Can COVID-19 Change the Big5 Personality Traits of HealthcareWorkers?

Akif AhmedAmerican International University

Bangladesh (AIUB)akif.ahmed5718@gmail.com

Md. Saddam Hossain MuktaUnited International University (UIU)

saddam944@gmail.com

Fardin MuntasirAmerican International University

Bangladesh (AIUB)fardinmuntasir@gmail.com

Shaguffta RahmanAmerican International University

Bangladesh (AIUB)shagufftarahman@gmail.com

A.K.M. Najmul IslamLUT University, Finlandnajmul.islam@lut.fi

Mohammed Eunus AliBangladesh University of Engineeringand Technology (BUET), Bangladesh

eunus@cse.buet.ac.bd

ABSTRACTTo curb the spread of the coronavirus, authorities around the worldimplemented lockdown measures for months. In these locked downdays, people as well as healthcare workers (HWs) are increasinglyrelying on social media platforms to socialize among themselvesand get the latest updates of this pandemic situation. As a result,social media users’ interaction data have become a key resource inanalyzing their behaviors. COVID-19 is the first global pandemicin the global scale after the appearance of social media. HWs suchas physicians, nurses, technicians, and hospital support staff pro-vide essential services to the patients with COVID-19 even afterhaving lack of sufficient equipment, poor preparation, and riskingtheir own lives. These HWs may experience sudden change in theirpersonality traits due to pandemic, which are not studied in theexisting literature. In this paper, we investigate how the pandemicinfluences Big5 personality traits of HWs using social media data.Our analysis shows that HWs’ neuroticism, conscientiousness, andextroversion have increased significantly due to pandemic. In con-trast, we observe a decrease of HWs’ agreeableness and opennessdimensions of Big5 personality.

CCS CONCEPTS• Human-centered computing → Social engineering (socialsciences).

KEYWORDSTwitter, Topics, Big5 Personality traits, COVID-19, Health Workers

ACM Reference Format:Akif Ahmed, Md. Saddam Hossain Mukta, Fardin Muntasir, Shaguffta Rah-man, A.K.M. Najmul Islam, and Mohammed Eunus Ali. 2020. Can COVID-19Change the Big5 Personality Traits of Healthcare Workers?. In 7th Inter-national Conference on Networking, Systems and Security (7th NSysS 2020),

Permission to make digital or hard copies of all or part of this work for personal orclassroom use is granted without fee provided that copies are not made or distributedfor profit or commercial advantage and that copies bear this notice and the full citationon the first page. Copyrights for components of this work owned by others than ACMmust be honored. Abstracting with credit is permitted. To copy otherwise, or republish,to post on servers or to redistribute to lists, requires prior specific permission and/or afee. Request permissions from permissions@acm.org.7th NSysS 2020, December 22–24, 2020, Dhaka, Bangladesh© 2020 Association for Computing Machinery.ACM ISBN 978-1-4503-8905-1/20/12. . . $15.00https://doi.org/10.1145/3428363.3428370

December 22–24, 2020, Dhaka, Bangladesh.ACM, NewYork, NY, USA, 6 pages.https://doi.org/10.1145/3428363.3428370

1 INTRODUCTIONTo combat the unprecedented global pandemic of COVID-19, thehealthcare workers (HWs) are fighting from the front-lines andproviding crucial services to the COVID-19 patients. Since theseHWs are also in a potential risk group of infecting others, they arealways in fear of infecting their family and friends. Consequently,in addition to posting different health-tips and advice, they oftenexpress their fears and frustrations 1, on social medias platforms,e.g., Twitter and Facebook, during COVID-19 crisis. Several stud-ies [14, 15] show that HWs experienced serious anxiety, depression,and distress in this crisis period. In our study, we investigate thelong-lasting effect of COVID-19 pandemic on HWs’ Big5 personalitytraits from their social media interactions.

Personality [13] is an important psychological attribute of hu-mans that greatly influences our decision making processes [24].Several socio-psychological studies [3, 25] show that personalitymight be changed due to different factors, i.e., event, technology [18],and pandemic [7]. HWs of COVID-19 are the front-line serviceproviders who remain isolated from their family for long time andmanage hundreds of bedridden patients and deceased bodies. Asa result, they are the potential risk group who may have experi-enced with the change of personality attributes. Due to the suddenshift on their personality traits, they may need immediate assis-tance for mental health rehabilitation. Moreover, the HWs, whoare less prone to the change of their personality traits may be rec-ommended to lead the health services due to their strong mentalhealth condition. Though after the advent of COVID-19, severalstudies [5, 12, 14, 15] have been conducted on investigating HWs’anxiety and depression, these studies largely ignore discoveringHWs’ change of personality traits in their studies, which is ourmain focus in this paper.

In the pandemic days, to conduct a survey asking psychologicalstate from the HWs is not a viable option as these people maynot show interests to share such data. Thus, to conduct our study,we resort social media as an alternative, yet a potential platformto inspect their psychological condition in these pandemic days.Several studies [9, 17, 20, 22] also suggest that social media can be a

1https://www.medscape.com/viewarticle/927861

12

7th NSysS 2020, December 22–24, 2020, Dhaka, Bangladesh Ahmed et al.

potential platform for delving into users’ Big5 traits, behavior andpreferences.

In this paper, first we identify 480 HWs who are active Twitterusers. Then, we collect the tweets of these users from their Twitterprofile. Later, we conduct psycholinguistic based analysis over thesetweets to understand users’ Big5 personality traits. Then, we con-duct topic modeling by using Latent Dirichlet Allocation (LDA) [4]to understand the major subject matters discussed by the Hws. Weanalyze HWs’ topic disseminating pattern over time by using Topicover time (TOT) [31] technique to understand the dynamics of thetopics during pandemic.

Next, we compute HWs’ personality traits by using IBM per-sonality insight API 2. Then, we statistically check their change ofpersonality traits by using paired T test [11] statistics. We find thatthe COVID-19 pandemic influences significant number of HWs’Big5 personality traits such as neuroticism, conscientiousness, andextraversion.

In summary, we have following contributions:• We build a dataset of Health Workers (HWs), i.e., doctors,nurses, management staff, etc. from Twitter.

• We extract users’ temporal emotional topics from the begin-ning of the pandemics by using LDA and TOT topic modelingtechniques.

• We also explore the changes of users’ Big5 personality traitsdue to the COVID-19 pandemic.

2 LITERATURE REVIEWSeveral studies [12, 14] find association between cognitive attributeand COVID-19. Lai et al. [14] conduct a survey among the physi-cians and nurses in hospitals in China. The study reveals that HWsserving COVID-19 patients have high rates of symptoms of de-pression, anxiety, insomnia, and distress. Tan et al. [28] observethat the pandemic not only made harm physically, but also left acatastrophic impact on HWs’ mental health. This study exposesHWs’ psychological fear can lead into result of depression, anxiety,and stigma. Lai et al. [12] conduct a study to highlight that HWsare very vulnerable group. They always give service with a strongfeeling of anxiety and depression. The doctors and nurses who aredirectly dealing with the active COVID-19 patients and deceased,are facing mental breakdowns and trauma. Rana et al. [23] conducta study on 1,257 HWs from 34 hospitals in China where they makea summary on the psychological condition of HWs. They observethat 50.4%, 44.6%, and 71.5% HWs are suffering from depression,anxiety, and insomnia, respectively.

Twitter has been a great source of communicating, transmitting,and disseminating disaster information including ebola [30], cy-clones [26], etc. Heverin et al. [10] show that analyzing tweets canbe of a great help during crisis moment to take central decision.Chew et al. [6] conduct a study on H1N1 pandemic 3 by analyzingtweets and find an instantaneous snapshot of the peoples’ opinionsand behavioral responses. Conducting survey on H1N1 by usingtraditional method is time consuming, labor intensive and expen-sive, thus analyzing tweets can leverage efforts of smooth strategicplanning. Dubey [8] extracts topics on the tweets of users from

2https://www.ibm.com/watson/services/personality-insights/3https://www.cdc.gov/flu/pandemic-resources/2009-h1n1-pandemic.html

France, china, Germany, and Italy and analyzes sentiment based on afew keywords such as virus, hospital, pandemic, political, quarantine,emergency, etc.

In our study, we analyze tweets of HWs and find insight abouttheir Big5 personality traits. We also investigate how a pandemicleft impact on HWs’ psychology. We inspect how the discussingtopics among the HWs change with the status of pandemic. Tothe best of our knowledge, our study is the first such study whichexplores the change of HWs’ personality where in current literatureresearchers largely investigate their depression and anxiety.

3 METHODOLOGYIn this section, we describe different steps of our methodology.

• Data Collection.We randomly search HWs’ profiles by usingTwitter advanced search technique.

• Topic Analysis.We investigate HWs’ posted tweets to extractimportant insight during the pandemic. We also explorehow the pattern of topics changed over time, Topic over time(TOT)[31], starting from the onset of pandemic to reachingits peak.

• Change of Big5 Personality traits. We investigate whetherHWs’ Big5 personality traits change during pandemic by us-ing IBM Personality insight API 4. We also test if these changesof Big5 personality traits are statistically significant [11].

4 DATA COLLECTION:We randomly search profiles of HWs by using Twitter’s advancedsearch technique. We search for Twitter users who write tweetsin English. We find users by words #healthworkers, #covid-fallen,#pandemic, corona virus, covid-19, #frontlines, and ppe. Then, wefind a number of HWs’ tweets. By using those tweets, we manuallyfind other HWs’ Twitter profile from their mentions. In this way,we find Twitter user_name of a total of 480 HWs. Then, we extractthe tweets of these users on July 20, 2020 by using tweepy 5 Pythonimplementation package. We find that these HWs are affiliated indifferent hospitals around the United States (US). Our collectedusers are HWs of a total of 23 different hospitals and health carecenters: Michigan Medicine, University of Utah Center for Drug Use& HIV, Northwestern Memorial Hospital, Boston Medical Center, TheMedical University of South Carolina, etc. Table 1 shows the statisticsof our data set.

5 DATA PREPROCESSINGIn this section, we describe the steps of our raw data pre-processingto conduct our analysis.We discard the user_name andmentions. Weremove hashtag, i.e., #COVID-19, and convert the text into COVID-19. We also remove urls and http links, because these text datacannot be analyzed by lexical method. We remove stop words byusing Python NLTK toolkit 6. We remove emoticon by using Pythondemoji implementation package 7. For example, we remove ‘:(’ bya text of ‘sad’. We also expand a shortened version of the writtenand spoken words, i.e., contractions. For example, contraction of

4https://personality-insights-demo.ng.bluemix.net/5https://www.tweepy.org/6https://www.nltk.org/7https://pypi.org/project/demoji/

13

Change of Big5 Traits of HWs during COVID-19 7th NSysS 2020, December 22–24, 2020, Dhaka, Bangladesh

Table 1: Statistics of our tweets

Number of HWs 480 Maximum Tweetsper HW 3,213

Numbers of Hospitals 23 Minimum tweetsper HW 14

Number of tweets 801,679Freq. of tweetsbefore pandemicper month

130

Average tweets per HW 2,166.7Freq. of tweetduring pandemicper month

217

‘I would’ is ‘I’d’. We write these expanded form by using pythonpycontractions implementation package 8. After performing theabove steps, users’ tweets become ready for further analysis.

6 EXPERIMENT AND RESULTSIn this section, we conduct our experiment to investigate the changeof HWs’ Big5 personality traits. We divide the sections in twodifferent parts: i) topic analysis of the HWs, and ii) understand thedynamics of Big5 personality scores, due to pandemics.

6.1 Topic analysisWe first find important topics that are shared by the HWs. To findtheir major topics, we run Latent Dirichlet Allocation (LDA) [4]technique. We also conduct a time-over-topic (TOT) [31] modelingover the tweets of the HWs in the months of February-April, 2020.The first COVID-19 patient was detected on February 29, 2020in the US. In our TOT model, we investigate the topics into twogroups: i) TOT before and during the pandemic, ii) TOT of the threeactive months (i.e., February, March, and April, 2020). Figures 1and 2 present the most used topics before and during the pandemic,respectively.

TOT of before and during pandemic: Before the pandemic situation,HWs used to do their regular jobs and post tweets about patients,new drugs, healthcare system, their regular life, etc. During thepandemic, they have been frequently tweeting about Coronavirus,Covid-19, health tips, to-dos during COVID-19, etc. We briefly de-scribe first three months’ updates of COVID-19 from the discussionof the HWs. Figure 3 presents the terms of 3 top topics in the firstthree months tweets of HWs.

• February 2020 (Figure 3-(d)): A small number of people havebeen affected with the virus in this month. HWs started totake into consideration that COVID-19 appears in US. HWstweeted related to the preparation of battle against COVID-19 such as icu management, tests, finding cases, etc.

• March 2020 (Figure 3-(e)): COVID-19 propagates in the com-munity level by the month of March. We find topics such asppe, healthcare, care, etc. in the tweets of HWs.

8https://pypi.org/project/pycontractions/

• April 2020 (Figure 3-(f)): COVID-19 now reaches at its peak.HWs are directly dealing with patients and deceased. HWstweet frequently with death, recovery, thank, etc. topics.

6.2 Change of Personality TraitsIn this subsection, we investigate how HWs’ Big5 personality traitschange by the influence of the pandemic. We observe that HWsincrease tweeting by 63.86% during pandemic than that of beforepandemic. To distinguish the dynamic nature of HWs’ personalitytraits [5, 25] due to pandemic, we divide each of the HWs’ tweetsin two different parts: before pandemic (i.e., before February 2020)and during pandemic (i.e., February 2020- April 2020). Then, wecompute Big5 personality traits from each part of HWs’ tweets.

We compute users’ Big5 personality traits by using IBM Watsonpersonality insights API 9. Arnoux et al. [2] show that IBM Person-ality Insight API outperforms the state of the art techniques forpersonality prediction. The underlying model of the API integratesword embedding features with Gaussian Process regression. TheAPI also requires 8 times fewer data than that of the previous ap-proaches. Several studies [16, 19] also used IBM Personality InsightAPI for personality inference and showed that such text based in-ferred personality can still predict real world behavior. The APIrequires at least 100 words to start analyzing. We observe that wecan compute personality by using the API for the users who haveon an average 12-15 tweets.

Significance test: After computing the Big5 scores of beforeand during COVID-19, we conduct Paired T Test [11] to check ifthe means of these two scores are significantly different by usingIBM SPSS. The results show that all five dimensions of Big5 havesignificantly been changed (see Table 2).

7 DISCUSSIONWe find that HWs’ discussed topics before and during pandemicreflect cue about the change of their behavior. Figure 1 shows thatHWs discuss about different topics before pandemic. For example,they discuss about affection related topics such as love, great us,thank, congratulations, best, amazing, etc. They also share their ca-reer building topics such as study, learn, know, etc. However, during

9https://www.ibm.com/watson/services/personality-insights/

14

7th NSysS 2020, December 22–24, 2020, Dhaka, Bangladesh Ahmed et al.

Figure 1: Topics among HWs before pandemic Figure 2: Topics among HWs during pandemic

Figure 3: Top 3 topics of the HWs before [a-c] and during [d-f] the pandemic.

Table 2: Results of Paired T Test on HWs’ Big5 personality

Big5 Traits Mean(Before COVID-19)

Mean(During COVID-19)

Mean Diff.(Significance)

Neuroticism 0.255 0.166 p=0.005Conscientiousness 0.151 0.101 p=0.002Extraversion 0.112 0.101 p=0.005Agreeableness 0.137 0.081 p=0.013Openness 0.143 0.099 p=0.001

pandemic (see Figure 2), HWs tend to focus on their efforts to com-bat against pandemic. They discuss awareness related topics suchas think, health, look, mask, know, ppe, use, etc. HWs also discusssympathetic and inspirational topics such as thank, agree, really,good, help, friends, love, amazing, etc. They also discuss terrifying

and frightening topics such case, horrible, pandemic, icu, death, etc.A recent study [1] also supports our findings related to HWs’ dis-cussed concerns, i.e., death, fear, stress, wearing masks, etc. Figure 3presents the TOT among the HWs’ discussed topics. Topics pre-sented in the Figures of 3-(a), (b), and (c) show HWs’ discussed

15

Change of Big5 Traits of HWs during COVID-19 7th NSysS 2020, December 22–24, 2020, Dhaka, Bangladesh

topics before the pandemic. Topics described in the Figures of 3-(d),(e), and (f) show the evolution of pandemic over three months,i.e., February, March, and April, 2020. Several studies [21, 27] alsodemonstrate that severity of a pandemic, i.e., severe cassava mosaic,H1N1, etc., primarily may increase over time.

Our results reveal that COVID-19 significantly change all theBig5 personality attributes of HWs. We think that some of thechanges might have severe negative impact on HWs’ mental health.We discuss the changes next. First, we find a significant increase inHWs’ neuroticism score. Higher neuroticism is often connected withmany negative psychological outcomes such as anxiety disorder,frustration, and depression, among others [29]. Therefore, our resultsupports the finding of prior studies [14, 15] that report the HWswho serve COVID-19 patients suffer from depression and anxiety.An example tweet supporting our finding is

"I know many folks are feeling anxious and worried. I just wantyou to know that you’re not alone.”

Second, we observe significant increase in conscientiousnessdimension due to COVID-19. Conscientiousness is related to beingcareful, self-disciplined and dependable [29]. Due to COVID-19, theHWs had to be more careful and kind in treating patients as wellas maintaining hygiene. Therefore, it makes sense that conscien-tiousness score of HWs is increased. An example tweet is

"Whatever you are feeling is a NORMAL HUMAN RESPONSETOANEXTRAORDINARYHUMANEVENT. Be extra kind to othersand yourself.”

Third, it is interesting to observe that HWs became extrovertsduring COVID-19 pandemic as their score is significantly increased.This can be explained by the fact that COVID-19 may have ledHWs to share information about the disease and their own workmore openly than ever before in order to raise awareness. Theybecame more vocal during the pandemic by sharing their feelingswith others.

Fourth, we observe a slight decrease in agreeableness dimensionof Big5 during the pandemic. Agreeableness is related to sympa-thetic, cooperative, warm and considerate [5]. Therefore, it is sur-prising to observe a decreased score on agreeableness. We thinka possible reason for this decrease is that although the HWs aresympathetic and cooperative with the patients, they may not befully satisfied with hospital supplies (e.g., protective equipment) orthe government preparation to fight the pandemic in general. Anexample tweet related to this is

"The shamefully inadequate amount of PPE threatens the veryhuman experience of medicine.”

Finally, we observe a decrease in openness dimension during thepandemic. High openness is related to seeking for new experience.As HWs are fighting COVID-19, they are more concerned aboutthe disease rather than any new experience.

8 CONCLUSIONIn this study, we have investigated the change of Big5 personalitytraits of the HWs. We have applied both static LDA and Time-over-Topic modeling in our study. We have noticed that HWs’ neuroti-cism, conscientiousness, and extroversion have increased signifi-cantly due to pandemic. In contrast, we observe a decrease of HWs’agreeableness and openness dimensions of Big5 personality. Our

research results reinforce the need of policies and relevant supportfor HWs’ mental health during the pandemic.

REFERENCES[1] Alaa Abd-Alrazaq, Dari Alhuwail, Mowafa Househ, Mounir Hamdi, and Zubair

Shah. 2020. Top concerns of tweeters during the COVID-19 pandemic: infoveil-lance study. Journal of medical Internet research 22, 4 (2020), e19016.

[2] Pierre-Hadrien Arnoux, Anbang Xu, Neil Boyette, Jalal Mahmud, Rama Akki-raju, and Vibha Sinha. 2017. 25 Tweets to Know You: A New Model to PredictPersonality with Social Media. arXiv preprint arXiv:1704.05513 (2017).

[3] Anat Bardi, Julie Anne Lee, Nadi Hofmann-Towfigh, and Geoffrey Soutar. 2009.The structure of intraindividual value change. Journal of personality and socialpsychology 97, 5 (2009), 913.

[4] DavidMBlei, Andrew YNg, andMichael I Jordan. 2003. Latent dirichlet allocation.JMLR 3, Jan (2003), 993–1022.

[5] Catherine Campbell and Flora Cornish. 2003. How has the HIV/AIDS pandemiccontributed to our understanding of behaviour change and health promotion.Learning from HIV and AIDS 15 (2003), 148.

[6] Cynthia Chew and Gunther Eysenbach. 2010. Pandemics in the age of Twitter:content analysis of Tweets during the 2009 H1N1 outbreak. PloS one 5, 11 (2010).

[7] Sara Y Del Valle, Susan M Mniszewski, and James M Hyman. 2013. Modeling theimpact of behavior changes on the spread of pandemic influenza. InModeling theinterplay between human behavior and the spread of infectious diseases. Springer,59–77.

[8] Akash Dutt Dubey. 2020. Twitter Sentiment Analysis during COVID19 Outbreak.Available at SSRN 3572023 (2020).

[9] Jennifer Golbeck, Cristina Robles, Michon Edmondson, and Karen Turner. 2011.Predicting personality from twitter. In PASSAT. IEEE, 149–156.

[10] Thomas Heverin and Lisl Zach. 2012. Use of microblogging for collective sense-making during violent crises: A study of three campus shootings. Journal of theAmerican Society for Information Science and Technology 63, 1 (2012), 34–47.

[11] Henry Hsu and Peter A Lachenbruch. 2005. Paired t test. Encyclopedia ofBiostatistics 6 (2005).

[12] RH Parlis J Lai. 2020. Every Clinician Is Also a Patient’: Study Shows PsychologicalImpact of COVID-19 Pandemic on Health Care Workers. https://bit.ly/3blJCCE.Last accessed 21 April 2020.

[13] Oliver P John. 1989. Towards a taxonomy of personality descriptors. In Personalitypsychology. Springer, 261–271.

[14] Jianbo Lai, Simeng Ma, Ying Wang, Zhongxiang Cai, Jianbo Hu, Ning Wei, JiangWu, Hui Du, Tingting Chen, Ruiting Li, et al. 2020. Factors associated with mentalhealth outcomes among health care workers exposed to coronavirus disease 2019.JAMA network open 3, 3 (2020), e203976–e203976.

[15] Guo Li, Jinfeng Miao, Hui Wang, Shabei Xu, Wenzhe Sun, Yebin Fan, ChenyanZhang, Suiqiang Zhu, Zhou Zhu, and Wei Wang. 2020. Psychological impacton women health workers involved in COVID-19 outbreak in Wuhan: a cross-sectional study. Journal of Neurology, Neurosurgery & Psychiatry (2020).

[16] Zhe Liu, Anbang Xu, Yi Wang, Jerald Schoudt, Jalal Mahmud, and Rama Akkiraju.2017. Does Personality Matter?: A Study of Personality and Situational Effectson Consumer Behavior. In Proceedings of the 28th ACM Conference on Hypertextand Social Media. ACM, 185–193.

[17] Md Saddam Hossain Mukta, Mohammed Eunus Ali, and Jalal Mahmud. 2016.Identifying and validating personality traits-based homophilies for an egocentricnetwork. Social Network Analysis and Mining 6, 1 (2016), 74.

[18] Md Saddam Hossain Mukta, Mohammed Eunus Ali, and Jalal Mahmud. 2019.Temporal modeling of basic human values from social network usage. Journal ofthe Association for Information Science and Technology 70, 2 (2019), 151–163.

[19] Md Saddam Hossain Mukta, Euna Mehnaz Khan, Mohammed Eunus Ali, andJalal Mahmud. 2017. Predicting movie genre preferences from personality andvalues of social media users. In Eleventh International AAAI Conference on Weband Social Media.

[20] Samia Nawshin, Md SaddamHossainMukta, Mohammed Eunus Ali, and AKMNa-jmul Islam. 2020. Modeling Weather-Aware Prediction of User Activities andFuture Visits. IEEE Access 8 (2020), 105127–105138.

[21] AM Presanis, RG Pebody, BJ Paterson, BDM Tom, PJ Birrell, A Charlett, MarcLipsitch, and D De Angelis. 2011. Changes in severity of 2009 pandemic A/H1N1influenza in England: a Bayesian evidence synthesis. Bmj 343 (2011), d5408.

[22] Md Mahabur Rahman, Md Taksir Hasan Majumder, Md Saddam Hossain Mukta,Mohammed Eunus Ali, and Jalal Mahmud. 2016. Can we predict eat-out prefer-ence of a person from tweets?. In Proceedings of the 8th ACM Conference on WebScience. 350–351.

[23] Waleed Rana, Sonia Mukhtar, and Shamim Mukhtar. 2020. Mental Health ofMedical Workers in Pakistan during the Pandemic COVID-19 Outbreak. AsianJournal of Psychiatry (2020).

[24] Antonio Rangel, Colin Camerer, and P Read Montague. 2008. A frameworkfor studying the neurobiology of value-based decision making. Nature ReviewsNeuroscience 9, 7 (2008), 545–556.

16

7th NSysS 2020, December 22–24, 2020, Dhaka, Bangladesh Ahmed et al.

[25] Brent W Roberts, Kate E Walton, and Wolfgang Viechtbauer. 2006. Patterns ofmean-level change in personality traits across the life course: a meta-analysis oflongitudinal studies. Psychological bulletin 132, 1 (2006), 1.

[26] Cheryll Ruth Soriano, Ma Divina Gracia Roldan, Charibeth Cheng, and NathanielOco. 2016. Social media and civic engagement during calamities: the case ofTwitter use during typhoon Yolanda. Philippine Political Science Journal 37, 1(2016), 6–25.

[27] AM Szyniszewska, C Busungu, SB Boni, R Shirima, H Bouwmeester, and JPLegg. 2017. Spatial analysis of temporal changes in the pandemic of severecassava mosaic disease in Northwestern Tanzania. Phytopathology 107, 10 (2017),1229–1242.

[28] Benjamin Y.Q. et al. Tan. 2020. Psychological Impact of the COVID-19 Pandemicon Health Care Workers in Singapore. Annals of Internal Medicine (04 2020).

[29] Edmund R Thompson. 2008. Development and validation of an internationalEnglish big-five mini-markers. Personality and individual differences 45, 6 (2008),542–548.

[30] Liza GG Van Lent, Hande Sungur, Florian A Kunneman, Bob Van De Velde, andEnny Das. 2017. Too far to care? Measuring public attention and fear for Ebolausing Twitter. Journal of medical Internet research 19, 6 (2017), e193.

[31] ChongWang, David Blei, and David Heckerman. 2012. Continuous time dynamictopic models. arXiv preprint arXiv:1206.3298 (2012).

17