‘It put me in their shoes’: challenging negative attitudes towards asylum seekers among Australian children

This paper evaluates a short school-based intervention run by Australian Red Cross, designed to reduce children’s prejudice towards asylum seekers. A total of 121 children aged between 10 to 12 in four schools in Perth, Western Australia, completed questionnaires at Time 1 (pre-intervention), Time 2 (immediately after the intervention), and Time 3 (8-9 months after the intervention). The intervention used a mixture of approaches: providing information, encouraging empathy, making positive social norms more explicit, and fostering imagined contact with asylum seekers. The intervention content was also reinforced by teachers throughout the school year. The study found that the intervention was effective in increasing the children’s positivity towards asylum seekers, reducing prejudiced attitudes, and increasing intentions to interact with asylum seekers. It also found that the intervention increased the children’s accuracy in defining ‘asylum seeker’ and ‘refugee’. These results occurred both in the short-and long-term, although there was some regression over time.


Introduction
At the end of 2019, there were over 26 million refugees and 4.2 million people seeking asylum worldwide (United Nations High Commissioner for Refugees [UNHCR], 2019). According to Article 1A of the 1951 United Nations Convention relating to the Status of Refugees and its 1967 Protocol: a refugee is any person who… owing to a well-founded fear of being persecuted for reasons of race, religion, nationality, membership of a particular social group or political opinion, is outside the country of his/her nationality and is unable, or owing to such fear, unwilling to avail himself/herself of the protection of that country. (UNHCR, 1951) An asylum seeker is someone who has sought protection as a refugee, but whose claim for refugee protection has not yet been finalised. While the vast majority of the world's refugees and asylum seekers reside in countries in the Global South (UNHCR, 2019), their arrival in countries of the Global North has become increasingly politicised. Refugees and asylum seekers are often met with intolerance, distrust, and contempt by political leaders and communities of these refugee-hosting states (Verkuyten, 2004). A growing number of studies has found high levels of negative attitudes towards refugees in countries such as the United States (Vollhardt, Nair, & Tropp 2016; Study 2), Israel (Canetti, Snider, Pedersen, & Hall 2016;Tartakovsky & Walsh, 2016), and the United Kingdom (Pehrson, Brown, & Zagefka 2009). This research demonstrates several strong perceptions of refugees that may drive reactions to them; however, there is also evidence that some cohorts of people from refugee backgrounds, particularly asylum seekers, may be viewed in particularly negative terms.
In Australia, where the current study was conducted, research using community surveys consistently finds high levels of negative attitudes towards asylum seekers (see Anderson & Ferguson, 2018). Research also finds that community members feel significantly higher levels of prejudice, anger, and fear towards asylum seekers who arrive without a valid visa by boat when compared with refugees who are resettled in Australia via the country's official resettlement programme (Hartley & Pedersen, 2015). In this regard, much research has focused on unpacking the factors that underpin such negative reactions in both the Australian context (see Anderson & Ferguson, 2018 for a metanalysis) and internationally (see Cowling, Anderson & Ferguson, 2019 for a metanalysis). This research raises important questions as to whether people's attitudes towards asylum seekers can be changed in a positive direction. Given that prejudice towards outgroups starts at an early age (e.g., Aboud, Tredoux, Tropp, Brown, Niens, & Noor, 2012) and that prejudiced attitudes have been linked to experiences of racial discrimination (Habtegiorgis, Paradies, & Dunn, 2014), this question becomes even more pertinent in relation to children. In this article, we explore whether children's attitudes towards asylum seekers might be changed more positively via a short prejudicereduction intervention undertaken in a school setting in Australia.
In particular, we undertook an evaluation of an intervention designed by practitioners at the Australian Red Cross, entitled the In Search of Safety programme; this is being used widely in schools in Australia as a means of reducing prejudice towards asylum seekers. In terms of positionality, Hartley, Fleay and Pedersen are human rights academics and asylum seeker advocates who were contracted by Australian Red Cross employees Cooke and Jeram to evaluate the efficacy of the programme in challenging children's attitudes towards asylum seekers in Australia. To develop an evaluation survey, Hartley, Fleay, and Pedersen conducted a literature review of evaluation tools used with children and effective ways of measuring attitude change (see below for more detail about the programme). Given the programme was designed by the Australian Red Cross, the evaluation was not theory driven, but practice/evidence-based driven.
In this regard, we found the prejudice-reduction framework proposed by Pedersen, Walker, Paradies, and Guerin (2011) particularly helpful in framing our evaluation. Based on an extensive evaluation of the relevant literature on how to effectively implement prejudice-reduction interventions, the framework proposes a number of different mechanisms and techniques that should be considered when conducting such interventions.  argue that effective interventions need to use multiple methods to create attitude change: these include providing accurate information, respectfully involving the audience, encouraging empathy, making positive social norms (e.g., anti-racist attitudes) explicit as these can legitimise attitudes, and fostering contact (real or imagined) with 'outgroup' members.

Prejudice-reduction interventions in school settings
There is a dearth of literature that theorises or evaluates how to effectively implement prejudice-reduction interventions (for a review, see . While definitions of prejudice may vary slightly in these studies, for the purposes of this article we use the term 'prejudice' to refer to 'negative evaluations of people on the basis of their group membership' (Aboud et al., 2012, p. 311). The focus in this article is also on education; that is, on a short-term individual-based prejudice-reduction intervention for children. Although beyond the scope of this article, it is important to note that there are structural elements of prejudice which play a role in creating and sustaining prejudice and are considered elsewhere (see Bonilla-Silva, 1997).
In terms of the literature focusing on children, some research suggests that prejudice-reduction interventions can be useful in creating attitudinal change and that different strategies can be useful. In the first instance, interventions that focus on facilitating contact between different groups of people (known as intergroup contact) appear to have some success. For example, interventions promoting intergroup contact in the form of a cooperative puzzle between urban residents and rural migrant children in China reduced negative attitudes towards the other group (Gu, Nielsen, Shachat, Smyth, & Peng, 2015). Imagining contact with a member of an outgroup has also been found to increase the likelihood that British children would approach an asylum seeker (Turner, West, & Christie, 2013) and refugees (Cameron, Rutland, Douch, & Brown, 2006), that Italian children would approach an immigrant peer (Vezzali, Stathi, & Giovannini, 2012), and that white children would approach a child from an ethnic out-group (Stathi, Cameron, Hartley, & Bradford, 2014). This research highlights the influential role that both imagined and actual intergroup contact can have on children's attitudes and behavioural intentions towards outgroups.
Other research has found that story-telling interventions depicting positive intergroup relationships between majority and minority children can improve some intergroup attitudes (Aronson et al., 2016). Similarly, Vezzali et al. (2012) found that by simply reading a book about intercultural topics, Italian adolescents showed improved intergroup attitudes towards immigrants, a reduction in stereotyping, and an increased desire to engage in future contact. Both intergroup (real and imagined) contact and story-telling based strategies are based on the premise that encouraging children to feel empathy towards an outgroup will make them feel more positive towards outgroup members.
Other prejudice-reduction interventions adopt a more cognitive, informational-based approach which focuses on the provision of correct information about an outgroup with the aim of changing attitudes. It has been proposed that for such strategies to have a sustainable effect, a number of minimum requirements must be met: the information and message must be received and paid attention to; the experience must be positive; and the information and message must be understood correctly, retained and internalised (Farley, 2005). While this approach appears to be particularly ineffectual with people whose prejudiced attitudes are deeply entrenched, it may be useful for people who might not have strongly established attitudes, such as children (Peucker, 2011).
It is important to note prejudice-reduction interventions do not always show positive effects for all participants and effectiveness varies over time. For example, Turner and Brown (2008) evaluated a programme designed to improve school children's attitudes toward refugees where the children received 4 weekly lessons, two of which focused on understanding what a refugee is and the other two on how it would feel to be forcibly displaced. Although the programme increased positive attitudes toward refugees at the beginning, those effects wore off over time. In another study, structured internet interactions have been found to create positive intergroup relations between Muslim and Christian adolescents in Australia; however, although ingroup bias was reduced for the Muslim children over a 12month period this was not the case with the Christian group (White, Abu-Rayya, Bliuc, & Faulkner, 2015). White, Abu-Rayya and Weitzel (2014) found that knowledge about outgroup members increased two weeks after an intervention; however, this effect was not found 12 months later.
It is also useful to report on metanalyses on school-based prejudice-reduction interventions. McGregor (1993) found that role playing reduced racism. Aboud et al. (2012) found that 50% of the interventions used in their meta-analysis produced non-significant results: 40% were positive and 10% were negative. Beelmann and Heinemann (2014) found low-moderate positive intervention effects, although it depended upon the participants' social status: there were larger effects for majoritygroup members (also see Feddes, Noack, & Rutland, 2009;White et al., 2014). Thus, it would seem that prejudice-reduction interventions in school settings, using methods such as intergroup contact, story-telling, and information-cognitive strategies, can be effective in a number of contexts and target groups. However, the strength of any changes after an intervention is likely to be low or moderate. Whether positive results are found also depends upon whether the target group is from a majority or minority group (it is more likely that positive effects will be seen with majority group members).
Interventions targeting prejudice towards asylum seekers in the Australian setting As outlined, there is a small number of studies that have evaluated the effectiveness of interventions that target the attitudes of majority group children towards asylum seekers (Turner et al., 2013) and refugees (Cameron et al., 2006;Turner & Brown, 2008). Some research has investigated the effect of prejudice-reduction interventions with Australian adults regarding asylum seekers (e.g., Pedersen, Paradies, Hartley, & Dunn, 2011;Hartley, Pedersen, & Dandy, 2012). However, to our knowledge there is no published research on interventions aimed at reducing Australian children's prejudice towards asylum-seekers. Given that one's context and location has been found to impact on the ways in which attitudes are formed and expressed (e.g. Pedersen, Watt, & Griffiths, 2007), addressing this gap in the literature is important.
To our knowledge, there are very few published studies which evaluate interventions designed by practitioners; most of those described above were designed by researchers and involve using a specific task (e.g., imagined intergroup contact) to achieve a particular outcome (e.g., reduced scores on a prejudice scale) (for an exception see Turner & Brown, 2008). By contrast, interventions designed by practitioners usually employ a diverse range of methods to foster attitudinal change, which makes it difficult to design and implement evaluations. In this case, it also makes it difficult to unpack any particular 'mechanism' unpinning changes that may occur as a result of the intervention. Given this, the fact that the In Search of Safety programme employs a range of strategies, and there is little control over how the practitioner actually implements the programme, it is important to have a wide array of measures that might be better able to capture any attitudinal change among the children. In the following section, we provide an overview of the programme evaluated in the current study and the rationale behind it. We will also outline the study's design and research aims and explain how we evaluated its impact.

'In Search of Safety' programme overview
The In Search of Safety programme, run by the Australian Red Cross, consists of three components: (i) a presentation, (ii) a game, and (iii) a DVD (note, in some of the other sessions of the programme, a talk by a person from a refugee background is included as a replacement for the DVD, but in this evaluation study only a DVD was used). The programme took approximately 1.5 hours to deliver in one day, and was led by a facilitator from the Australian Red Cross. In the presentation, the facilitator gave the children an overview of human rights, information about the situation of asylum seekers in Australia, and definitions of asylum seekers and refugees that addressed notions that asylum seekers were illegal immigrants and other related myths. The game consisted of the facilitator putting the children in a hypothetical situation of being an asylum seeker; they were asked to imagine that soldiers were coming to harm them and their family in 15 minutes, so they needed to pack six items and decide which six people they would take with them. With regard to the DVD, the children watched Seeking Refuge, a British DVD featuring children's voices narrating the experiences of a number of asylum-seeker children. However, one class did not see the video and another class only saw part of it, due to the extended time taken in the presentation component of the programme in these sessions.
In terms of Pedersen et al.'s (2011) framework, we note a number of mechanisms employed in this intervention: the provision of information and knowledge, the encouraging of empathy (through the DVD and the game), the making of positive social norms explicit (through the teacher reinforcing what was learned in the intervention), and the provision of imagined contact with asylum seekers (through the DVD and the game).

Research design and aims
The research employed a quasi-experimental design that involved asking the children and teachers to complete a questionnaire at Time 1 (immediately before the programme), Time 2 (immediately after the programme), and Time 3 (8-9 months after the programme). Following Greene's (2006) mixed-method framework, quantitative and qualitative data were collected to provide an in-depth understanding of the findings and the effectiveness of the programme over these three time periods. There were two primary research questions. 1.
The first involved quantitative data. It examined whether there were statistically significant differences between Time 1, Time 2, and Time 3 for (a) the single question about how positive a student felt towards asylum seekers, (b) an attitude scale measuring prejudice towards asylum seekers, and (c) behavioural intention scales regarding the student's likelihood of interacting with an asylum seeker in three different settings.

2.
The second research question involved qualitative data. It asked whether there were differences in accuracy with respect to the children's definitions of refugee and asylum seekers at Times 1, 2, and 3. We used NVivo11 to synthesise the most relevant themes.
There were also two minor research aims involving data taken from the teachers that acted as a backdrop to the two research questions regarding the children. i. The first was a qualitative question at Time 2, asking teachers how they found the In Search of Safety programme. ii.
The second question was also a qualitative one, this time at Time 3. It asked whether the issue of asylum seekers had been discussed with the children over the preceding eight months. If it had been, they were asked in what way.

Participants
The children were aged from 10 to 12, and from four schools in Perth, Western Australia. There were 142 children who participated at Time 1, 136 who participated at Time 2, and 125 who participated at Time 3. We were interested in the children (n = 121 children) who participated at all three times. From this point onwards, only the data of this latter group of children is analysed. There were 121 children from four different schools (School 1 = 38 children; School 2 = 21 children; School 3 = 28 children; School 4 = 34 children). There were more boys than girls (boys = 65 children; girls = 54 children; did not disclose = two children).
There were eight teachers who participated at Time 2 and seven teachers who participated at Time 3. Given that the questions for the teachers at Times 1 and 2 differed, and the numbers were so small, we used all teacher data.

Procedure
After approval from the Curtin University Human Research Ethics Committee and piloting of the questionnaire, all children were given information and consent forms by their teachers to be taken home for their parents to consider their participation. On the day of the intervention, only students who provided a signed parental consent form were given a survey to complete. Those who did not have a signed consent form were asked to sit quietly while others completed the surveys at three times (immediately before and after the programme, and 8-9 months later). To minimise the possibility of the teacher's presence influencing the children's response to the survey all questionnaires were circulated and collected by Hartley and Fleay, who were not related to the school in any way. All student and teacher responses were anonymous. With regard to the children, their data were linked by way of a nonidentifying code that they themselves created.

Measures: children
The following questions were asked in the questionnaire at Times 1, 2, and 3.

Qualitative questions
First, the children were asked: 'Please describe what you know or understand about asylum seekers'. They were then asked: 'Please describe what you know or understand about refugees'. There were three categories used in this analysis: correct, half correct, and incorrect -the same coding was used at Times 1, 2, and 3. We awarded a correct response when words like 'safe' or 'flee' were used and there was no mention of 'legality' (the incorrectness of such a notion had been stressed in the presentation). Because of the age of the children, we tried not to get too caught up in technicalities.
For example, an incorrect response about asylum seekers was 'asylum seekers are from another country and they are tourists' (Student 18). A half-correct response was 'I know that asylum seekers are migrants that are seeking for [sic] safety. They are travelling illegally' (Student 37). A correct response was 'asylum seekers are people who are threatened by something and is searchin(g) [sic] for safety' (Student 66).

Positivity
We used an attitude thermometer that had successfully been employed to measure adult prejudice against asylum seekers (e.g.,  and amended it so that it was more suitable for children. Specifically, we asked: 'Please rate how positive you feel about asylum seekers by circling a number from 1 to 7 to show your thoughts.' The children could respond from 1 (not at all positive) to 4 (neutral) to 7 (extremely positive).

Attitudes
We used five items adapted from the Turner and Brown (2008) attitude scale. The prefacing question asked: 'Please rate how much you agree or disagree with each of the following sentences by circling a number from 1 to 7 to show your thoughts' (the midpoint being 'unsure'). We used a 7-point Likert scale ranging from 1 (strongly disagree) to 4 (unsure) to 7 (strongly agree). The questions were: 'It must be scary for asylum seekers when they arrive in a new country'; 'We should be nice to asylum seekers and help them settle in'; 'Based on what I know about asylum seekers, I like them'; 'I would like to have asylum seekers in my class'; and 'I would like to play a game with asylum seekers in the playground'. High scores indicated more positive attitudes towards asylum seekers.

Behavioural intention
We used three items adapted from Vezzali et al. (2012). The prefacing question asked: 'Next, we would like you to imagine you are at a park meeting someone who is an asylum seeker. He/she is the same age as you. Thinking about this, please answer the following questions by circling a number from 1 to 7 to show your thoughts'. The first question asked: 'How happy are you to meet him/her?' We used a 7-point Likert scale ranging from 1 (Very unhappy ) to 4 (unsure) to 7 (Very happy ). The second and third questions asked: 'Would you hang out with him/her at the park?' and 'Would you play a game with him/her at the park?'. Both used a 7point Likert scale ranging from 1 (not at all) to 4 (unsure) to 7 (very much). High scores indicated stronger intentions to interact with an asylum seeker.

Measures: teachers
At Time 2, teachers were asked the qualitative question: 'What are your impressions of the In Search of Safety Programme?' At Time 3, teachers were asked the qualitative question: 'Did you discuss the issue of asylum seekers with the children over the last 8 months? If yes, in what way?' Possible responses were 0 (no), 1 (unsure) or 2 (yes).

Research question 1: using quantitative data, were there changes in positivity, attitude, and behavioural intention at Times 1, 2, and 3?
The reliability of the scales was satisfactory, as far as data for the children who completed all three questionnaires was concerned. For the attitude scale, item 1 at Times 1, 2, and 3 ('It must be scary for asylum seekers when they arrive in a new country') lowered reliability, which meant that this item did not appear to measure prejudiced attitudes in the same way as the other items. Because of this, the item was removed, giving Cronbach alphas of .87 at Time 1, .86 at Time 2, and .85 at Time 3. With the behavioural intention scale, no questions were removed to increase reliability, giving Cronbach alphas of .93 at Time 1, .93 at Time 2, and .87 at Time 3.
As can be seen in Table 1, for positivity, attitudes, and behavioural intentions, all scores were above the midpoint (that is, 4). There was an increase in the single positivity item, the attitude scale, and the behavioural intention scale from Time 1 to Time 2, and a slight decrease in scores at Time 3 (but not as low as Time 1). Using three repeated measure ANOVAs, we investigated whether the average scores changed across different times.
Using a Greenhouse-Geisser correction, the mean scores for the positivity items were statistically significant F(1.89, 226.81) = 68.04, p < .001 with post hoc tests revealing that there was a significant difference between Times 1 and 2 (p < .001), between Times 1 and 3 (p < .001), and between Times 2 and 3 (p < .001). In other words, the intervention elicited a statistically significant increase in positivity over time, even though mean scores dropped significantly from Time 2 to Time 3 (see Table 1).
Using a Greenhouse-Geisser correction, the mean scores for the attitude scales were statistically significant(1.86, 223.17) = 58.98, p < .001, with post hoc tests revealing that there was a significant difference between Times 1 and 2 (p < .001), between Times 1 and 3 (p < .001), and between Times 2 and 3 (p < .001). In other words, the intervention elicited a statistically significant increase in positive attitudes towards asylum seekers over time, even though scores dropped significantly from Time 2 to Time 3.
Using a Greenhouse-Geisser correction, the mean scores for the behavioural intention scales were statistically significant F(1.70, 203.09) = 30.88, p < .001, with post-hoc tests revealing that there was a significant difference between Times 1 and 2 (p < .001) and between Times 1 and 3 (p < .001). There was no significant difference between Times 2 and 3 (p = .550). In other words, the intervention elicited a statistically significant increase in behavioural intentions to interact with asylum seekers over time. Even though mean scores dropped from Time 2 to Time 3, this drop was not significant.

Research question 2: Were there differences in accuracy with respect to the children's definitions of asylum seekers and refugees at Times 1, 2, and 3? Asylum Seekers
At Time 1, most of the children did not know how to define an asylum seeker correctly, left the question blank, or were incoherent in their responses (n = 86; 71%). The next most prevalent category was the children who were correct (n = 24; 20%), followed by children who were half-correct (n = 11; 9%). At Time 2, most of the children knew how to define an asylum seeker (n = 93; 77%). The second most prevalent category was those who did not know how to define an asylum seeker or left the question blank (n = 19; 16%), followed by those who were half-correct (n = 9; 7%). At Time 3, over half of the children knew how to define an asylum seeker (n = 68; 56%). The second most prevalent category was children who did not know how to define an asylum seeker or left the question blank (n = 32; 26%), followed by children who were half-correct (n = 21; 17%). See Figure 2 for a graphical representation of these figures.

Refugees
As was the case with the asylum seeker question, many children did not know how to define a refugee at Time 1. At Time 1, over half the children did not know how to define a refugee (n = 70; 58%). The next most prevalent category was the children who were correct (n = 40; 33.1%), followed by children who were half-correct (n = 11; 9%). At Time 2, almost four-fifths of the children knew how to define a refugee (n = 79; 65%). The second most prevalent category was children who did not know how to define a refugee or left the question blank (n = 30; 25%), followed by children who were half-correct (n = 12; 10%). At Time 3, just under half of the children knew how to define a refugee (n = 55; 46%). The second most prevalent category was children who did not know how to define a refugee or left the question blank (n = 44; 36%), followed by children who were half-correct (n = 22; 18%). See Figure 2 for a graphical representation of these figures. As can be seen by Figures 1 and 2, some of the children's memories of these definitions became less accurate over time. However, the responses were still more accurate at Times 2 and 3 compared to Time 1. Correct

Half-correct Incorrect
Teachers' data At Time 2, most of the teachers reported finding the programme useful (86%). However, one teacher (14%) indicated some ambivalence. Furthermore, all teachers who were surveyed indicated at Time 2 that they intended to discuss the content of the programme with the children in the future, and all teachers who were surveyed 8-9 months later (Time 3) reported that they had. The asylum seeker issue was integrated into other units of study such as Literacy, History, Migration/Immigration, and the Humanities and Social Sciences Curriculum.

Discussion
The aim of this study was to investigate whether a short school-based prejudicereduction intervention would have a positive effect on children's attitudes towards people seeking asylum in Australia. Our results suggest that children who participated in the In Search of Safety programme held more positive attitudes towards asylum seekers, and this change lasted over a period of 8-9 months, when Time 3 testing took place. While these effects regressed over time, scores at Time 3 were still significantly higher than at Time 1. Because the teachers' data act as a backdrop to the children's data, we outline the two minor research questions first. First, most teachers reported finding the programme useful. For example, Teacher 2 stated: 'the children responded well. They posed questions and were reflective at the end of the movies'. Teacher 4 stated: 'A very important programme for young people that presents complete focus [sic] in an easy-to-understand way'. However, Teacher 5 said: 'Suitable for age -but slow' indicating both positive and negative aspects of the programme. Second, all teachers surveyed reported 8-9 months later that they had incorporated the asylum-seeker issue into a variety of other units (the most prevalent being Migration/Immigration). Thus, overall, the teachers who were surveyed were satisfied with the programme and integrated asylum seeker issues into their teaching over the next 8-9 months.
When we focus on the impact of the intervention on the children's positivity, attitudes, behavioural intentions, and knowledge about asylum seekers and refugees, we see that scores regarding the positivity item and the attitude scale dropped significantly from Time 2 to Time 3. However, the children reported significantly more positive feelings and attitudes towards asylum seekers directly after their participation in the programme, as well as 8-9 months later. Furthermore, they were significantly more likely to intend to interact with an asylum seeker at the end of the programme (although there was a small drop in mean scores on the behavioural intention scale from Time 2 to Time 3, they did not drop significantly).
The results of the evaluation suggest the intervention was successful, which supports some previous research. For example, past studies have found that prejudice-reduction interventions can increase attitudes and/or positivity when using a combination of empathy and cognitive-informational strategies (e.g. Cameron et al., 2006), increase behavioural intentions to interact with an outgroup member (e.g. Stathi et al., 2014;Vezzali et al., 2012), and increase knowledge about an outgroup (White & Abu-Rayya, 2012). However, our findings do not support other research findings which suggest that while interventions lead to more positivity afterwards, these effects do not last over time (e.g. Turner & Brown, 2008). In our study, although there were drops at Time 3, significant improvement from Time 1 was still evident. This may be due to teachers discussing the issue of asylum seekers with the children in the 8-9 months after the intervention. While we are unable to comment definitively on this, as we did not measure this in our survey, the role of teachers' fostering open-discussion about race-related issues has been found to be a critical factor in reducing racism in schools (Spencer, 1998). In the current study, the fact that the teachers integrated what had been learnt in the intervention into their teaching programme over the following months is likely to have been an important component in shifting attitudes in a more positive direction. One reason for the potential importance of continued classroom discussions is that prejudice and discrimination is known to thrive in environments in which they are perceived to be the norm, but perish when the existing social norms do not allow it Sechrist & Stangor, 2001).
There was an increase in knowledge about the definitions of asylum seeker and refugee immediately following the programme. While there was a decline in this knowledge gain 8-9 months later, children still reported an increase in knowledge at Time 3 compared with the knowledge they had at Time 1. For example, Student No 37 stated before the programme: 'I don't know what a [sic] asylum seeker is'. Immediately after the programme, it was remarked that asylum seekers were 'people running from their homeland from war or other things. But they may not be able to have safety and be sent back'. 8-9 months later, we read that: 'asylum seekers are people that leave their country and go to another place far away to be safer'.
The same pattern was found with the definitions of 'refugee'. Before the programme, we have comments such as those made by Student No 26: 'I do not know much about them'. Immediately after the programme, we read that 'A refugee is someone who once was an asylum seeker but has been living somewhere [sic] safe for a long time'. Eight-9 months later, the comment is made that: 'I know that they move or evacuate from their home when something really bad happens. So they come to a different country like Australian [sic] for safety'.
As noted previously, some leeway was given to the children with their responses. This was because of their age -it was not expected that children of this age would have complex understandings about asylum seekers and refugees. Indeed, many Australian adults do not have a good understanding of the issue, even though they are given correct information. For example, despite participants being given a clear description of a refugee at the beginning of a questionnaire in a study on refugees in the Australian community, many participants confused the categories of 'refugee' and 'asylum seeker', so that a qualitative category was established with the title: 'confused refugee with immigrant or asylum seeker' (Turoy-Smith, Kane, & Pedersen, 2013).
Furthermore, the fact that only one fifth of the children in the current study could correctly define 'asylum seeker' prior to the programme raises the question of who they were thinking about when they rated their attitudes and behaviours towards asylum seekers at Time 1. Nevertheless, what is important is that whoever they were thinking about beforehand was later thought of in more positive terms.
It is also important to highlight that all aspects of the programme were rated as very important by the children. For example, with regard to the game, Student 43 said that 'it put me in their shoes and it showed me how terrifying it is'. Further, Student No 6 said: 'It told a true story so I could understand how it would feel'. This suggests that the role of story-telling in the game and the DVD in eliciting empathy may be a useful strategy to foster understanding towards asylum seekers (Aboud et al. 2012). The idea that aspects of the intervention helped the students identify with asylum seekers supports previous research which finds that people will be more favourably disposed towards and concerned about other people when they learn to see they are closer and more similar to themselves (Aronson et al. 2016). The intervention strategies that focused on showing what it would be like to seek asylum arguably enabled students to identify with asylum seekers and their plight.
With regard to the information provision, Student 8 said 'I didn't know too much about asylum seekers but the presentation was full of information so it helped me learn more'. While it is difficult to pinpoint the specific role that the provision of information played in increasing the scores on our evaluation survey, this does suggest that for some children information was an important factor, as highlighted by Peucker (2011). As noted previously, the teachers' discussion of the issue throughout the next 8-9 months is also likely to have been important.
Our evaluation supports theory on how to run a prejudice-reduction strategy . In particular, it is argued by  that effective prejudice-reduction strategies need to use multiple methods and techniques, as was the case in the current study. These included the provision of accurate information (i.e., knowledge), the encouraging of empathy (through the use of the game and watching the DVD), the making of positive social norms explicit as these can legitimise attitudes (i.e., the fact that the teachers reinforced pro-refugee sentiment throughout the school year), and the provision of real or imagined intergroup with the target group which could involve either guest presentations or appropriate DVDs. Finally,  suggest that interventions be evaluated. This is important because it is possible for interventions to backfire (see Trevors, Muis, Pekrun, Sinatra, & Winne, 2016) or simply have no effect (described in Aboud et al. 2012).

Limitations and future research
Despite the interesting results, it is important to reflect on the potential limitations of the current study. A shortcoming may be that part of the intervention relied on the provision of correct information. Peucker (2011) makes the valid point that strategies focusing on providing correct information are often ineffective when negative attitudes towards the target group are entrenched because they are resistant to persuasive communication. They argue that information provision is more likely to be effective for people with mildly prejudiced attitudes, or even positive attitudes towards diversity and outgroups. In our study, scores on the positivity, attitudes, and behavioural intentions towards asylum seekers at Time 1 were all above the midpoint, suggesting that, on average, the children had somewhat positive inclinations towards asylum seekers before the intervention. This suggests that using cognitive-informational strategies may be useful in solidifying this positivity.
Another consideration with informational-based strategies is the ability to create structural change.  argue that for social change to occur, the structural elements of oppression must be examined as well as individual attitudes. As noted above, there is serious structural discrimination against asylum seekers; for example, they are subjected to a range of punitive polices in Australia, including indefinite detention. That said, we believe that Australia needs top-down and bottom-up change; that is, changes to the structure that affects asylum seekers and refugees as well as to individual attitudes and actions. As argued by Duckitt (2001) and Paradies (2005), we need to look at cognitive (e.g., how we categorise), individual (e.g., stereotypes), interpersonal (e.g. social influencing such as persuasion and intergroup contact), and societal (e.g. government policies) levels of racism (also see Jensen, Cismaru, Lavack, & Cismaru, 2010). We hope that, in the long term, personal attitudes and actions might affect social structures. One place to start is the school system.
We also need to consider whether the results of the intervention will last, given the negative societal norms surrounding asylum seekers. Our study shows that while prejudice-reduction scores did decrease over time, they were still more positive than they were initially. The teachers' continuing use of the topic throughout the school year was no doubt helpful in maintaining positive attitudes. Only future research can inform us how long these changes will last, and indeed whether they last at all without the teachers' input. We note that we did not test the actual behaviour of the children but their attitudes and intended behaviour. Future research could investigate the relationship between intended behaviour and actual behaviour.
While we acknowledge that the In Search of Safety programme is an individualistic intervention that does not seek to address the structural and discriminatory factors facing people seeking asylum, we concur with others that oppression involves multiple layers (e.g. Bronfenbrenner, 1979). We argue that for oppression against asylum seekers to be dismantled, it is necessary to tackle oppression at many levels: both individual and systemic. Furthermore, on an individualistic level, a change in children's attitudes to a more positive one would be helpful to refugee children that they come in contact with.
As occurs in many real-life situations, it is also important to note that there was no control group, which makes it difficult to exactly locate any particular mechanisms or method underpinning the changes we found. (see Paluck & Green, 2009). For example, were changes to the prejudice scale a result of watching the DVD, viewing the presentation, participating in the game, or a combination of these? Future research looking to experimentally manipulate these conditions could follow up on these questions. That said, such 'real-world research' is a very important contribution. Indeed, as Paluck and Green (2009) note, we should be sceptical of interventions conducted only within laboratory settings. To our knowledge, there are few published studies which evaluate interventions designed by practitioners in the field (see Turner & Brown, 2008). Similarly, there are few interventions that test participants three times (e.g. White & Abu-Rayya, 2012): the current study did this, using reliable and valid instruments.

Overall conclusions
Regarding the student questionnaire data, both the qualitative and quantitative results indicated that the programme was effective in increasing the children's positivity, attitudes, and intended behaviour towards asylum seekers in the short term and long term. There was also a positive change in children's knowledge of how to define a refugee and asylum seeker both in the short-term and long-term. However, given the decrease in scores from Time 2 to Time 3, it would be helpful for teachers to remind children of these definitions when bringing up the issue in future classes. As White et al. (2014) also found, knowledge acquisition does not always last.
Regarding the teacher data, the Time 3 results indicate that the teachers generally felt very positive about the In Search of Safety programme although some reported negative issues (e.g., that there was not enough time). Given that all of the teachers surveyed at Time 3 had discussed the issue of asylum seekers with their children in the 8-9 months following the intervention, we would argue this contributed to the positive results of the student data over the long term. This study indicates that school-based interventions using a mixture of approaches to fostering attitude change-providing information, encouraging empathy, making positive social norms more explicit, fostering imagined contact with asylum seekers, and reinforcing the intervention content by teachers throughout the school year-can positively influence children's attitudes towards asylum seekers.