Compliance with COVID-19 prevention measures during the onset of the pandemic in Australia: investigating the role of trust in federal and state governments and scientists

ABSTRACT

Objective

The current study explored (1) changes in trust in federal and state governments and scientists across representative Australian national samples from 2003–2020; and (2) the extent to which trust in these sources predicted compliance with COVID-19 prevention measures at the onset of the pandemic.

Method

Using a nationally representative samples (N = 1000), we asked participants to rate their trust in federal and state government and in scientists, their extent of compliance with COVID-19 prevention measures, and to provide demographic information.

Results

We found that trust in federal and state governments had significantly increased, while trust in scientists was at a high level matched by only three other time-points. Higher levels of trust in state government and scientists uniquely predicted greater compliance with COVID-19 prevention measures. Women and older respondents also reported greater compliance.

Conclusion

The current findings reinforce those from Australia and other countries indicating that trust increased during the onset of the COVID-19 pandemic, and those identifying trust in government and in scientists as important predictors of compliance. Importantly, our findings highlight the role of trust in state government, which potentially reflects the role played by Australian state governments in enacting and enforcing COVID-19 prevention measures.

Key points

What is already known about this topic:

1. Older individuals, women and those who voted for the government in power tended to display greater compliance with COVID-19 measures.

2. Trust in government appears to have generally increased in accordance with governmental actions to address COVID-19.

3. Older individuals, women and those who voted for the government in power tended to display greater compliance with COVID-19 measures.

What this topic adds:

1. The current study provides the most comprehensive longitudinal overview of trust in federal and state government and in scientists from 2003 to 2020. Trust in both forms of government increased with the onset of COVID-19, whereas trust in scientists remained steady.

2. In Australia, greater compliance was significantly predicted by increased trust in scientists and state government, but not federal government. Older respondents and women also reported increased compliance.

3. The current findings indicate that scientists and public health experts should play an important role in shaping and delivering policy where public compliance is required.

KEYWORDS:

• Australia
• compliance
• COVID-19
• government
• scientists
• trust

The global COVID-19 pandemic associated with the spread of the 2019 novel coronavirus (SARS-CoV-2) necessitated compliance with a range of directives in order to “flatten the curve”. In Australia, these included social distancing, handwashing, wearing masks and restriction of movement in the community (i.e., lockdowns). However, there was a lack of uniform compliance with public health directives and restrictions by the Australian public (Murphy et al., 2020). Issues of compliance with behaviour intended to limit the spread of COVID continue to be important in the context of new and emerging omicron subvariants, which are driving new waves of COVID-19 infections in Australia (McMillan & Aubusson, 2022), even if these are now framed as recommendations rather than directives.

While past research has explored the role and psychological characteristics of preparedness for disasters and national crises (e.g., Notebaert et al., 2016; Paton, 2019), the current research focuses on compliance with health initiatives at the onset of a public health crisis. In this study, we examine two distinct but related research questions: First, we document changes in trust in federal and state governments and in scientists preceding the COVID-19 pandemic, and during the onset of the pandemic – the time at which COVID-19 prevention measures were being implemented. Second, we investigate the extent to which trust in each of these entities predicts self-reported compliance with COVID-19 prevention measures at the onset of the pandemic.

Several studies have documented high trust in governments at the onset of the pandemic. Findings from Germany (Leininger & Schaub, 2020) and Denmark and Sweden (e.g., Bækgaard et al., 2020; Esaiasson et al., 2021; Nielsen & Lindvall, 2021) indicated that heightened trust in government may have been reflective of a “rally around the flag” effect (i.e., an increase in support for political leaders in the face of a crisis; Mueller, 1973). This effect has been found in a range of countries (Hegewald & Schraff, 2022). For example, in a large sample of unemployed Danes, Bækgaard et al. (2020) found that trust in the Prime Minister’s administration increased in conjunction with the announcement of a COVID-19-related lockdown. The announcement of lockdowns was also associated with enhanced political trust across seven Western European countries (Bol et al., 2021) and in Canada (Harell, 2020). On the other hand, similar announcements were associated with reduced government support in the state of Florida (Shino & Binder, 2020), and the rallying effect was short-lived in the USA more broadly (Enten, 2020), likely a result of the highly politicised and partisan nature of support for COVID-19 prevention-related health behaviours and restrictions (Motta et al., 2020). More generally, it has been suggested that the rallying effect is short-lived (Johansson et al., 2021), and depends on partisanship (Shino & Binder, 2020) and on the response of political leaders to the pandemic. However, there has been little evidence of the trajectory of trust in governments over the longer term, which would provide important context for understanding this rallying effect. In Australia and New Zealand, Goldfinch et al. (2021) found that trust in governments was higher in 2020 than in a 2009 study conducted by the same authors (see Goldfinch et al., 2009). However, it is possible that trust was already increasing between 2009 and 2020. The aim of the current study was to provide clearer evidence of whether there was a rallying effect in Australia at the onset of the pandemic, as observed with increased trust over time.

A clear understanding of the trajectory of trust is important because several nationally or regionally focused studies have identified that trust in government was a major predictor of compliance with COVID-19 prevention measures. For example, Ayalon (2021) found that Israelis with lower trust in government tended to report reduced compliance with COVID-19 restrictions. Throughout Europe, a reduction in non-essential outings and travel tended to be more pronounced in areas marked by greater trust in government (Bargain & Aminjonov, 2021). In the US, willingness to be vaccinated against COVID-19 was predicted by trust in the government (Viskupič et al., 2022). In Australia, Murphy et al. (2020) reported that trust, based on perceived competence of authorities (i.e., federal government, state government and related health authorities) and the perceived integrity of the federal government, was correlated with greater compliance with COVID-19 lockdown restrictions. In Australian and New Zealander samples, Goldfinch et al. (2021) reported that trust in government – a variable combining trust in both state and federal governments – and trust in health experts were the largest predictors of the perceived efficacy of a mobile phone application intended to warn users if they may have been exposed to individuals with COVID-19.

In addition to trust in government, it also appears that trust in science or scientists is an important predictor. In the United States, greater trust in scientists was associated with greater compliance with social distancing (Fazio et al., 2021). In a similar vein, Freeman et al. (2020) found that people who hold conspiratorial beliefs (in general or specifically about the COVID-19 pandemic) reported reduced trust in the United Kingdom government and in scientists, and reduced compliance with COVID-19 measures. Plohl and Musil (2021) also identified that trust in science is a particularly important predictor of compliance with COVID-19 prevention guidelines. In Australian research, the role of medical experts (i.e., scientists) has been shown to positively predict health promotion intentions (Hardie, 2011). Collectively, the literature to date indicates that trust in experts (e.g., health experts or scientists) plays an important role in compliance, whereas trust in federal and state governments may have differing effects on compliance depending on who is in power.

Demographic factors have also been associated with greater compliance with COVID-19 restrictions. In an Australian sample, older respondents and women reported greater compliance with lockdown measures (Murphy et al., 2020). In a US sample, Fazio et al. (2021) similarly found that, on average, older respondents reported higher levels of compliance. In addition, women indicated greater awareness of social distancing advice than men. Women and older respondents also tended to report greater trust in scientists than men and younger respondents, with men tending to report greater trust in the government than women. In Australian and New Zealander samples, Goldfinch et al. (2021) found that higher levels of education were associated with greater trust in government, but education was only associated with increased confidence in public health scientists in the Australian sample. Income was associated with a greater likelihood of using a COVID-19-tracking phone application in both samples, with higher levels of education also a significant predictor in New Zealand. Political affiliation was also included as a covariate by Goldfinch et al., such that having voted for the government in power was associated with greater trust in government and public health scientists. Similarly, political affiliation in the USA (i.e., Republican or Democrat) was found to be an important component of trust and compliance by Goldstein and Widemann (2021). In a recent latent class analysis, five distinct behavioural profiles, each associated with a number of demographic and psychological characteristics including trust, were characterised by different degrees of compliance with different health behaviours (Oldmeadow et al., 2023); these findings underscore the importance of a contextualised approach to trust and its effect on compliance.

The current study

The current study has two distinct but related aims: first, to track trust in federal and state governments and scientists over time; second, to determine the extent to which trust in these sources predicted compliance with COVID-19 prevention measures at the onset of the pandemic. In the present research, we expected to replicate and extend findings reported in Australian studies by Goldfinch et al. (2021) and Murphy et al. (2020). The current study makes several novel contributions to this line of research. First, the current study examines broader compliance with COVID-19 prevention measures such as social distancing, extending the narrower focus of Goldfinch et al. on the perceived efficacy of using a COVID-19 tracking phone application. Second, whereas Goldfinch et al. explored changes in trust across two time points in 2009 and 2020, the current study utilises a unique longitudinal dataset – the Swinburne National Technology and Society Monitor – to provide a more robust analysis by exploring changes in trust across multiple timepoints from 2003 to 2020. Third, rather than collapsing trust in state and federal governments into a single trust measure, the current study explores trust in state and federal governments separately, thus providing a more granular and health policy-relevant assessment of the unique influence of trust in federal and state government on compliance.

It is important to consider the unique roles of trust in federal and state governments due to the manner in which the pandemic was handled. For example, there had been widespread criticism of how the Australian Prime Minister, Scott Morrison, and the federal government more generally handled the unprecedented bushfires that affected much of the country in late 2019 and early 2020 (Wheeler et al., 2022), as well as other distrust-promoting missteps such as attending a sporting event shortly after announcing a ban on large outdoor gatherings (see O’Sullivan et al., 2020). Considering that perceptions of leaders can affect trust in political parties (Dassonneville & McAllister, 2021), distrust in the Prime Minister and the federal government conceivably contributed to an increased reliance on the perceived need for state governments to effectively manage the pandemic response. Furthermore, as of March 2020, disagreements had emerged between the federal government and state governments on how to quarantine individuals entering Australia from overseas (O’Sullivan et al., 2020). Quarantine and border restrictions were largely managed by state governments, thus the need to trust one’s state government and its directives may have been particularly salient. In support of likely differences between residents of different states, a recent study by Nickel et al. (2022) found perceptions of confidence and trust differed from state to state in a nationally representative sample of younger Australians in their COVID-19 response study.

In accordance with previous findings outside the United States, where compliance is more uniformly encouraged by governments, it is hypothesised that greater trust in federal government and state government, as well as in scientists, will contribute to a greater level of compliance with COVID-19 prevention measures. Additionally, the extent to which trust in each of federal government and state government contributes to enhanced compliance will be explored. Demographic variables of age, gender, education and political orientation will be controlled for, and we expect that older respondents and women will report greater compliance, based on findings from other recent Australian studies (Goldfinch et al., 2021; Murphy et al., 2020). The effect of education and political orientation will also be examined.

Method

Participants and procedure

Data were obtained through two sources. Historical data on trust in federal government, state government and scientists, covering most years from 2003 to 2017, was collected through the Swinburne National Technology and Society Monitor. Respondents were members of the Australian general public aged 18 years or older who responded to a survey. From 2003 to 2015, all respondents were recruited by telephone calls to landlines. In 2017, half of respondents completed the survey online, with one-quarter of respondents called via landline and the remaining quarter called on their personal mobile phone. The most recent sample was obtained in June 2020 by having respondents complete an online survey via Qualtrics, with the intent of obtaining a sample that was as nationally representative as possible. Each sample is summarised in Table 1 for the purpose of displaying the demographic similarity of the 2020 sample to previous samples, thus ensuring suitability for comparisons between years.

Table 1. Demographics for each sample year.

	2003	2004	2005	2006	2007	2008	2009	2010	2011	2012	2013	2015	2017	2020
N	1044	1013	1013	1000	1000	1000	1001	1001	1000	1000	1000	1000	1004	1000
Age range	18–24: 2.0% 25–34: 9.3% 35–44: 10.0% 45–54: 17.0% 55–64: 18.4% 65–74: 19.6% 75+: 23.0%	1910s: 1.1% 1920s: 5.2% 1930s: 9.0% 1940s: 19.2% 1950s: 20.9% 1960s: 22.2% 1970s: 14.2% 1980s: 8.1%	1910s: 6.0% 1920s: 6.0% 1930s: 11.2% 1940s: 17.8% 1950s: 21.0% 1960s: 19.5% 1970s: 13.7% 1980s: 9.9%	18–92	18–91	18–93	18–96	18–94	18–94	18–97	18–92	18–92	18–96	18–88
Age M (SD)	-	-	-	52.05 (16.96)	52.32 (16.48)	54.64 (16.65)	54.64 (16.79)	55.40 (16.40)	55.50 (15.77)	56.17 (15.98)	52.60 (17.59)	50.67 (17.36)	47.65 (18.93)	48.79 (18.97)
Gender (% female)	61.6%	50.0%	57.7%	60.6%	61.3%	62.6%	59.5%	62.0%	62.9%	62.4%	53.6%	51.7%	51.4%	50.4%
Education (%)
Did not complete high school	27.9%	26.9%	30.1%	27.2%	28.9%	26.3%	25.2%	24.5%	23.9%	24.0%	18.0%	15.0%	15.1%	14.2%
Completed high school	24.4%	21.7%	23.8%	22.5%	19.9%	19.9%	19.5%	20.5%	17.5%	16.5%	18.3%	18.2%	19.1%	17.1%
Vocational diploma or certificate	14.7%	14.4%	13.4%	17.6%	14.7%	18.3%	17.3%	18.3%	19.4%	20.9%	19.3%	19.7%	23.8%	28.3%
Undergraduate degree	25.1%	28.3%	22.5%	22.4%	24.7%	23.7%	27.0%	25.5%	28.0%	26.8%	29.5%	34.3%	31.3%	30.1%
Postgraduate degree	7.2%	8.1%	9.7%	9.5%	11.6%	11.3%	10.7%	11.1%	11.1%	11.3%	14.5%	12.2%	10.7%	10.3%
Political orientation M (SD)	-	-	-	-	-	-	-	5.26 (1.86)	5.44 (2.14)	-	-	-	-	5.75 (2.10)
State/Territory (n)
ACT	27	20	20	22	20	20	20	20	21	21	25	28	16	20
NSW	311	338	338	326	328	328	328	323	321	323	322	314	322	314
NT	8	20	20	20	20	20	20	20	20	20	21	21	19	21
QLD	203	188	188	190	190	190	191	192	192	192	189	189	204	199
SA	99	77	77	78	78	78	78	81	80	80	81	90	68	76
TAS	33	24	24	24	24	24	24	24	25	24	26	26	21	19
VIC	272	248	248	240	240	240	240	241	240	240	244	227	254	251
WA	91	98	98	100	100	100	100	100	101	100	92	105	100	100

ACT = Australian Capital Territory, NSW = New South Wales, NT = Northern Territory, QLD = Queensland, SA = South Australia, VIC = Victoria, WA = Western Australia.

According to the Australian Bureau of Statistics (ABS, 2021) as of June 2020, the average age of Australians was 39.32 (median age = 37.77), with 50.41% of the population being female. In terms of education, 35% of Australians aged between 20 and 64 had completed a bachelor's degree or higher. Therefore, the 2020 sample was approximately representative of the broader Australian community for gender but included a slightly higher proportion of university-educated individuals. The mean age of the current sample was higher on account of not including anyone under the age of 18. However, given that 75.7% of Australians are aged 20 or older (ABS, 2021), the average age in the current sample is likely to reflect that of the broader adult population.

Measures

Demographics

Data on age were collected each year. In 2003 only, respondents were asked to identify their relevant age group (i.e., 18–24, 25–34 and so on until “75+”). In 2004 and 2005, respondents were asked to select the decade in which they were born. For all other survey years, age was recorded directly.

In each yearly survey, gender (0 = Male, 1 = Female) and education (1 = Did not complete high school, 2 = Completed high school, 3 = Vocational diploma or certificate, 4 = University degree or diploma, 5 = Postgraduate degree) were obtained. The state or territory that respondents lived in at the time of completion was also recorded each year.

In 2010, 2011 and 2020, political orientation was based on a single item (“In political matters people talk of ‘the left’ and ‘the right’. How would you place your views on this scale, generally speaking?”) that has previously been used in the World Values Survey (see Rockey, 2014). Responses were on a scale from 1 (left-wing) to 10 (right-wing). Respondents were able to select that they were unsure or could refuse to answer. These responses were not included in the analyses.

Trust

Each year, participants were separately asked how much they trusted, “the federal government”, “the state government” and “scientists” on a scale from 0 (Don’t trust at all) to 5 (Trust a very great deal). Participants were also able to select that they were unsure or could refuse to answer. These two responses were excluded from the analyses.

Compliance with COVID-19 prevention measures

A single item (“Overall, to what extent are you following the government’s advice and mandates for reducing the spread of COVID-19 [i.e., social distancing, staying home as much as possible]”) was measured on an 11-point scale (0 = Not at all, 10 = All the time). Respondents were also able to select that they were unsure, and these responses were also omitted from the analyses.

Results

Changes in trust over time

Mean scores for trust in each yearly sample are shown in Figure 1. A one-way analysis of variance (ANOVA) indicated that there were significant differences in trust across all time periods for each of trust in federal government, F(13, 13810) = 51.41, p < .001, state government, F(13, 13841) = 59.78, p < .001 and in scientists, F(13, 13478) = 30.33, p < .001. SNK post-hoc tests revealed that trust in both federal government and state government was at their highest in 2020 (p < .05). Trust in scientists in 2020 was at a similar level as in 2007, 2015 and 2017, with reported trust in those 4 years being significantly (p < .05) higher than in other years.

Figure 1. Average levels of trust in federal and state governments and in scientists in each yearly sample.

Additional analyses to examine method effects and potential differences between states/territories

Response method

To clarify whether differences in levels of trust in 2020 compared to previous years were genuine versus due to people responding to an online survey differently than when asked via a landline telephone call (i.e., years 2003–2015), we compared responses in 2017 that were obtained via landline, mobile phones or online survey. Mean scores for each trust variable for the different response methods are shown in Table 2. A one-way ANOVA found that there were significant differences across the three response methods for trust in federal government, F(2, 991) = 7.06, p = .001, trust in state government, F(2, 994) = 6.69, p = .001 and trust in scientists, F(2, 991) = 4.83, p = .008. SNK post-hoc tests indicated that respondents contacted by mobile phone significantly (p < .05) reported the highest mean scores on each trust measure. No significant differences were found between respondents who completed the survey via landline or online survey, indicating that any observed difference between years was measuring differences in trust and not an artefact of the response method.

Table 2. Mean levels of trust in federal government, state government and in scientists across the different response methods in the 2017 sample.

	Online survey			Landline			Mobile phone
Trust	M	SD	n	M	SD	n	M	SD	n
Federal government	1.98^a	1.44	497	2.04^a	1.36	280	2.40^b	1.22	217
State government	2.13^a	1.42	498	2.09^a	1.32	282	2.49^b	1.21	217
Scientists	3.70^a	1.05	498	3.73^a	1.04	278	3.95^b	.91	218

Superscript letters indicate significant differences between each response method for each form of trust.

Differences between Australian states and territories

We also examined potential differences between each Australian state and territory. This was due to the possibility that large increases in trust of federal and/or state governments in a small number of states/territories may have been responsible for increasing mean levels of trust in the whole sample in 2020. Mean scores for each state and territory on the three trust variables and compliance are shown in Table 3.

Table 3. Mean scores for trust in federal government, state government and scientists, and mean compliance with COVID - 19 prevention measures in each Australian state and territory.

	Trust in federal government M (SD)	Trust in state government M (SD)	Trust in scientists M (SD)	Compliance M (SD)
Australian Capital Territory*	3.47(1.68)	3.26(1.56)	3.75(1.25)	8.11(2.05)
New South Wales	3.24(1.28)	3.25(1.23)	3.73(1.07)	8.73(1.58)
Northern Territory*	2.33(1.53)	2.35(1.66)	3.50(1.19)	7.75(2.29)
Queensland*	3.35(1.28)	3.09(1.34)	3.78(.91)	8.74(1.64)
South Australia	3.16(1.41)	3.25(1.29)	3.90(.94)	8.86(1.49)
Tasmania	2.32(1.57)	2.83(1.30)	3.67(1.37)	8.11(2.28)
Victoria*	3.13(1.30)	3.23(1.29)	3.77(1.01)	8.80(1.63)
Western Australia*	3.43(1.23)	3.60(1.05)	3.70(.95)	8.63(1.86)

* = The state or territory was governed by the Labor Party at the onset of the pandemic. All other states were governed by the Liberal Party.

Separate one-way ANOVAs with a Bonferroni corrected alpha level of p = .013 (i.e., .05/4) to reduce the chance of Type 1 errors were used to examine differences between each state and territory on the three trust variables and on compliance. Results for Levene’s test indicated that the assumption of homogeneity of variance was violated for the measure of trust in state government, F(7, 967) = 2.14, p = .04, and compliance with COVID-19 prevention measures, F(7, 974) = 2.16, p = .04. As a result, for these two measures, Welch test results which are robust to violations of normality are reported, and post-hoc tests utilise Tamhane’s T2 test.

There were significant differences across the states and territories on trust in federal government, (F(7, 967) = 3.56, p = .001) and state government (Welch’s F(7, 110.84) = 2.96, p < .01). A SNK post hoc test indicated that respondents from Tasmania and the Northern Territory reported significantly lower trust in federal government than other states and territories. The Tamhane’s T2 post hoc test for trust in state government indicated that the only significant difference was observed between respondents from Queensland and Western Australia. No significant differences were observed for trust in scientists and levels of compliance with COVID-19 prevention measures among the states and territories.

Due to the inherent limitation of making comparisons where some samples are very small due to obtaining a representative national sample (e.g., Tasmania n = 19) and heterogeneity of variance is evident, we created two groups to assist with making potentially more meaningful comparisons: states governed by the Liberal Party or Coalition (i.e., the Liberal and National Parties; n = 409) and states/territories governed by the Labor party (n = 591). An independent samples t-test revealed no significant differences between these groups on any variables in the current study. As a result, we chose to continue analysing the whole sample rather than consider differences between states and territories.

Compliance with COVID-19 prevention measures

Correlations

Bivariate correlations between compliance with COVID-19 prevention measures and all independent variables in the whole sample are shown in Table 4. Greater compliance was associated with higher levels of trust in each of federal government, state government and scientists. As hypothesised, older respondents and women also reported greater compliance in the current study. In addition, higher levels of trust in federal government were associated with greater trust in state government, and higher levels of trust in each form of government were associated with greater trust in scientists. Older respondents and those with a right-leaning political orientation reported higher trust in federal government. Higher levels of educational attainment and right-leaning political orientation were significantly associated with greater trust in state government.

Table 4. Relationships between self-reported compliance with COVID-19 prevention measures and all predictors being examined in the current study.

	1	2	3	4	5	6	7	8
1. Compliance	-
2. Trust in federal government	.22***	-
3. Trust in state government	.24***	.67***	-
4. Trust in scientists	.22***	.30***	.39***	-
5. Age	.34***	.16***	.04	−.01	-
6. Gender	.09**	−.02	.00	−.00	−.15***	-
7. Education	−.05	.02	.08**	.05	−.22***	−.05	-
8. Political orientation	−.02	.30***	.08*	−.06	.07*	−.11**	.04	-

* p < .05, ** p < .01, *** p < .001.

Regression analysis

Hierarchical linear multiple regression was used to examine predictors of compliance with COVID-19 prevention measures. Age, gender, education and political orientation were controlled for by including each in the first step of the analysis. Trust in federal government, state government and in scientists (from the most recent survey data) was added to the second step of the regression. Multicollinearity was not evident among any of the variables based on the criteria of Tolerance < .10 and VIF > 10 (Hair et al., 2019) As shown in Table 5, at the first step of the regression the demographic variables significantly accounted for 15% of the variance in compliance (Adjusted R² = .15, F(4, 846) = 38.39, p < .001), indicating that older respondents and women were more likely to comply with COVID-19 restrictions. In the second step of the analysis, the addition of trust in federal and state governments and in scientists significantly (p < .001) accounted for an additional 7% of variance, accounting for 22% of variance in compliance overall, F(7, 843) = 35.17, p < .001; Adjusted R² = .22. Again, the results indicated that women and older respondents were more likely, on average, to comply with COVID-19 prevention measures. Greater compliance was also significantly predicted by higher levels of trust in scientists and state government (as predicted), but not federal government.

Table 5. Standardised beta coefficients for each variable predicting compliance with COVID - 19 prevention measures.

	β	t
Step 1
Age	.39***	11.99
Gender	.15***	4.50
Education	.05	1.41
Political orientation	−.03	−1.01
Step 2
Age	.39***	12.08
Gender	.14***	4.63
Education	.02	.72
Political orientation	−.04	−1.11
Trust in federal government	.02	.42
Trust in state government	.13**	3.06
Trust in scientists	.18***	5.42

*p < .05, **p < .01, ***p < .001.

Discussion

The current study examined changes in trust in federal and state governments and in scientists across most years from 2003 to 2020. Trust in both forms of government was at its highest recorded point in 2020. This replicates the findings of Goldfinch et al. (2021) in which trust during the COVID-19 pandemic, in 2020, was higher than in 2009. This is notable given that political trust declined in Australia in the years prior to 2020 (Dassonneville & McAllister, 2021). In the current study, it is unlikely that the increases in trust were due to the use of an online survey or demographic factors. Therefore, the results are indicative of a rally around the flag effect, which was also widely observed in European countries during the pandemic (e.g., Bækgaard et al., 2020; Bol et al., 2021; Esaiasson et al., 2021; Nielsen & Lindvall, 2021). However, as suggested by Goldfinch et al., the increase in trust in Australia may reflect the initial effectiveness of the governmental response to the pandemic.

The regression analyses used to identify predictors of compliance with COVID-19 prevention measures also aligned with those from previous studies. Older respondents and women tended to report higher levels of compliance (Fazio et al., 2021; Murphy et al., 2020), and age was the most important predictor overall. The findings also clearly highlighted the importance of trust in scientists (Fazio et al., 2021; Plohl & Musil, 2021) as a predictor of compliance with COVID-19 prevention measures. The results also provide some support for those indicating that trust in government was an important predictor of compliance (e.g., Ayalon, 2021; Bargain & Aminjonov, 2021; Fazio et al., 2021; see also Murphy et al., 2020). Notably and contrary to previous work, the current study provided an indication that trust in state and federal governments may have contributed differently to compliance with COVID-19 prevention measures in Australia.

Whereas Goldfinch et al. (2021) summed together trust in federal and state governments as they were highly correlated, in our larger 2020 sample, trust in federal and state governments was strongly correlated but not multicollinear, which is why we chose not to combine them. While trust in federal government was initially correlated with increased compliance, this reduced to non-significance when included in the regression model alongside trust in state government and scientists. It is uncertain, however, why this is the case. For example, when comparing Labor states/territories and Liberal states, no significant difference was found in the mean trust of state governments or levels of compliance. Furthermore, while political orientation was correlated with trust in the federal government, it did not predict compliance, suggesting COVID and compliance behaviours were not politically polarised in Australia at the time, unlike in the US (e.g., Carson et al., 2021; Painter & Qiu, 2020; Viskupič et al., 2022). It, therefore, appears unlikely to be due to the party in charge in any state or territory. This aligns with a national poll from the middle of June 2020 (approximately the same time that data was collected in the current study) which found that people reported similar levels of approval for state government responses to the COVID-19 outbreak across Labor and Liberal states (Essential Research, 2020). As theorised in the Introduction, the result may instead reflect the public’s perception of the differential role of state and federal governments in managing the pandemic (see O’Sullivan et al., 2020). Additionally, as the perception of political leaders can affect political trust (Dassonneville & McAllister, 2021), the perceived quality of leadership at state and federal governmental levels at the time may have been a factor.

It is, therefore, worthwhile for future research to continue to examine levels of trust in federal and state governments in Australia. While the current findings potentially reflect the differing roles played in Australia by the federal and state governments in the early stages of the pandemic (see O’Sullivan et al., 2020), there is the potential for trust in federal and state governments to have changed since the collection of data for the current study (i.e., June 2020). From 2020 to the present, state governments have continued to take a primary role in matters such as border restrictions and quarantine, with the federal government criticised for a rollout of vaccines that was slower than comparable countries (see Stobart & Duckett, 2022). As a result, it could be the case that any rallying effects contributing to increased trust may have been driven mainly by trust in state governments. Additionally, differences in trust between states/territories may have become more pronounced from the collection of data in the current study due to the perceived and actual effectiveness of COVID-19 restriction measures.

Limitations

Despite the current findings generally aligning with and replicating those of other studies, a limitation was the lack of breadth of measures. For example, additional items asking about perceptions of political leaders could have further elucidated trust in state and federal governments. A further limitation of the current study is the representativeness of the sample. While efforts were made to ensure the 2020 sample was demographically representative of the general population, as an online survey was provided to individuals who had signed up to Qualtrics’ participant pool, the sample may not be truly reflective of the average Australian. That is, although Qualtrics aims to deliver samples that align with population data derived from national censuses, people who sign up as panel participants may differ from those in the general population who do not choose to participate in such research.

Second, this study focused on predicting compliance with COVID regulations, rather than on the end goals of compliance (e.g., COVID infection or mortality rates). That said, large-scale surveys have found strong links between compliance and mortality (e.g., Margraf et al., 2021) suggesting that models predicting compliance, in the absence of mortality and infection metrics, are nonetheless useful. Finally, the statistical models tested in this study assumed linear relationships between trust and compliance measures. While there is compelling evidence of linear relationships between trust and compliance (e.g., Fazio et al., 2021; Goldfinch et al., 2021; Murphy et al., 2020), emerging evidence suggests that the relationship may be nonlinear (Zaki et al., 2022a). Although this study found no strong evidence for nonlinear relationships, future studies could explore whether nonlinear relationships identified in European countries also hold in the Australian context.

Third, given that data for this study was collected at a single point in time (i.e., June 2020, 3 months into the pandemic), we were unable to examine whether the relationship between trust in federal and state governments and compliance changed over time. Given that the rally around the flag effects is typically short-lived and given that trust in government peaked in Australia in 2021, declining thereafter (Edelman, 2022), it is conceivable that there would be variations in the relationships between trust in federal and state governments and compliance. It is also conceivable that other variables, such as stringency of government public health measures, would have an appreciable effect on the relationship between trust in government and compliance. Moreover, although we found no relationship between such education and compliance, other socioeconomic factors may conceivably have had an effect on the compliance, as suggested by research by Zaki et al. (2022b) on the relationship between economic disparities across 300 regions and 25 countries in Europe and excess mortality during the COVID-19 pandemic.

Implications

To the extent that community trust in public institutions is critical for public perceptions of their integrity and social licence to operate, these results have important implications for the proposed Australian Centre for Disease Control (Department of Health and Aged Care, 2022). Specifically, the importance of trust in scientists in predicting compliance with COVID-19 prevention measures suggests the need for a prominent role of scientific and clinical experts, rather than political actors, in the leadership, operation and community engagement activities of the proposed Centre. It is vital that population health advice is seen by the public as reflecting sound information rather than mis- or dis-information. Second, the importance of trust in state government in predicting compliance with COVID-19 prevention measures, and the negligible influence of federal government, suggests, from the vantage point of the public, that any new public institution should be seen as much a state as a federal institution in order to sustain public trust and confidence in the context of the ongoing COVID-19 and future crises.

In conclusion, the current findings clarify that older individuals, women and those with greater trust in scientists tend to be more compliant with COVID-19 reduction measure. The findings also provide further evidence of the increase in political trust associated with the onset of the COVID-19 pandemic. Additionally, we extend on previous studies by highlighting the importance of trust in state rather than federal government as a predictor of greater compliance.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data are publicly available and published on the Open Science Framework: https://osf.io/sjp6g.

Additional information

Funding

This study was conducted in accordance with the approved guidelines of the Swinburne University Human Research Ethics Committee, in accordance with the Australian National Health and Medical Research Council’s (NHMRC) National Statement on Ethical Conduct in Human Research. Informed consent was obtained from all respondents included in the current study.