Social Media Influencers as Message Source for Public Health Communication in

Health Emergencies/crisis like COVID-19

ABSTRACT

COVID-19 pandemic has put one and all under anomalous circumstances. Such situation calls

for a public health message design which reiterates the preventive recommendations in an

innovative and tailored manner for repetitive priming. Since, dynamics of Social Media

Influencers (SMIs) in Health and Fitness related information sharing are gaining progressive

attention, they can play an important role in communicating persuasive, customised and

region-specific public health messages. The aim of study was to identify key factors in

persuasive message design for shaping public's intention to follow the advice/information

when source of message is a Social Media Influencer. The theoretical underpinning of this

study come from Yale’s attitude change model and Theory of Reasoned Action. A sequential

Mixed-Method research approach was adopted for the study. Firstly, three focus group

discussions were conducted with Pre-Final and Final year Medical students. This set of

experts was identified to be closely related to health sector alongside being Second

Generation Digital Natives. The findings from this initial exploratory phase helped in

development of survey instrument for main phase of the research. Data were collected

digitally in the form of questionnaire from 332 respondents. The data analysis was done using

Confirmatory Factor Analysis and Structural Equation Modelling. An integrated model has

been presented that identifies how components of persuasive message and attitude towards

information shared by SMI antecedence an individual’s intention to follow advice. The

findings illustrated that the influence of SMI’s Credibility, Information Quality and

Homophily on intention to follow advice is direct as well as partially mediated by attitude.

Knowledge of proposed relationships and focus group recommendations can work as a guide
in designing a persuasive public health message for online communication in current and

similar times.

Keywords: Public Health Communication; Social Media Influencers; Persuasive Message;

Behavioural Intention; COVID-19 Pandemic

1. Introduction

COVID-19 pandemic has put the world in the most unprecedented situation. Scientists and

researchers globally are striving to find an effective cure for the viral disease. In order to

break the chain of rapid spread of the virus, lockdowns have been imposed in most of the

countries. While COVID-19 continues hammering the world, tenacious efforts are being put

to make people across the globe to cling to the CDC-recommended Non-pharmaceutical

interventions (NPIs). Various global, national and regional health campaigns have been

designed in order to increase public awareness about virus transmission and prevention (Olum

& Bongomin, 2020). Mediums of mass communication, including Social Media Platforms

have been employed for disseminating pandemic related updates (Chan et al., 2020).

Inclusion of Wireless media alongside traditional mass media for communicating NPI’s in

present situation can be accredited to "Shelter in Place” or “Stay at Home" orders.

Among many other lifestyle changes virus has brought, a notable change has been

observed in overall internet usage and use of internet for seeking health related information

(Wai, 2020). There are over 4.57 billion active internet users world wide and 3.96 active

social media users as of July, 2020 https://www.statista.com/statistics/617136/digital-

population-worldwide/). Engagements on posts have significantly increased, especially for

healthcare, consumer goods, and entertainment industries. (Wai, 2020

https://www.tributemedia.com/blog/how-covid-19-is-changing-social-media-usage-and-

platforms). A survey conducted in 2010 which included more than 12,000 Internet users
across 12 countries concluded that the people belonging to emerging economies such as India

(39%), Russia (39%), China (32%) and Brazil (29%) used online modes to seek health

information more frequently (McDaid & Park, 2011). Refer HC 3 Seow Ting Lee & Julian

Lin (2019). Deloitte (2016) mentions that millennials tend to obtain more information from

social media than from traditional media like TV and social media is an integral part of

people’s life.

In recent years, a lot of innovative ideas and strategies have been suggested and implemented

for Public Health Communication (PHC) over social media which address public as single

monolithic group (HC SM and PH 8).

There are evidences which support that social media platforms like Twitter and Facebook

have been used in past to spread information related to health emergencies like Swine Fue in

2009 https://academic.oup.com/eurpub/article/29/Supplement_4/ckz185.681/5624362

, Haiti Cholera outbreak in 2010, etc. These social platforms provided facility to engage with

public and spread awareness information. They also worked as myth-breakers and quick

source to correct any misinformation being spread. With these events, (HC SM and PH 3)

mention that in case of sudden outbreaks, social media could facilitate communication and

report public health related concerns faster and more reliably than traditional methods. On

social media platforms, a rising role of Social Media Influencers has been identified by

various researchers ( Borchers, 2019; Khamis, 2016; Lim et al.,2017. Social media influencers

(SMIs) can be seen as independent endorsers who have the ability to persuade and shape

attitude of their audience through their content (Freberg et al., 2010). Suitably of SMIs in

health and finiteness communication has been advocated by various researchers like Pilgrim

& Joschko (2019) (HC SMI2); Byrne et al., 2017(HC SMI 3). Andrews et al., 2020 (HC SMI

4). HC SMI4 mentions that influencers can help to sail through current times by encouraging

people, to abide by NPIs .

Hence, the purpose of this research was to explore the suitability of SMIs in communicating

public health messages in view of public's intention to follow the message. Specifically, the

study identifies key determinants of a persuasive message and verifies their impact on

individual’s intention to follow it. The mediating role of attitude towards a persuasive

message and their behavioural intentions has also been explored. This study contributed to

both practice and literature. First, this study fills the literature gap associated with SMIs in

healthcare sector. Second, an empirically tested model of persuasive communication on social

media platforms using SMIs has been presented. The model may find applicability in diverse

contexts. Third, a sequential mixed method approach has been used with can be adopted while

carrying out similar studies across disciplines. Lastly, it has direct practical implication in

current health crisis situation for designing persuasive health messages. Also, this study may

draw attention of practitioners towards SMIs as potential communicators. This work is

structured as follows: first, literature related to persuasive message design and theories which

support the argument have been presented. Second, the focus groups ideas and discussion

have been summarised to design the research instrument in for next step. Third, hypotheses

have been postulated and quantitative research methodology has been discussed. Finally,

results and conclusions are presented and some possibilities for future leads have been

outlined.

2. Theoretical background

2.1 Advances in Public Health Communication (PHC)

Health communication got a formal recognition in 1960s. Since then, medical science, health

communication, social sciences and communication technologies have evolved significantly.

PHC in today’s digital era, synchronises the modern study methods and new paradigms with
optimism for social impact. It tends to serve as a ray of hope for the mankind (Charles T.

Salmon & Thanomwong Poorisat, 2019). PHC became important to make public aware about

various diseases and to disseminate knowledge about preventive measures. The knowledge

dissemination was aspired to save lives and to create a a world which would be disease-free

(Tesh, 1988). A culture-centered approach to health communication is necessary to develop

strategies that are relevant for a specific community’s cultural framework (Airhihenbuwa,

1995). Some studies have been dedicated to create messages which are aimed at approaching

people with a suitable message design partnered with trusted local leaders. This could spread

the word in a localised manner for clear, relaxed and relatable communication (HC QP 4).

Recently a study by Lindsay Ems, Amy L Gonzales, 2015 contributed on usage of social

media at organizational level. The study suggested on how efficient, culture centered and

scalable communication strategies which are intended to improve health outcomes can be

designed by public health communicators. HC QP 5 ‘No one has really xed the model of

health communications for the 21st century,’ he said.” Refer HC QP 4

2.2 Designing Persuasive messages

Persuasion is a term which can be used in many contexts of influence. It can influence other

person's believes, attitude, intentions, motivation or even behaviour (Gass & Seiter, 2010).

Persuasion is one of the most broadly studied and researched areas in social psychology

(Fiske et al., 2010). The persuasiveness of a message in online context is said to possess to

qualities which are: message’s ability to be perceived as reliable information source by the

receivers and its capability of influencing receiver's attitude (Refer HC IAT 1 Nunes R.H. et

al (2018)).

Yale's attitude model describes the conditions which tend to change people’s attitude

when they come across a persuasive message. These conditions can be described as “who said
what to whom” where (i) who: the source of communication, (ii) what: nature of

communication and (iii) whom: the audience, all three are important for a persuasive

communication (Hovland, Janis, & Kelley, 1953). Various important features have been

associated to these conditions like: (i) Credibility (Hovland & Weiss, 1951; Kumkale &

Albarracín, 2004) and Attractiveness (Eagly & Chaiken, 1975; Petty et al., 1997) of the

message source. The messages seem more persuasive when they come from a credible source

(ii) Subtleness, Quality, Timing and Frequency of the communication (Miller & Campbell,

1959) have been stated to be important features which define the nature of communication

and medium used for sending the message and finally (iii) Attention (Festinger & Maccoby,

1964), Intelligence, Age (Krosnick & Alwin, 1989) and Self-esteem (Rhodes & Wood, 1992),

of the audience. It is easier to persuade people who pays attention, have lower intelligence,

moderate self-esteem and are younger in age (Rhodes & Wood, 1992). This model has lead to

enormous volumes of research into the nature of persuasion.

2.3 SMIs for Public Health Communication

(HC SMI 6) SMIs, can be seen as human brands on social media who have perceived

authenticity to attract other social media users and expand their social reach exponentially

(Kay et al., 2020). SMIs have been established their presence in various fields ranging from

advertising, paid partnerships, entertainment, infotainment, health and fitness communication,

lifestyle blogging, etc. Using SMIS for health communication is suggested and supported by

various studies. Pilgrim & Joschko (2019) studied influencers in context of health and fitness

communication. They said that Influencers on social media design their content in a way that

wins them the trust of their followers and their followers perceive them to be their friends

(HC SMI2). A study by Byrne et al., 2017 mentioned that Social Influencers, especially

qualified dieticians and nutritionists are the best source of credible and accurate information
on social media as they can motivate their followers to choose a healthier diet and stay fit (HC

SMI 3). Andrews et al., 2020 have suggested that Public health bodies should enrol and even

incentivise the social media influencers who have strong online presence to diffuse

information regarding social distancing and risk of COVID-19. They believe that loyal

adolescent audience of these influencers can be targeted in most naive manner through them

(HC SMI 4). Various Health experts and practitioners have suggested the necessity of pushing

localised and region specific messages though local opinion leaders in situation of crisis like

COVID-19 pandemic (HC SM and PH 8), HC QP3,). These views harmonise with the views

of focus group discussion outcomes (mentioned in Table 1)

2.4 Theory of Reasoned Action (TRA)

The TRA explains the relationship between attitude, subjective norms, behavioural intentions,

and actual behaviour of individuals (Fishbein & Ajzen, 1975; Madden et al.,1992; REFER

HC IAT 2 Erkan & Evans, 2016 ). Theory considers individual’s intention to be main

predictor of their actual behaviour. Attitude and subjective norms are considered as

antecedents of behavioural intentions. TRA and its extension: Theory of Planned behaviour

TPB (Ajzen, 1991 HC IAT 4) are the most widely accepted and used theories while predicting

the behavioural intentions. TPB postulates the same relationships as TRA with Intention

Behavioural Control (PBC) as an additional antecedent to Behavioural Intention.

Current study considers only two components of TRA, which are: attitude and

behavioural intention as used by Erkan &Evans (2006) for studying eWOM influence in

social media context. Intention has been selected instead or actual behaviour since present

context suggests a model for future applicability. None of the study so far has empirically

tested the the use of SMIs in communicating public health messages. The other antecedent of
intention i.e. subjective norms, which suggests the approval of other people or group in case

behaviour performed has been overlooked since, following an influencer or opinion leader

implies conformity to a group/community of individuals who think alike/share similar

interests. Therefore, only the two constructs have been borrowed and applied to understand

the public's attitude towards the health message communicated by SMI and to predict their

intention to follow the message in context of COVID-19. For conceptual framework, the

constructs were finalised and relationships were hypothesised only in accordance with the

focus group discussion outcomes highlighted in next section.

3. Study 1

Since, strategies for PHC on social media platforms, specially involving SMIs are at nascent

stage (rarely implemented), and next to nil in case of health crisis like COVID-19, it was

necessary to seek advice from various experts. The focus group studies were conducted in

pursuit of: understanding the current status of NPI adoption in public; evaluating existing

channels of PHC; obtaining views on using SMIs as information source and discovering

factors pertinent to shape public's intention to follow the NPIs.

3.1. Participants and Methodology

The participants of focus group were pre-final and final year students belonging to a medical

college in Jammu, India. This set of contributors was espoused firstly, as they are presumed

proficient of taking cognisance of current health crisis and secondly they are “Second

Generation Digital Natives”, since they were born after 1990, they were exposed to Web 2.0

while growing up (Helsper & Enyon, 2009). The prospects were invited over a WhatsApp

group for voluntary participation in the discussion. Due to regulative guidelines, physical

gathering size was restricted to four individuals (including one moderator), while others

participants and a second moderator joined via video conferencing to ensure descriptive
validity. Three focus groups were carried out in this fashion with a total of 16 participants.

The term of each gathering ranged from 55 to 70 mins in length. The discussion was guided

by a few questions appurtenant to the current health crisis including what PHC channels are

currently in place and their message design, views on adoption of NPIs, suggestions to boost

current strategies and ideas to implement new ones and finally views on inclusion of SMIs in

communication process. Following discussion points were literature-driven, including factors

affecting intention to follow NPI’s, attributes of a persuasive message and strategies for quick

and effective implementation, etc. While the main points were tapped upon, the moderator

had multiple opportunities to steer conversation towards emerging themes and capture

additional insights. The comments acquired from all three gatherings were noted and broken

down in light of existing literature accessible. Tables were created using Numbers 10.1

(spreadsheets) for each key point and responses of each participant were arranged. Since,

devising relevant themes was of paramount importance, each response was studied by two

authors independently to remove possible biases and ensure interpretative validity. The

themes derived are presented in Table 1.

[Table 1 near here]

3.2 Results of Study 1

As a result of analysis, three overarching attributes to design a persuasive message in case of

SMIs emerged: 1) Source related attributes and 2) Content related attributes and 3)

Relatibility attributes. More precisely, eight noteworthy sub-categories that appeared were:

Trust, Expertise, Bonding, Visual Aesthetics, Language, Engagement, Similarity and Attitude.

Among these sub-themes, trust, expertise, and engagement were categorised under SMIs’/

Source related attributes, while visual aesthetics and language were associated to Content

related attributes. Similarity and Bonding were categorised as relatibility attributes and
Attitude was kept as a separate construct since some members proposed the aforementioned

factors to be antecedents to attitude as well.

In addition to key attributes of a persuasive message, following excerpts illustrate inclusion of

Regional influencers & opinion leaders in PHC to be worth:

“influential person in a region like Sarpanch (head of a village) are being used to communicate

precautionary messages. They are considered trustworthy and their opinions are accepted by others.

This strategy has worked as an example in some regions of India.”

“Regional customisation of messages is necessary and local people who are famous online should be

contacted as everyone is indoor and online due lockdown…addressing mass gatherings online can

work in current scenario”

“ it is much easier for people to relate with people coming from same setting/ background. They seem

more relatable”

Lastly, the data analysis revealed that ways in which SMIs can present the message to their

audience, once designed and approved by the authorities:

“they are great story tellers so message shared that way stays in mind for longer”

“how-to videos which would guide people to understand the proper way of following NPIs like

wearing masks and hand washing / sanitising can be shared”

“interesting content, likes funny stories could be posted addressing how they are avoiding meeting

friends and relatives and making them understand the true meaning of social distancing”

4. STUDY 2

Integrating the findings of Study 1 with Yale’s attitude change model and TRA , Study 2 was

designed to present a model that depicts the the Public’s intention to follow COVID-19 related
Information/advice sent through a SMI. Rationales for each postulated hypothesis are

described below.

4.1 Hypotheses Development

4.1.1 Source Credibility, Information Quality & Homophily

The aforementioned theories and focus group outcomes suggest that the factors or conditions

which make a message persuasive would act as antecedents of Individual’s attitude and

behaviour, when source is a SMI. Out of three conditions suggested by Yale’s model, two

were accepted for studying the SMI’s without alteration i.e. the source of communication,

which can be measured by credibility and characteristic of the message, which would be

measured by information quality. The third component, nature of the audience was replaced

by Homophily. Homophily has been used to measure the similarity between information

source and message receiver (McCroskey et al. 1975). This was suggested by focus group,

since, audience qualities mentioned for mass communication by the Yale’s approach were

perceived to be unfit in case of SMI’s. Firstly, the audience in case of SMIs would be

addressed online and not in a physical gathering. So, it lacked important factor of association

with SMI, moreover, measuring audience’s intelligence or self-esteem (as suggested in Yale's

model) over survey was considered irrelevant in the present context.

Source credibility has been defined by various researchers as measure to assess the

communicator or message sent by them. It has been used to check competence and reliability

of a message (Sussman & Siegal, 2003; Cheung et al., 2008). (Refer HC IAT). A model by

McGuire’s (1985) mentions that effectiveness of message depends upon the source credibility

determinants: like attractiveness & likability of the source and familiarity & similarity with

source. (REFER HC IC 1) From the perspective of assessing a communicator, source

credibility has been used as a measure to judge & believe a communicator (O’Keefe, 1990)
(REFER HC IC 1). People tend to easily believe and accept messages received from sources

that they consider more credible (Cheung et al., 2009). (HC IAT 1). Since, the study focuses

on communicating COVID-19 related information and advice through SMIs. Credibility of

the SMI seems to be an important factor for creating a more acceptable and persuasive

message.

The quality of information is related to how message is perceived and evaluated. Refer

HC IAT 1 Nunes R.H. et al (2018). Based on this quality evaluation, the receiver can judge,

whether or not an information is suitable for them to accept and use (Wang & Strong, 1996).

Refer HC IAT. For processing an information or message, the quality of information/message

is one of the most examined factor. (Cheung et al., 2012; Hussain et al., 2017; Lederman et

al., 2014; Mun et al., 2013; Zhou et al., 2016 add name of REFER HC IC 1). In online

context, visually enriched and good quality content seem more appealing and acceptable

(Sohn, 2014);REFER HC IAT 2. For present study, the quality of information/ advice has

been measured in terms of understandability, clarity, overall perceived quality (Erkan &

Evans (2016), and language. check scale for more authors REFER HC IAT 2)

Lastly, Homophily represents the similarity between information source and the

receiver (McCroskey et al., 1975). (REFER HC IC 1) The idea behind Homophily is that

people prefer associating or grouping themselves with other people who seem similar

(McPherson, Smith-Lovin, & Cook, 2001) (HC HOMOP1) or who have similar interests as

them (Teng et al., 2014a). ( HC IAT 1). A study by Casalo et al., 2018 (HC SIT 2) which

evaluated the impact of Instagram influencers on customers found that, the congruence of

influencers and contents shared by them with consumer's thoughts and personality, generated

a psychological closeness between the two. Also, in advertising context, it is seen that

consumers positively assess an advertisement and have higher purchase intention when they

perceive a fit between self-image and that of the endorser (Choi & Rifon, 2012).HC SIT 2 So,
for the current study, homophily would be the similarity between SMIs and the message

receivers (general public).

4.1.2 Attitude and Intention

Attitude defines an individual's feelings towards a particular behaviour. It is evaluated to be

positive, negative or neural based on the assessment of outcomes to a favourable/unfavourable

(Fishbein & Ajzen, 1980; Dillon & Morris, 1996). Attitude is considered to be an important

variable in behavioural theories like Theory of Reasoned Action (Fishbein & Ajzen, 1980),

Theory of Planned Behavior (Ajzen, 1991), Technology Acceptance Model (Davis et al.,

1989), etc. REFER HC IAT 4. Several studies have examined consumer’s attitude in online

marketing context like e-Word-of-Mouth (Park et al., 2007; Prendergast et al., 2010). REFER

HC IAT 2. Attitude towards information was used in a study by Casaló et al.,2010 (HC IAT

4) to examine consumer’s intention to follow travel advice obtained through an online

community. Similarly, the present study investigates the attitude of an individual towards

COVID-19 related information/ advice received through SMIs.

The intention to follow is a significant behavioural intention as it predicts future

behaviour of an individual (McKnight et al., 2002). REFER HC IAT 4. Various researches

have investigated the influence of opinion leaders/ influencers/ celebrities on consumer's

intention to follow their advice (Thakur et al., 2016, Rahman et al., 2014). People tend to

follow advise of these influencers since they find them experienced and early adopters who

are capable of providing dependable advice/suggestions. In online context, Intention to follow

advice has been used in various studies like, McKnight et al. (2002) investigated antecedents

of the intention to follow e-vendor advice, and Casaló (2011) REFER HC IAT 4, studied

motivating factors which shape an individual’s intention to follow the advice obtained though

an online community. In present study, investigating the individual's intention to follow

COVID-19 related advice/information becomes a relevant construct which can closely predict

their future behaviour.

4.1.3 Persuasive message components, Attitude & Intention

The Yale’s attitude change model used for designing the persuasive message components in

for present study describes that an individual is likely to change his/her attitude in response to

a persuasive message (Hovland, Janis, & Kelley, 1953). A few studies which have

investigated each persuasive component in different areas are mentioned next. Source

credibility has been found to be an important antecedent of of attitude change (Ehrlich and

Cash, 1996; Ellis, 1992; Petty et al.,1981). The effect of information quality on attitude

formation has been tested in a study by Priyadarshini et al., (2017) which concluded that the

information quality dimensions positively impacted the attitude towards website. Hahn and

Lee (2014) HC SIT 2, conducted a study that indicated that consumer's psychological

homophily with fashion blogger positively impacted their attitude toward the blogger as well

as products promoted by them. So, for current study where source are SMIs and message are

COVID-19 related information /advice, it is postulated that:

H1: Source credibility has a positive impact on individual's attitude.

H2: Quality of information positively impacts individual's attitude.

H3: Homophily has a positive influence on individual's attitude.

A lot of studies have mentioned persuasive message components to have influence over

receiver's behavioural intentions (Kong & Goh (2015)HC ELM 1, Teng et al. (2014a) HC

IAT 1, Zhang et al. (2014) check HC HSM 2/ HC IC 1). Impact of credibility of source on

influencing consumers intention have been mentioned by researchers like Thakur et al. (2016)

and Rahman et al. (2014). Source credibility positively influenced users’ acceptance of

recommendations in case of information systems research (Mak and Lyytinen 1997, Mak et
al. 1997 REFER HC IAT 3). Similarly, quality of information has been identified as a

determinant of consumers' purchase intentions in social media context (Erkan & Evans (2016)

REFER HC IAT 2). A study by Amriel (2018) concluded information quality to have

significant influence on consumer’s purchase intentions in context of mobile advertising.

Lastly, Choi & Rifon (2012) HC SIT 2 found homophily to antedate consumer’s intention to

purchase. Congruent of opinion leaders ideas with those of followers was suggested to impact

the follower’s intentions & behaviour (Casaló et al., 2018 (HC SIT 2)). So, the following

hypotheses are proposed:

H4: Source credibility has a positive impact on individual’s intention to follow advice.

H5: Quality of Information has a positive influence on individual's intention to follow advice.

H6: Homophily positively influences individual's intention to follow advice.

As mentioned afore, various theories like TRA, TPB & Technology Acceptance Model

(TAM) indicate relationship between attitude and behavioural intentions. Attitude has been

used to explain online consumer behaviour Casaló et al.(2011) REFER HC IAT 4, predict

online purchase intentions (Hausman & Siekpe, 2009), use of Internet (Porter & Donthu,

2006), intention to use an information system (Taylor & Todd, 1995 REFER HC IAT 4). etc.

Hence, it is proposed that attitude towards COVID-19 related information/ advice shared by

SMIs will positively impact Individuals intention to follow:

H7: Individual's attitude positively influences their intention to follow advice.

4.1.4 Attitude as mediator

Ample studies have mentioned attitude as mediator between: perceived credibility and

purchase intention in context of product placement (Adis et al.); trust in eWOM source and

intention to follow in tourism industry (Zainal et al., 2017); ease of use and behavioural

intention (Sinha & Verma, 2017); cognitive dissonance and intention to repurchase (Ishak &
Harun, 2020); characteristics of social networking advertising and behavioural intention (Jung

et al., 2015). In case of source credibility, attitude was also found to mediate the relationship

between endorser’s credibility and online purchase (Samat et al., 2016). Relation between

endorsement strategy using celebrity and purchase intention was also mediated by brand

attitude (Vidyanata et al., 2018) and attitude towards advertising factors (Hasanah, 2019). In

reference to information quality, attitude variable was found to mediate consumer knowledge

influence on purchase intention of green product. (Wulandari et al., 2015). Similarly,

homophily was to play a mediating role in word-of-mouth behaviours among users of Social

networking sites (Farías2017). Therefore, following hypothesis have been proposed in context

of present study:

H8: Individual's attitude mediates the relation between Source credibility and Intention to

follow advice.

H9: Individual's attitude mediates the relation between Quality of Information and Intention

to follow advice.

H10: Individual's attitude mediates the relation between Homophily and Intention to follow

advice.

4.2 Methodology

Survey method based on an online questionnaire developed using Google Forms was used for

data collection from at PAN India. The questionnaire was floated for a limited time period

using convenience/ judgemental sampling technique due to restrictive movement measures.

Questionnaire was divided into three sections: Section 1, related to demographics, Section 2

starting with two screening questions: (i) that respondent has social media account on at least

one of the following platforms: Facebook/Instagram. (ii) that the respondent knows/ follows

at least on SMI. In introduction, the definition of SMIs and brief on NPIs was mentioned

along with several examples, so that the respondents could understand the context of study.
The two platforms Facebook and Instagram were recommended by the focus group as these

platforms have personal messaging facilities associated to them, unlike Youtube and also

word limit is not restricted, unlike Twitter. Section 2 further carried questioned related

publics’ general perception about SMIs and finally in Section 3, questions related to use of

SMIs as medium for communicating COVID-19 related information/advice were asked.

Questionnaires were distributed to 425 respondents from Jun 01-20, 2020, out of which only

332 fully complete questionnaires were considered for analysis after eliminating few based

on filter questions and rest of partially filled responses.

4.3 Measurements

All the constructs were measured using 7-point Likert multiple-item scale, ranging from

Strongly Disagree (1) to Strongly Agree (7). A total of 20 items adapted from the literature

were used for capturing 5 latent constructs. Precisely, five items measuring SMI’s credibility

have been adopted from Sokolova & Kefi (2019 REFER SMI 6 (1)), which was an adapted

and reduced version of scale by McCroskey and Teven (1999). To measure Quality of

message, 3 items were adapted from Erkan & Evans (2016, REFER HC IAT 2) and an

additional item was added on focus group’s which checked the language aspect as: “SMI(s) I

follow, post content in a language which is easy for me to understand”. Similarly, Homophily

was measured by adapting scale of McCroskey et al. (1975) us used by Xiao et al. (2018 HC

IC1) to measure follower’s homophily with YouTube SMIs. Three scale items used to

measure attitude towards the message were borrowed from Casaló et al. (2010 REFER HC

IAT 4). Finally, to measure intention to follow the message, 4 items were adapted from

McKnight et al. (2002). The overall items are reported in Table 3. To ensure then face

validity, the questionnaire was passed to three members from the focus group panel and three

members of community medicine studies and three members of marketing academics, who’s
comments on item comprehensibility, readability, wording and ambiguity were considered

while designing the final questionnaire (Hair et al., 2010).

4.4 Results

The respondents were nearly equally distributed with 51% of respondents being male while

49% were female. In terms of age the respondents were divided as per the generations

(Berkup, 2014; Jopling, 2004). Generation was described as cohorts or groups who had or

will have similar life experiences due to same age group (Hung & Gu & Yim, 2008). The five

generations considered are Generation Z, Generation Y (Millennials), Generation X, Boomers

and Silent. The respondents were also divided based on the amount of time they spend on

Social Networking Sites (SNS) and the SMI they prefer getting information. The average time

spent was 2 hours and 33 minutes. Most of the respondents expressed that they would prefer

getting information from a SMI they follow and who belong to their region/ state/ community

or whom they know personally. The details of the respondents are presented in Table 2.

4.4.1 Measurement Model Evaluation

The research model and proposed relationships were tested using IBM SPSS Amos 20.0. A

two-step Structural Equation Modelling technique, was adopted (Anderson & Gerbing, 1988).

The measurement model results using Confirmatory Factor Analysis (CFA) are displayed in

Table 3. The goodness-of-fit indices for tested model were satisfactory with the values: χ2 =

212.016, df=160, χ2 /df= 1.325, CFI=.988, TLI=.985, NFI= 0.952, GFI=0.942 and RMSEA=

0.031. The cutoff values being CFI, TLI, GFI, and NFI >.90; RMSEA <.08 (Bentler &

Bonnett, 1980). Two covariances were identified within the error terms. Since they belong to

same latent constructs, it can be attributed to close placement of these statements in the survey

tool or synonymous sentence structure as cautioned by Bollen & Lennox(1991). All factor
loadings (Table 3) were between 0.75 and 0.87 which are higher than the suggested threshold

value of 0.70 (Fornell and Larcker, 1981).

4.4.2 Reliability and Validity

Validation tests were conducted to check the convergent and discriminant validity. The value

of Average Variance Extracted (AVE) ranged from 0.617 to 0.699 (Table 4) for all constructs,

which are higher than the recommended minimum value, 0.50. This satisfies the convergent

validity criteria. For testing discriminant validity, Fornell and Larcker (1981)’s suggested

guidelines were used to check the correlations among the constructs against the square root of

AVE for each construct. It was achieved since, the value of square-rooted AVEs for all

constructs were higher than other correlation coefficients as shown in Table 4. Finally, the

composite reliability was tested . The values of composite reliability for all constructs were

between 0.854 to 0.903 (Table 4), which were above the cutoff value of .70, as suggested by

Fornell & Larcker (1981).

4.4.3 Structural Model Evaluation

[Figure 1 near here]

Figure 1 represents the Structural model and Model fit indices obtained using Maximum

Likelihood Estimation in AMOS. Overall model achieved a good fit with the model fit indices

in acceptance to the recommended values by Bentler & Bonnett (1980): Chi-Square (smaller

the better); P-Value > 0.05; GFI, AGFI, CFI> 0.9 & REMSA< 0.08). Hypothesised

relationships H1-H7 were tested and found to be statistically significant. The results of testing

have been shown in Table 5. More specifically, H1,H2 &H3 which represent the effect of

source credibility, information quality and homophily, appeared to have a positive significant

2
impact on individual’s attitude; (respectively) H1 (β =.45, R = .34, p < .001), H2 (β = .45,
 2 2
R = .23, p < .01) and H3 (β = .45, R = .23, p < .001). Further, individual’s intention to

2
follow was found to be positively influenced by Source credibility H4 (β = .67, R = .27, p < .

2 2
001), information quality H5 (β = .67, R = .19, p < .001) and homophily H6(β = .67, R = .

27, p < .001). Finally, the attitude towards the information/advice was found to positively

2
influence intention to follow H7(β = .67, R = .28, p < .001).

4.6. Mediating effect of Attitude

The study tests the influence of three independent variables namely, source credibility,

information quality and homophily (independent variable) on Individual’s intention to follow

(dependent variable) through Individual’s attitude towards the information/advice shared by

SMI (mediator). To examines the mediation: direct, indirect, and total effects for all three

variables have been calculated (Table 6). All indirect effects, have been tested using

bootstrapping procedures (2,000 bootstrap samples) and bias-corrected confidence level

(95%) (Hayes & Scharkow, 2013). Indirect effects for all three relationships appeared

statistically significant. In addition, all the paths with and without mediator stayed significant

and confidence level: upper and lower limit does not include 0 for all three constructs (Table

6), it is said that the Individual's attitude partially mediates the relationships. Overall

explained variance of Intention to follow increased from without mediation (β = 0.534) to

with mediation (β =0.578). Since, all conditions were fulfilled, it is concluded that

Individual's attitude partially mediates the effect of source credibility, information quality and

homophily on intention to follow, supporting H8, H9 and H10.

5. Discussion
The study tested the influence of SMI-sourced persuasive health messages on public’s

intention to follow. The persuasive components: source credibility, information quality and

homophily were tested to have positive influence over attitude and intentions. The direct

impact of source credibility and homophily on intention to follow was found to be statistically

more significant than of information quality. This finding coordinates with the Dual process

theories of persuasion (Chaiken, 1980; Petty & Cacioppo, 1986) which mention use of two

types/routes of persuasion i.e. either Systematic route or Peripheral route or both. In case of

influencers, when people given experts more priority than self-evaluation, the peripheral route

was found to be dominant in persuasion ( (Tang et al., 2012; Fu et al. (2018) refer HC HSM

2) Peripheral route includes factors other than those related to information itself. In present

context, these were credibility & homophily. All three persuasive message factors, were

found to significantly influence the individual’s attitude towards information shared by SMI.

However, credibility of the source, was found to be the leading factor in this relationship. This

matches with the recent studies that identify credibility to be key factor of follower's attitude

formation in case of influencers and online other opinion leaders (Pick, 2020; Singh &

Banerjee, 2018; Xiao et al., 2018; refer email ). The impact of attitude on behaviour was

found to be positively significant as mentioned in past theories like TRA, TBP and TAM and

recent studies as well (Casaló et al.,2011; Hausman & Siekpe, 2009); Porter & Donthu,

2006). The three independent factors i.e. source credibility, information quality and

homophily could explain the variance in individual's intention to follow more significantly

with attitude as mediator. All three relationships from independent to dependent factors were

partially mediated by attitude and devised a better model than those with direct effects. These

results are compatible with past studies which identified mediating role of attitude (Jung et al.,

2015; Samat et al., 2016; Sinha & Verma, 2017; Zainal et al., 2017).
 This study has both theoretical and practical implications. First, the study attempts to

fill the literature gap associated with SMIs in healthcare sector. Based on Yale’s attitude

model and Theory of reasoned actions, a model for communicating persuasive messages for

public health has been developed in accordance with expert suggestions. The model can be

used in further researches pertinent to use of SMIs for persuasive communication in various

sectors. Second, a sequential mixed method approach has been adopted in study with opens

doors for verifying the propositions in a more comprehensive manner. The results of first

study, qualitative in nature (using focus groups) served as basis for the second study, which

was quantitative in nature (questionnaire based survey). This can be viewed as an approach

where results of one study could be substantiated with another for more reliable inferences.

Finally, for practitioners, health officials, health organisations and other concerned

authorities, this study has a direct practical implication, were the suggested model can work as

a guide while designing persuasive health messages in social media context.

6. Conclusion

Today in the age of inter webs, social media platforms play a discernible role as mode for

mass communication. Researchers across various disciplines have vindicated potential of

these platforms to capture the interests of avid users with help of online opinion

leaders/influencers. Present study proposed a model and empirically verified these claims in

context of public health communication. The model recognises credibility, information

quality and homophily as significant components of a persuasive message which positively

affect individual's intention to follow it, when source is a SMI. A mediating role of

Individual’s attitude towards the message was also identified. The study was conducted in

view of current pandemic situation, to investigate the suitability of SMIs in communicating

COVID-19 related information/advice by assessing public's intention to follow. The results

revealed that people have a positive attitude and intention towards SMIs as source of health

related communication. These results were consistent with the focus group conclusions. The

focus group recommended the use of preferably local/ regional SMIs to communicate health

messages in native languages. They advocated that local influence can persuade public

stronger than any other aspect in crisis situation. Hence, it is concluded that SMIs rather

regional/ local SMIs can be adopted for health communication in current pandemic situation

as well as for future.

7. Limitations and Future Research

Irrespective of practical as well as theoretical implications, the study has a few limitations

which can be seen with a lens of future research leads . Though the present study has been

designed in guidance of experts and tested empirically, it is suggestive and predictive in

nature. Since, the SMIs have not really been used in context of public health communication

((HC SM and PH 3) Amy Burnett Heldman, MPH, 1 Jessica Schindelar, MPH,1 James B.

Weaver (HC SM and PH 4)), the model validates the publics intentions and not the actual

behaviour. Although behavioural intention is a strong predictor of the actual behaviour

(Fishbein & Ajzen, 1975; Madden et al.,1992; REFER HC IAT 2), future studies can be

experimental in nature. This could be done by considering few local SMIs and studying the

actual behaviour of the follows or people of that specific region. This would validate the

claims further. Secondly, though the results supported all the hypotheses and achieved a good

R2 level for endogenous variable, an interesting path to extend this work could be analysing

additional dimensions to create a persuasive message and consider more constructs which can

shape public’s intention. Another lead can be designing the model for PHC in view of dual

process models of persuasion like Elaboration Likelihood Model (ELM) and Heuristic-
Systematic Model (HSM) (Chaiken, 1980; Davis & Tuttle, 2013; Yang, 2015; Zha et al., 2018

HC ATT & PERSUASION). These models suggest two routes of persuasion: systematic and

peripheral. This would help in understanding receiver’s motivation and behaviour in response

to persuasion via chosen path(s). Finally, researchers like (HC QP 4; Duffett, 2017) have

mention the influence of SMIs more on generation X & generation Y. Future works can test

the results for different generations based on age groups (Hung & Gu & Yim, 2008). This

would help to predict the right audience for health communication in case of SMIs.

References

Figure 1 :
Table 1 : Categorise identified by Focus Group

Overarching
Sub
Categories Examples of Participants’ quotes
Categories
Identified
P 04 …trusting the information source.

P 08 ...they are considered trustworthy and their opinions matter to

Trust others.
P 13 …build the trust and connection with followers…

P 14 ...prominent personalities whom people of a particular region trust

and follow can use their social media accounts…

Source Related P 02 …Fitness experts like dieticians, sportsperson,

Attributes yoga/zumba/aerobics/pilates and other fitness trainers.

Expertise P 09 ...popularity & expertise can be essential factors.

P 13 …health and fitness micro-influencers who are experts…

P 15 …professionals share their expert views

P 01 …just putting information on website is not sufficient, engagement

is important.
P 03 ...health coaches who engage others in fitness activities…
 P 09 People who can engage with audience and open doors for
interaction…
Engagement P 14 …people with high engagement in social media setting

P 02 …message should be clear and presented in an interesting manner

P 05 …content shared by them should be attractive, interesting and new.

Visual
Aesthetic P 10 …digital content creators who create impactful and attractive
content.
Content P 12 …videos/ photos shared should be eye catching
Related
P 03 correct choice of words in description as well as spoken message
Attributes
P 07 ...language used should be easy to understand and preferably
Language regional
P 14 Regional customisation of messages.

P 16 ...healthcare workers who can share stories in their native language.

P 06 ...specially people from similar backgrounds or similar settings

Similarity P 10 ...influencer and follower fit would matter

P 15 …how much their interests align or how similar they are

Relatibility
Attributes P 09 ...feel that they share a bond with them

Bonding P 11 ... create a bond or a para social relationship

P 13 ...potential to form personal contact and persuade others

P 04 Positive attitude towards what is being advised.

P 07 …attitude towards the influencer/message alone won’t be enough. It

should be studied as attitude towards the message shared by the
influencer
Attitude P 10 …real stories can change anyone’s attitude.

P 11 …clicks people in a way that impacts their attitude and behaviour.

P 16 …if they agree to adopt it, will be mediated by their attitudes

Table 2

             
 Demographics Numbers Percentage     Numbers Percentage
Gender Age (years)
Male 170 51 Gen X: below 23 084 25
Female 162 49 Gen Y: 24-39 155 47
Time spent on
SNS Gen Z: 40-55 063 19
(hours) Boomers: 56-74 027 08
<1 025 08 Silent: above 75 003 01
1-2 104 31 Preferred SMI
2-3 115 35 Any SMI 014 04
3-4 067 20 SMI they follow 131 40
4-5 017 05 Follow & Know 154 46
>5 004 01 Don't prefer SMI 033 10

Table 4

           
SM
Constructs SL TV M SD
C
SMI's Credibility (CRED)
0.80 15.09 4.98 1.11 0.64
CRED1: I think SMIs are expert in their field.
5 5 8 6 8
0.78 14.60 5.08 1.05 0.61
CRED2: I find SMIs to be efficient in their job.
2 7 1 0 2
0.83 15.69 5.10 1.08 0.69
CRED3: I find SMIs to be Trustworthy.
3 0 8 6 4
0.80 15.04 4.95 1.03 0.64
CRED4: I think SMIs care about their followers.
2 3 2 0 4
CRED5: SMIs regularly update content on their 0.76 5.52 0.95 0.57
N/A
social media account. 0 1 7 8
Information Quality (IQ)
0.79 13.95 5.34 0.81 0.63
IQ1: SMI(s) post content which is understandable.
5 6 0 3 3
0.79 13.98 5.38 0.80 0.63
IQ2: SMI(s) post clear content.
7 5 9 6 5
IQ3: SMI(s) post content in a language which is 0.79 13.99 5.41 0.89 0.63
easy to understand. 8 9 0 0 7
1Q4: In general, content posted by SMI(s) is of 0.75 5.34 0.73 0.56
N/A
good quality. 0 0 0 2
Homophily (HOM)
0.82 18.51 5.22 0.98 0.67
HOM1: The SMI(s) I follow is like me.
1 8 6 0 5
0.81 18.29 5.29 0.90 0.66
HOM2: The SMI(s) I follow is similar to me.
5 3 5 2 4
0.77 17.00 5.10 0.91 0.60
HOM3: The SMI(s) I follow, thinks like me.
8 4 5 2 5
 0.87 5.37 0.98 0.76
N/A
HOM4: The SMI(s) I follow, behaves like me 4 7 3 3
Attitude (ATT)
ATT1: Following CRI/A obtained through SMI(s) 0.79 15.57 5.14 1.18 0.63
would be good for me. 7 8 2 4 6
ATT2: I think CRI/A obtained through SMI(s) 0.79 15.58 5.10 1.14 0.63
would be beneficial for me. 7 2 5 9 6
ATT3: I will have a positive opinion about the 0.83 5.13 1.11 0.70
N/A
CRI/A shared by SMI(s). 8 0 8 2
Intention to Follow Advice (IFA)
IFA1: I would feel comfortable in following the 0.86 19.37 5.09 1.05 0.75
CRI/A obtained through SMI(s). 8 6 0 6 3
IFA2: I would not hesitate to take into account the 0.79 17.03 5.05 1.03 0.63
CRI/A shared by SMI(s). 7 6 1 1 5
IFA3: I would feel secure in following the CRI/A 0.84 18.52 5.09 1.02 0.70
obtained through SMI(s). 2 2 6 1 9
IFA4: I would rely on recommendations made by 0.83 5.21 1.07 0.70
N/A
the SMI(s). 7 4 7 1

Table 5

               
Construct CR AVE CRED IQ HOM ATT IFA
CRED 0.897 0.635 0.797
IQ 0.865 0.617 0.497 0.785
HOM 0.893 0.677 0.675 0.293 0.823
ATT 0.854 0.657 0.615 0.467 0.535 0.811
IFA 0.903 0.699 0.723 0.539 0.658 0.685 0.836

Table 6

               
      SRW C.R. P Value Supported DE IDE TE
H1 CRED  ATT 0.339 3.898 *** Yes 0.339 NA 0.339
H2 IQ  ATT 0.238 3.705 0.002 Yes 0.238 NA 0.238
H3 HOM  ATT 0.233 3.053 *** Yes 0.233 NA 0.233
H4 CRED  IFA 0.271 3.803 *** Yes 0.271 0.096 0.367
H5 IQ  IFA 0.192 3.634 *** Yes 0.192 0.067 0.259
H6 HOM  IFA 0.27 4.325 *** Yes 0.27 0.066 0.335
H7 ATT  IFA 0.282 4.639 *** Yes 0.282 NA 0.282
Table 7

           

Independen Total Direct Indirect Mediation

IE: LB IE: UB
t Variables Effects Effects Effects Type
CRED .359*** .280*** .079*** 0.046 0.131 Partial
IQ .244*** .190*** .054*** 0.037 0.126 Partial
HOM .310*** .251*** .058*** 0.033 0.11 Partial