Abstract
AUSTRALIAN JOURNAL OF PSYCHOLOGY 2023, VOL. 75, NO. 1, 2180299 https://doi.org/10.1080/00049530.2023.2180299 Is it time to stock up? Understanding panic buying during the COVID-19 pandemic a a,b Karina T. Rune and Jacob J. Keech a b School of Health, University of the Sunshine Coast, Sippy Downs, Australia; School of Applied Psychology, Griffith University, Brisbane, Australia ABSTRACT ARTICLE HISTORY Received 19 July 2022 Background: Lockdowns to reduce the spread of COVID-19 have triggered sharp increases Accepted 9 February 2023 in consumer purchasing behaviour, labelled panic buying. Panic buying has detrimental consequences as it leads to product shortages and disrupts supply chains, forcing retailers KEYWORDS to adopt quotas to manage demand. Developing an understanding of the psychological COVID-19; behavioural correlates of panic buying can provide targets for public messaging aimed at curbing the triggers; hoarding behaviour. Objective: The study aimed to identify the psychological, individual difference, and demo- graphic factors associated with increased purchasing of non-perishable, cleaning, and hygiene products during COVID-19 lockdowns in Australia. Methods: The study used a cross-sectional design (N = 790) with online survey measures adminis- tered to community members in Australia during April and May 2020. Data were analysed using structural equation modelling. Results: Structural equation models revealed that 1) attitudes, subjective norms, and risk perceptions predicted increased purchasing of non-perishable products; 2) attitudes, risk perceptions, social anxiety sensitivity, and the non-impulsivity facet of trait self-control pre- dicted increased purchasing of hygiene products; and 3) attitudes and risk perceptions pre- dicted increased purchasing of cleaning products. Conclusion: Findings provide an understanding of the factors that were associated with panic buying during COVID-19 lockdowns in Australia. Future studies should investigate whether messages designed to influence risk perceptions, attitudes, and subjective norms are effective in curbing the behaviour. KEY POINTS What is already known about this topic: (1) Lockdowns to curb the spread of COVID-19 prompted substantial increases in consumer purchasing behaviour, labelled panic buying. (2) Prior research had identified a range of individual difference factors as being associated with panic buying, including intolerance of uncertainty and distress intolerance. (3) Identification of modifiable psychological processes, which are associated with the beha- viour, is needed to inform public messaging aimed at curbing the behaviour. What this topic adds: (1) The study provides information from a large national sample of Australians who regularly purchase groceries. (2) Our results suggest that potentially modifiable social cognition factors were most closely associated with increases in consumer purchasing behaviour when COVID-19 lockdowns were announced. (3) Public messaging should target attitudes, subjective norms, and risk perceptions regarding increased purchasing behaviour and future research should evaluate the effect of such messaging. Introduction coronavirus (COVID-19) and the resultant pandemic. Since December 2019, the world has lived through The COVID-19 pandemic impacted consumer behaviour “uncharted territory” (World Health Organization throughout the world with reports of panic buying and [WHO], 2020) with the spread of the SARS-CoV-2 novel stockpiling of household commodities leading to CONTACT Jacob J. Keech jkeech@usc.edu.au Supplemental data for this article can be accessed at https://doi.org/10.1080/00049530.2023.2180299. © 2023 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. 2 K. T. RUNE AND J. J. KEECH temporary shortages (O’Connell et al., 2020; Sim et al., consumers perceive the probability and consequences 2010). The increase in consumer purchasing behaviour, of contracting a disease as high, self-motivated beha- labelled as panic buying or hoarding in the media, refers viours, such as panic buying, increase to mitigate per- to buying and stockpiling more of everyday household ceived risk. In 2021, Labad et al. (2021) published items than is necessary to sustain a household during a systematic review on 14 recent studies that had routine life, and in anticipation of a potential future investigated toilet paper hoarding prior to the shortage (Bentall et al., 2021; Shou et al., 2013; Yoon COVID-19 pandemic. The review found that social et al., 2018). Panic buying becomes problematic when it media and social cognitive biases were potential con- disrupts production and supply chains (Shou et al., tributors to toilet paper stockpiling. Further, psycholo- 2013), creating demand-side scarcity, which forces retai- gical factors such as situational stress, fear of contagion lers into adopting quotas and price increases (Woertz, and personality traits (e.g., conscientiousness and 2010). emotionality) were associated with panic buying beha- During the first wave of the COVID-19 pandemic viour. These studies suggest that panic buying is in Australia (April – May 2020), the media actively a complex behaviour, however, there is currently lim- attempted to discourage consumers from engaging ited literature on the social cognition variables under- in panic buying. However, research indicates that pinning panic buying. Understanding these influences the media may have exacerbated the situation and is important as they may support the development of made people more anxious by using sensationalist targeted public campaigns, framed with personally headlines, showing pictures of people engaged in relevant information to help alleviate the perceived panic buying, long queues, and empty supermarket need to engage in panic buying. shelves (Arafat et al., 2020; Dubey et al., 2020; The COVID-19 pandemic has been a time of uncer- Nicola et al., 2020). This phenomenon has been tainty for many with extended periods of increased referred to as the scarcity effect whereby people social isolation. In contrast to disaster preparation observe the behaviour of others to estimate the (e.g., floods or hurricanes), where people are more seriousness of a crisis (Arafat et al., 2020; Pantano aware which items to purchase, panic buying appears et al., 2020). Additionally, a key reason for the inef- to be largely impulsive and driven by uncertainty. ficacy of media messaging for behaviour change is While conceptually different, hoarding and panic buy- that these messages are often largely atheoretical ing behaviours have been theorised to be driven by and do not target the psychological processes that the same psychological mechanisms fermented in an underpin the behaviours (Randolph & Viswanath, instinctive evolutionary reaction to a perceived threat 2004; Webb et al., 2010). Understanding theoreti- (Cameron & Shah, 2015) and motivated by a fear of cally based psychological processes will support being caught unprepared (Frost & Gross, 1993). In the identification of behaviour change methods a survey of 54 countries, Keane and Neal (2021) that have been shown to be effective for changing found that bursts of panic buying typically lasted 7 to human behaviour (Hagger et al., 2020). 10 days and were linked to government announce- Panic buying is an unusual and rare event that is ments of movement restrictions (e.g., lockdowns). difficult to study retrospectively. A study conducted in Research into hoarding behaviour more broadly sug- Singapore during the 2003 SARS outbreak reported gests that individual difference factors such as distress substantial increases in psychiatric morbidities includ- intolerance and intolerance of uncertainty may influ - ing somatic disorders, anxiety, depression, and social ence panic buying (Grisham et al., 2018; Norberg et al., dysfunction (Sim et al., 2010). Participants in the study 2015). In an attempt to reduce feelings of insecurity expressed concerns about a loss of control, fear of and uncertainty, people may gravitate towards things contagion, unpredictability of the situation, impact they can control, such as purchasing products that on the economy, and family health concerns. In addi- fulfill basic needs (Dubey et al., 2020; Prentice et al., tion, younger age and being female were associated 2020; Sim et al., 2010; Yoon et al., 2018). with greater anxiety. A systematic review by Yuen et al. Individuals have reported up to a 49% increase in (2020) identified that panic buying in health crises anxiety concerning their safety and livelihood during prior to the COVID-19 pandemic was influenced by the COVID-19 pandemic (Evidation, 2020). Such unease four factors: (1) perception of threat, (2) fear of the may turn individuals to coping strategies aimed at unknown, (3) coping behaviour, and (4) social psycho- exerting control over an uncontrollable situation (Hori logical factors. The authors concluded that when & Iwamoto, 2014). Taylor et al. (2020) reported that AUSTRALIAN JOURNAL OF PSYCHOLOGY 3 people who are extremely worried about COVID-19 are in the behaviour. Theoretical messages are widely more likely to engage in panic buying. These people supported as being more effective and more straight- also tend to be high in intolerance to uncertainty. forward to evaluate than atheoretical messages for Similarly, research conducted during the very early changing behaviour (Webb et al., 2010). Social cogni- stages of the COVID-19 pandemic in the UK (N = 2,025) tion theories have a long tradition of being applied and Ireland (N = 1,041) reported that psychological dis- to changing behaviour relevant to health and in tress and anxiety (death anxiety and threat sensitivity) health-related contexts (Hagger et al., 2020). predicted over-purchasing across a wide range of pro- A prototypical social cognition model is the theory ducts (Bentall et al., 2021). Weismuller et al. (2020) of planned behaviour (Ajzen, 1991). Within the the- investigated correlates of COVID-19 adherent and dys- ory, behavioural intention (a person’s readiness to functional (e.g., panic buying) safety behaviours in over engage in a given behaviour), is the most proximal 15,000 German participants. The study found that panic determinant of behaviour. Behavioural intention is buying was a psychological reaction to a current crisis determined by attitudes towards the behaviour (posi- and the fear of an interruption to the supply chain. tive or negative evaluations of the behaviour), sub- Similarly, Herjanto et al. (2021) found that situational jective norms (perceived social pressure to engage in ambiguity and evaluation-based thinking style the behaviour), and perceived behavioural control increased perceived risk, which in turn generated (perceived ability to perform the behaviour). panic buying. Given these findings, it is essential to While the theory of planned behaviour is a well- further investigate factors associated with panic buying established and parsimonious approach to explaining behaviour in an Australian context as this will aid in the health-related behaviour (Armitage & Conner, 2001; development of campaigns aimed at decreasing such McEachan et al., 2011), it has some notable limitations. behaviour so that supply chains can manage stock First, the theory constructs consistently explain con- levels. siderably more variance in intention than behaviour. Emerging research suggests that social psychologi- That is, those who intend to engage in a behaviour cal and demographic variables may also impact panic often do not follow through with their intention buying behaviour. For example, Bentall et al. (2021) (Orbell & Sheeran, 1998). This is known as the inten- found that household income and children in the tion-behaviour gap. Second, the theory only seeks to home predicted increased panic buying behaviour. account for conscious and reasoned processes. Finally, Similarly, Dinić and Bodroža (2021) reported that gen- a large portion of variance in behaviour remains unex- der, age, and educational level (N = 545) were not plained by theory constructs. To overcome these lim- related to stockpiling, whereas household size posi- itations, researchers have sought to integrate tively correlated with stockpiling. Results from complementary constructs from other social cognition a Brazilian study found that panic buying happens in theories. One such approach is to include an index of every income class, but that a positive correlation behavioural automaticity to account for the potential exists between average income per capita and panic impulsive nature of behaviour. This draws from dual- buying (Yoshizaki et al., 2020). Further, social influence process models of cognition and behaviour (Evans & (e.g., via the media, online, or in person) and distrust Stanovich, 2013; Strack & Deutsch, 2004), which posit have been proposed to influence consumer behaviour that behaviour is a function of two independent sys- in reaction to legislation and restrictions put in place tems working in parallel – a reflective system whereby by the government (Loxton et al., 2020; Zheng et al., behaviour is regulated through reasoned conscious 2021). This is proliferated by mimicking influential decision-making, and an impulsive system whereby others and knee jerk reactions to social media posts behaviour is regulated through automatic and non- of panic buying and empty supermarket shelves conscious decision-making. (Zheng et al., 2021). Including behavioural automaticity in integrated social cognition models has been found to explain unique variance in a range of health-related beha- An integrated social cognition approach viours (Brown et al., 2020; Hamilton et al., 2020; While the research conducted to date has provided Phipps et al., 2021), including transmission preven- an indication of a range of factors that are associated tion behaviours during the COVID-19 pandemic with panic buying behaviour, using theory to identify (Hagger, Smith, et al., 2021). Another complementary modifiable psychological processes underpinning construct that has been commonly included in inte- these behaviours can help to ascertain the optimal grated social cognition models is the affective con- content of messages aimed at reducing engagement struct known as risk perceptions. Risk perceptions are 4 K. T. RUNE AND J. J. KEECH beliefs regarding personal risk or susceptibility to through measurement of behaviour in relation to spe- certain outcomes or conditions if engaging or not cific product categories can provide important insight engaging in a particular behaviour. Risk perceptions into the optimal content for public messaging to pre- form a part of two theories that have widely been vent panic buying during future similar health-related applied to understanding health-related behaviour, events in Australia. namely the health belief model (Rosenstock, 1974) and the health action process approach (Schwarzer, The present study 2008). Applying an extended theory of planned behaviour, The current study aimed to identify the psychological Lehberger et al. (2021) found that attitudes, subjective and individual difference factors that are associated with norms, and fear of future unavailability were the main increased purchasing of (1) non-perishable food items, predictors of stockpiling of non-perishable foods in (2) personal hygiene products, and (3) household clean- Germany. Similarly, Roșu et al. (2021) examined stock- ing products during the COVID-19 pandemic of 2020. piling during COVID-19 lockdowns in Romania. Their First, based on prior research into similar behaviours findings suggested that attitudes and social norms (e.g., Bentall et al., 2021; Grisham et al., 2018), it was predicted both intentions to stockpile and actual hypothesised that individual difference factors includ- stockpiling behaviours. While this research has started ing intolerance of uncertainty, distress tolerance, anxiety to map the social cognition factors associated with sensitivity, trait self-control, hoarding rating, and COVID- panic buying, no research to date has explored the 19 risk perceptions, will be associated with increases in predictors of panic buying in an Australian context. In each purchasing behaviour. Second, drawing upon an addition, no research has examined and compared the integrated social cognition theoretical approach, it was predictors of panic buying across product categories. hypothesised that attitudes, subjective norms, risk per- Increasing the precision of measuring the behaviour to ceptions, and behavioural automaticity will be asso- specify specific product categories provides the oppor- ciated with increases in purchasing behaviour. Third, it tunity to determine whether different factors are asso- was hypothesised that demographic factors including ciated with changes in different types of products. To minutes to the supermarket, household size, household date, no research has made this distinction in the con- income, age, and gender, will predict each purchasing text of panic buying during the COVID-19 pandemic. behaviour. Refer to Figure 1 for a conceptual map of the This information from the Australian context attained study hypotheses. Figure 1. Conceptual map of the research hypotheses. AUSTRALIAN JOURNAL OF PSYCHOLOGY 5 Method Design and procedure Participants The University Human Research Ethics Committee approved the study (protocol: A201375). The study A total of 821 participants were recruited across the used a cross-sectional survey design. Participants were general Australian population. Participants were eligible recruited across Australia through methods which to participate in the study if they regularly purchased included the research team speaking about the study food or other household items from the supermarket on television and radio news broadcasts. The research- and were over 18 years of age. A total of 15 participants ers were also interviewed about the study and quoted were excluded due to not meeting eligibility criteria (2 in online and print newspaper articles. All media stories based on age and 13 based on not regularly purchas- about the study included a link to participate in the ing). Due to inattentive responding, 31 participants survey. Participants were also recruited online using were also systematically excluded, leaving a final sample social media and snowball sampling. Data were col- of 790 participants. The age range of participants was 18 lected during the height of the COVID-19 pandemic in to 86 years (M = 48.89 years SD = 13.23), with 613 par- age Australia (April and May 2020). Participants were ticipants identifying as female, 173 as male and 4 as informed the study was investigating attitudes and a different gender. Participation was voluntary and beliefs towards stocking up on groceries during the responses were anonymous. No incentives were offered COVID-19 pandemic. The survey package included two to participate in the study. Detailed demographic infor- screening questions (i.e., “do you regularly purchase mation can be found in the Table 1. Table 1. Demographic information relating to frequency and percentage of participants in the study (N = 790). Frequency Percentage Frequency Percentage Ethnicity Education level Australian 562 71.1 Year 10 67 8.5 Aboriginal Australian 2 .3 Year 12 82 10.4 European/Caucasian 71 9.0 TAFE, Certificate/Diploma, trade 255 32.3 Asian 12 1.5 or VET Qualification Other 13 1.6 Bachelor’s degree 196 24.8 Country of birth Post graduate degree 187 23.7 Australia 559 70.8 Currently studying a degree at university Africa 14 1.8 Yes 98 12.4 Asia 19 2.4 No 690 87.3 Canada 7 .9 Employment Europe 28 3.5 Full-time 309 39.1 Middle East 3 .4 Part-time 111 14.1 New Zealand 24 3.0 Unemployed/home duties 58 7.3 United Kingdom 60 7.6 Unemployed looking for work 20 2.5 United States of 28 .5 Unemployed not looking for work 11 1.4 America Retired 129 16.3 Marital status Full-time student 23 2.9 Never married 113 14.3 Studying and working 29 3.7 In a relationship 53 6.7 Disabled 26 3.3 Married 398 50.4 Currently not working due to COVID-19 52 6.6 De-facto 81 10.3 Weekly (annual) household income Separated/divorced 115 14.6 Nil income 14 1.8 Widowed 28 3.5 $1–$199 ($1-$10.399) 11 1.4 Children $200–$299 ($10,400-$15,599) 14 1.8 Yes 538 68.1 $300–$399 ($15,600-$20,799) 41 5.2 No 246 31.1 $400–$599 ($20,800–31,199) 53 6.7 Children living at home $600–$799 ($31,200-$41,599) 67 8.5 0 328 41.5 $800–$999 ($41,600-$51,999) 49 6.2 1 123 15.6 $1000–$1249 ($52,000-$64,999) 70 8.9 2 145 18.4 $1250–$1499 ($65,000-$77,999) 69 8.7 3 67 8.5 $1,500–$1,999 ($78,000-$103,999) 131 16.6 4 18 2.3 > $2000 (>$104,000) 256 32.4 5+ 9 1.0 Number of people in Minutes to shop household 1–10 minutes 620 86.1 1 132 16.7 1–20 minutes 76 9.6 2 293 37.1 21–30 minutes 25 3.2 3 118 14.9 31–40 minutes 2 .3 4 130 16.5 41–50 minutes 3 .4 5 69 8.7 51–60 minutes 1 .1 6+ 35 4.5 >60 minutes 3 .4 Missing responses are not recorded. Information is based on the final sample of 790 participants after careless respondents were removed. 6 K. T. RUNE AND J. J. KEECH food or other household items from the supermarket” <category> products that I usually buy”; scales 2–7 and “age>18 years”). Ineligible participants were exited started with “I have increased my purchasing to buy from the survey. Eligible participants were asked to enough <category> products for . . . ” 2 = “an extra few complete the online survey hosted by Qualtrics survey days”, 3 = “an extra week”, 4 = “an extra two weeks”, 5 = “an extra three weeks”, 6 = “an extra month”, 7 software. The survey remained open for 5 weeks and = “more than an extra month”. Given that an estab- sample size was determined based on this recruitment lished scale did not exist for this purpose, we applied window. best practice principles for self-report measurement of behaviour such as ensuring behavioural definitions Measures and rating scales are clearly worded in a manner that is specific regarding target, action, context, and time Item wording for all measures is provided in (Ajzen, 2006a). Single item measures of behaviour have Supplementary Appendix A. Where item wording been found to be valid in other relevant contexts such could not be reproduced due to copyright, the supple- as retrospective recall and self-reporting of physical mentary material contains information regarding activity (Hamilton et al., 2012). where the items can be accessed. Revelle’s ω was calculated using the userfriendlyscience (Peters, 2017) Social cognition constructs package in R (R Core Team, 2019) as a measure of Measures of social cognition constructs were adapted reliability for each scale (McNeish, 2018; Peters, 2017). to the current behavioural context based on estab- Reliability for two-item scales was calculated using lished guidelines (Ajzen, 1991, 2006b; Gardner et al., Spearman rank order correlations. All scales exhibited 2012). satisfactory reliability. See Supplementary Appendix B for reliability coefficients. Attitudes. Attitudes towards each behaviour were assessed using three items preceded by a common Demographic variables stem: “If I were to buy more non-perishable products The following demographic variables were assessed: than I would use based on my usual frequency of gender; marital status; number of children; number of shopping, it would be”:. Responses were provided on children living at home; number of people living in semantic differential scales (e.g., 1 = bad and 7 = good). household; education levels; currently studying an The scale was administered for each behaviour. undergraduate degree; employment status; hours in paid employment; field of study and/or work; and Subjective norms. Subjective norms for each beha- weekly (annual) income. viour were measured using five items (e.g., “Most peo- ple who are important to me would approve of me Purchasing behaviour buying more non-perishable products”.). The scale was Participants’ increases in purchasing behaviour was administered for each behaviour. Responses were pro- assessed using a single item measure for each of the vided on 7-point scales (1 = strongly disagree and 7 = following three behaviours (referred to as a category in strongly agree). the sample measurement wording below). The first behaviour was defined as increased purchasing of non- Risk perceptions. Risk perceptions for each behaviour perishable food items (e.g., pastas, rices, drinks, were measured using two items (e.g., “It would be risky canned, flour, sugar, frozen vegetables, pet food etc.). for me not to buy more non-perishable products”). The The second behaviour was defined as increased pur- scale was administered for each behaviour. Responses chasing of hygiene products (e.g., hand sanitiser, were provided on 7-point scales (1 = strongly disagree bleach, wipes, disinfectant, washing powder etc.). The and 7 = strongly agree). third behaviour was defined as increased purchasing of cleaning products (e.g., toilet paper, tissues, nap- Behavioural automaticity. Behavioural automaticity pies, nappy wipes, etc.). Participants were advised that was measured using the four-item behavioural auto- the questions would ask them to indicate the extent to maticity subscale (Gardner et al., 2012) of the Self- which they have bought more products than they Report Habit Index (Verplanken & Orbell, 2003). The would use based on their usual frequency of shopping, measure asks respondents to reflect on their agree- since the COVID-19 pandemic began this year, regard- ment with statements regarding their enactment of ing each category. Responses were provided on the behaviour automatically and without the need for 7-point scales (1 = “I have bought only the amount of conscious thought. The scale was administered for AUSTRALIAN JOURNAL OF PSYCHOLOGY 7 each behaviour. Responses were provided on 7-point measures inhibition of action or experiences (e.g., “I scales (1 = strongly disagree and 7 = strongly agree). must get away from all uncertain situations”). Higher scores are indicative of greater intolerance of uncer- Individual difference constructs tainty. Both subscales have been found to have simi- Anxiety sensitivity. The Anxiety Sensitivity Index-3 larly high internal consistency, α = .85 (Carleton et al., (ASI-3; Taylor et al., 2007) is an 18-item questionnaire 2007). The IUS-12 has also been found to be highly measuring fear of arousal-related sensations across correlated (r = .96) with the full version in two studies three empirically established subscales with 6 items with both student (Carleton et al., 2007) and clinical each. The subscales relate to physical (e.g., “when my samples (McEvoy & Mahoney, 2011). stomach is upset, I worry that I might be seriously ill”), cognitive (e.g., “when my thoughts seem to speed up, Self-control. The Brief Self-Control Scale (BSCS; I worry that I might be going crazy”, and social con- Tangney et al., 2004) was developed to assess disposi- cerns (e.g., “it is important for me not to appear ner- tional self-control across 13 items rated on a 5-point vous”). The scale utilises a 5-point Likert scales from 0 Likert scale from 1 (not at all) to 5 (very much). Due to (very little) to 4 (very much). Total scores are summed underlying concerns with the validity of its standard and range from 0 to 72 with higher scores indicating unidimensional structure, an updated approach to greater arousal-related sensation. The scale has scoring the Brief Self-Control Scale was utilised demonstrated excellent psychometric properties, with (Maloney et al., 2012). Maloney's et al. (2012) two- coefficient alpha across the subscales ranging from .73 factor model includes 4-items measuring restraint to .91 in cross-cultural norm groups (Taylor et al., 2007). (i.e., items 1, 2, 7, 8 from the original BSCS) and 4 items measuring impulsivity (i.e., items 5, 9, 12, 13 Distress tolerance. The Distress Tolerance Scale (DTS; from the original BSCS). Examples of subscale items Simons & Gaher, 2005) is a 15-item scale examining are “I am good at resisting temptation” and “I have ability to tolerate psychological distress rated on trouble concentrating”, respectively. Scores are aver- a 5-point Likert scale with response options ranging aged, with higher scores indicative of greater self- from 1 (strongly agree) to 5 (strongly disagree). The scale control. Hagger, Zhang, et al. (2021) found that the has four subscales: tolerance (3 items e.g., “I can’t two-factor model had the best psychometric proper- handle feeling distressed or upset”), appraisal (6 ties across four international samples. items e.g., “my feelings of distress or being upset scare me”), absorption (3 items e.g., “my feelings of distress are so intense that they completely take over”), Hoarding. The Hoarding Rating Scale (HRS; Tolin and regulation (3 items e.g., “I’ll do anything to avoid et al., 2010) is a brief self-administered scale that feeling distressed or upset”). Lower scores indicate assesses the features of compulsive hoarding, which a tendency to experience psychological distress as includes five questions covering clutter, difficulty dis- intolerable. Scores on the DTS have been shown to carding, acquisition, distress, and impairment. Each be negatively correlated with measures of negative question is measured on a 9-point Likert scale ranging affectivity and lability and positively correlated with from 0 (none) to 8 (extreme). When scores are averaged, measures of positive affectivity (Simons & Gaher, a score of 4 represents moderate symptoms. The HRS 2005). The scale has demonstrated test-retest stability correlates strongly with the interview version of the over six months (Simons & Gaher, 2005). scale and both scales have excellent psychometric properties (internal consistency, test-retest reliability, Intolerance of uncertainty. The Intolerance of and interrater reliability; Tolin et al., 2010). Uncertainty Scale, Short From (IUS-12; (Carleton et al., 2007) is a short-form of the original 27-item Intolerance of Uncertainty Scale (Buhr & Dugas, 2002; Perceived risks surrounding COVID-19 Freeston et al., 1994), rated on a 5-point Likert scale COVID-19 risk perceptions. A measure of COVID-19 from 1 (not at all characteristic of me) to 5 (entirely risk perceptions was developed based on the core characteristic of me), which measures reactions to components of risk perceptions identified by Brewer uncertainty, ambiguous situations, and the future. et al. (2007): risk likelihood, susceptibility, and severity The IUS-12 has two subscales. The 7-item prospective (e.g., “If I got COVID-19, there is a good chance that anxiety subscale measures anxiety in anticipation of I would have trouble”). Responses were provided on uncertainty (e.g., “I can’t stand being taken by sur- 7-point scales (1 = strongly disagree and 7 = strongly prise”), whereas the 5-item inhibitory anxiety subscale agree). 8 K. T. RUNE AND J. J. KEECH Data quality questions more . . . ”. For behavioural automaticity, residuals for Two questions were used to detect inattentive items 3 and 4 were allowed to covary. Items 3 and 4 responding (e.g., please select option two to ensure were conceptually similar in that they assessed the you are paying attention; Maniaci & Rogge, 2014; extent to which a behaviour is performed without Schroder et al., 2016). The 31 participants who did thinking or having to consciously remember. For not answer the two questions correctly were excluded COVID-19 risk perceptions, residuals for items 1 and 2 prior to data analysis. were allowed to covary. Items 1 and 2 were concep- tually similar in that they were the items within the scale assessing perceived susceptibility (as opposed to Data analysis perceived severity which was assessed by the other The three hypothesised models were evaluated using two items). For anxiety sensitivity (cognitive), residuals latent variable structural equation modelling in the were allowed to covary for items 2 and 5. Items 2 and 5 lavaan package (Rosseel, 2012) in R (R Core Team, were conceptually similar in that they both assess fear 2019). Due to the presence of multivariate skewness and worry around inability to keep one’s mind on (determined based on ratio of skewness to skewness a task. For hoarding ratings, residuals for items 4 and SE>3.29), the MLR estimator was used for all analyses. 5 were allowed to covary. Items 4 and 5 were concep- The MLR estimator is a maximum likelihood estimator tually similar in that they both assess the impact of that implements robust standard errors based on the hoarding on the individual. Residual covariances were Huber-White method (Rosseel, 2012). There were no applied consistently across the models for each of the missing data on behavioural outcome variables, how- three behaviours. The data file, analysis scripts, and ever, 5.3% of participants had a small amount of miss- analysis output are available on the Open Science ing data on the social cognition predictor variables. Framework: https://osf.io/cznj6/ Missing data were estimated using the full information maximum likelihood (FIML) procedure consistent with Results best practice (Enders & Bandalos, 2001; Enders, 2021). Goodness of fit of the hypothesised models was eval- Standardised path coefficients, standard errors, and uated using multiple criteria which compare the pro- 95% confidence intervals for each of the final models posed model to the baseline model. This included are presented in Table 2. Means, standard deviations, Tucker-Lewis index (TLI) and comparative fit index and bivariate correlations between study variables are (CFI), which should have values close to or exceeding presented in Supplementary Appendix D and E. Factor .95; standardised root mean squared residual (SRMR), loadings for each model are presented in which should have a value less than .08; and root mean Supplementary Appendix F. All factor loadings were squared error of approximation (RMSEA), which should satisfactory and statistically significant. The final models have a value less than .06 (Hu & Bentler, 1999). In including factor loadings, covariances, and path coeffi - addition, Bentler (1990) suggested that TLI and CFI cients are graphically depicted in Supplementary statistics between .90 and .95 are indicative of accep- Appendix G. table model fit. An initial model was estimated for each of the three behaviours with all predictors entered. Model 1 – predictors of increased purchasing of Because the default assumption of uncorrelated resi- non-perishable food items duals may be a source of misfit for similarly worded items (Brown, 2015), a subsequent final model for each Initial analysis of the hypothesised structure for the behaviour was estimated where residuals were model with increases in purchasing of non-perishable allowed to covary for similarly worded and concep- food items as the outcome variable yielded poor model tually similar items that were identified as a major fit, χ (2365) = 5493.98, p < .001, CFI =.89, TLI =.88, SRMR source of misfit by modification indices. For subjective =.05, RMSEA =.04 (90% CI [.04, .04]. Following the spe- norms, residuals for items 2 and 3, and items 4 and 5, cification of residual covariances as described above, were allowed to covary. Items 2 and 3 were compo- the final model exhibited a good fit to the data, χ nents of the scale assessing injunctive norms, and used (2359) = 4508.77, p < .001, CFI =.93, TLI =.92, SRMR similar wording, i.e., “Those people who are important =.05, RMSEA =.03 (90% CI [.03, .04]. Specifically, CFI to me would want me to/think that I should buy more . . and TLI values were close to .95, the SRMR value was . ”. Items 4 and 5 were components of the scale asses- below .08, and the RMSEA value was below .06, which sing descriptive norms, and used similar wording, i.e., together can be considered an indication of acceptable “people who are similar to me/like me would buy model fit based on Hu and Bentler’s (1999) guidelines. AUSTRALIAN JOURNAL OF PSYCHOLOGY 9 Table 2. Summary of standardised path coefficients, standard errors, and 95% confidence intervals for the final models for each behaviour. Model 1 - Non-Perishable Foods Model 2 - Hygiene Items Model 3 - Cleaning Items Effect β p SE 95% CI β p SE 95% CI β p SE 95% CI Attitude+ .30* <.001 .05 .19, .38 .29* <.001 .04 .19, .35 .24* <.001 .05 .13, .33 Subjective Norm+ .12* .024 .05 .02, .22 .07 .166 .05 −.03, .17 .03 .639 .05 −.08, .13 Risk Perception+ .32* <.001 .05 .21, .41 .36* <.001 .05 .25, .46 .39* <.001 .05 .29, .49 Behavioural Automaticity+ .03 .509 .04 −.05, .11 .02 .683 .04 −.07, .10 .02 .714 .05 −.07, .11 COVID Risk Perception .07* .044 .07 .00, .29 .02 .536 .08 −.11, .21 .04 .275 .08 −.07, .25 Intolerance of Uncertainty – P .04 .565 .14 −.20, .36 −.02 .842 .17 −.37, .30 .04 .636 .17 −.25, .41 Intolerance of Uncertainty – I .07 .493 .19 −.24, .51 .22 .052 .22 −.00, .85 .05 .659 .21 −.32, .50 Distress Tolerance .01 .829 .09 −.16, .20 .02 .644 .10 −.14, .23 −.05 .373 .10 −.28, .10 Anxiety Sensitivity – Physical −.05 .360 .11 −.32, .12 .05 .301 .11 −.11, .34 .04 .427 .12 −.13, .32 Anxiety Sensitivity – Cognitive .05 .319 .13 −.12, .38 .03 .573 .146 −.21, .37 −.04 .485 .15 −.40, .19 Anxiety Sensitivity – Social −.06 .324 .15 −.45, .15 −.14* .012 .16 −.73, −.09 −.02 .798 .18 −.40, .31 Self-control – Retraint .01 .835 .13 −.22, .27 −.02 .681 .131 −.31, .20 .05 .382 .13 −.14, .38 Self-control – Nonimpulsivity .02 .722 .15 −.23, .34 .12* .038 .05 .02, .61 .07 .224 .15 −.11, .49 Hoarding Rating −.09* .041 .05 −.20, −.00 −.04 .369 .05 −.16, .06 −.03 .476 .05 −.13, .06 Minutes to Supermarket .05* <.001 .00 .00, .00 −.03* <.001 .00 −.00, −.00 −.03* .002 .00 −.00, −.00 Household Size −.05 .068 .03 −.11, .00 −.03 .327 .04 −.10, .04 −.02 .582 .04 −.09, −.02 Income .03 .321 .02 −.11, .00 .01 .630 .02 −.03, .05 −.01 .687 .02 −.05, .03 Age .05 .083 .00 −.00, .01 .07* .020 .00 .00, .02 .06 .054 .00 −.00, .02 Gender .04 .215 .04 −.03, .12 .03 .364 .11 −.12, .33 .00 .990 .12 −.23, .23 + = referenced to each behaviour. Intolerance of Uncertainty – P = IUS-12 Prospective Anxiety Subscale; Intolerance of Uncertainty – I = IUS-12 Inhibitory Anxiety Subscale; * = statistically significant based on p < .05. The modifications resulted in a significant improvement specification of residual covariances as described to the model fit, χ (6) = 985.21, p < .001; however, infer- above, the final model exhibited a good fit to the ences based on the statistical significance of path esti- data, χ (2359) = 4511.35, p < .001, CFI =.93, TLI =.93, mates remained largely unchanged. The exception was SRMR =.05, RMSEA =.03 (90% CI [.03, .04]. Specifically, that subjective norms only significantly predicted beha- CFI and TLI values were close to .95, the SRMR value viour in the final model; however, and the effect size for was below .08, and the RMSEA value was below .06, subjective norms was consistently small across both which together can be considered an indication of models. The predictors in the final model accounted acceptable model fit based on Hu and Bentler’s for 49.8% of the variance in increased purchasing of (1999) guidelines. The modifications resulted in non-perishable food items. Attitudes and subjective a significant improvement to the model fit, χ (6) = norms regarding increased purchasing of non- 1265.42, p < .001; however, inferences based on the perishable food items, and risk perceptions regarding statistical significance of path estimates remained not increasing purchasing of non-perishable food unchanged. The predictors in the final model items, significantly predicted increased purchasing of accounted for 49.8% of the variance in increased pur- non-perishable food items products. Risk perceptions chasing of hygiene products. Attitudes towards regarding COVID-19 and the number of minutes it takes increased purchasing of hygiene products, risk percep- to get to the supermarket also significantly predicted tions regarding not increasing purchasing of hygiene increased purchasing of non-perishable food items. products, and the non-impulsivity facet of trait self- Finally, hoarding rating scores significantly negatively control, significantly predicted increased purchasing predicted increased purchasing of non-perishable food of hygiene products. Social anxiety sensitivity also items. It should be noted that effect sizes were small for negatively predicted increased purchasing of hygiene minutes to the supermarket and hoarding rating. No products. Age also significantly positively predicted, other variables significantly predicted increased pur- and the number of minutes it takes to get to the super- chasing of non-perishable food items. market significantly negatively predicted, increased purchasing of hygiene products; however, the effect sizes were small. No other variables significantly pre- Model 2 – predictors of increased purchasing of dicted increased purchasing of hygiene products. hygiene products Initial analysis of the hypothesised structure for the Model 3 – predictors of increased purchasing of model with increases in purchasing of hygiene pro- cleaning products ducts as the outcome variable yielded poor model fit, χ (2365) = 5776.77, p < .001, CFI =.89, TLI =.88, SRMR Initial analysis of the hypothesised structure for the =.05, RMSEA =.04 (90% CI [.04, .04]. Following the model with increases in purchasing of cleaning 10 K. T. RUNE AND J. J. KEECH products as the outcome variable yielded poor model attitudes towards increasing purchasing and risk per- fit, χ (2365) = 5450.61, p < .001, CFI =.90, TLI =.89, ceptions regarding not increasing purchasing, were the SRMR =.05, RMSEA =.04 (90% CI [.04, .04]. Following strongest of all predictors in each model. This is consis- the specification of residual covariances as described tent with past research (e.g., Lehberger et al., 2021; Roșu above, the final model exhibited a good fit to the data, et al., 2021), and suggests that social cognition con- χ (2359) = 4513.61, p < .001, CFI =.93, TLI =.92, SRMR structs may play an important role in increasing pur- =.05, RMSEA =.03 (90% CI [.03, .04]. Specifically, CFI and chasing behaviour during national crises. Additionally, TLI values were close to .95, the SRMR value was below risk perception regarding COVID-19, the number of .08, and the RMSEA value was below .06, which minutes it takes to get to the supermarket, and hoard- together can be considered an indication of acceptable ing rating were shown to predict purchasing of non- model fit based on Hu and Bentler’s (1999) guidelines. perishable food items; anxiety sensitivity, the non- The modifications resulted in a significant improve- impulsivity of trait self-control, age, and minutes to ment to the model fit, χ (6) = 937.00, p < .001; how- supermarket predicted purchasing of hygiene products; ever, inferences based on the statistical significance of and minutes to supermarket predicted purchasing of path estimates remained unchanged. The predictors in cleaning products, albeit with considerably smaller path the final model accounted for 41.5% of the variance in coefficients compared to the social cognition con- increased purchasing of cleaning products. Attitudes structs. Together, these findings suggest that social towards increased purchasing of cleaning products, cognition constructs may have played a more promi- and risk perceptions regarding not increasing purchas- nent role in influencing panic buying during COVID-19 ing of cleaning products significantly predicted lockdowns than psychological and individual difference increased purchasing of cleaning products. The num- factors. This is an important finding, given that social ber of minutes it takes to get to the supermarket also cognition factors are potentially modifiable through significantly negatively predicted increased purchasing public messaging and interventions. of cleaning products; however, the effect size was While the majority of the social cognition con- small. No other variables significantly predicted structs, and a small number of individual difference increased purchasing of cleaning products. factors and demographic factors, were supported as predictors of increased purchasing behaviours, several variables were not found to predict purchasing beha- Discussion viour. Specifically, demographic variables (i.e., house- The current study identified the psychological and hold size, income, age, and gender) did not individual difference factors associated with increased significantly predict any type of purchasing behaviour purchasing behaviours during COVID-19 lockdowns in in our study. This contrasts with findings by Bentall Australia. The study was informed by social cognition et al. (2021), who observed associations between theories including the theory of planned behaviour income and panic buying in the UK and Ireland. Our and dual-process models, which posit that constructs findings also contrast with Dinić and Bodroža (2021), such as attitudes, subjective norms, behavioural auto- who observed associations between household size maticity, and risk perceptions are key determinants of and panic buying in Serbia. This suggests that income behaviour. Campaigns targeting such constructs are and household size may have been less important in likely to be more effective for changing behaviours predicting panic buying in Australia than in other con- than atheoretical campaigns (McEachan et al., 2011). texts, and that the behaviour was occurring across the In addition, individual difference factors suggested as income spectrum and across household sizes. determinants of panic buying (Bentall et al., 2021; Therefore, public messages and campaigns aimed at Labad et al., 2021; Norberg et al., 2015; Yuen et al., decreasing panic buying during national crisis events 2020), and demographic factors were examined as would benefit from targeting the population at large, predictors of panic buying. rather than focussing on specific sub-groups. Consistent with our hypotheses, attitudes, and risk Additionally, several individual difference factors (e.g., perceptions were supported as predictors of increased hoarding, self-control, anxiety sensitivity, distress intol- purchasing of non-perishable food items, hygiene pro- erance, intolerance of uncertainty) suggested by pre- ducts, and cleaning products. Subjective norms also vious research to increase hoarding (Norberg et al., predicted increased purchasing of non-perishable food 2015) and panic buying (Bentall et al., 2021; Taylor items, although to a lesser extent. The path coefficients et al., 2020) and the social cognition constructs of for the social cognition constructs, and in particular behavioural automaticity and COVID-19 risk AUSTRALIAN JOURNAL OF PSYCHOLOGY 11 perception, were not found to be significant predictors relation to attitudes towards panic buying, an example of increased purchasing behaviour, or if significant, of a behaviour change method is the use of messaging displayed relatively small path coefficients. These find - that prompts individuals to shift their perspective (Kok ings indicate that individual difference factors pro- et al., 2016). This could include the use of public messa- posed by current research, and variables linked to ging and campaigns that prompt people to imagine hoarding, may not be applicable to increased purchas- themselves in the shoes of frontline line works or vul- ing behaviour during COVID-19 lockdowns. Similar to nerable persons unable to buy products that have been recent findings by Taylor (2021), our results suggest sold out due to panic buying. Second, regarding target- that when aiming to decrease panic buying at the time ing subjective norms in relation to panic buying, an of announcing lockdowns, focusing public messaging example of a behaviour change method is providing on variables such as self-control (e.g., “don’t panic buy” information about others’ approval (Kok et al., 2016). and “have some self-control”), distress, and anxiety This could include the use of public messaging that (e.g., “calm down”) are not likely to be effective as suggests that family, friends, and peers do not approve they do not target the theoretical constructs asso- of the behaviour, and would prefer that everyone plays ciated with the behaviour. Finally, despite having their part in not contributing to shortages by only buy- bivariate associations with the behaviours, the finding ing what they need. Third, an example of a behaviour that behavioural automaticity and COVID-19 risk per- change technique targeting risk perceptions in relation ceptions did not predict increased purchasing beha- to panic buying, is providing scenario-based risk infor- viour suggests that automatic and impulsive mation (Kok et al., 2016). This could include the use of purchasing may play a less prominent role in explain- public messaging indicating that supply chains which ing increased purchasing behaviour surrounding lock- support grocery stores across the country are strong down announcements compared to other constructs and that there is no risk of running out of stock if people in the model. The results of the study have yielded continue to only buy what they need. To date there a range of theoretical and practical implications. have been no published studies on the use of behaviour change methods in the context of panic buying. However, given the risk of future health-related crises, Theoretical and practical implications moving from formative research to experimental inter- The current study advances theory in two important vention research is an important future direction. ways. First, the study highlights that social cognition constructs were the most important correlates of the Strengths, limitations, and future directions behaviour in the context of panic buying. This is a valuable finding, as identification of social cognition The current study has several strengths that enhance our constructs can directly inform the development of understanding of the individual difference and social theory-based messages and interventions for chan- cognition constructs underlying increased purchasing ging behaviour (Hagger et al., 2020). Second, theory behaviours during COVID-19 lockdowns. First, the study is also advanced by our integration of social cognition applied an integrated theoretical approach grounded in constructs for explaining consumer behaviour during the theory of planned behaviour and dual-process mod- a novel public health crisis. This specifically includes els to explain increased purchasing behaviour. Second, reasoned processes from the theory of planned beha- the study recruited participants from a large national viour such as attitudes and subjective norms, non- sample, thus enhancing generalisability of the findings. conscious processes such as behavioural automaticity, Third, the use of structural equation modelling to analyse and affective processes such as risk perceptions. Such the data allowed for specification of latent variables and a theoretical approach can inform the prediction of the modelling of measurement error. behaviour, and the development of targeted messa- The findings of the current study must also be con- ging in the event of future COVID-19 lockdowns, or sidered in the light of some limitations. First, while the future novel public health crises. study used a large national sample, the study did not The current study also has several important implica- use a stratified sample and therefore cannot be tions for practice. In particular, the results have identi- assumed to be a true representation of the Australian fied three modifiable psychological processes that can population at large. Future research should also con- be targeted in public messaging and campaigns aiming sider examining panic buying across countries to deter- to reduce engagement in panic buying behaviour. This mine whether cultural factors are associated with the could be done by mapping behaviour change methods behaviour. Second, the study looked at increased pur- to social cognition constructs (Kok et al., 2016). First, in chasing of non-perishable, hygiene, and cleaning 12 K. T. RUNE AND J. J. KEECH ORCID products; however, these products are not exhaustive examples of products that have been purchased during Karina T. Rune http://orcid.org/0000-0002-6851-4644 COVID-19 lockdowns. In addition, while we selected Jacob J. Keech http://orcid.org/0000-0003-2504-9778 a range of individual difference factors to examine, future research could consider examining the associa- tion between further constructs such as personality Data availability statement factors and panic buying behaviour. Further, the study The data, analysis code, analysis output, and study materials are looked at only a specific time-point and did not account available at the Open Science Framework: https://osf.io/cznj6/. purchasing behaviours during prolonged lockdowns as have been seen in the Australia states of Victoria and New South Wales during 2021 and many European Open scholarship countries during 2020. Future research is needed to replicate the findings in response to living with COVID- 19 in the future and possible short local lockdowns as This article has earned the Center for Open Science badges more states and countries meet minimum target vacci- for Open Data and Open Materials through Open Practices nation rates. Finally, relying on a cross-sectional retro- Disclosure. The data and materials are openly accessible at spective design does not allow for causal inferences to https://osf.io/cznj6 be made. Future research should use experimental designs to evaluate the efficacy of a targeted interven- tion message aimed at changing underlying social cog- References nition constructs to determine a causal effect, which Ajzen, I. (1991). The theory of planned behavior. Organizational could support public messaging campaigns aimed at Behavior and Human Decision Processes, 50(2), 179–211. decreasing the behaviour and could inform media https://doi.org/10.1016/0749-5978(91)90020-T reporting guidelines. Ajzen, I. (2006a). Behavioral interventions based on the theory of planned behavior. Ajzen, I. (2006b). Constructing a theory of planned behavior questionnaire. Conclusion Arafat, S., Kar, S. K., Menon, V., Alradie-Mohamed, A., Mukherjee, S., Kaliamoorthy, C., & Kabir, R. (2020). The current study aimed to identify the individual Responsible factors of panic buying: An observation from difference factors and social cognition constructs online media reports. Frontiers in Public Health, 8, 747. underlying increased purchasing behaviour during https://doi.org/10.3389/fpubh.2020.603894 COVID-19 lockdowns in Australia. The study provides Armitage, C. J., & Conner, M. (2001). Efficacy of the theory of insight into the processes associated with panic buying planned behaviour: A meta-analytic review. British Journal of Social Psychology, 40(4), 471–499. https://doi.org/10. behaviours. Specifically, the results suggest that mod- 1348/014466601164939 ifiable social cognition constructs including attitudes, Bentall, R. P., Lloyd, A., Bennett, K., McKay, R., Mason, L., subjective norms, and risk perception were closely Murphy, J., McBride, O., Hartman, T. K., Gibson-Miller, J., & associated with increased panic buying during Levita, L. (2021). Pandemic buying: Testing a psychological COVID-19 lockdowns. It is recommended that these model of over-purchasing and panic buying using data from the United Kingdom and the republic of Ireland social cognition constructs be considered as targets during the early phase of the COVID-19 pandemic. PLoS when seeking to decrease future panic buying during ONE, 16(1), e0246339. https://doi.org/10.1371/journal. health-related crises and other national crises. Future pone.0246339 research should seek to replicate the findings during Bentler, P. M. (1990). Comparative fit indexes in structural panic buying events and investigate if interventions models. Psychological Bulletin, 107(2), 238–246. https://doi. targeting the constructs identified in this study result org/10.1037/0033-2909.107.2.238 Brewer, N. T., Chapman, G. B., Gibbons, F. X., Gerrard, M., in causal effects on intentions to panic buy and panic McCaul, K. D., & Weinstein, N. D. (2007). Meta-analysis of buying behaviour. Such findings will enhance our the relationship between risk perception and health beha- understanding of panic buying during pandemics vior: The example of vaccination. Health Psychology, 26(2), and may inform future public messaging campaigns. 136–145. https://doi.org/10.1037/0278-6133.26.2.136 Brown, T. A. (2015). Confirmatory factor analysis for applied research. Guilford Publications. Brown, D. J., Hagger, M. S., & Hamilton, K. (2020). The mediat- Disclosure statement ing role of constructs representing reasoned-action and No potential conflict of interest was reported by the author(s). automatic processes on the past behavior-future behavior AUSTRALIAN JOURNAL OF PSYCHOLOGY 13 relationship. Social Science & Medicine, 258, 113085. Hagger, M. S., Zhang, C. Q., Kangro, E. M., Ries, F., Wang, J. C., https://doi.org/10.1016/j.socscimed.2020.113085 Heritage, B., & Chan, D. K. (2021). Trait self-control and Buhr, K., & Dugas, M. J. (2002). The intolerance of uncertainty self-discipline: Structure, validity, and invariance across scale: Psychometric properties of the English version. national groups. Current Psychology, 40(3), 1015–1030. Behaviour Research and Therapy, 40(8), 931–945. https:// https://doi.org/10.1007/s12144-018-0021-6 doi.org/10.1016/S0005-7967(01)00092-4 Hamilton, K., Gibbs, I., Keech, J. J., & Hagger, M. S. (2020). Cameron, L., & Shah, M. (2015). Risk-taking behavior in the Reasoned and implicit processes in heavy episodic drink- wake of natural disasters. The Journal of Human Resources, ing: An integrated dual-process model. British Journal of 50(2), 484–515. https://doi.org/10.3368/jhr.50.2.484 Health Psychology, 25(1), 189–209. https://doi.org/10.1111/ Carleton, R. N., Norton, M. P. J., & Asmundson, G. J. (2007). bjhp.12401 Fearing the unknown: A short version of the intolerance of Hamilton, K., White, K. M., & Cuddihy, T. (2012). Using a single- uncertainty scale. Journal of Anxiety Disorders, 21(1), item physical activity measure to describe and validate 105–117. https://doi.org/10.1016/j.janxdis.2006.03.014 parents’ physical activity patterns. Research Quarterly for Dinić, B., & Bodroža, B. (2021). “My precious . . . toilet paper”: Exercise and Sport, 83(2), 340–345. https://doi.org/10.1080/ Stockpiling during the COVID-19 pandemic is related to 02701367.2012.10599865 selfishness, but not to fear. Primenjena Psihologija, 13(4), Herjanto, H., Amin, M., & Purington, E. F. (2021). Panic buying: 489–504. https://doi.org/10.19090/pp.20.4.489-504 The effect of thinking style and situational ambiguity. Dubey, S., Biswas, P., Ghosh, R., Chatterjee, S., Dubey, M. J., Journal of Retailing and Consumer Services, 60, 102455. Chatterjee, S., Lahiri, D., & Lavie, C. J. (2020). Psychosocial https://doi.org/10.1016/j.jretconser.2021.102455 impact of COVID-19. Diabetes & Metabolic Syndrome: Hori, M., & Iwamoto, K. (2014). The run on daily foods and Clinical Research & Reviews, 14(5), 779–788. https://doi. goods after the 2011 Tohoku earthquake: A fact finding org/10.1016/j.dsx.2020.05.035 analysis based on homescan data. The Japanese Political Enders, C. K. (2021). Applied missing data analysis. The Economy, 40(1), 69–113. https://doi.org/10.2753/JES2329- Guilford Press. 194X400103 Enders, C. K., & Bandalos, D. L. (2001). The relative perfor- Hu, L. T., & Bentler, P. M. (1999). Cutoff criteria for fit indexes mance of full information maximum likelihood estimation in covariance structure analysis: Conventional criteria ver- for missing data in structural equation models. Structural sus new alternatives. Structural Equation Modeling: Equation Modeling, 8(3), 430–457. https://doi.org/10.1207/ A Multidisciplinary Journal, 6(1), 1–55. https://doi.org/10. S15328007SEM0803_5 1080/10705519909540118 Evans, J. S. B., & Stanovich, K. E. (2013). Dual-process theories Keane, M., & Neal, T. (2021). Consumer panic in the COVID-19 of higher cognition: Advancing the debate. Perspectives on pandemic. Journal of Econometrics, 220(1), 86–105. https:// Psychological Science, 8(3), 223–241. https://doi.org/10. doi.org/10.1016/j.jeconom.2020.07.045 1177/1745691612460685 Kok, G., Gottlieb, N. H., Peters, G. J. Y., Mullen, P. D., Evidation. (2020). COVID-19 pulse: Delivering regular insights Parcel, G. S., Ruiter, R. A., Fernández, M. E., Markham, C., on the pandemic from a 150,000+ person connected cohort. & Bartholomew, L. K. (2016). A taxonomy of behaviour Freeston, M. H., Rhéaume, J., Letarte, H., Dugas, M. J., & change methods: An intervention mapping approach. Ladouceur, R. (1994). Why do people worry? Personality Health Psychology Review, 10(3), 297–312. https://doi.org/ and Individual Differences , 17(6), 791–802. https://doi.org/ 10.1080/17437199.2015.1077155 10.1016/0191-8869(94)90048-5 Labad, J., González-Rodríguez, A., Cobo, J., Puntí, J., & Frost, R. O., & Gross, R. C. (1993). The hoarding of possessions. Farré, J. M. (2021). A systematic review and realist synth- Behaviour Research and Therapy, 31(4), 367–381. https:// esis on toilet paper hoarding: COVID or not COVID, that is doi.org/10.1016/0005-7967(93)90094-B the question. PeerJ, 9, e10771. https://doi.org/10.7717/ Gardner, B., Abraham, C., Lally, P., & de Bruijn, G. J. (2012). peerj.10771 Towards parsimony in habit measurement: Testing the Lehberger, M., Kleih, A. K., & Sparke, K. (2021). Panic buying in convergent and predictive validity of an automaticity sub- times of coronavirus (COVID-19): Extending the theory of scale of the self-report habit index. The International planned behavior to understand the stockpiling of non- Journal of Behavioral Nutrition and Physical Activity, 9(1), perishable food in Germany. Appetite, 161, 105118. https:// 1–12. https://doi.org/10.1186/1479-5868-9-102 doi.org/10.1016/j.appet.2021.105118 Grisham, J. R., Roberts, L., Cerea, S., Isemann, S., Svehla, J., & Loxton, M., Truskett, R., Scarf, B., Sindone, L., Baldry, G., & Norberg, M. M. (2018). The role of distress tolerance, anxi- Zhao, Y. (2020). Consumer behaviour during crises: ety sensitivity, and intolerance of uncertainty in predicting Preliminary research on how coronavirus has manifested hoarding symptoms in a clinical sample. Psychiatry consumer panic buying, herd mentality, changing discre- Research, 267, 94–101. https://doi.org/10.1016/j.psychres. tionary spending and the role of the media in influencing 2018.05.084 behaviour. Journal of Risk and Financial Management, 13 Hagger, M. S., Cameron, L. D., Hamilton, K., Hankonen, N., & (8), 166. https://doi.org/10.3390/jrfm13080166 Lintunen, T. (Eds.). (2020). The handbook of behavior Maloney, P. W., Grawitch, M. J., & Barber, L. K. (2012). The change. Cambridge University Press. multi-factor structure of the brief self-control scale: Hagger, M. S., Smith, S. R., Keech, J. J., Moyers, S. A., & Discriminant validity of restraint and impulsivity. Journal Hamilton, K. (2021). Predicting physical distancing over of Research in Personality, 46(1), 111–115. https://doi.org/ time during COVID-19: Testing an integrated model. 10.1016/j.jrp.2011.10.001 Psychology & Health, 37(12), 1–21. https://doi.org/10. Maniaci, M. R., & Rogge, R. D. (2014). Caring about careless- 1080/08870446.2021.1968397 ness: Participant inattention and its effects on research. 14 K. T. RUNE AND J. J. KEECH Journal of Research in Personality, 48, 61–83. https://doi. Rosseel, Y. (2012). Lavaan: An R package for structural equa- org/10.1016/j.jrp.2013.09.008 tion modeling and more. Version 0.5–12 (BETA). Journal of McEachan, R. R. C., Conner, M., Taylor, N. J., & Lawton, R. J. Statistical Software, 48(2), 1–36. https://doi.org/10.18637/ jss.v048.i02 (2011). Prospective prediction of health-related beha- Roșu, M. M., Ianole-Călin, R., Dinescu, R., Bratu, A., viours with the theory of planned behaviour: A Papuc, R. M., & Cosma, A. (2021). Understanding consumer meta-analysis. Health Psychology Review, 5(2), 97–144. stockpiling during the COVID-19 outbreak through the https://doi.org/10.1080/17437199.2010.521684 theory of planned behavior. Mathematics, 9(16), 1950. McEvoy, P. M., & Mahoney, A. E. (2011). Achieving certainty https://doi.org/10.3390/math9161950 about the structure of intolerance of uncertainty in a Schroder, H. S., Dawood, S., Yalch, M. M., Donnellan, M. B., & treatment-seeking sample with anxiety and depression. Moser, J. S. (2016). Evaluating the domain specificity of Journal of Anxiety Disorders, 25(1), 112–122. https://doi. mental health–related mind-sets. Social Psychological and org/10.1016/j.janxdis.2010.08.010 Personality Science, 7(6), 508–520. https://doi.org/10.1177/ McNeish, D. (2018). Thanks coefficient alpha, we’ll take it from here. Psychological Methods, 23(3), 412. https://doi. Schwarzer, R. (2008). Modeling health behavior change: How org/10.1037/met0000144 to predict and modify the adoption and maintenance of Nicola, M., Alsafi, Z., Sohrabi, C., Kerwan, A., Al-Jabir, A., health behaviors. Applied Psychology, 57(1), 1–29. https:// Iosifidis, C., Agha, M., & Agha, R. (2020). The doi.org/10.1111/j.1464-0597.2007.00325.x socio-economic implications of the coronavirus pandemic Shou, B., Xiong, H., & Shen, X. (2013). Consumer panic buying (COVID-19): A review. International Journal of Surgery, 78, and quota policy under supply disruptions. City University of 185–193. https://doi.org/10.1016/j.ijsu.2020.04.018 Hong Kong. Norberg, M. M., Keyan, D., & Grisham, J. R. (2015). Mood Sim, K., Chan, Y. H., Chong, P. N., Chua, H. C., & Soon, S. W. influences the relationship between distress intolerance (2010). Psychosocial and coping responses within the and discarding. Journal of Obsessive-Compulsive and community health care setting towards a national out- Related Disorders, 6, 77–82. https://doi.org/10.1016/j. break of an infectious disease. Journal of Psychosomatic jocrd.2015.06.005 Research, 68(2), 195–202. https://doi.org/10.1016/j.jpsy O’connell, M., de Paula, Á., & Smith, K. (2020). Preparing for chores.2009.04.004 a pandemic: Spending dynamics and panic buying during Simons, J. S., & Gaher, R. M. (2005). The distress tolerance the COVID-19 first wave. Fiscal Studies, 42(2), 249–264. scale: Development and validation of a self-report https://doi.org/10.1111/1475-5890.12271 measure. Motivation and Emotion, 29(2), 83–102. https:// Orbell, S., & Sheeran, P. (1998). ‘Inclined abstainers’: doi.org/10.1007/s11031-005-7955-3 A problem for predicting health-related behaviour. British Strack, F., & Deutsch, R. (2004). Reflective and impulsive Journal of Social Psychology, 37(2), 151–165. https://doi. determinants of social behavior. Personality and Social org/10.1111/j.2044-8309.1998.tb01162.x Psychology Review, 8(3), 220–247. https://doi.org/10.1207/ Pantano, E., Pizzi, G., Scarpi, D., & Dennis, C. (2020). s15327957pspr0803_1 Competing during a pandemic? Retailers’ ups and downs Tangney, J. P., Baumeister, R. F., & Boone, A. L. (2004). High during the COVID-19 outbreak. Journal of Business self-control predicts good adjustment, less pathology, bet- Research, 116, 209–213. https://doi.org/10.1016/j.jbusres. ter grades, and interpersonal success. Journal of 2020.05.036 Personality, 72(2), 271–324. https://doi.org/10.1111/j. Peters, G. J. Y. (2017). Userfriendlyscience: Quantitative analysis 0022-3506.2004.00263.x made accessible. R package version 0.7. 0. R Foundation for Taylor, S. (2021). Understanding and managing Statistical Computing. pandemic-related panic buying. Journal of Anxiety Disorders, Phipps, D. J., Hagger, M. S., & Hamilton, K. (2021). Evidence 78, 102364. https://doi.org/10.1016/j.janxdis.2021.102364 that habit moderates the implicit belief-behavior relation- Taylor, S., Landry, C. A., Paluszek, M. M., Fergus, T. A., ship in health behaviors. International Journal of Behavioral McKay, D., & Asmundson, G. J. (2020). COVID stress syn- Medicine, 29(1), 1–6. https://doi.org/10.1007/s12529-021- drome: Concept, structure, and correlates. Depression and 09975-z Anxiety, 37(8), 706–714. https://doi.org/10.1002/da.23071 Prentice, C., Quach, S., & Thaichon, P. (2020). Antecedents Taylor, S., Zvolensky, M. J., Cox, B. J., Deacon, B., and consequences of panic buying: The case of COVID-19. Heimberg, R. G., Ledley, D. R., Abramowitz, J. S., International Journal of Consumer Studies, 46(1), 132–146. Holaway, R. M., Sandin, B., & Stewart, S. H. (2007). Robust https://doi.org/10.1111/ijcs.12649 dimensions of anxiety sensitivity: Development and initial Randolph, W., & Viswanath, K. (2004). Lessons learned from validation of the anxiety sensitivity index-3. Psychological public health mass media campaigns: Marketing health in Assessment, 19(2), 176–188. https://doi.org/10.1037/1040- a crowded media world. Annual Reviews of Public Health, 3590.19.2.176 25(1), 419–437. https://doi.org/10.1146/annurev.pub Tolin, D. F., Frost, R. O., & Steketee, G. (2010). A brief interview lhealth.25.101802.123046 for assessing compulsive hoarding: The hoarding rating R Core Team. (2019). R: A language and environment for scale-interview. Psychiatry Research, 178(1), 147–152. statistical computing. R Foundation for Statistical https://doi.org/10.1016/j.psychres.2009.05.001 Computing. https://www.R-project.org/ Verplanken, B., & Orbell, S. (2003). Reflections on past beha- Rosenstock, I. M. (1974). The health belief model and preventive vior: A self-report index of habit strength 1. Journal of health behavior. Health Education Monographs, 2(4), Applied Social Psychology, 33(6), 1313–1330. https://doi. 354–386. https://doi.org/10.1177/109019817400200405 org/10.1111/j.1559-1816.2003.tb01951.x AUSTRALIAN JOURNAL OF PSYCHOLOGY 15 Webb, T. L., Joseph, J., Yardley, L., & Michie, S. (2010). Using remarks-at-the-media-briefing-on-covid-19—2-march- the internet to promote health behavior change: 2020 A systematic review and meta-analysis of the impact of Yoon, J., Narasimhan, R., & Kim, M. K. (2018). Retailer’s sour- theoretical basis, use of behavior change techniques, and cing strategy under consumer stockpiling in anticipation mode of delivery on efficacy. Journal of Medical Internet of supply disruptions. International Journal of Production Research, 12(1), e4. https://doi.org/10.2196/jmir.1376 Research, 56(10), 3615–3635. https://doi.org/10.1080/ Weismuller, B., Schweda, A., Dorrie, N., Musche, V., Fink, M., 00207543.2017.1401748 Kohler, H., Skoda, E. M., Teufel, M., & Bauerle, A. (2020). Yoshizaki, H. T., de Brito Junior, I., Hino, C. M., Aguiar, L. L., & Different correlates of COVID-19 related adherent and Pinheiro, M. C. R. (2020). Relationship between panic buy- dysfunctional safety behavior. Frontiers in Public Health, ing and per capita income during COVID-19. Sustainability, 8, 625664. https://doi.org/10.3389/fpubh.2020.625664 12(23), 9968. https://doi.org/10.3390/su12239968 Woertz, E. (2010). The Gulf food import dependence and trade Yuen, K. F., Wang, X., Ma, F., & Li, K. X. (2020). The psycholo- restrictions of agro exporters in 2008. In S. J. Evenett (Ed.), Will gical causes of panic buying following a health crisis. stabilisation limit protectionism? The 4th GTA report: A focus on International Journal of Environmental Research and the Gulf region (pp. 43–56). Centre for Economic Policy Public Health, 17(10), 3513. https://doi.org/10.3390/ Research. https://www.files.ethz.ch/isn/113611/Global_ ijerph17103513 Trade_Alert_4th_report_3636.pdf#page=52 Zheng, R., Shou, B., & Yang, J. (2021). Supply disruption World Health Organization. (2020, March 8). World Health management under consumer panic buying and social Organization Media Release. https://www.who.int/director- learning effects. Omega, 101, 102238. https://doi.org/10. general/speeches/detail/who-director-general-s-opening- 1016/j.omega.2020.102238
Journal
Australian Journal of Psychology
– Taylor & Francis
Published: Dec 31, 2023
Keywords: COVID-19; behavioural triggers; hoarding
