• Open Access

Public perceptions of the threat of terrorist attack in Australia and anticipated compliance behaviours


Correspondence to:
Garry Stevens, Senior Research Fellow, School of Medicine, University of Western Sydney, Building EV, Parramatta Campus, Locked Bag 1797, Penrith NSW DC1797. Fax: (02) 9685 9554; e-mail: g.stevens@uws.edu.au


Objective: To determine the perceived threat of terrorist attack in Australia and preparedness to comply with public safety directives.

Methods: A representative sample of 2,081 adults completed terrorism perception questions as part of the New South Wales Population Health Survey.

Results: Overall, 30.3% thought a terrorist attack in Australia was highly likely, 42.5% were concerned that self or family would be directly affected and 26.4% had changed the way they lived due to potential terrorist attacks. Respondents who spoke a language other than English at home were 2.47 times (Odds Ratios (OR=2.47, 95% CI:1.58-3.64, p<0.001) more likely to be concerned self or family would be affected and 2.88 times (OR=2.88, 95% CI:1.95-4.25, p<0.001) more likely to have changed the way they lived due to the possibility of terrorism. Those with high psychological distress perceived higher terrorism likelihood and greater concern that self or family would be directly affected (OR=1.84, 95% CI:1.05-3.22, p=0.034). Evacuation willingness was high overall but those with poor self-rated health were significantly less willing to leave their homes during a terrorism emergency.

Conclusion: Despite not having experienced recent terrorism within Australia, perceived likelihood of an attack was higher than in comparable western countries. Marginalisation of migrant groups associated with perceived terrorism threat may be evident in the current findings.

Implications: This baseline data will be useful to monitor changes in population perceptions over time and determine the impact of education and other preparedness initiatives.

If a terrorist attack should occur, it is essential that public health authorities are prepared to act. Understanding how the public perceives the threat of a terrorist attack, and is likely to respond, is integral to incident preparedness and planning.1–2

Studies following large-scale terrorist attacks show that the perceived risk of further attacks is linked with changes in behaviour, such as restricting travel, avoiding places of perceived high risk and increased substance use.3–5 There is evidence that some of these changes will persist in the medium term and be associated with negative health and economic outcomes.6–8 There is also evidence that some of these population impacts may be mitigated through public information campaigns and other preparedness initiatives.5

Information is also emerging from populations that have not experienced a major attack but are at risk of such events.9–11 In one Canadian study, only 20% of the population thought a terrorist attack was extremely or very likely to occur. Perceived threat to themselves as individuals was even lower. Most respondents worried little about terrorism; preparatory behaviours were consistent with these perceptions, with less than 5% establishing a family emergency plan for terrorism or avoiding public places of perceived risk. While specific preparedness was low, there was high willingness to follow public safety directives during such incidents.9

Australia, like Canada, is a country that has experienced little terrorism within its borders. However, in recent years Australians have been affected by large-scale terrorist attacks within the region, including bombings in Bali in 2002 and 2005, and the 2005 Jakarta bombing presumed to have targeted the Australian Embassy. Australians were also among those killed in the Mumbai attacks of 2008. In this environment of increasing international threat, there is a need for planning to protect public health and safety should terrorist attacks occur within Australia.12

Disaster planning requires information about community perceptions of risk, associated behaviours and information about vulnerable sub-populations.13 Researchers and planners stress the importance of developing specific measures of risk perception, and their behavioural correlates, as an integral part of disaster planning.14 Such instruments can support the establishment of baseline data against which the trajectory of psychosocial recovery can be measured. They can also be used to monitor other shifts in community perceptions over time, including the effectiveness of risk communication strategies and education programs. For these reasons, the relative absence of baseline data has been cited as a significant impediment to disaster planning.14–15

The aim of this paper is to determine perceptions of the threat of terrorist attack in Australia and preparedness to comply with evacuation directives in the event of such incidents. A further aim of the wider study has been to establish a surveillance tool and source of baseline data to allow ongoing monitoring of terrorism risk perception and behavioural correlates within the Australian population.


Question design

A literature search was conducted to identify existing tools for collecting information on perceptions of terrorist attack with the underlying themes of likelihood, effect on family (vulnerability and risk), life changes and compliance with government authorities.

The primary reference was a study by Canadian researchers on anticipated public response to terrorism.9–10 Questions on threat likelihood, effect on family and behavioural compliance were adapted, with permission, by subject matter experts and survey methodologists. Each proposed question was considered for clarity, ease of administration and possible biases. A set of five questions was developed for field-testing as well as the additional open question “Do you have any comments that you would like to make on any of the questions or any other issues?”

Field testing

The terrorist attack questions were field tested for test–retest reliability using the protocol of the New South Wales Health Survey Program.16 A detailed description of its application in this study is presented elsewhere.17 Data manipulation and analysis were conducted using SAS Version. 9.2.16 Kappa values for the indicators derived from the questions ranged between 0.27 and 0.64 in the second field test. There were low rates of ‘don't know’ responses (0.0–3.9%) and refusal (0.0–0.5%).

The survey

The New South Wales Adult Population Health Survey is a continuous telephone survey of the health of the state population using the in-house CATI facility of the New South Wales Department of Health.16

The terrorist attack question module was administered as part of the survey between 22 January and 31 March, 2007. The terrorist attack questions were submitted to the ethics committees of the NSW Population Health and Health Services and the University of Western Sydney, for approval prior to use. The survey also included other modules on health behaviours, health status (including psychological distress, using the Kessler K10 measure, and self-rated health status) and access to health services, as well as the demographics of respondents and households. As field test data had indicated high assumed knowledge regarding the concept of terrorism and presumptions this typically involved bombings or shootings (i.e. ‘conventional’ terrorism), a specific definition of terrorism was not outlined in the preamble. The target population for the survey was all state residents living in households with private telephones. Up to seven calls were made to establish initial contact with a household and five calls were made in order to contact a selected respondent.

Response categories were dichotomised into indicators of interest and responses of ‘don't know’ or ‘refused’ were excluded. For the hypothetical questions (i.e. likelihood of a terrorist attack in Australia, concern that self or family would be directly affected, willingness to comply with evacuation of home, willingness to comply with evacuation of workplace or public facility) the responses of extremely likely and very likely were combined into the indicator of interest. For the non-hypothetical question “changed the way you live because of the possibility of a terrorist attack”, the responses ‘a little’, ‘moderately’, ‘very much’ and ‘extremely’ were combined into the indicator of interest: that is, changed way of living.

The survey data were weighted to adjust for probability of selection and for differing non-response rates among males and females and different age groups.17 Data were manipulated and analysed using SAS version 9.2.16 The SURVEYFREQ procedure in SAS was used to calculate point estimates and 95% confidence intervals.

Odds ratios were calculated as described by Bland.18 All calculations were performed using the ‘SVY’ commands of Stata version 9.2 (Stata Corp, College Station, TX, USA), which allowed for adjustments for sampling weights.


In total 2,081 state residents aged 16 years and over completed the module on terrorist attack. The overall response rate was 65%. The demographics of the weighted survey population were comparable with the Australian Bureau of Statistics 2006 Census.19 These comparisons are reported elsewhere.20

Overall, 30.3% of the population thought a terrorist attack was extremely or very likely, 42.5% were extremely or very concerned that they or their family would be affected by a terrorist attack and 26.4% had made some (small to extreme) level of change to the way they lived their life because of the possibility of an attack. Table 1 shows the prevalence estimates for all of the survey questions by response category.

Table 1.  Prevalence estimates for each question by response category including don't know and refused.
QuestionResponse%95% LCI95% UCI
  1. Source: New South Wales Health Survey Program. Sydney: New South Wales Department of Health, 2008.

How likely do you think it is that a terrorist attack will occur in Australia?Not at all8.87.010.6
 A little23.120.425.8
 Don't know3.42.24.6
If a terrorist attack happened in Australia, how concerned would you be that you or your family would be directly affected by it?Not at all10.68.812.5
 A little22.419.725.0
 Don't know2.11.33.0
How much have you changed the way you live your life because of the possibility of a terrorist attack?Not at all71.468.674.2
 A little14.712.416.9
 Don't know1.60.92.3
In case of an emergency situation such as a terrorist attack, how willing would you be to evacuate your home?Not at all6.34.87.9
 A little7.05.28.8
 Don't know2.41.63.2
How willing would you be to evacuate your workplace or a public facility?Not at all1.81.12.5
 A little3.82.65.0
 Don't know1.50.92.0

Table 2 presents prevalence estimates for the likelihood, concern and changed way of living variables by demographic and socio-economic characteristics, and the indicators of level of psychological distress and general self-rated health status. When these variables were combined, the greatest proportion of the population (37.0%) thought a terrorist attack was unlikely to occur, were not concerned that they or family members would be directly affected and had not changed the way they lived their life because of the possibility of a terrorist attack. Less than 1 in 10 people (9.0%) thought a terrorist attack was likely, were concerned that they or family members would be directly affected and had made changes to the way they lived their life due to the prospect of an attack.

Table 2.  Prevalence and Odds Ratios (95% confidence intervals) of terrorist attack likely, concern for self/family, changed way of living and combined indictors.
Independent variableTerrorist attack likelyConcerned self or family directly affectedChanged way of living due to possibility of terrorismTerrorist attack likely and concerned for self/familyTerrorist attack likely and concerned and changed way of living
 Weighted (%)OR95% CIP valueWeighted (%)OR95% CIP valueWeighted (%)OR95% CIP valueWeightedOR95% CIP valueWeightedOR95% CIP value
  1. Notes: Psychological distress was measured using the K10. Values range from 10-50, with ‘high’ psychological distress considered as being geqslant R: gt-or-equal, slanted22. Source: New South Wales Health Survey Program. Sydney: New South Wales Department of Health, 2008.

NSW pop'n Gender31.6---43.7---27.0---18.9---9.0---
Male30.81.00  38.51.00  24.71.00  16.81.00  7.91.00  
Urban31.21.00  45.71.00  29.21.00  19.11.00  9.71.00  
High psychological distress (geqslant R: gt-or-equal, slanted 22)
No29.91.00  45.21.00  25.11.00  17.31.00  7.71.00  
16-2417.41.00  39.51.00  27.91.00  14.11.00  7.61.00  
65-7433.62.401.37, 4.18<0.0151.71.641.02,2.660.0423.80.800.46,1.390.4224.01.991.06,3.720.0310.41.420.60,3.390.43
Children in household
No29.71.00  45.71.00  26.31.00  19.11.00  8.71.00  
Born in Australia
No32.41.00  50.41.00  36.01.00  20.11.00  10.51.00  
Yes31.30.940.68,1.290.7041.30.690.52,0.930.0223.80.560.40,0.77<0.0118.40.900.61, 1.330.618.40.830.48,1.450.52
Speak language other than English at home
No32.11.00  40.11.00  23.21.00  17.81.00  8.01.00  
Living alone
No31.21.00  43.41.00  27.91.00  18.51.00  9.21.00  
Highest formal qualification
University degree/equivalent27.41.00  35.31.00  24.31.00  11.51.00  5.61.00  
TAFE certificate/Diploma34.91.430.97,2.100.0741.01.280.89,1.840.1926.61.130.74,1.710.5817.41.610.96,2.690.017.91.440.66,3.140.36
High school certificate25.80.920.60,1.430.7244.01.440.95,2.180.0928.91.270.80,2.010.3117.61.630.95,2.800.,2.850.55
School certificate35.21.440.99, 2.080.0552.11.991.41,2.82<0.0128.41.240.84,1.830.2926.72.681.68,4.27<0.0113.42.531.32,4.870.01
No32.51.00  49.31.00  26.71.00  22.41.00  10.81.00  
Household income (before tax)
<$20k34.51.00  47.61.00  27.41.00  21.91.00  9.21.00  
Health self-rated as good
Yes33.41.00  43.41.00  26.21.00  18.31.00  8.31.00  
Marital status
Married34.71.00  42.81.00  27.91.00  18.91.00  8.61.00  
Never married21.60.520.36,0.74<0.0143.51.030.75,1.420.8523.70.800.55,1.170.2515.90.810.53,1.230.327.80.910.50,1.640.75

The results of univariate analyses identified a number of statistically significant factors associated with threat likelihood, concern and changed way of living (see Table 2). For the latter variable, a sensitivity analysis comparing the current indicator (‘a little’, ‘moderately’, ‘very’ and ‘extremely’) and the more conservative indicator (‘moderately’, ‘very’ and ‘extremely’) showed that the association did not change for the co-variates and therefore the current indicator was retained for the analysis.

The univariate analysis indicated that young people (16-24 years) were significantly less likely to report high terrorism likelihood compared to all other age categories. Those with no formal qualifications were 2.09 times (Odds Ratios (OR=2.09, 95% CI:1.32-3.31, p=0.002) more likely to think that a terrorist attack was very or extremely likely compared to those with a university degree or equivalent and women were significantly more likely (OR =1.54, 95% CI:1.20-1.99, p=0.001) to be very or extremely concerned for themselves or family members in the event of an attack.

Those who spoke a language other than English at home were 2.47 times (OR=2.47, 95% CI=1.58-3.64, p<0.001) more likely to be concerned for self or family and 2.88 times (OR=2.88, 95% CI:1.95-4.26, p<0.001) more likely to have made changes in living due to the risk of terrorism attack. When these indicators were combined, those with high psychological distress were found to perceive higher terrorism likelihood and to have greater concern that they or family members would be directly affected (OR=1.84, 95% CI:05-3.22, p=0.034).

Table 3 shows the prevalence estimates for willingness to evacuate home, workplace/public facility, and both home and workplace/public facility, by demographic characteristics and the indicators psychological distress and health status. Table 3 also shows combined indicators of evacuation willingness for those concerned about self and family and who also thought a terrorist attack was likely. Overall, if an emergency such as a terrorist attack were to occur, the majority of the population would be willing to evacuate their home (67.4%), their workplace or a public facility (85.2%), or both location types (65.8%). Conversely, 12.5% would be willing to evacuate neither of these locations. Fewer than 20% would be willing to evacuate their workplace or a public facility but not their home, whereas far fewer (1.9%) would be willing to evacuate their home but not their workplace or public facility.

Table 3.  Prevalence and Odds Ratios (95% confidence intervals) of willingness to evacuate home, willingness to evacuate office/public facility and combined indicator.
Independent variableWilling to evacuate homeWilling to evacuate office/public facilityWilling to evaluate home and office/public facility
 Weighted (%)OR95% CIPvalueWeighted (%)OR95% CIPvalueWeighted (%)OR95% CIPvalue
  1. Notes: Psychological distress was measured using the K10. Values range from 10-50, with ‘high’ psychological distress considered as being geqslant R: gt-or-equal, slanted22. Source: New South Wales Health Survey Program. Sydney: New South Wales Department of Health, 2008.

NSW population67.4---85.2---65.8---
Male63.41.00  82.21.00  62.21.00  
Urban66.31.00  84.41.00  64.61.00  
High psychological distress (geqslant R: gt-or-equal, slanted22)
No68.51.00  86.71.00  67.41.00  
16-2466.11.00  80.01.00  63.01.00  
Children in household
No64.91.00  82.61.00  63.31.00  
Born in Australia
No70.01.00  82.31.00  67.81.00  
Speak language other than English
No67.01.00  86.41.00  65.51.00  
Living alone
No67.71.00  85.51.00  66.01.00  
Highest formal qualification
University degree/equivalent70.71.00  88.91.00  70.01.00  
TAFE certificate/Diploma67.70.870.59,1.290.4987.00.830.45,1.530.5666.70.850.58,1.260.42
High school certificate63.50.720.46,1.120.1583.20.610.32,1.170.1463.00.680.44,1.050.08
School certificate66.00.810.55,1.180.2680.90.530.30,0.920.0261.80.690.47,0.990.05
Work (paid or unpaid)
No62.71.00  80.71.00  60.31.00  
Household income (before tax)
<$20k61.01.00  77.21.00  60.01.00  
Health self-rated as good
Yes69.61.00  85.61.00  68.11.00  
Marital status
Married68.31.00  86.21.00  66.51.00  
Never married65.80.890.64,1.250.5182.50.760.49,1.180.2262.50.840.60,1.170.30

Univariate analyses of evacuation intentions, also presented in Table 3, indicated that females were 1.45 times (OR=1.45, 95% CI:1.11-1.89, p=0.007) more willing to evacuate their homes than males, employed Australians were 1.37 times (OR=1.37, 95% CI:1.06-1.79, p=0.018) more willing to evacuate their homes than those unemployed and respondents with poor self-rated health were significantly less willing (OR=0.63, 95% CI: 0.42-0.96, p=0.032) to evacuate their homes than those with good or excellent self-rated health.

With regard to evacuation of work/public facilities, females were 1.62 times (OR1.62, 95% CI:1.14-2.32, =0.008) more willing than males, people with children were 1.84 times (OR=1.84, 95% CI:1.10-3.05, p=0.019) more willing than those without children and those with higher household incomes ($A80,000 and over) were 2.57 times (OR=2.57. 95% CI:1.48-4.44, p=0.001) more willing than those with lower incomes.


Almost one-third of the population perceive a high likelihood of terrorist attack within Australia, with a greater proportion expressing high levels of concern that they, or a family member, could be directly affected should a terrorist attack occur. More than one-quarter had made some level of accommodation in the way they live due to this possibility. In the context of an immediate threat, the majority of the population would be willing to follow public safety directives to evacuate homes, workplaces and public facilities should a terrorist attack occur.

To date, there have not been recent substantial acts of terrorism within Australia. Our findings indicate a lower level of terrorism risk perception than that observed in countries such as the US and UK where significant terrorist events have occurred.3,5 However, the level of perceived risk within the Australian population is notably higher than in comparable western countries that have not experienced recent attacks. Although comparisons with other studies are difficult to make, only 20% of Canadians perceived a high likelihood of domestic terrorism compared with 30.3% of Australians. Possible reasons for this may include Australia's regional exposure to terrorism, such as the Bali bombings in 2002 and 2005. These and associated events such as the naming of Australia as a specific target by terrorist organisations such as Al-Qaeda are likely to have increased the awareness of Australians to such threats domestically.12

Concurrent high concern and changed way of living were noted among those born outside Australia and/or who spoke a language other than English in the home. Given that these groups did not perceive a higher likelihood that an attack would occur, these results may reflect a perceived ‘secondary’ threat from within the wider population. There is broad evidence that heightened community threat perception is associated with increased ethnocentrism and xenophobia.21 In terrorism affected countries, culture, appearance and religion have been found to be strong predictors of high terrorism-related distress and appear to reflect increased stigmatising of these groups.5 Further studies are needed to explore the possible reasons for this in the Australian context. While the Australian government has produced recent population-level information campaigns to address terrorism concerns, the current results highlight potential vulnerabilities in these sub-populations and a possible need for tailored risk communication to address unease in these groups.

There is evidence that mental health factors such as stressful life events and exposure to trauma are associated with increased fear of terrorism. In this study, those with high levels of psychological distress where almost twice as likely to perceive high terrorism likelihood and greater concern that they or a family member would directly affected. This is consistent with recent data from the Australian Unity Wellbeing Index, which showed that those with the highest ratings of perceived terrorism likelihood also had significantly lower levels of personal wellbeing compared to the population average.22 There is evidence from the therapeutic field that those with greater risk appraisal and vigilance tendencies are at greater risk of negative mood states.1,23 The findings occurred in the absence of specific domestic terrorist incidents and suggest increased vulnerability may exist even with the general threat of terrorism. Recent practices in Australia such as the issuing of national terror alerts may have adverse impacts on this sub-group that may warrant further examination.22

The findings regarding willingness to evacuate were quite emphatic and indicate high levels of intent where the specific threat also appears to be high. The proportion of people with low willingness to evacuate based on terrorism threat (12%) was found to be about half the ‘non-compliance’ rate observed during mandatory natural disaster evacuations.24 Little is known about the observed lower willingness of those with poor self-rated health. This may reflect lower confidence about receiving or recognising warnings as well as perceptions about the physical attributes needed for escape. This finding is consistent with protection motivation theory which posits that low perceived self-efficacy reduces the motivation for protective acts even when these actions are regarded as effective. Proactive disaster planning and education can potentially overcome these limitations, which highlights the importance of identifying and engaging these vulnerable groups early in the process.


There are several limitations of our study. The question “Have you changed the way you live your life because of the possibility of a terrorist attack?” was intentionally broad, since current evidence indicates that where specific incidents have not occurred, preparatory changes for terrorism are limited and general in focus.9,10 We sought to determine broader markers of change (experiential as well as behavioural) that may be sensitive to public health messages or varied threat status over time. The decision to adopt the full response set (a little, moderately, very and extremely) as positive indicators for this question was made on related conceptual grounds: that is, lower level change reflecting change of some kind. The more conservative range (very, extremely), produced a response rate of 4%, which is consistent with the rate of preparatory response noted in the Lemyre et al. study.9 As anticipated, the broader range produced a notably higher response rate (27%). This may indicate that more specific behaviours are being endorsed at the upper end of the range, with more subtle or even ‘felt’ changes being endorsed by a larger group at the lower end of the range.

The aim of this paper is to explore population threat perceptions of terrorist attack in Australia and some anticipated responses in the acute context. The cross-sectional design of this study captures only a snapshot view of these frequencies and no firm conclusions can be made regarding causes. Also, OR's reported in this study may lead to bias due to a failure to account for multiple relationships, which may lead to inflation of type 1 error and the over interpretation of any apparent positive findings. As this is a baseline analysis, further studies can examine trends over time and the consistency of these findings.

It is also important to consider whether recent terrorism information campaigns significantly affected this baseline data, which was gathered in early 2007. The ‘Be alert, not alarmed’ campaign was conducted in late 2002 and again, in a modified form, from July 2005.25 The National Security Hotline was also launched as part of the initial campaign and has been advertised periodically. Given that the 2005 campaign ran for a three week period 18 months prior to the survey, it is unlikely that any specific shifts in threat perception would have been maintained so as to have significantly affected the current data. Nonetheless, it remains possible that the cumulative effects of these public awareness campaigns have contributed to longitudinal change e.g. as one significant factor in the rate differences observed between Australian and Canadian population surveys.

Finally, the questions regarding evacuation only provide a measure of behavioural intent. While its specific translation to evacuation compliance is unclear, evidence from experimental psychology shows that concurrent high intent and high perceived positive outcomes predict high levels of behavioural translation.26 The data establish high intent, while perceived positive outcomes in this situation (presumably safety and survival) could reasonably be assumed based upon these results.


Perceived terrorism likelihood and associated concerns were moderately high, with some groups notably affected. Community terrorism concerns may have increased ‘out group’ social dynamics for some ethnic sub-populations and this has implications for the framing of risk communications. In the context of immediate threat, the majority of Australians would follow terrorism-related evacuation directives and at higher rates than is typical of natural disasters. The study has also established a source of pre-event baseline data and is one of the few available sources of such information internationally.


This study was funded by Emergency Management Australia and supported by the New South Wales Department of Health. This analysis is part of the first author's thesis for a doctoral dissertation with the College of Health and Science at the University of Western Sydney. Our thanks to Matthew Gorringe, Centre for Epidemiology and Research, New South Wales Department of Health, who assisted with question development and data collection.