Predicting community attitudes towards asylum seekers: A multi‐component model

Abstract

Objective

The current study investigated the role of cognitive, affective, and behavioural information in the prediction of overall attitudes towards asylum seekers.

Method

A sample of 98 Australian adults participated in an online self‐report questionnaire where participants generated their cognitive, affective, and behavioural factors towards asylum seekers and then rated those factors on a continuum from ‘positive’ to ‘negative’.

Results

Multiple regression analysis confirmed the primary role of cognitive, then affective, factors in predicting attitudes towards asylum seekers. Cognitive information accounted for a moderate, significant 31.92% of the variance in overall attitudes towards asylum seekers. The unique variance contributed by affective information accounted for a small but significant 3.46% of the variance in overall attitudes; the unique variance contributed by behavioural information was not significant.

Conclusions

The results provide a holistic theoretical basis for the assertion that community attitudes towards asylum seekers are based primarily on cognitive evaluations of the minority group. These findings have implications for changing community attitudes towards people seeking asylum in Australia.

Key words:

• affect
• asylum seekers
• attitudes
• attitudes towards asylum seekers
• beliefs
• social issues

What is already known about this topic

• Asylum seeking and asylum‐seeking policies are contentious issues in the Australian community.

• Community attitudes towards, and portrayal about, asylum seekers tend to be negative.

• The role of cognitive, affective, and behavioural factors in attitude formation remains under‐explored.

What this topic adds

• Age, education level, and gender were not significantly associated with attitude.

• Individuals base their attitudes toward asylum seekers on their beliefs over and above their affective or behavioural evaluations.

• Attempts to promote positive attitudes towards asylum seekers require a focus on changing the negative beliefs about people seeking asylum in Australia.

Within Australia, prejudicial attitudes and negative portrayals of asylum seekers are routinely expressed by the media and political figures (Every & Augoustinos, 2007; Klocker & Dunn, 2003; McKay, Thomas, & Blood, 2011; Suhnan, Pedersen, & Hartley, 2012) as well as within the wider community (Klocker & Dunn, 2003; McKay, Thomas, & Kneebone, 2012; Muller, 2016; Pedersen, Watt, & Hansen, 2006). These attitudes are accompanied by legislative policies that remain incongruent with the United Nations Refugee Convention of 1951, to which Australia is a signatory, and have been criticised for their severity (Corbett, Gunasekera, Maycock, & Isaacs, 2014; Haslam & Holland, 2012; Woodhead, 2015). Long‐term physical and psychological effects are evident within asylum seeker populations (Briskman, Goddard, & Latham, 2008) that are likely to be a consequence of these policies, and prolonged detention is linked to an increase in the incidence of depression, post‐traumatic stress disorder, and suicide (Davidson, Murray, & Schweitzer, 2008; Schweitzer, Melville, Steel, & Lacherez, 2006).

A growing body of literature has noted several antecedent factors that predict negative attitudes towards asylum seekers (Louis, Duck, Terry, Schuller, & Lalonde, 2007; McKay et al., 2012). These factors range from the role of false beliefs to the role of nationalism and positive contact with the asylum seeker group (Augoustinos & Quinn, 2003; Louis et al., 2007; Pedersen, Attwell, & Heveli, 2005). Given the current prominence of asylum seekers in current discourse and legislative policies, a theoretically underpinned study investigating the antecedent factors that predict attitudes towards asylum seekers is required. A number of studies have examined antecedent factors from varying theoretical standpoints, such as dual process theory (Duckitt, 2001; Duckitt & Sibley, 2010) and social dominance theory (Sidanius & Pratto, 1993). However, most studies to date have focused on either a cognitive or affective orientation. As such, a holistic theoretical account of the antecedent factors in prediction of attitudes towards asylum seekers is warranted.

THE MULTI‐COMPONENT MODEL OF ATTITUDES

A holistic and empirically supported model of attitudes was developed by Zanna and Rempel (1988) and Eagly and Chaiken (1993). They propose a tripartite model of attitude formation that articulates an attitude as a predisposition to evaluate an attitude object with varying degrees of favour or disfavour. This evaluation is comprised of affective, cognitive, and behavioural reactions towards the attitude object (Eagly, Mladinic, & Otto, 1994)—in this case, asylum seekers. The affective component is the valanced feelings that are associated with the attitude object, the cognitive component is the beliefs or attributes associated with the attitude object, and the behavioural component of the model relates to past or intended behaviours that are associated with the attitude object (Eagly et al., 1994; Huskinson & Haddock, 2006; Zanna & Rempel, 1988). When evaluating an attitude object, it is theorised that individuals take into account their affective, cognitive, and/or behavioural reactions towards the attitude object. These three components are related yet empirically distinguishable (Eagly & Chaiken, 1993) and account for unique and shared proportions of variability in elicited attitudes (Abelson, Kinder, Peters, & Fiske, 1982; Breckler, 1984; Eagly et al., 1994; Esses, Haddock, & Zanna, 1993; Haddock & Zanna, 1997). Furthermore, the overall attitude towards an attitude object (whether negative or positive) can comprise all three factors or any combination of factors (Eagly & Chaiken, 1993; Zanna & Rempel, 1988).

Utility of the multi‐component model

While affective information was found to be the best predictor of attitudes towards national politicians (e.g., Abelson et al., 1982) and certain ethnic groups (e.g., Hispanics, Asians, Arabs, Americans) in some studies (e.g., Stagnor, Sullivan, & Ford, 1991), other studies indicated that cognitive information was a better predictor than affective information when evaluating political candidates (e.g., Haddock & Zanna, 1997) and attitudes towards men, women, Democrats, Republicans, abortion, affirmative action, and welfare (Eagly et al., 1994). Within the context of the abovementioned literature, the construct of behaviour has largely been omitted due to the complexity of behaviour and the difficulty of measuring it with semantic differential scales. Haddock, Zanna, and Esses (1994) used open‐ended measures of cognition, affect, and behaviour to assess overall attitudes to Native peoples in Canada and found that the quality and magnitude of past experiences accurately predicted overall attitudes towards particular minority groups.

The equivocal findings highlighted above can be explained by the individual features of the attitude object in question (Eagly et al., 1994). For example, within their study, Esses et al. (1993) discussed the fact that their high cognitive predictability, measured as symbolic beliefs about the target groups, emerges when the target attitude object is perceived in a less favourable light (i.e., native Indian, homosexual, and/or Pakistani). Moreover, individual differences may explain the wide range of results reported using the tripartite model.

PREDICTORS OF NEGATIVE ATTITUDES

For the purposes of the current study, the findings of previous research examining predictors of attitudes towards asylum seekers have been re‐conceptualised as the three predictor variables in the tripartite model to elicit the appropriateness of the model in the current context.

Cognitive factors

Cognitive factors are the thoughts or ideas that emerge about a particular stimulus or community group, mostly defined as a set of beliefs that have associations with particular characteristics of that stimulus or group (Fishbein & Ajzen, 1975). Whilst a whole overview of the dual process model is beyond the scope of the current article, the model posits that prejudice is derived by dual processes related to threat and competition (Duckitt, 2001; Duckitt & Sibley, 2010). The threat‐based processes manifest as right‐wing authoritarianism and is seen as an individual factor where individuals who score high in right‐wing authoritarianism tend to be more intolerant of societal deviance and align themselves with social control and conformity. This leads to more negative attitudes of particular out‐groups, perceived to be a threat to the social order and the security of their in‐group. The competition‐based processes manifest as social dominance orientation, which posits that society functions optimally with a hierarchal system, with a few dominant groups in a superordinate position and a number of subordinate groups below them (Sidanius & Pratto, 1993).

Australian research has found support for the link between right‐wing authoritarianism, social dominance orientation, and negative attitudes towards asylum seekers (e.g., Louis et al., 2007; Nickerson & Louis, 2008). In accordance with the threat‐based process (right‐wing authoritarianism), the internalisation of false beliefs that asylum seekers are queue jumpers, financially well off, a threat to border and national security, and were provided with government grants were strongly correlated with negative attitudes towards asylum seekers (Augoustinos & Quinn, 2003; Louis et al., 2007; McKay et al., 2012; Pedersen et al., 2005; Pedersen et al., 2006; Suhnan et al., 2012). Nickerson and Louis (2008) found that individuals who aligned with authoritarian beliefs and strongly identified as Australian were more likely to hold unwelcoming attitudes towards asylum seekers. Similarly, Louis et al. (2007) demonstrated that participants who preferred a hierarchical structure (social dominance orientation) were more likely to endorse the restriction of asylum seekers' access to Australia. More recent research has supported these earlier studies and found that individuals who score high on social dominance ratings have more negative attitudes towards asylum seekers (Anderson, Stuart, & Rossen, 2015; Perry, Paradies, & Pedersen, 2014; Trounson, Critchley, & Pfeifer, 2015).

Affective factors

It has been demonstrated that Australians report significantly higher levels of anger, fear, and threat towards asylum seekers compared to resettled refugees (Hartley & Pedersen, 2015). The perception of threat and false beliefs posed by asylum seekers directly affects individuals within a community setting through an affective evaluation of the cognitive information. The generation and internalisation of cognitive information, such as false beliefs, perception of threats, and stereotypes, has been recognised to occur through Australian media and government (Augoustinos & Quinn, 2003; Klocker & Dunn, 2003; McKay et al., 2011). Negative portrayals inherent in the use of terms such as ‘boat people’ and ‘queue jumpers’ have resulted in a social construction of asylum seekers as an out‐group who are perceived as a realistic and symbolic threat to Australia (Every & Augoustinos, 2007; McKay et al., 2011). This affective evaluation leads to feelings of fear, anger, and disgust, which increase negative attitudes towards the asylum seeker minority (Louis et al., 2007; McKay et al., 2012; Suhnan et al., 2012). Eagly et al. (1994) noted that the connection between affective and cognitive evaluations in attitude formation provides explanatory power of the antecedent factors that predict the attitude.

In terms of viewing affective information as a direct predictor of attitudes towards asylum seekers, Pedersen and Thomas (2013) and Hanson‐Easey and Augoustinos (2011) found that individuals who empathised and/or sympathised with asylum seekers had significantly less prejudicial attitudes towards the group compared with individuals who did not. Currently, the predictive value of affective factors has been limited by the use of highly specific factors as predictors of attitudes towards asylum seekers rather than a holistic approach of examining several factors in combination and interaction.

Behavioural factors

A small number of studies have looked at the role of behavioural factors in the prediction of attitudes towards asylum seekers and other groups. Barlow et al. (2012) found that negative contact (past experiences) resulted in an increase in negative attitudes towards asylum seekers. Moreover, studies have shown that positive contact increased positive attitudes towards asylum seekers. For instance, Turoy‐Smith, Kane, and Pedersen (2013), in their survey on attitudes towards refugees, found that the quality of contact was more important than the quantity of contact in changing attitudes. These findings indicate that contact is associated with, and can predict, attitudes towards out‐group members in situations where the quality of the contact is optimised. This finding has been established in a meta‐analysis that demonstrated the effect of positive contact on decreasing negative attitudes towards out‐group minorities (Pettigrew & Tropp, 2006).

A special case: The role of threat

Threat is seen as symbolic, realistic, intergroup, or stereotypical, and it is assumed that the perceived threat of a stimulus arises from cognitive and affective evaluations (Stephan & Stephan, 1996). Schweitzer, Perkoulidis, Krome, Ludlow, and Ryan (2005) found that individuals who perceived a threat towards their physical health held negative attitudes towards the asylum seeker population. A small but significant relationship was also found in terms of symbolic threat (i.e., threats to the Australian way of life). More recent findings support the work of Schweitzer et al. (2005) by using contemporary threat perceptions, such as threats to border and national security (Louis et al., 2007; McKay et al., 2012; Suhnan et al., 2012).

THE CURRENT STUDY

Overall, while the literature base has identified a number of factors that predict attitudes towards asylum seekers, these factors have primarily been examined in isolation. Pedersen and Thomas (2013) found that both cognitive and affective information predicted attitudes. Behavioural information was omitted, which the authors highlighted as a limitation. Nickerson and Louis (2008) included the three components, although attitudes were used as the cognitive component. As such, we applied a holistic theoretical framework, the tripartite model of attitudes, which has been used extensively in minority group research (Eagly & Chaiken, 1993; Zanna & Rempel, 1988), to the study of attitudes towards asylum seekers.

The aim of the current study was to measure and predict community attitudes towards asylum seekers, drawing on the tripartite model of attitudes and utilising an open‐ended template of attitudes where participants generated their cognitive, affective, and behavioural factors towards asylum seekers and then quantitatively rated those factors on a continuum—from ‘positive’ or ‘negative’ (Eagly et al., 1994). Two hypotheses were proposed. First, cognitions, emotions, and past experience will each predict a significant proportion of unique variance in attitude towards asylum seekers after controlling for gender, age, and educational attainment; Second, based on research demonstrating that cognitive information is a better predictor than affective information, especially for highly salient groups (Eagly et al., 1994), cognitions will predict a ‘large’ proportion of the variance in attitude towards asylum seekers, whereas emotions and past experience will each predict a ‘small’ proportion of the variance.

METHOD

Research design

The study used a cross‐sectional correlation design. The criterion variable was the valanced attitude(s) towards asylum seekers. The predictor variables were the cognitive, affective, and behavioural factors. Age, education, and gender were control variables.

Participants

An a priori power analysis using G*Power (version 3.1.9.2; Faul, Erdfelder, Buchner, & Lang, 2009) determined that 77 participants would be necessary to achieve adequate power of .80 for a medium effect size (f ² = .15) at an alpha level of .05. The effect size was based on previous research with tripartite model studies (e.g., Haddock et al., 1994; medium effect size, N = 83). Convenience and snowball sampling were used to optimise the diversity and number of responses (Lewis, Watson, & White, 2009).

Measures

A modified template constructed by Eagly et al. (1994) was utilised, which comprised five sections. The original measure has adequate internal consistency reliability (0.62–0.94) and high discriminant validity. Reliability of our modified version was high, with Cronbach's alpha for the cognitive, affective, and behavioural scales being 0.88, 0.81, and 0.80, respectively. Section 1 measured the criterion variable through rating the overall attitude towards asylum seekers along a single 7‐point Likert‐type scale ranging from very negative (‐3) to very positive (+3). Sections 2, 3, and 5 measured the affective, cognitive, and behavioural components, respectively. Section 4 measured control variables of age (categories of 18–24, 25–30, 31–40, 41–50, 51–60, and 61–70), gender, and level of education (ranging from year 10 through to postgraduate qualification). For sections 2, 3, and 5, participants were asked to list up to seven feelings (affect), thoughts (cognition), and past experiences (behaviour) they have in relation to asylum seekers. An example response, unrelated to the issue of asylum seekers, was given prior to each section. For instance, ‘For the issue water pollution you might respond: I believe it is wrong’. Upon generating each item, participants were asked to rate responses on a 7‐point scale ranging from very negative (–3) to very positive (+3).

Procedure and data collection

Ethics approval was obtained from Curtin University's Human Research Ethics Committee. An event page on social media was constructed to invite participation in the study, with a link to the questionnaire hosted on Qualtrics®. All participants were required to confirm that they were aged 18 or over and living in Australia.

RESULTS

The final sample was 98 adults (44 men, 54 women) aged 18–70 years; thus, adequate power was achieved. In terms of age, 42.9 % were in the 18–24 age category. In terms of educational attainment, 40.4% had a high school level of education, 35.4% had an undergraduate qualification, and 9.1% had a postgraduate qualification; 15.2% stated other.

Data screening and scoring

A total of 129 people commenced the questionnaire; 12 were deleted as spam, 17 were deleted due to incomplete data, and 2 cases were excluded due to influence on the regression model. Case‐wise diagnostic analysis of these two cases showed that covariance ratio for these individual cases were .599 and .524, which is less than the requirement of CVR < 1‐[3(k + 1)/n]. As such, the removal of these cases was likely to improve the precision of some of the model's parameters (Belsley, Kuh, & Welsch, 2005).

The self‐response format of the survey required participants to generate beliefs, emotions, and behavioural factors that were idiosyncratic to them. As such, there was variation in the number of responses generated between participants. The number of cognitions expressed by participants ranged from 1 to 7, with the majority listing 7. There were fewer emotions and behaviours provided by participants; while the number of emotions and behaviours expressed by participants ranged from 1 to 7, the majority listed 3 and 2 for each construct, respectively. Example items generated by participants for the cognitive, affective, and behavioural scales are provided in Table 1.

Table 1. Community beliefs, emotions, and behaviours in relation to asylum seekers

Construct	Example response
Beliefs
Authenticity of asylum seekers	I believe….not all are genuine refugees I believe….most asylum seekers are genuine refugees I believe…they are queue jumpers I believe…no one really knows who they are
Resources should go to Australians first	I believe….we can't even look after our own, let alone someone else coming in
People have the right to seek asylum	I believe…we should welcome asylum seekers to Australia I believe…people have the right to seek asylum
Asylum seekers are escaping terrible experiences	I believe…they come from terrible countries I believe…people who are in fear of their lives should be able to seek asylum I believe…most have faced extreme adversity
Alternatives to processing/accepting asylum seekers	I believe…that we should help in other ways I believe…that people need to enter Australia in official ways I believe…they deserve to have claims processed in Australia I believe…that it's not the right thing to lock up children I believe…asylum seekers are treated terribly
Everyone deserves freedom and respect	I believe…asylum seekers deserve respect as everyone else in the society I believe…that everyone in the world has the same rights to freedom and a happy life
Consequences of accepting asylum seekers	I believe…it could eventually lead to overcrowding
Asylum seekers and crime/terrorism	I believe…asylum seekers are terrorists I believe…they are all illegal I believe…they are innocent I believe…that they are not terrorists
Emotions
Anger/frustration	I feel…angry that the media is lying on behalf of the government I feel…angry when asylum seekers are phony I feel…angry that people assume they're greedy I feel…frustrated by the need for these people to escape their own countries
Empathy/pity	I feel…pity for those genuine asylum seekers I feel…very sorry for their situation, having to leave their homes and families
Shame/disappointment	I feel…ashamed to be an Australian because of the Australian government I feel…disappointed in what I thought was a country that looks after their own
Helplessness	I feel…helpless as I can do nothing
Sadness	I feel…sad they are locked up I feel…sad they need to flee their home I feel…sad that people who are not real asylum seekers just come to our country willy nilly
Worry/Fear	I feel…worried about how much is being spent on them and not on our own elderly, poor and homeless that we already have here I feel…scared they may have terrorists among them I feel…some are murderers terrorists which makes me feel unsafe
Gratitude	I feel…lucky that I am born and raised in Australia
Behaviours
Activism	An experience I had was…attending a rally for asylum seekers An experience I had was…a friend was being super xenophobic and racist about it, and I had to say something
Contact with asylum seekers	An experience I had was…a conversation with asylum seeker An experience I had was…I live in a town with the asylum seeker camp An experience I had was…I saw an asylum seeker in detention An experience I had was…I have worked with people who came to Australia as asylum seekers An experience I had was…I have taught a girl who was a refugee from Africa which I think is a different situation. She was gorgeous and I really liked her
School/university	An experience I had was…I learnt about them in school An experience I had was…case studies in psychology
Media	An experience I had was…I heard about them on the news An experience I had was…I heard it on the radio An experience I had was…hearing about the Christmas Island Boat Disaster An experience I had was…I viewed a television program that revealed the atrocities asylum seekers face
No experience	An experience I had was…I have no experience with asylum seekers An experience I had was…I haven't; they are all locked up

Individual participant scores were obtained by summing the valence ratings and dividing this by the number of listed responses. Mean ratings could range from ‐3 to 3. Little's MCAR test produced a non‐significant result, χ²(724, N = 101) = 732.784, p < .402, indicating that all data were missing completely at random.

Hypothesis testing

Neither age (F(5, 92) = 1.212, p = .310) nor education level (F(4, 93) = 1.844, p = .127) had a significant effect on attitude towards asylum seekers. Similarly, gender (r = .133, p = .095, N = 98) had no relationship with attitude towards asylum seekers, and therefore, none of the three variables was used as controls in subsequent analyses.

The mean, standard deviation, and Spearman's rho correlations for each of the retained predictor and control variables are displayed in Table 2. Overall attitude to asylum seekers was positive, as were cognitive and behavioural ratings; the affective factor was negative. The cognitive and behavioural factors both had a significant and moderate correlation with overall attitude (0.484 and 0.321, respectively), while the affective factor was not significantly correlated with overall attitude towards asylum seekers. All three antecedent factors were correlated with one another.

Table 2. Descriptive statistics and spearman's rho correlation showing relationships between overall attitude and each of the predictors

Construct	M	SD	1	2	3	4
1. Overall Attitude	0.92	1.66	1
2. Cognitive	0.83	1.55	.484^**	1
3. Affective	−0.26	1.54	.045	.425^**	1
4. Behavioural	0.23	1.57	.321^**	.346^**	.225^**	1

Note. N = 98.

** p < .01 level (2‐tailed)

* p < .05 level (2‐tailed).

Regression analysis

Assumptions of normality, linearity, and homoscedasticity of residuals were met. A standard multiple regression was performed, with Cohen's (1988) conventions used to interpret effect sizes. Cognitive, affective, and behavioural variables accounted for an additional significant 40% of the variability in attitudes towards asylum seekers, R ² = .42, adjusted R ² = .40, F(3,94) = 22.658, p < .001 (large effect f ² = .67). Unstandardised (B) and standardised (β) regression coefficients and squared semi‐partial correlations (sr ² ) for each variable are reported in Table 3. The unique variance contributed by cognitive information accounted for a moderate and significant 31.92% (sr ² = .565) of the variance in overall attitudes towards asylum seekers. The unique variance contributed by affective information accounted for a small but significant 3.46% (sr ² = ‐.186) of the variance in overall attitudes. The unique variance contributed by behavioural information accounted for a small and non‐significant 1.28% (sr ² = .113) of the variance in overall attitudes.

Table 3. Unstandardised (B) and standardised (β) regression coefficients and squared semi‐partial correlations (sr²) for each predictor variable in a regression model predicting attitudes towards asylum seekers (N = 98)

Variable	B [95% CI]	(β)	sr ^2
Cognitive	.720 [.468, .862]^***	.669	.565
Affective	−.230 [−.408, −.039]^*	−.212	−.186
Behavioural	.129 [−.095, .253]	.122	.113

Note. CI = Confidence interval.

*** p < .001 level

** p < .01 level

* p < .05 level (2‐tailed).

DISCUSSION

The purpose of this study was to determine the underlying structure of attitudes towards asylum seekers in the wider Australian community by applying the tripartite model of attitudes (Eagly & Chaiken, 1993; Zanna & Rempel, 1988). This adds to the literature utilising the tripartite model as well as being the only study to date to apply this theoretical model to the issue of community attitudes to people seeking asylum in Australia. The hypothesis that cognitive, affective, and behavioural factors would each predict a significant proportion of unique variance in attitudes towards asylum seekers was partially confirmed as cognitive, affective, and behavioural factors contributed a significant proportion of the variance with a large effect; however, even though the behavioural factor was significantly correlated with overall attitude, it contributed a small, non‐significant proportion of the variance in the predictor model. The hypothesis that cognitive information would predict a ‘large’ proportion of the variance in attitude towards asylum seekers, compared to the affective and behavioural factors, was confirmed.

Cognitive information

The significant correlation between cognitive information and overall attitude towards asylum seekers provides support for the current study's hypothesis that individuals base their attitudes towards asylum seekers on their beliefs over and above their affective or behavioural evaluations. The finding that cognition explained a much larger proportion of the variance than did affect or behaviour is consistent with previous studies showing that cognitive information is a better predictor of attitudes than affective information (Eagly et al., 1994; Haddock & Zanna, 1997), especially when the attitude object in question is highly salient, such as ethnic minorities, political figures, and specialist groups (e.g., pro‐choice supporters [Eagly et al. (1994]). Further validation comes from Pedersen and Thomas's (2013) study which found that priming participants as being either ‘similar’ or ‘different’ to asylum seekers resulted in decreased or increased prejudicial attitudes, respectively.

The current study also provides evidence for the link between beliefs and attitudes towards asylum seekers. Individuals who hold negative and false beliefs about asylum seekers have more a negative attitude towards the group (McKay et al., 2012; Suhnan et al., 2012). This is explained by the current research findings that individuals may base their attitude on a cognitive evaluation of asylum seekers. The fact that an increase in the positive nature of cognitive evaluations results in an increase in overall positive attitudes towards asylum seekers implies that individuals may be basing their attitudes towards asylum seekers on cognitively based information. This suggests that there is a strong internalisation of beliefs and cognitive processes.

Negative false beliefs, especially the notion that asylum seekers are not genuine refugees (Hartley & Pedersen, 2007), has been linked to a greater acceptance of the Australian government's harsh policies towards asylum seekers (e.g., mandatory detention; Hartley & Pedersen, 2007). Given the association between detention and serious mental health concerns (Briskman et al., 2008; Davidson et al., 2008; Schweitzer et al., 2006), targeting cognitive processes and beliefs may be a cornerstone of influencing and changing negative attitudes towards this minority group.

Affective information

Some studies have identified affective information as the best predictor of overall attitudes towards certain groups (Abelson et al., 1982; Esses et al., 1993; Stagnor et al., 1991). Despite the moderate correlation between cognition and affect, the current research findings suggest that attitudes towards asylum seekers may be best predicted by cognitive rather than affective evaluations. This is evidenced by the positive β coefficient for the cognitive factor and negative β coefficient for the affective factor, suggesting that individuals with negative affective evaluations but positive cognitive evaluations may endorse more positive attitudes towards asylum seekers. This implies a dominant reliance on cognitive rather than affective evaluation and supports the study's second hypothesis. In addition, this is testament that attitude objects are idiosyncratic and can comprise one or more antecedent factors in combination (Eagly & Chaiken, 1993; Esses et al., 1993).

A further explanation of this complex relationship may be that, whilst empirically distinguishable, individuals may be affectively evaluating the cognitive appraisals of the attitude object or cognitively appraising the affective evaluations (Eagly & Chaiken, 1993). This has been reported in previous studies where individuals internalise beliefs about asylum seekers and form beliefs that they represent symbolic or realistic ‘threats’. This internalisation is then affectively evaluated and may predict attitudes towards asylum seekers (Louis et al., 2007; McKay et al., 2012; Suhnan et al., 2012).

Affect produced a small proportion of variance in the current model, but it still added a significant unique proportion, which suggests that affective information is a significant predictor of overall attitudes towards asylum seekers. The negative (β) suggests this as negative attitude evaluations were associated with overall negative attitudes towards asylum seekers. This partially supports the first hypothesis and mirrors findings by Pedersen and Thomas (2013) and Hanson‐Easey and Augoustinos (2011), which highlighted the role of affective predictor factors in the prediction of attitudes towards asylum seekers.

Behavioural information

The contribution of behavioural information to the formation of attitudes towards asylum seekers has been examined in only a limited number of studies (e.g., Barlow et al., 2012), and mostly, these studies are about attitudes towards refugees (e.g., Turoy‐Smith et al., 2013). Furthermore, the difficulty in assessing behavioural information has limited the role that behavioural information has played in studies that utilise the tripartite model. It was therefore not completely clear if behavioural information would correlate significantly with overall attitude. Whilst there was a significant correlation between past behaviour and overall attitudes, the relationship explained a small and non‐significant proportion of the variance in attitudes towards asylum seekers.

This finding contrasts with findings by Haddock et al. (1994) who used open‐ended measures of cognition, affect, and behaviour to assess overall attitudes and found that the quality and magnitude of past experiences accurately predicted overall attitudes towards particular minority groups. Furthermore, Barlow et al. (2012) and Turoy‐Smith et al. (2013) found that negative contact resulted in an increase in negative attitudes towards asylum seekers and refugees, respectively, and positive contact increased positive attitudes. The current research findings need to be interpreted with caution as the smaller number of behavioural items generated by our sample, and some of the items themselves, suggests they had not had a great deal of contact with asylum seekers, which is likely to have affected results.

LIMITATIONS AND FUTURE RESEARCH

The non‐standard structure of the questionnaire, which required participants to generate their own beliefs, affects, and behaviours, meant that participants generated an unequal number of construct responses, which was conceptually difficult to manage in traditional quantitative analysis. However, the unique structure of the self‐report format meant that participants were able to tailor their cognitive, affective, and behavioural evaluations in relation to asylum seekers. Social desirability on the overall attitude rating item was a concern, but this was mitigated by the mode of data collection; research indicates that online data collection involving issues that are highly sensitive decreases social desirability responses (Kreuter, Presser, & Tourangeau, 2008). With approximately 45% of the sample having a university qualification, the sample was more highly educated than the general Australian population, where approximately one fourth has a bachelor degree (Australian Bureau of Statistics, 2012). Given the role that education may play in attitudes towards asylum seekers (Pedersen et al., 2005; Suhnan et al., 2012), the generalisability of the results to the wider Australian community may be limited due to self‐selecting bias (Bowden, 1986); importantly, however, there was no difference in attitude according to education level in our sample. The strengths of the current study were that it was theoretically based and used a measure that was internally consistent to investigate the predictors of community attitude towards asylum seekers. Additionally, the inclusion of a selection of participant‐generated items allows insight into how the attitude processes emerge.

The inclusion of political affiliation as a control variable would be prudent in future research given that individuals who align themselves with authoritarian or right‐wing political views have a higher incidence of negative attitudes towards asylum seekers compared with left‐wing or centralist politics (Pedersen, Griffiths, & Watt, 2008; Pedersen et al., 2005; Suhnan et al., 2012). A further future research option would be to incorporate the dual process model concepts of right‐wing authoritarianism and social dominance orientation (Duckitt & Sibley, 2010) as a means of informing how the cognitive processes occur and are internalised.

CONCLUSION

The present study has expanded the understanding of how attitudes towards asylum seekers are formed with the Australian wider community. The adaptation of the tripartite model has shown that individuals may base their attitudes towards asylum seekers more readily on the beliefs that they hold rather than on their affective or behavioural evaluations of asylum seeker groups. The current study also found that affective factors predicted overall attitudes towards asylum seekers. The study presented empirical support for the efficacy of the tripartite model of attitudes. These findings have applicability in that any attempts to restructure negative attitudes towards asylum seekers would need to focus on influencing the beliefs held by the wider Australian population in order to change community attitudes to people seeking asylum in Australia. The knowledge that the Australian population bases attitudes of asylum seekers on the beliefs that they have towards them may be used to shape how political policy is formulated and affect the way the media refers to this highly salient minority group. The understanding that Australian community attitudes are driven in large part by people's beliefs provides a way to target negative attitudes. This could be achieved by directly addressing the prejudiced beliefs held, and the distorted presentations presented, by the Australian government and media.