number of asylum claimants that in recent years have reached the Italian coast has increased the fear of invasion among the native population. Together, the rhetoric of the Italian media has documented the arrival of the asylum claimants mainly by contraposing images of landings and images of deaths at sea (Barretta 2015). These images have probably mixed fears that asylum seekers are deviant or competitive with fears that asylum seekers are disadvantaged people who need the help of western countries. Therefore, it might be that both fears of threat and invasion (realistic threat) and feelings of compassion and piety towards people from a different culture (symbolic threat) have been stimulated among the native population, making the perception of the sources of threat less clear and more mixed, as shown in recent Italian studies (e.g., Caricati et al. 2017; Mancini et al. 2018b).

Another unexpected result was that conservative ideologies had no direct effects on behavioural intentions because their effects were completely mediated by prejudice. In other words, the present results suggest that conservative ideologies (i.e. SDO and RWA) could affect people’s behavioural intentions (i.e., discrimination or favouritism towards a group) only through an increased or decreased attitudinal disposition towards an out-group, highlighting that the effect of conservative ideologies may be limited to the attitudinal domain (e.g., Saunders et al. 2016). This might suggest that prejudice and intention to help asylum seekers are not expressions of the same underlying psychosocial process. However, as we have discussed above, this result could be target specific, sample specific, and culture specific, so other studies with other group targets and with other populations should be done in order to confirm our results.

From a practical point of view, these results provide a road to understanding how to ameliorate prosocial behaviours, i.e., the practical ways in which people demonstrate behavioural intentions, through hospitality, money or volunteerism. According to the studies that show that behavioural intentions predict actual behaviour (e.g., Sutton 1998; Hutchison and Rosenthal 2011), we can suppose that the intention to act pro-socially can stimulate behaviors of effective help towards asylum seekers. The results of this study show that, in order to increase pro-social behaviours, the perception of the host population’s worries about asylum seekers should be reduced. This is the reason why political and social interventions, as well as media communication, should promote and sustain the idea that asylum seekers are more of a resource than a threat for the host population. European research on the economic impact of refugees who arrived as asylum seekers is only at the early stage. The International Monetary Fund staff (IMF Staff 2016) estimated that, on a Gross Domestic Product-weighted basis, average budgetary expenses for asylum seekers in EU countries will add up to around 0.19% in 2016. However, in time, as the new arrivals integrate into the workforce, they are expected to boost annual output by 0.1% for the EU as a whole and, given the relative youth of the migrants, to help (in a small way) to reverse the upward creep of the cost of state pensions. There is evidence in Australia, a country that has been facing the problem of forced migration for a long time, that refugees can be beneficial to the general economy (e.g., Stilwell 2003).

The study described in this paper has some important limitations. First, it must be emphasized that only the explicit, classical component of prejudice has been measured (Anderson 2017; Anderson and Cheers 2017) in this study. Thus, a generalization of our results to other components of prejudice – such as modern prejudice (Akrami et al. 2000) and conditional prejudice (Anderson 2017) – is not possible. The second rests on the correlational nature of the research design. This, as is known, suggests that caution should be used when inferring causal relationships among variables. Moreover, the results may be affected by common method bias. Overall, these limitations may have affected the results and limited the generalizability of the findings. Replicating our model in other samples and/or in a more controlled manner, for example manipulating in-group threat, could strengthen the results. In general, more research is necessary in order to demonstrate how threats, conservative ideologies and prejudice affect pro-social behavioural intentions in different samples: for example, in the policemen who work at the national borders and along the Mediterranean migrant routes, in people who live close to the reception centres, or in professionals working with forced migrants. Further studies could explore other predicting variables such as the inclination of people to conform to social norms, for example, the norm of helping a person in need. As to professionals working with forced migrants, it could be very useful to study how these pro-social behavioural intentions could drive the quality of their professional actions. It would also be interesting to conduct interviews with laypeople, service providers, and others to gain a deep understanding of why they engage in pro-social behaviour towards asylum seekers. Another question that needs to be explored is how religion might promote or limit one’s pro-social behavioural intentions. Due to the fact that 83.5% of this sample group were Catholic, no conclusion about this question was possible in this study. As one study conducted in Australia showed (Perry et al. 2015), Christians were less likely to hold prejudice towards asylum seekers compared with non-Christians participants, but this occurred only when RWA was taken into account. Therefore, exploring the effects of religious identity in Italy would be an interesting new line of research. Finally, in light of the Syrian refugee crisis, it would be interesting to examine the rhetoric of ‘terrorism/terrorists’ as another layer of prejudice or in-group threat perception towards this group of asylum seekers and whether or not this rhetoric adversely affects behavioural intentions. However, despite these shortcomings, we believe that the results are robust enough to offer some useful suggestions for dealing with prejudice against asylum seekers – or at least with the explicit component of it – and for improving the reception of asylum seekers in host countries. The results of this research suggest that reducing the fear of threat and invasion among the native population, a threat that the Italian media and political rhetoric always contribute to reinforcing (Barretta 2015), is an important goal to achieve in order to encourage pro-social attitudes towards asylum seekers.

This research has some practical implications not only for laypeople, but also for those who provide professional assistance to asylum seekers. We must not forget that encouraging prosocial behaviours in a native population is not only an expected action in a civilized country, but also a necessity at this particular historical moment. The migration crisis and the large number of asylum claimants that some European countries have to manage means that, today, more and more people are involved with asylum seekers. Some of these individuals provide services, others offer or refuse political and material support, and others are engaged in charity. Whatever the reasons people interact with asylum seekers, it is their feeling of perceived threat and their prejudice that improve or hinder their pro-social actions. How these actions could be driven is not without consequence for the quality of the reception of asylum systems, for the integration of asylum seekers in the host society, and for asylum seekers’ wellbeing, as some studies (e.g., Mancini et al. 2018a) and reviews have shown (Rossi and Mancini 2016). This is the reason why European countries should promote positive and prosocial behaviours both in the local population and in professionals connected with asylum systems. Raising awareness in the local population seems to serve this aim, as some reception projects in Italy seem to show (see for example the projects in which refugees are hosted in Italian families).