key: cord-0321507-wbmtav14
authors: Kaine, G.; Wright, V.; Greenhalgh, S.
title: Motivation, intention and action: wearing masks to prevent the spread of COVID-19
date: 2022-05-26
journal: nan
DOI: 10.1101/2022.05.25.22275599
sha: e41c2730ed57515145ed7a6893268378895e6938
doc_id: 321507
cord_uid: wbmtav14

Governments around the world are seeking to slow the spread of COVID-19 by implementing measures that encourage, or mandate, changes in peoples behaviour. These changes include the wearing of face masks, social distancing, and testing and self-isolating when unwell. The success of these measures depends on (1) the willingness of individuals to change their behaviour and (2) their commitment and capacity to translate that intention into actions. Consequently, understanding and predicting the willingness of individuals to change their behaviour, and their enthusiasm to act on that willingness, is critical in assessing the likely effectiveness of these measures in slowing the spread of the virus. In this paper we analyse responses to two separate regional surveys about peoples intentions and behaviour with respect to preventing the spread of COVID-19 in New Zealand. While motivations and intentions were largely similar across the regions, there was marked difference in action across the regions, specifically with respect to the frequency of wearing face masks. Our analysis suggests that the translation of intention (preventing the spread of COVID-19) into action (as measured by self-reported frequency of face mask use) was strongly associated with perceptions of the risk of infection (as measured by regional case numbers). The results highlight the importance to policy design of distinguishing the factors that might influence the formation of behavioural intentions from those that might influence the implementation of those intentions.

their behaviour, and their enthusiasm to act on that willingness, is critical in assessing the 23 likely effectiveness of these measures in slowing the spread of the virus.

In this paper we analyse responses to two separate regional surveys about people's intentions 26 and behaviour with respect to preventing the spread of COVID-19 in New Zealand. While 27 motivations and intentions were largely similar across the regions, there was marked 28 difference in action across the regions, specifically with respect to the frequency of wearing 29 face masks. Our analysis suggests that the translation of intention (preventing the spread of 30 COVID-19) into action (as measured by self-reported frequency of face mask use) was 31 strongly associated with perceptions of the risk of infection (as measured by regional case 32 numbers). 6 112 113 In two studies [6, 7] in different regions, behavioural intentions with respect to wearing face 114 masks, self-isolating and getting tested for COVID-19 were investigated together with self- 115 reports of actual behaviour with respect to wearing face masks and getting tested. The results 116 in these two studies indicated that, despite similarities across regions in behavioural 117 intentions, there were marked dissimilarities in actual behaviour [7] . In particular, although 118 the willingness to act to prevent the spread of COVID-19 was similar across the regions, the 119 wearing of face masks was dramatically different across the regions [7] . Where, as in this instance, diversity in actual behaviour occurs in a context of shared 122 behavioural intentions with respect to a novel behaviour, it is necessary to identify the cause. 123 First, because, if the claimed integrity of the I 3 model [6, 7, 9, 16, 17, 18, 19, 20] , and other 124 models of behavioural intentions [10, 11] , is to be sustained, it is necessary to be able to 125 identify a plausible and active cause for hesitancy to implement an intention. In a basic sense, 126 behavioural intention is not the policy target if it is detached in some major way from actual 127 behaviour. Second, because, for any additional actions to be taken to accelerate adoption of 128 the behaviour to be appropriate, the causes of the hesitancy need to be identified. consequences for understanding what factors can properly be said to act as 'barriers' [10] to 140 desired behaviour changes. The notion of a 'barrier' is often unhelpfully broad because, as 141 here, the set of factors serving to create a favourable or unfavourable attitude to novel 142 behaviour can, and in this case must, be different to the set of factors seeming to impede 143 behaviour change. Particularly, 'barriers', in normal usage, usually refers to things or 144 situations impeding movement in an intended direction: intended by the subject, not some 145 observer. 'Barriers' has relevant meaning, therefore, to factors causing behavioural intention 146 to not lead to the behaviour in question. Its use with respect to the forming of behavioural 147 intentions reveals more about observer preferences than impediments the subject confronts. In the case of mask wearing, the barrier to continuous action may be non-availability of 150 masks, unanticipated social opprobrium when they are worn or unexpected discomfort (both 151 of which reflect poor judgement in arriving at behavioural intention). The most obvious, and 152 logically the first, 'barrier' to seek out is the absence of a behavioural trigger. Identifying 153 behavioural intention by questioning subjects rather than tracking behaviour, is to discover a 154 predisposition to act. A failure to act implies the absence of a trigger to activate the 155 predisposition. In this case, if masks are readily available, socially acceptable, and reasonably 156 comfortable, the missing trigger will presumably be related to perceived need: the perceived 8 162 wearing. Infection incidence may be the critical catalyst that triggers action. The adoption of 163 behaviours such as the wearing of face masks has been associated with a range of variables 164 including perceptions of the perceived risk of infection, the local incidence rate of COVID-19 165 and feelings of stress in relation to . 166 167 In the next section we provide a brief description of the history of COVID-19 in New 168 Zealand to place the subsequent analysis in its proper context. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) 

Data from two surveys were used in this study. The first survey, the 'Auckland' survey, was 220 of Auckland residents and was conducted over two weeks from 7 September to 22 September 221 2020. Auckland was chosen for the survey because it is New Zealand's largest city and is the 222 mostly likely place for community transmission to occur, given the greater number of MIQ 223 facilities and frontline border workers in the city. At the time of the survey, Auckland 224 residents were mostly under Alert Level 2, which meant that they were expected to maintain 225 social distancing when outside their homes and to wear masks in public places. They were 226 also expected to keep track of their movements and to self-isolate and seek testing for 227 COVID-19 if they felt unwell and experienced symptoms associated with COVID-19.

The second survey, the 'regional' survey, was of residents in five regions outside Auckland 230 with, or near, MIQ facilities (Hamilton, Rotorua, Tauranga, Wellington, and Christchurch) 231 and was conducted during the first and second week of March 2021, before the Delta variant 232 was detected in New Zealand and before vaccinations were available to the general public.

When the survey commenced, residents in these regions were under Alert Level 2, which 234 meant that they were expected to maintain social distancing when outside their homes and to 235 wear masks in public places. They were also expected to keep track of their movements and 236 to self-isolate and seek testing for COVID-19 if they felt unwell and experienced symptoms . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) Kaine [27] , with respondents rating two statements on each of the five components of 247 involvement (functional, experiential, identity-based, risk-based, and consequence-based).

Attitudes were measured using a simple, evaluative scale (the questionnaire is reproduced in 249 S1). The ordering of the statements in the involvement and attitude scales was randomised to 250 avoid bias in responses. Respondents indicated their agreement with statements in all the 251 involvement, attitude and belief scales using a five-point rating, ranging from strongly 252 disagree (1) to strongly agree (5).

Respondents' propensity to wear face masks was obtained by asking them if they had worn a 255 face mask when out in public the previous week and whether they had to go out to work the 256 previous week. Respondents answered both questions using a five-point scale ranging from 257 'always' to 'never'. Their propensity to self-isolate was obtained by asking them, 'Thinking 258 about the next few days, would you stay home if you were unwell or had any of the following 259 symptoms: a dry cough, fever, loss of sense of smell, loss of sense of taste, shortness of 260 breath or difficulty breathing?'. We also asked, 'If you were advised to do so by a healthcare . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) were answered using a five-point scale ranging from 'definitely' to 'definitely not'.

Information was also sought on the demographic characteristics of respondents, including 265 age, education, and ethnicity, and whether they wore masks, would self-isolate and had been The Auckland questionnaire had been piloted with a small random sample of residents 276 (n=30), and subsequently completed by a larger random sample of residents (n=1001) who 277 were members of a large-scale, commercial consumer internet panel. The regional 278 questionnaire, which was identical to the Auckland questionnaire, was completed by a large 279 random sample of residents (n=2000), stratified by regional population, who were also and were not minors. To reduce respondent fatigue, a split-sample approach was taken, with . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 26, 2022. ; https://doi.org/10.1101/2022.05.25.22275599 doi: medRxiv preprint

As the data on involvement, attitudes, behavioural intentions, and behaviour were collected 312 using the same survey but at different times, a number of confounding factors may give rise To begin with, although the samples were broadly similar with regard to their age, education, 318 and income composition, they differed substantially with respect to gender and ethnicity (see 319 Tables 1, 2, 3 and 4). Approximately 53% of respondents to the Auckland survey were 320 women whereas approximately 65% of respondents to the regional survey were women.

There were statistically significant but weak associations [30] between the demographic 323 characteristics and willingness to take responsibility for eliminating COVID-19 and 324 willingness to change normal behaviour, make sacrifices and work with others to eliminate 325 COVID-19 (see Table 5 ). There were also some statistically significant but weak associations 326 [30] between demographic characteristics and the wearing of face masks and willingness to 327 self-isolate (see Table 6 ). Differences in demographic characteristics such as age, ethnicity, 328 gender, and education may influence perceptions of the danger to health posed by COVID-329 19. This suggests that differences in the demographic composition of the two surveys could 330 partly explain differences in intentions and behaviour in the two surveys.

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 26, 2022. ; https://doi.org/10.1101/2022.05.25.22275599 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) We investigated the effect of differences in Alert Level restrictions by classifying regional We found a statistically significant, but extremely small, difference between the categories in 368 the mean scores for willingness to change normal behaviour (ƞ2=0.007), make sacrifices 369 (ƞ2=0.006) and work with others to eliminate COVID-19 (ƞ2=0.007). We also found a 370 statistically significant, but extremely small, difference between the categories in the mean 371 scores for wearing face masks in public (ƞ2=0.012) and wearing face masks at work 372 (ƞ2=0.11). This suggests that the difference in Alert Levels between the two surveys may 373 also partly explain differences in intentions and behaviour in the two surveys. Approximately 570 cases had been detected in the MIQ regions prior to conducting the 379 regional survey [28] . As noted earlier, the pandemic received extensive media coverage, 380 including daily daily government briefings, in New Zealand during the six months preceding . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 26, 2022. ; https://doi.org/10.1101/2022.05.25.22275599 doi: medRxiv preprint the Auckland survey. The Delta variant of COVID-19, a more highly infectious and severe 382 variant of the virus [31, 32], emerged during the five months between the Auckland and 383 regional surveys but was not present in New Zealand at the time the regional survey was 384 conducted. Given the emergence of the Delta variant, and that the regional survey occurred 385 some five months after the Auckland survey, awareness of COVID-19 among regional 386 respondents could reasonably be expected to be at least as great as it was among Auckland 387 respondents.

Intentions and behaviour 390 The purpose of this analysis was to explain differences in the propensity of Auckland and 391 regional respondents to comply with wearing face masks in public and at work, given their 392 behavioural intentions were similar. Note that satisfactory reliabilities [33] were obtained in 393 both surveys for the involvement and attitudinal scales with respect to eliminating COVID-394 19, wearing face masks, self-isolating when unwell and getting tested for COVID-19 (see 395   Table 7 ).

The means for all the involvement, attitude, intention, and behaviour variables for both 398 surveys are reported in Table 8 . Where the means for the two surveys were statistically 399 significantly different, the magnitude of the differences, as measured by effect size [30] , is 400 also reported in the table. An inspection of the table reveals that wearing face masks in public 401 and at work were the only variables for which the means were statistically significantly 402 different, and the magnitude of the difference was large, for the two surveys.

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) These results suggest that, on average, respondents in both surveys were similar in their 416 motivations (as measured by involvement) and attitudes towards eliminating COVID-19, 417 wearing face masks, self-isolating and getting tested for COVID-19. They were similar, on 418 average, regarding their intentions to take some responsibility for eliminating COVID-19, 419 and their intentions to change their normal behaviour, work with others and make sacrifices 420 to eliminate COVID-19 from New Zealand. They were also similar, on average, regarding 421 their intentions to self-isolate and get tested for COVID-19. Relatedly, Auckland and regional The only substantive difference between the two samples relates to the wearing of face 427 masks, with the means for the regional sample being substantially lower than the means for 428 the Auckland sample. The difference, on average, roughly corresponds to regional 429 respondents reporting that they only wore face masks sometimes (at best) when they were out 430 in public whereas Auckland respondents reported they often wore face masks when they were 431 out in public. Table 9 . The regressions were statistically significant and, for cross-sectional data, a Again, following Kaine et al. [6, 7] , we hypothesised that respondents' propensity to self-460 isolate and wear face masks would be a function of their involvement with, and attitude 461 towards, self-isolating and wearing face masks, respectively. Consequently, we estimated 462 regressions with respondents' self-reported willingness to self-isolate and wearing of face 463 masks as dependent variables. The explanatory variables for willingness to self-isolate were 464 respondents' involvement with, and attitude towards, wearing face masks and self-isolating.

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 26, 2022. were also included in the regressions as well as dummy variables representing when regional 470 respondents were at Alert Level 2 and the first week at Alert Level 1.

Willingness to self-isolate when unwell, or instructed to do so by a health authority, was 473 strongly and positively influenced by involvement with, and attitudes towards, self-isolating 474 (see Table 10 ). As with the other behavioural intention variables, the variation in 475 respondents' intentions to self-isolate were only weakly related, if at all, to their demographic 476 characteristics or, for regional respondents, changes in Alert Levels.

The explanatory variables for wearing face masks were respondents' involvement with, and 479 attitude towards, wearing face masks and, as with previous regressions, respondents' 480 demographic characteristics were also included to account for the possibility that the 481 demographic differences might be correlated with relevant omitted variables. As before, we 482 also included dummy variables representing when regional respondents were at Alert Level 2 483 and the first week at Alert Level 1.

We included additional explanatory variables in the regressions for wearing masks which 486 were intended to account for the differences observed previously in the wearing of face 487 masks by Auckland and regional respondents. We had attributed this difference in behaviour 

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) To begin with, we assumed that respondents' perception of the risk of infection would be 502 proportional to the total number of COVID-19 cases reported in their region prior to the 503 survey, and that their perception of the risk of infection would become increasingly sensitive 504 to the total number of cases, as that total increased. Consequently, the two additional 505 variables included in the regressions were the total number of cases in a respondent's region, 506 and the total number of cases in their region squared (and centred to avoid multi-collinearity).

Note that respondents might also use Alert level settings in judging the risk of infection. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 26, 2022. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 26, 2022. ; https://doi.org/10.1101/2022.05.25.22275599 doi: medRxiv preprint Willingness to wear face masks in public was strongly and positively influenced by 535 involvement with, and attitudes towards, wearing face masks. As was the case with 536 behavioural intentions, the variation in respondents' wearing of face masks in public was 537 only weakly related to their demographic characteristics. Changes in Alert Levels did appear 538 to have some influence on the wearing of face masks in public by regional respondents, 539 though this influence was weakened when the variables intended to account for respondents' These results support our hypothesis that the difference between Auckland and regional 556 respondents in the wearing of face masks can be attributed to differences in perceptions of the 557 risk of infection. If regional respondents did perceive the risk of infection from COVID-19 to 558 be lower than Auckland respondents, then they should be less likely than Auckland . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 26, 2022. ; https://doi.org/10.1101/2022.05.25.22275599 doi: medRxiv preprint respondents to seek testing for COVID-19 unless they felt unwell. That is to say, the 560 proportion of respondents who felt unwell when they were tested for COVID-19 should be 561 significantly higher among regional respondents than Auckland respondents. We found this 562 to be the case with approximately 69.7% of regional respondents feeling unwell when they 563 were tested for COVID-19 compared to 50.4% of Auckland respondents (p<0.001). The results are consistent with our hypothesis that the difference between Auckland and The willingness of people to adopt behaviours to prevent the spread of COVID-19 such as 593 wearing face masks, self-isolating and getting tested for COVID-19 has been the subject of 594 numerous studies [6, 7, 34, 35, 36, 37, 38, 39, 40, 41] . These studies have shown that . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) by COVID-19 through social media could lead to measures that disrupt daily routines being 661 adopted more slowly, and abandoned more rapidly, than is desirable. This is because such 662 misinformation may provide a self-serving rationale for failing to continue to comply with Our findings are subject to a several qualifications including the following. First, as the 672 survey samples were drawn from internet-based consumer panels there may be selection bias.

While the nature and severity of this bias in relation to the attitudes and involvement we 674 investigated is unknown, it does seem reasonable to suppose, ceteris paribus, that people with 675 low-to-mild involvement may be under-represented in the sample.

Second, as the scales measuring the wearing of face masks, willingness to self-isolate and 678 being tested for COVID-19 were self-reported, our measurements of these behaviours may desirability bias appeared to be consistent across gender, age, and education categories, the 681 potential for bias in self-reporting of socially desirable behaviours (or opinions) in the context . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted May 26, 2022. ; https://doi.org/10.1101/2022.05.25.22275599 doi: medRxiv preprint of the differing degrees of involvement is less clear. However, while there may be a 683 correlation between intensity of involvement and social desirability bias, the dramatic 684 difference between Auckland and regional respondents in their self-reported frequency of 685 wearing face masks suggests that the degree of social desirability bias in our study is small.

See Kaine et al. [6] for a more detailed discussion of this matter. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted May 26, 2022. ; https://doi.org/10.1101/2022.05.25.22275599 doi: medRxiv preprint their behaviour, and their enthusiasm to act on that willingness, is critical in assessing the 706 likely effectiveness of these measures in slowing the spread of the virus.

In this paper we analysed data from two regional surveys about people's intentions and 709 behaviour with respect to preventing the spread of COVID-19 in New Zealand. While 710 motivations and intentions were similar across the regions, there was a marked difference in 711 action across the regions, specifically with respect to the frequency of wearing face masks. 712 We found that the translation of intention (preventing the spread of COVID-19) into action 713 (as measured by self-reported frequency of face mask use) was strongly associated with 714 perceptions of the risk of infection (as measured by regional case numbers).

The results serve as a reminder of the importance when designing policies of distinguishing 717 the factors that might influence the formation of behavioural intentions from those that might 718 influence the implementation of those intentions.

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted May 26, 2022. ; https://doi.org/10.1101/2022.05.25.22275599 doi: medRxiv preprint

Public 727 attitudes, behaviors, and beliefs related to COVID-19, stay-at-home orders, 728 nonessential business closures, and public health guidance -United States

Knowledge, attitude, and perceptions towards the 2019 Coronavirus Pandemic: A bi-733 national survey in Africa

Attempt to understand public health 736 relevant social dimensions of COVID-19 outbreak in Poland 2020

Compliance with Covid-19 measures: Evidence from 743 New Zealand

Willingness to wear masks, self-isolate and be tested for Covid-19 in 746 MIQ regions of New Zealand

Involvement and use of the Covid-19 tracer app

Consumer Action: Automaticity, Purposiveness and Self-Regulation

Review of Marketing Research

Attitude-behaviour relations: A theoretical analysis and review of 755 empirical research

Technology acceptance model: a literature review from 1986 to 760 2013. Universal access in the information society

Vaccine hesitancy: the next challenge in the fight against COVID-19

A framework for understanding individual 769 response to regulation

An application of the I 3 framework to rat control in Hawke's Bay. Landcare 771 Research Contract Report LC3646

Predicting people's 773 motivation to engage in urban possum control

An application of the I 3 framework to rat control

Using 1080 to control possums and rats: An 778 application of the I 3 framework

Stress and worry in the 780 2020 coronavirus pandemic: relationships to trust and compliance with preventive 781 measures across 48 countries in the COVIDiSTRESS global survey

The End of the Elimination 791 Strategy: Decisive Factors towards Sustainable Management of

Measuring consumer involvement profiles

A pilot application of the I 3 framework to compliance behaviour in farming

New Zealand Ministry of Health, NZ COVID-19 data, cases

Eta squared and partial eta squared as measures of effect size in 803 educational research

Severity of illness in 806 persons infected with the SARS-CoV-2 Delta variant vs Beta variant in Qatar

Hospital admission and emergency

An 816 evidence review of face masks against COVID-19

Public perceptions, anxiety, and the perceived efficacy of health-protective 819 behaviours to mitigate the spread of the SARS-Cov-2/COVID-19 pandemic

Negative attitudes about facemasks during the COVID-19 822 pandemic: The dual importance of perceived ineffectiveness and psychological 823 reactance

What predicts attitudes about mask

Willingness to Self-Isolate When Facing a Pandemic 829 Risk: Model, Empirical Test, and Policy Recommendations

Factors Affecting Voluntary Self-Isolation 832

Behavior to Cope with a Pandemic: Empirical Evidence from Colombia vs

Attitudes, behaviours and barriers to public 839 health measures for COVID-19: a survey to inform public health messaging. BMC 840 Public Health

Applying principles of behaviour change to reduce 842 SARS-CoV-2 transmission

845 Wearing one for the team: views and attitudes to face covering in New 846

Aotearoa during COVID-19 Alert Level 4 lockdown

A bioweapon or a hoax? The link between distinct conspiracy 849 beliefs about the Coronavirus disease (COVID-19) outbreak and pandemic behavior

The causes 852 and consequences of COVID-19 misperceptions: Understanding the role of news and 853 social media. The Harvard Kennedy School (HKS) Misinformation Review

Compliance with containment 865 measures to the COVID-19 pandemic over time: Do antisocial traits matter?

The Emotional Path 869 to Action: Empathy Promotes Physical Distancing and Wearing of Face Masks During 870 the COVID-19 Pandemic

Psychological Correlates of News Monitoring

Disinfecting, and Hoarding Behaviors Among US Adolescents During the 874 COVID-19 Pandemic

 721 We would sincerely like to thank those people throughout New Zealand who completed our 722 questionnaires. Thanks also to our two anonymous referees for their time, patience, 723 constructive advice.