Background

In March 2020, the UK imposed lockdown measures because of the coronavirus pandemic. These measures have been of unprecedented scope in the context of UK governance. These rules were developed over a series of weeks, culminating in the Prime Minister’s statement on the 23rd March that people must stay at home at all times unless they were engaging in one of activities allowed under the law, which were: shopping for necessities as infrequently as possible; one form of outdoor exercise per day, only with people within one’s own household; any medical need or providing care for vulnerable people; and travelling to and from work where it was absolutely necessary and could not be done from home (Gov.uk, 2020). Nonetheless, despite these restrictions being unprecedented in the context of the UK, they were less strict than comparable rules from some other states that were similarly as badly affected. That left space for some ambiguity about the rules, and potentially made enforcement of the rules by the police more complex. As such, the success or failure of this lockdown depended to a significant degree on the extent to which people were prepared to voluntarily comply with the restrictions.

In late May and in late July/early August we conducted two large surveys (n > 1000) of the UK population on attitudes to and beliefs around coronavirus. Within the survey we asked a range of questions on people’s understanding of and compliance with the lockdown rules. This provides us with detailed data on factors that correlate with compliance.

Data collection

We collected data using Prolific, an online platform where people can opt-in to receive invites to surveys and research studies. We conducted two surveys each taking around 20-40 minutes for respondents to complete. Survey 1 recruited 1001 respondents on 20th May 2020 and 200 respondents on the 27th May 2020. Survey 2 recruited 515 respondents on 30th July 2020 and 636 respondents on 7th August 2020. These data collections aim to approximate a random sample of the people in the country to allow general inferences to be made. Like all surveys, those who opt-in to such studies will be slightly different from those who did not; not least because it is a condition of doing so that one has access to the internet. Nonetheless there is no reason to believe that associations between factors are adversely affected by the selection method and our distribution of respondents is similar to what we would expect to see in a truly representative sample.

UK compliance rates

Those surveyed were asked whether they believe they always followed the guidance social distancing and ‘Stay at Home’ or whether they behaved in ways inconsistent with the guidance. In the latter case we asked whether they had behaved in ways inconsistent with the guidance to ‘help a neighbour, family member etc.’, because of ‘personal need, e.g. paid, non-essential work’, or because they ‘do not agree with aspects of the policy’. Table 1 shows the proportion answering each question over time.

Table 1, Lockdown compliance

Survey 1Survey 2
 20th May 2020 (n=1001)27th May 2020 (n=200)30th July- (n=515)  7th August 2020 (n=636)
Followed guidance at all times69.5%63.5%56.1%58.9%
Broke guidance to help someone24.2%26.5%31.9%29.9%
Broke guidance because of personal need6.8%9.0%10.5%9.3%
Broke guidance because did not agree with aspects of guidance4.2%9.0%9.6%7.8%
Note: respondents could select more than one option to describe why they acted in ways potentially contradicting the guidance.

You can see that the proportion of respondents complying with guidelines has dropped from around 70% to 60% over the last 8 weeks. This drop is highly statistically significant. There is a statistically significance increase in those saying they broke the guidance for all three options – to help, personal need, and because of disagreeing with the policy.

The Dominic Cummings scandal

Particularly noteworthy in our data is the drop in compliance during the week 20th May – 27th May 2020. The Dominic Cummings scandal broke in the media on the 22nd May and is the most likely explanation for this drop. Note that the drop could be because of decreased compliance or because of a greater willingness to report past non-compliance. Either way, we can see that compliance has not returned to the previous high level and has likely declined a little further. Our findings are consistent with complementary research with a larger sample (n = 40,597) that demonstrated that the scandal can be plausibly thought of as the cause of a general decline in confidence in the government to handle the pandemic. It is uncertain whether the Cummings scandal is the cause of the continued drop in compliance, though this is plausible given the general importance of leadership and ‘tone from the top’ seen in many areas of public life.

Factors related to the compliance rate

Age

Older people are more likely to report following the guidance at all times. This finding is highly significant in both surveys. Nonetheless, the predictive power of age in isolation for understanding why someone complied with the lockdown is relatively low. Age explains less than 2% of the variation between people’s reporting of always following the guidance or not. 

Older people are also less likely to report breaking the guidance for each of the reasons we questioned. Yet, while this effect was consistent, it was not reliably statistically significant. Also, in no case did age explain more than 1% of the variation in answers on breaking the lockdown for any particular reason.

As such, while it is fair to say that older people are more likely to adhere to guidance at all times, it is not fair to suggest that age is a singularly important factor in explaining reported behaviour.

Gender

While men and women on average have different patterns of paid work and are on average differentially involved in household labour, we do not find any evidence of gender on lockdown compliance.

BAME status

From relatively early on in the pandemic, reports have suggested that black and minority ethnic people are more likely to have serious consequences from catching Coronavirus. Nonetheless, in neither of our surveys do we find any systematic differences in adherence to the guidance between white and non-white people. The only marginal exception is that BAME people are around half as likely to report breaking the guidance because they do not agree with aspects of the policy. A summary of the data is in Table 2. Our results are consistent with either (1) there being no effect of ethnicity on lockdown compliance, or (2) there being substantive differences between different non-white ethnic groups which we cannot reliably detect due to small sample sizes within these groups.

Table 2, Compliance by BAME status

Survey 1Survey 2
 White (n=1035)BAME (n=164)White (n=977)  BAME (n=174)
Followed guidance at all times67.9%72.0%57.0%59.2%
Broke guidance to help someone25.0%22.5%31.5%28.2%
Broke guidance because of personal need6.8%9.1%9.3%13.8%
Broke guidance because did not agree with aspects of guidance5.4%2.4%9.4%5.2%

Knowledge of lockdown guidance

One important issue to consider with individual compliance with any lockdown is the extent to which members of the public know what the guidance is and what they are expected to do. If people do not understand the guidance then it is possible that people will incorrectly believe they are following the guidance while not doing, and simultaneously others may believe they are breaking the guidance when they are not. While the latter may seem less problematic, it remains possible that believing that the guidance is more restrictive than it is may undermine a belief in the legitimacy or efficacy of the guidance for those individuals, which may ultimately challenge compliance. In Survey 2, we asked respondents ‘Which of the following best represents your knowledge of the current government guidance on social distancing, ‘Stay at Home’ and ‘Stay Alert’?’. The results are shown in Table 3.

Table 3, Understanding of government guidance on social distancing, ‘Stay at Home’ and ‘Stay Alert’

I am confident I know what the guidance says58.5%
I have not paid any attention to what the guidance says1.7%
I have tried to become informed of the guidance but it is confusing37.0%
I look to what friends and family are doing and learn from them2.9%

Clearly, most people feel confident that they know what the current guidance says. This is encouraging, even though the guidance becomes increasingly nuanced as time goes by and lockdowns are gradually eased. Nonetheless, over 40% of respondents are either ignorant of the guidance, or have not attempted to learn about the guidance first-hand at all.

Confidence in knowing what the guidance says was strongly correlated with compliance to the lockdown rules. In particular, those who were confident they knew the rules were significantly more likely to have complied than those who were confused. This suggests that it may be necessary to consider how the public are engaged with the actual government guidance, and how this communication can enhance the clarity of the guidance.

Education, income and employment

We find no evidence of a statistically significant effect of either education level, income level, or employment on compliance. The only marginal effect worthy of note is that those in full time employment were slightly more likely to have broken the guidance for personal need.

There is, though, a highly significant effect of there being a keyworker in the householder. Householders with a key worker are significantly less likely to have complied with the lockdown and more likely to have broken the guidance to help another.  Similarly health-care workers are significantly less likely to have complied with the lockdown and more likely to have broken the guidance to help another and for personal need.

We offer three potential interpretations for this result: (1) Key workers and health-care workers may perceive they have broken the guidance because they did not ‘stay at home’. (2) Because key workers were ‘out and about’ they were less averse to performing other tasks, such as helping friends. (3) It was more difficult for key-workers and health-care workers to obey the lockdown restrictions.

Household and region

We find no statistically significant effect due to region. Nor do we find any effect of household size or number of children. There is some evidence that married people were more likely to obey the guidance than single people but this effect is relatively small.

Experience of the pandemic

A strong predictor of compliance with the lockdown restrictions is whether the pandemic is having a negative/positive effect on a personals personal/family relationships and mental health. The more negative the effect the less likely to have followed the lockdown restrictions. This is illustrated in Figure 1 where we can see that those for whom lockdown has had a strongly negative effect are far less likely to have complied with lockdown restrictions than those with a neutral to positive experience. Compliance, therefore, seems to be driven by how positively the pandemic is being ‘experienced’ and to a lesser extent the impact on ‘material’ outcomes (income and career).  

Figure 1: Compliance with lockdown in relation to effect of the pandemic on life

Attitudes to coronavirus

A very strong predictor of compliance with the lockdown restrictions is positive/negative attitudes about the restrictions. For instance, those who agree there has ‘been an over-reaction’ are far less likely to comply than those who do not agree. Similarly those who disagree ‘the economic cost is justified to save lives’ are far less likely to comply. While this may not be a surprise Figure 2 illustrates that the effect is very stark – those strongly agreeing that there has been ‘an over-reaction’ are 25% less likely to have complied than those who strongly disagree. 

Figure 2: Compliance and reasons for breaking the lockdown depending on agreement with statement ‘there has been an over-reaction to coronavirus’.

Face Masks

In survey 2 we explored people’s support for wearing face masks. We see that those following the lockdown guidance at all times are positive about masks. Trust in scientists and newspaper journalists is positively related while age is negatively related (with older people being less supportive). Keyworkers are more supportive of masks.

Summary

In interpretation we suggest the following key reasons why individuals are not obeying the lockdown rules:

1. Those who disagree with the rules are less likely to comply – presumably because of preference.

2. Keyworkers are less likely to comply – presumably because of constraint.

3. Those who are unsure of the guidance are less likely to comply.

4. Those for whom the pandemic has had a negative psychological effect are less likely to have complied.

Unfortunately, as time goes on all four of these factors are likely to become more prevalent – there may be fatigue with the rules, more people are returning to work, and the guidance is becoming ever more complex.