Question

Article

Narrowband Influencers and Global Icons: Universality and Media Compatibility in the Communication Patterns of Political Leaders Worldwide

Renana Peres, Sunali Talwar, Liav Alter, Michal Elhanan, and Yuval Friedmann

Abstract This article analyzes how political leaders communicate with their target audiences and examines whether they adopt a country- specific communication persona, or react to the global media-intensive environment by offering more universal communication. Politicians communicate through presentational (e.g., social media) and representational (e.g., press) outlets, and the compatibility between these outlets represents the leader’s effectiveness in transmitting the desired messages to the audience. The authors of this study suggest a theoretical framework that classifies public figures’ communication along two dimensions: universality (particular–universal) and media compatibility (low–high). The authors used language processing tools to study the sentiment, topic mixture, and use of pronouns by 61 global world leaders in more than 300,000 messages from the leaders’ Twitter accounts and press articles. The results show a high level of universality across political leaders in sentiment, topic mixture, and pronoun usage. The media compatibility is high, with Twitter being slightly more positive. Most leaders fall within the categories of Cosmopolitan Antagonist (high universality, low media compatibility) and Global Icon (high universality, high media compatibility). Overall, the sentiment of their communications is positive. Popular topics include diplomacy, economy, corruption, and the Arab world. No significant relationship was found between the sentiment or communication topics and country characteristics.

Keywords political marketing, social media, press, politicians, presentational media, representational media, universality, media compatibility, globalization, content analysis, topic analysis, LDA, sentiment analysis, communication of political leaders

Much of the brand building of political leaders is done through

communication with their audiences (Bennet and Iyengar

2008; Dahlgren 2009). The current age of global communica-

tion platforms, access to information, and global press cover-

age poses new challenges to political leaders who want to

design, develop, manage, and control their public persona.

We examine these challenges using two communication

dimensions. The first dimension is the level of universality.

On one hand, the political leader of a country can be perceived

as a salient, visible national icon. In branding terminology, a

country’s leader, at least while in office, can be viewed as a

distinct national brand or as a recognizable brand element of

the country, similar to the flag, major monuments, and famous

citizens (French and Smith 2010). A leader’s communication is

mainly with the local population. Therefore, one would expect

country leaders to adopt a country-specific communication

persona that reflects the interests and characteristics of their

country and people. On the other hand, many current topics of

interest are related to global issues. Moreover, politicians oper-

ate in a multichannel, media-intensive environment, which, as

suggested by recent trends in communication theory, is a strong

driver in enhancing globalization (Krotz 2007) and influences

public figures (Van Aelst et al. 2008) to take a more global

view of their media identity and adopt a more universal com-

munication style.

Renana Peres is an Associate Professor of Marketing, Hebrew University of

Jerusalem, Israel (email: [email protected]). Sunali Talwar is an

independent researcher working in Germany (email: sunalitalwar@gmail.

com). Liav Alter is a research student, Hebrew University of Jerusalem,

Israel (email: [email protected]). Michal Elhanan is a research student,

Hebrew University of Jerusalem, Israel (email: [email protected]).

Yuval Friedmann is a research student, Hebrew University of Jerusalem, Israel

(email: [email protected]).

Journal of International Marketing 2020, Vol. 28(1) 48-65

ª American Marketing Association 2020 Article reuse guidelines:

sagepub.com/journals-permissions DOI: 10.1177/1069031X19897893

journals.sagepub.com/home/jig

The other dimension is the extent to which the leaders

can communicate the identity they selected. Politicians com-

municate with their audiences through a mix of presentational

channels (e.g., social media), where the leader has the full

freedom of self-presentation, and representational channels

(e.g., press, television), which are based on journalism and are

supposed to be more objective (Hepp, Breiter, and Hasebrink

2018). High compatibility in information and sentiment

between these two types of channels implies that the represen-

tational media generally reflects the leader’s self-presentational

goals. Low media compatibility indicates a high level of inde-

pendence and investigative motivation by the press.

We organize these two dimensions into a 2 � 2 matrix (Figure 1) to suggest the following prototypical communication

patterns. In this article we focus on political leaders, but the

classification can apply to any public figure.

� The Peculiar Notorious communicates about specific topics that deviate from the average distribution of cur-

rent affairs. The media coverage of the leader is not

compatible with the leader’s own discussion of these

topics on social media. An example (from the nonpoli-

tical world) is Bill Cosby, who is connected to the spe-

cific subject of TV shows and has received negative

media coverage recently (i.e., large gap between his

representation and self-presentation).

� The Narrowband Influencer is interested in uncom- mon, specialized subjects, and the press coverage of

his/her ideas is highly compatible with his/her opinion.

Thus, the leader controls the message and leads public

opinion in the subject of interest. Examples are Kate

Middleton, whose media coverage deals mostly with

fashion and royal family activity and is generally posi-

tive, and Greta Thunberg, a climate activist whose

media coverage usually agrees with her opinions.

� The Cosmopolitan Antagonist talks about diverse topics resembling the topics generally discussed, but the

media’s view of these topics is incompatible with the

leader’s. Examples are Kim Jong Un, who as a head of

state speaks about common topics such as economy,

diplomacy, and military, but the media (outside North

Korea) is critical of his opinions and claims, and Rush

Limbaugh, who expresses strong right-wing opinions on

many topics and is not regarded positively by many

other media outlets.

� The Global Icon communicates about topics that are commonly discussed by others, and the media’s opinion

and perception of the leader are highly aligned with the

leader’s opinion. Examples are Michelle Obama, who as

the first lady of the United States advocated for poverty

awareness, education, and healthy lifestyles and

received positive coverage from the media, and Emma

Watson, an actress who is also a women’s rights activist

and was named a UN Women Goodwill Ambassador.

Studying these two dimensions is interesting for three rea-

sons. First, brand researchers have long wondered how globali-

zation processes affect brand perceptions (Cayla and Arnould

2008; Diamantopoulos et al. 2019; Gürhan-Canli, Sarıal-Abi,

and Hayran 2018; Steenkamp 2019). Because political leaders

are salient national brands, their globalization process might

provide early insights as to what one can expect for commercial

brands. Second, the citizens of the world’s nations are increas-

ingly facing challenges of a global nature, such as immigration,

environmental issues, and the future of the European Union and

other regional treaties. More than ever, these issues influence the

lives of individual citizens. Coping with these challenges

requires world leaders to create a communication infrastructure

that goes beyond national boundaries, and therefore it is of inter-

est to determine whether such infrastructure has already been

created. Third, one of the outcomes of intensive, multichannel

media usage is the continuous spillover of contents between

media platforms. Specifically, the presentational and repre-

sentational channels increasingly overlap. Still, politicians

often complain about media bias (Brants et al. 2010; D’Ales-

sio and Allen 2000; Geis 1987). Hence, studying the compat-

ibility in communication styles between presentational and

representational communication outlets is important to under-

stand the real variety and objectivity in the information that

citizens receive on their leaders (Shapiro and Hemphill 2017).

From an international branding perspective, political leaders

are an interesting case of personal branding in a media-

intensive environment. Similar to commercial brands, they

cope with the interplay of their local and international identi-

ties. However, they are people, and as personal brands, they use

communication in ways that are much more elaborate than the

communication methods of commercial brands. In addition,

their ability to communicate their identity depends not only

on their own marketing communications, but also strongly on

their press coverage. As personal branding gains an increasing

importance in art, entertainment, sports (Chadwick and Burton

Universality

Par�cular Universal

M ad

ia c

o m

p a�

b ili

ty

H ig

h Lo

w Peculiar Notorious e.g. Bill Cosby

Cosmopolitan Antagonist

e.g. Kim Jong Un, Rush Limbaugh

Narrowband Influencer

e.g. Kate Middleton, Greta Thunberg

Global icon e.g. Michelle Obama, Emma Watson

Figure 1. Four prototypical patterns of public communication.

Peres et al. 49

2008), high tech (Anderson 2013), and the organizational envi-

ronment (Ulrich and Smallwood 2007), insights on communi-

cation patterns are of importance.

The goal of this article is to provide a first data-intensive,

systematic, and comparative investigation of universality in

the communication styles of the world’s major political

leaders. We classify the leaders into the four quadrants of

the universality–compatibility matrix and ask how univers-

ality and media compatibility relate to the leader’s country

characteristics. To do so, we collected a massive data set of

the communications of 61 major political leaders. We

extracted the entire Twitter communication from their Twit-

ter accounts, and retrieved from the LexisNexis database the

press articles in which they were mentioned between Janu-

ary 2018 and July 2019. Thus, we captured the leader’s

communication in a presentational social media outlet

(Twitter) and in a representational media outlet (printed

press), which is supposed to be a more objective,

journalism-based medium. We used text processing tools,

such as translation, elimination of stop words, and cleaning

to preprocess the data, and we implemented automated lan-

guage processing tools to analyze the sentiment, topic mix-

ture, and usage of “I,” “we,” “you,” and “they” messages.

We tested the dependence of these elements on country

characteristics, such as the gross domestic product (GDP)

per capita, the continent, and Hofstede’s cultural dimensions

(Hofstede 2011).

Our results indicate a striking universality among political

leaders. Leaders are similar to each other in their sentiment,

topic mixture, and usage of pronouns. Overall, the sentiment is

positive, and the discussed topics are mainly diplomacy, econ-

omy, corruption, and the Arab world. “We” messages are used

the most, followed by “I” messages. The results are robust

across communication channels, with Twitter being slightly

more positive on sentiment score than the press articles. With

a few exceptions, we saw almost no significant dependence on

country characteristics. Relative to Africa, leaders from Asia,

Europe, and Central America use first-person messages more

than second- and third-person messages while tweeting. In

addition, leaders from countries with shorter-term orientation

were found to use first-person pronouns, relative to second- or

third-person pronouns, more often on Twitter than leaders from

countries with longer-term orientations do, but these effects are

small. Of the four quadrants on the universality–compatibility

matrix, most leaders were classified into the rightmost quad-

rants: Cosmopolitan Antagonists (topics similar to those of the

average leader, with relatively low media compatibility) and

Global Icons (topics similar to those of the average leader, with

high media compatibility).

The rest of the article is organized as follows. First, we

present the conceptual background. Then we describe the

data, explain the preprocessing, and describe the sentiment

and content analysis procedures. Finally, we discuss the

results of the sentiment, topic mixture, and pronoun fre-

quency analysis and present the leaders’ classification on

the communication matrix.

Conceptual Background

Politicians as Brands

Research in political marketing relates to political candidates

as one component in a trinity consisting of the party as the

brand, the politician as its tangible characteristics, and policy

as core service offerings (Henneberg and O’Shaughnessy

2007). Recently, voters’ consideration has shifted from focus-

ing on parties to focusing on the candidates (Guzman and

Sierra 2009), and, consequentially, researchers have started to

study political candidates as standalone brands (Guzman and

Sierra 2009; Hockett 2005; Phipps, Brace-Govan, and Jevons

2010; Smith 2009). This focus shift is in line with the increas-

ing interest in personal branding in other domains, such as art,

entertainment, social media, and high tech (Anderson 2013;

Chadwick and Burton 2008; Ulrich and Smallwood 2007).

The expansion and proliferation of communication outlets

in the past decades has created profound changes in the way

people, especially politicians, manage their personal brand.

The main concept in this discussion is “mediatization” (Coul-

dry and Hepp 2017). The term relates to the omnipresence of

multiple communication channels, which interact with each

other and collectively transform (i.e., “mediatize”) many

aspects of our everyday lives (Hjarvard 2013; Krotz and Hepp

2013). In this article, we discuss two challenges faced by per-

sonal brands in general, and political leaders in particular, in a

mediatized world.

Particularism Versus Universalism

The increased complexity of the communication landscape in

the past decades is both an outcome and a driver of globaliza-

tion. In one direction, globalization facilitates the creation of

global communication infrastructures and digital communica-

tion platforms. For example, the availability of internet infra-

structures, usage of the English language, and communalities in

consumer tastes facilitate the operation of social media chan-

nels, such as Instagram and Twitter. However, a recent stream

of research focuses on the other direction to argue that the

variety and ubiquity of communication channels is not only

an outcome but also an important driver of globalization. The

main argument of this research stream is that mediatization is a

core driver of globalization (Krotz 2007): the exposure to mul-

tiple communication channels helps connect people, creates

ties that cross national boundaries, and helps to create a global

culture.

Political leaders depend heavily on communication with

their citizens. Therefore, mediatization processes have deeply

affected the political arena (Enli and Skogerbø 2013; Howard

and Parks 2012). Marshall (2015), shows how people change

many of their everyday behaviors to fit with their online per-

sona; Strömbäck and Van Aelst (2013) demonstrate how polit-

ical parties adapt their structure and organization to respond to

the increasing complexity of communication; Van Aelst et al.

(2008) claim, using the case study of the 2003 election cam-

paign in Belgium, that the media tend to promote certain

50 Journal of International Marketing 28(1)

politicians, and therefore politicians adapt their actions and

personality to follow the “media logic” in order to receive a

higher level of media attention.

The fact that political leaders operate in a global, mediatized

world raises an intriguing question pertaining to their commu-

nication styles. On one hand, the political leader of a country is

a brand that is strongly associated with the country and is

assumed to have a strong country-of-origin effect. Also, much

of their target audience is local. Therefore, we can expect

strong local flavor in the leader’s communication and, conse-

quently, large differences in the communication styles of lead-

ers. On the other hand, mediatization calls for globalization: the

intensive usage of communication outlets of all types, together

with the global nature of many of these outlets and the exposure

to new audiences, might drive leaders to adopt a more global

communication persona and therefore employ a more homo-

genous communication style that depends less on local

characteristics.

Commercial brands also cope with the particularism–uni-

versalism balance. Brand researchers have intensively studied

how globalization influences brand perceptions (Cayla and

Arnould 2008; Diamantopoulos et al. 2019; Gürhan-Canli,

Sarıal-Abi, and Hayran 2018; Steenkamp 2019). This influence

is intricate. On one hand, brands use global positioning to

appeal to their consumers (Diamantopoulos et al. 2019; Nijssen

and Douglas 2011), and consumers develop global orientation

that influences their brand perception (Guo 2013). However, on

the other hand, the country of origin of the brand has still an

effect on attitudes, preferences, and purchasing decisions

related to the brand (Herz and Diamantopoulos 2017; Kock,

Josiassen, and Assaf 2019), and cultural dimensions and coun-

try characteristics affect consumer-brand interactions (Cayla

and Arnould 2008; Özsomer 2012).

This intensive research done for commercial brands has not

yet extended to the study of personal brands. Specifically,

research on communication in political science is mostly con-

ceptual (e.g., Krotz and Hepp 2013) or uses case studies (e.g.,

Enli and Skogerbø 2013), and researchers have not system-

atically monitored leaders’ communication to estimate the

level of universality. Personal brands differ from commercial

brands in, among other things, the complexity of their com-

munication with the target audience. Their usage of commu-

nication is much more elaborate than that of commercial

brands, and their ability to communicate their identity

depends not only on their own marketing communication but

also strongly on their press coverage.

Compatibility of Media Outlets

Political leaders often complain that their ability to communi-

cate their identity, goals, and ideas is oppressed by the press,

which they often claim is biased against them. Brants et al.

(2010) surveyed politicians in the Netherlands and found that

politicians regard journalists as too sensationalist, too event

driven in their coverage, and too focused on power struggles

rather than substance. In the United States in the 1980s,

American conservatives founded watchdog groups to focus

attention on what they perceived as biased reporting by the

American press (Geis 1987), although proof of such bias is

difficult to find (D’Alessio and Allen 2000).

One of the outcomes of mediatization is the constant inter-

action and spillovers across communication channels and

media environments (Hepp, Breiter, and Hasebrink 2018). Tra-

ditionally, presentational media, meaning communication ini-

tiated directly by the transmitting party (e.g., social media,

announcements, press releases), was distinct from representa-

tional media, defined as reporting done by a third party (e.g.,

television or press). Presentational media typically differs from

representational media in that it is more interactive, is inexpen-

sive, and is generated by ordinary users instead of professionals

(e.g., journalists; Hepp, Hjarvard, and Lundby 2015).

The massive growth of social media has reduced depen-

dency on the press for information (Shang, Wu, and Li 2017)

and has enabled ongoing direct communication between pub-

lic figures and their audiences. Realizing the opportunities

offered by new media settings, politicians across the world

have eagerly adopted social media (Bennett and Iyengar 2008;

Bos, Van der Brug, and De Vreese 2011; Wattal et al. 2010)

and are using a mix of presentational and representational

media platforms to communicate with their citizens and vot-

ers (Chadwick 2006).

In mediatized environments, presentational and representa-

tional media are becoming increasingly interwoven. For exam-

ple, Twitter affects political news reporting (Shapiro and

Hemphill 2017), and, at the same time, user-generated commu-

nicative patterns on social media may involve references to

news articles. Strömbäck and Aelst (2013) claim that social

media affects actual political behaviors, which, in turn, are

reported by the press media. Thus, the boundaries between

media environments are blurred. Hence, a question of interest

is the extent to which the two media types differ and whether

leaders are able to communicate their identity consistently

across presentational and representational outlets.

Integrating the Dimensions: Four Communication Types

This paper aims to explore how political leaders operate in a

mediatized word, with respect to the two dimensions discussed

previously. As described in Figure 1, public figures (in our

case, political leaders) can be classified on a 2 � 2 matrix in which the horizontal dimension is the level of similarity of the

person to the “average” public figure in the same domain (par-

ticular–universal) and the vertical dimension is the level of

compatibility between the presentational and representational

media concerning the leader (low–high).

In what follows, we describe the data collection and the

analysis.

Data

We collected data on the presentational and representational

communication of 61 political leaders from January 2018

Peres et al. 51

T a b

le 1 .

L e ad

e rs

’ D

at a.

N am

e C

o u n tr

y A

ge M

o n th

s in

P o si

ti o n

T w

it te

r F o llo

w e rs

(M ill

io n s)

T o ta

l T

w e e ts

Ja n u ar

y 2 0 1 8 –

Ju ly

2 0 1 9

T o ta

l P re

ss Ja

n u ar

y 2 0 1 8 –

Ju ly

2 0 1 9

G D

P P e r

C ap

it a

(U S D

) P o p u la

ti o n

D e m

o cr

ac y

P D

IN D

M A

S U

A L T

IN D

A b d e l F at

ta h

al -S

is i

E gy

p t

6 4

2 7 2

2 .6

1 2 ,5

0 0

3 ,7

7 4

2 ,5

4 9 .1

9 8 ,4

2 3 ,5

9 8

1 7 0

2 5

4 5

8 0

7 4

A b d u lla

h II

Jo rd

an 5 7

2 4 6

1 .5

2 1 1 5

2 ,4

2 0

4 ,2

4 7 .8

9 ,9

6 5 ,3

1 8

0 7 0

3 0

4 5

6 5

1 6

4 3

A b d u lla

h b in

Z ay

e d

A l N

ah ya

n U

n it e d

A ra

b E m

ir at

e s

4 7

1 6 2

4 .5

1 3 ,2

0 0

1 ,8

0 8

4 3 ,0

0 4 .9

9 ,6

3 0 ,9

5 9

0 —

9 0

2 5

5 0

8 0

— A

d e l al

-J u b e ir

S au

d i A

ra b ia

5 7

9 2 .8

3 4 4 7

3 ,1

8 4

2 3 ,2

1 9 .1

3 3 ,7

0 2 ,7

5 6

0 9 5

2 5

6 0

8 0

3 6

5 2

A le

k sa

n d ar

V u či

ć S e rb

ia 4 9

2 6

.2 9 9

2 ,5

7 1

2 ,4

4 4

7 ,2

3 4

8 ,8

0 2 ,7

5 4

1 8 6

2 5

4 3

9 2

5 2

2 8

A n d ré

s M

an u e l L ó p e z

O b ra

d o r

M e x ic

o 6 5

8 5 .9

1 3 ,2

4 2

1 ,8

4 8

9 ,6

9 8 .1

1 2 6 ,1

9 0 ,7

8 8

1 8 1

3 0

6 9

8 2

2 4

9 7

A ri

f A

lv i

P ak

is ta

n 6 9

1 1

2 .0

3 3 ,1

9 8

1 ,6

4 2

1 ,4

7 2 .9

2 1 2 ,2

2 8 ,2

8 6

1 5 5

1 4

5 0

7 0

5 0

0 B e n ja

m in

N e ta

n ya

h u

Is ra

e l

6 9

1 2 5

1 .6

6 3 ,2

0 2

1 ,6

7 0

4 1 ,6

1 4

8 ,3

8 1 ,5

1 6

1 1 3

5 4

4 7

8 1

3 8

— D

an ilo

M e d in

a D

o m

in ic

an R

e p u b lic

6 7

8 4

.6 9 4

3 ,2

3 8

3 6 5

7 ,6

5 0 .1

1 0 ,6

2 7 ,1

4 1

1 6 5

3 0

6 5

4 5

1 3

5 4

D m

it ry

M e d ve

d e v

R u ss

ia 5 3

8 7

4 .8

3 1 ,5

9 1

3 ,5

3 4

1 1 ,2

8 8 .9

1 4 5 ,7

3 4 ,0

3 8

1 9 3

3 9

3 6

9 5

8 1

2 0

D o n al

d T

ru m

p U

n it e d

S ta

te s

7 3

3 1

6 3 .4

3 ,1

9 4

1 ,6

4 6

6 2 ,6

4 1

3 2 7 ,0

9 6 ,2

6 5

1 4 0

9 1

6 2

4 6

2 6

6 8

D o n al

d T

u sk

E u ro

p e an

U n io

n 6 2

2 4 6

1 .0

7 5 4 3

1 ,6

3 0

3 6 ,5

3 1 .7

— 1

— —

— —

— —

E d ga

rs R

in k ē vi

čs L at

vi a

4 5

9 4

.0 0 2 4 5 5

3 ,2

4 8

9 2 6

1 8 ,0

8 8 .9

1 ,9

2 8 ,4

5 9

1 4 4

7 0

9 6 3

6 9

1 3

E m

m an

u e l M

ac ro

n F ra

n ce

4 1

2 7

4 .0

7 3 ,2

4 8

1 ,5

2 7

4 1 ,4

6 3 .6

6 4 ,9

9 0 ,5

1 1

1 6 8

7 1

4 3

8 6

6 3

4 8

E rn

e st

o A

ra ú jo

B ra

zi l

5 2

7 .3

6 6

1 2 4

9 8 6

8 ,9

2 0 .8

2 0 9 ,4

6 9 ,3

2 3

1 6 9

3 8

4 9

7 6

4 4

5 9

E vo

M o ra

le s

B o liv

ia 5 9

1 6 3

.5 7 2

3 ,2

4 0

3 ,0

4 0

3 ,5

4 8 .6

1 1 ,3

5 3 ,1

4 2

1 —

— —

— —

— G

iu se

p p e

C o n te

It al

y 5 5

1 4

.2 6 1

4 5 5

2 ,6

8 1

3 4 ,3

1 8 .4

6 0 ,6

2 7 ,2

9 1

1 5 0

7 6

7 0

7 5

6 1

3 0

H ill

ar y

C lin

to n

U n it e d

S ta

te s

7 1

— 2 5 .2

3 ,2

0 4

1 ,5

9 3

6 2 ,6

4 1

3 2 7 ,0

9 6 ,2

6 5

1 4 0

9 1

6 2

4 6

2 6

6 8

Im ra

n K

h an

P ak

is ta

n 6 6

1 2

1 0 .2

3 ,1

9 8

1 ,6

7 8

1 ,4

7 2 .9

2 1 2 ,2

2 8 ,2

8 6

1 5 5

1 4

5 0

7 0

5 0

0 Iv

án D

u q u e

C o lo

m b ia

4 3

1 3

.9 2 4

3 ,2

4 3

1 ,3

9 9

6 ,6

5 1 .3

4 9 ,6

6 1 ,0

4 8

1 6 7

1 3

6 4

8 0

1 3

8 3

Ja ir

B o ls

o n ar

o B ra

zi l

6 4

7 4 .9

3 3 ,1

7 6

1 ,8

7 0

8 ,9

2 0 .8

2 0 9 ,4

6 9 ,3

2 3

1 6 9

3 8

4 9

7 6

4 4

5 9

Ja va

d Z

ar if

Ir an

5 9

7 2

1 .2

2 6 0 4

1 9 0 0

5 6 2 7 .7

8 1 ,8

0 0 ,1

8 8

0 5 8

4 1

4 3

5 9

1 4

4 0

Jo k o

W id

o d o

In d o n e si

a 5 8

5 8

1 1 .9

1 ,7

9 3

3 ,2

0 5

3 ,8

9 3 .6

2 6 7 ,6

7 0 ,5

4 3

1 7 8

1 4

4 6

4 8

6 2

3 8

Jo rg

e A

rr e az

a M

o n ts

e rr

at V

e n e zu

e la

4 6

2 4

1 .6

3 ,1

8 9

1 ,4

8 8

1 6 ,0

5 4 .5

2 8 ,8

8 7 ,1

1 8

1 8 1

1 2

7 3

7 6

1 6

1 0 0

Ju an

O rl

an d o

H e rn

án d e z

H o n d u ra

s 5 0

6 7

.3 7 6

3 ,1

7 9

1 ,7

5 2

2 ,4

8 2 .7

9 ,5

8 7 ,5

2 2

1 8 0

2 0

4 0

5 0

— —

Ju st

in T

ru d e au

C an

ad a

4 7

4 5

4 .5

5 3 ,2

2 5

9 7 6

4 6 ,1

2 4 .7

3 7 ,0

7 4 ,5

6 2

1 3 9

8 0

5 2

4 8

3 6

6 8

K h al

id b in

A h m

e d

B ah

ra in

5 9

1 6 7

.5 2 5

3 ,1

9 0

1 ,3

2 5

2 4 ,0

5 0 .8

1 ,5

6 9 ,4

4 6

0 —

— —

— —

— K

o n o

T ar

o Ja

p an

5 6

2 4

.7 5 2

3 ,2

2 7

2 ,3

2 3

3 9 ,2

8 6 .7

1 2 7 ,2

0 2 ,1

9 2

1 5 4

4 6

9 5

9 2

8 8

4 2

L e e

N ak

-y e o n

S o u th

K o re

a 6 7

2 6

.1 2 8

3 ,2

0 2

1 ,4

4 9

3 1 ,3

6 2 .8

5 1 ,1

7 1 ,7

0 6

1 6 0

1 8

3 9

8 5

1 0 0

2 9

M ah

at h ir

M o h am

ad M

al ay

si a

9 4

1 5

1 .0

7 3 ,1

9 8

2 ,0

1 6

1 1 ,2

3 9

3 1 ,5

2 8 ,0

3 3

0 1 0 4

2 6

5 0

3 6

4 1

5 7

M ah

m o o d

Q u re

sh i

P ak

is ta

n 6 3

1 2

2 .6

2 1 ,5

2 7

1 ,3

4 8

1 ,4

7 2 .9

2 1 2 ,2

2 8 ,2

8 6

1 5 5

1 4

5 0

7 0

5 0

0 M

ar ce

lo E b ra

rd M

e x ic

o 5 9

8 1 .6

2 3 ,2

1 4

1 ,1

2 9

9 ,6

9 8 .1

9 7 ,3

3 8 ,6

5 7

1 8 1

3 0

6 9

8 2

2 4

9 7

M ar

t́ı n

V iz

ca rr

a P e ru

5 6

1 7

.6 0 7

1 ,6

4 2

1 ,4

0 1

6 ,9

4 7 .3

3 1 ,9

8 9 ,2

6 0

1 6 4

1 6

4 2

8 7

2 5

4 6

M au

ri ci

o M

ac ri

A rg

e n ti n a

6 0

4 4

4 .9

1 3 ,2

4 0

3 ,6

7 5

1 1 ,6

5 2 .6

4 4 ,3

6 1 ,1

5 0

1 4 9

4 6

5 6

8 6

2 0

6 2

M e vl

ü t

Ç av

u şo

ğl u

T u rk

e y

5 1

4 5

1 .3

5 3 ,2

2 8

1 ,9

1 3

9 ,3

1 1 .4

8 2 ,3

4 0 ,0

8 8

1 6 6

3 7

4 5

8 5

4 6

4 9

M ig

u e l D

ı́a z-

C an

e l

C u b a

5 9

1 6

.1 5 6

7 7 6

2 ,0

0 0

8 ,5

4 1 .2

1 1 ,3

3 8 ,1

3 4

0 —

— —

— —

— M

ik e

P o m

p e o

U n it e d

S ta

te s

5 5

1 6

.7 6 6

9 4 2

1 ,6

8 7

6 2 ,6

4 1

3 2 7 ,0

9 6 ,2

6 5

1 4 0

9 1

6 2

4 6

2 6

6 8

M o h am

m e d

A l M

ak to

u m

U n it e d

A ra

b E m

ir at

e s

7 0

1 6 2

9 .7

7 3 ,2

3 9

1 ,8

9 7

4 3 ,0

0 4 .9

9 ,6

3 0 ,9

5 9

0 —

9 0

2 5

5 0

8 0

— M

o h am

m e d

b in

Z ay

e d

A l

N ah

ya n

U n it e d

A ra

b E m

ir at

e s

5 8

1 7 7

2 .5

6 3 ,2

0 9

4 ,1

3 7

4 3 ,0

0 4 .9

9 ,6

3 0 ,9

5 9

0 —

9 0

2 5

5 0

8 0

—

M o o n

Ja e -i n

S o u th

K o re

a 6 6

2 7

1 .7

7 3 ,0

7 4

1 ,7

3 8

3 1 ,3

6 2 .8

5 1 ,1

7 1 ,7

0 6

1 6 0

1 8

3 9

8 5

1 0 0

2 9

(c o n ti n u ed

)

52

T a b

le 1 .

(c o n ti n u e d )

N am

e C

o u n tr

y A

ge M

o n th

s in

P o si

ti o n

T w

it te

r F o llo

w e rs

(M ill

io n s)

T o ta

l T

w e e ts

Ja n u ar

y 2 0 1 8 –

Ju ly

2 0 1 9

T o ta

l P re

ss Ja

n u ar

y 2 0 1 8 –

Ju ly

2 0 1 9

G D

P P e r

C ap

it a

(U S D

) P o p u la

ti o n

D e m

o cr

ac y

P D

IN D

M A

S U

A L T

IN D

M u h am

m ad

u B u h ar

i N

ig e ri

a 7 6

5 1

2 .2

2 3 ,2

3 7

1 ,9

5 3

2 ,0

2 8 .2

1 9 5 ,8

7 4 ,6

8 3

1 8 0

3 0

6 0

5 5

1 3

8 4

N ar

e n d ra

M o d i

In d ia

6 8

6 3

4 9 .5

3 ,2

1 4

1 ,9

0 0

2 ,0

1 5 .6

1 ,3

5 2 ,6

4 2 ,2

8 0

1 7 7

4 8

5 6

4 0

5 1

2 6

N ic

o lá

s M

ad u ro

V e n e zu

e la

5 6

7 6

3 .6

5 3 ,2

2 4

1 ,6

3 7

1 6 ,0

5 4 .5

2 8 ,8

8 7 ,1

1 8

1 8 1

1 2

7 3

7 6

1 6

1 0 0

P au

l K

ag am

e R

w an

d a

6 1

2 3 2

1 .5

1 2 ,7

8 6

3 ,3

2 0

7 7 3

1 2 ,3

0 1 ,9

7 0

1 —

— —

— 1 8

3 7

P e d ro

S án

ch e z

S p ai

n 4 7

1 4

1 .0

7 3 ,2

4 8

9 8 9

3 0 ,5

2 3 .9

4 6 ,6

9 2 ,8

5 8

1 5 7

5 1

4 2

8 6

4 8

4 4

P e tr

o P o ro

sh e n k o

U k ra

in e

5 3

5 8

1 .2

3 ,2

4 8

1 ,8

2 5

3 ,0

9 5 .2

4 4 ,2

4 6 ,1

5 6

1 9 2

2 5

2 7

9 5

5 5

1 5

P o p e

F ra

n ci

s V

at ic

an 8 2

7 7

1 8 .1

2 ,0

0 7

1 ,7

6 7

— 8 0 1

0 —

— —

— —

— R

an ia

A l A

b d u lla

h Jo

rd an

4 8

2 4 6

1 0 .5

2 ,1

2 6

8 2 2

4 ,2

4 7 .8

9 ,9

6 5 ,3

1 8

0 7 0

3 0

4 5

6 5

1 6

4 3

R e ce

p T

ay yi

p E rd

o ğa

n T

u rk

e y

6 5

6 0

— 2 ,6

2 7

1 ,7

8 0

9 ,3

1 1 .4

8 2 ,3

4 0 ,0

8 8

1 6 6

3 7

4 5

8 5

4 6

4 9

R ic

ar d o

R o ss

e lló

P u e rt

o R

ic o

4 0

3 1

.1 6 1

3 ,2

1 1

2 ,8

5 3

3 1 ,6

5 1 .3

3 ,0

3 9 ,5

9 6

1 6 8

2 7

5 6

3 8

1 9

9 9

S aa

d H

ar ir

i L e b an

o n

4 9

3 2

1 .5

4 3 ,2

3 1

2 ,8

9 8

8 ,2

6 9 .8

6 ,8

5 9 ,4

0 8

1 7 5

4 0

6 5

5 0

1 4

2 5

S al

m an

b in

A b d u la

zi z

A l S au

d S au

d i A

ra b ia

8 3

5 5

7 .6

5 3 3 0

3 ,2

4 9

2 3 ,2

1 9 .1

3 3 ,7

0 2 ,7

5 6

0 9 5

2 5

6 0

8 0

3 6

5 2

S e b as

ti an

K u rz

A u st

ri a

3 2

1 7

.3 4 7

3 ,2

1 9

2 ,7

7 4

5 1 ,5

1 2 .9

8 ,8

9 1 ,3

8 8

1 1 1

5 5

7 9

7 0

6 0

6 3

S e b as

ti án

P iñ

e ra

C h ile

6 9

1 7

2 .2

4 3 ,2

2 0

1 ,7

7 8

1 5 ,9

2 3 .4

1 8 ,7

2 9 ,1

6 0

1 6 3

2 3

2 8

8 6

3 1

6 8

S h in

zo A

b e

Ja p an

6 4

7 9

1 .5

1 ,6

5 6

1 ,7

4 3

3 9 ,2

8 6 .7

1 2 7 ,2

0 2 ,1

9 2

1 5 4

4 6

9 5

9 2

8 8

4 2

S te

ff e n

S e ib

e rt

G e rm

an y

5 9

1 0 8

.9 2 2

2 ,8

9 5

1 ,5

5 4

4 8 ,1

9 5 .6

8 3 ,1

2 4 ,4

1 8

1 3 5

6 7

6 6

6 5

8 3

4 0

T am

im b in

H am

ad Q

at ar

3 9

2 6

.5 8

1 0 1

3 ,2

0 5

6 9 ,0

2 6 .5

2 ,7

8 1 ,6

8 2

0 9 3

2 5

5 5

8 0

— —

T e o d o ro

L o cs

in P h ili

p p in

e s

7 0

1 0

.6 5

3 ,2

0 3

6 4 9

3 ,1

0 2 .7

1 0 6 ,6

5 1 ,3

9 4

1 9 4

3 2

6 4

4 4

2 7

4 2

T h e re

sa M

ay U

n it e d

K in

gd o m

6 2

3 6

.8 9 2

1 ,6

1 4

1 ,6

1 4

4 2 ,4

9 1 .4

6 7 ,1

4 1 ,6

8 4

1 3 5

8 9

6 6

3 5

5 1

6 9

U h u ru

K e n ya

tt a

K e n ya

5 7

7 6

.0 0 5 5 4

1 6

1 ,8

4 3

1 ,7

1 0 .5

5 1 ,3

9 2 ,5

6 5

1 7 0

2 5

6 0

5 0

- V

la d im

ir P u ti n

R u ss

ia 6 6

8 7

.7 3

3 ,2

0 8

1 ,5

1 2

1 1 ,2

8 8 .9

1 4 5 ,7

3 4 ,0

3 8

1 9 3

3 9

3 6

9 5

8 1

2 0

N o te

s: P D ¼

H o fs

te d e

p o w

e r

d is

ta n ce

; ID

V ¼

H o fs

te d e

in d iv

id u al

is m

; M

A S ¼

H o fs

te d e

m as

cu lin

it y;

U A ¼

H o fs

te d e

u n ce

rt ai

n ty

av o id

an ce

; L T ¼

H o fs

te d e

lo n g-

te rm

o ri

e n ta

ti o n ; IN

D ¼

H o fs

te d e

in d u lg

e n ce

.

53

through July 2019. We constructed the list of leaders from

listings in the popular press and Twitter reports (e.g., Twiplo-

macy; https://twiplomacy.com/ranking/50-effective-world-

leaders/#). As illustrated in Table 1, the list contains leaders

from Europe, Asia, North, Central and South America, Africa,

and the Middle East, all of whom either held an active political

position or were politically active (e.g., Hillary Clinton) during

the data collection period. For each political leader, we col-

lected two types of communication data:

1. Social media: Twitter is a commonly used social media

communication outlet for political leaders (Golbeck

et al. 2010) and has been recently used in research on

political leaders (Bovet and Makse 2019; Bovet, Mor-

one, and Makse 2018). We used Twitter API to down-

load the tweets from each political leader’s public

account. We included everything posted under the lead-

er’s Twitter handle, including original tweets as well as

retweets, with the exception of Angela Merkel, for

whom we used her spokesman’s Twitter account

because her own Twitter account was not sufficiently

active. Altogether, we collected a total of 150,696 Twit-

ter posts. Note that although the political leaders often

employ people who do some of the actual tweeting on

their behalf, they mostly are personally involved in their

Twitter account, and the leaders’ tweets are regarded as

the closest approximation to direct communication with

their audience (Draper 2018).

2. Press articles: We mined the LexisNexis database

(using its API) to collect the entire press coverage of

the leaders on our list, resulting in a total of 120,689

press articles. The data set contains most of the world’s

press publications, including online and offline daily

newspapers and magazine articles as well as press

releases and newswire articles.

In addition, because we wanted to examine the communi-

cation style with respect to country and leader characteristics,

we collected, for each leader, a list of personal characteristics:

age, role, and tenure in position. For each country, we collected

data on the country’s GDP, social inequality (Gini index), and

its political regime. We also used, for each country, the latest

measures of Hofstede’s cultural dimensions (Hofstede 2011),

namely, uncertainty avoidance, masculinity versus femininity,

long-term orientation versus short-term orientation, individu-

alism versus collectivism, indulgence versus restraint, and

power distance.

Preprocessing of Communication Data

Conducting content analysis on Twitter and on press articles is

challenging, with each data source bearing its own challenges:

First, the Twitter accounts of most global leaders from non-

English-speaking countries are not in English, and press arti-

cles also come in many languages. Content analysis algorithms,

however, are mostly trained on English dictionaries; hence, the

entire data set needs to be translated. Second, while the press

articles are usually edited before publication and use standard

language, Twitter is characterized by a low degree of regula-

tion, and tweets use spoken language (Crystal 2006). Process-

ing Twitter data requires coping with contracted auxiliaries

(e.g., “I’ll,” “won’t,” “isn’t,” “ain’t,” “gotta,” “wanna”), slang

expressions (e.g., “lol”), exclamations (e.g., “oh,” “yay,”

“aww”), terms coined explicitly by Twitter (e.g., “retweet” or

“RT”), use of the “@” and “#” symbols to mention other users

and topics, and errors of spelling, capitalization, and

punctuation.

To prepare the data, all tweets were decoded to natural

language (some of them were coded as bytes) and were trans-

lated to English using Google translation libraries. All linguis-

tic characteristics of a tweet (i.e., @ usernames, hashtags,

retweet abbreviations) were retained. LexisNexis preproces-

sing included cleaning all the metadata (e.g., length, load date,

language, publication type, journal code, section, byline). Non-

English publications were translated to English as part of the

data corpus. The entire text from both sources was converted to

lowercase, and extra blank spaces were removed.

Content Analysis

We conducted two types of content analysis on the communi-

cation data:

1. Sentiment analysis: We applied sentiment analysis to

measure the level of positivity or negativity of the

tweets and the news articles (Pang and Lee 2004; Zhao,

Qin, and Liu 2010). We used two sentiment analysis

tools. The first is the polarity measure of the TextBlob

Python library, which returns a sentiment scaled from

�1 to 1. The second sentiment score (VADER of Python) returns four different scores: positive (0 to 1),

neutral (0 to 1), negative (0 to 1), and compound (�1 to 1). The first three are scores of positivity, neutrality, and

negativity, respectively, and the compound score is a

weighting of all three. We used the weighted score

(compound) as an alternate measure for the sentiment

analysis. Both measures (hereafter termed “polarity”

and “sentiment”) intend to capture the same construct

and are highly correlated; we use them both for robust-

ness purposes.

For the Twitter data, we used the entire text of the

post for the sentiment analysis. The news articles are

longer and in many cases only partially relate to the

political leader. Therefore, the overall sentiment of the

article might not represent the sentiment attached to the

political leader. To overcome this problem, we con-

ducted the sentiment analysis of news articles only on

paragraphs that include the name of political leader.

2. Topic analysis: We used latent Dirichlet allocation

(LDA) to elicit the topics of the Twitter and news posts.

This method is regarded as an effective way to identify

topics (i.e., subjects discussed in one or more

54 Journal of International Marketing 28(1)

documents) and topic categories (i.e., groups of topics

belonging to a common subject area) in large document

collections (Krippendorff 2004). It has been recently

applied in marketing research to analyze user-

generated content (e.g., Büschken and Allenby 2016;

Tirunillai and Tellis 2014). We trained the LDA model

on the Twitter data, where each tweet was considered as

a single document. In a standard preprocessing stage for

the LDA, we removed exclamation marks, emoticons,

common slang words, URLs, hashtags, and Twitter-

A: Sentiment Per Leader: Twitter Versus Press Articles

B: Polarity Per Leader: Twitter Versus Press Articles

−.40

−.20

.00

.20

.40

.60

.80

A bd

el F

at ta

h al

-S is

i A

bd ul

la h

bi n

Za ye

d A

l N ah

ya n

A de

lA lju

be ir

A le

ks an

da rV

uc ic

A nd

ré s

M an

ue lL

óp ez

O br

ad or

S te

ffe n

S ei

be rt

A rif

A lv

i B

en ja

m in

N et

an ya

hu D

an ilo

M ed

in a

D m

itr y

M ed

ve de

v D

on al

d T

ru m

p D

on al

d T

us k

M ah

a t hi

r M

oh am

ad E

dg ar

s R

in ke

v i c s

E m

ir Ta

m im

bi n

H am

ad E

m m

an ue

l M ac

ro n

E rn

e s to

A ra

új o

E vo

M or

al es

G i u

se pp

e C

o n te

H ill

a r y

C lin

to n

Im ra

n K

ha n

Iv án

D u q

ue Ja

ir B

ol so

na ro

Ja va

d Za

rif Jo

ko W

id od

o Jo

rg e

A rr

ea za

M on

t s er

ra t

Ju an

O rla

nd o

H er

ná nd

ez Ju

st in

Tr ud

ea u

K ha

lid bi

n A

hm e d

K in

g A

bd ul

la h

II K

in g

S al

m an

b i n

A bd

ul az

iz A

lS au

d K

on o

Ta ro

Le e

N ak

-y eo

n M

ar c e

lo E

br ar

d M

ar tín

V iz

ca rr

a M

au ric

io M

ac ri

M ev

lü t Ç

av uş

oğ lu

M ig

ue lD

í a z -

C an

el M

ik e

P om

p e o

M oh

am m

ed bi

n Za

ye d

A l N

ah ya

n M

oo n

Ja e-

in M

uh am

m ad

u B

uh ar

i N

ar en

dr a

M od

i N

ic ol

ás M

ad ur

o P

a u lK

ag am

e P

e d ro

S án

ch ez

P et

ro P

or o s

he nk

o P

op e

Fr an

ci s

R an

ia A

lA bd

ul la

h R

e c ep

T a yy

ip E

rd oğ

an R

ic ar

d o R

o s se

lló S

aa d

H ar

iri S

eb as

tia n

K ur

z S

e b as

tiá n

P iñ

e r a

S ha

h M

ah m

oo d

Q ur

es hi

S he

ik h

M oh

am m

ed A

lM ak

to um

S hi

nz o

A be

Te od

or o

Lo cs

in Th

er es

a M

ay U

h u ru

K en

ya tta

V la

di m

ir P

ut in

Twitter sentiment Press articles sentiment

−.05

.00

.05

.10

.15

.20

.25

.30

A bd

el F

at ta

h al

-S is

i A

bd ul

la h

bi n

Za ye

d A

l N ah

ya n

A de

lA lju

be ir

A le

ks an

da rV

uc ic

A nd

ré s

M an

ue lL

óp ez

O br

ad or

S te

ffe n

S ei

be rt

A rif

A lv

i B

en ja

m in

N et

an ya

hu D

an ilo

M ed

in a

D m

itr y

M ed

ve de

v D

on al

d T

ru m

p D

o n al

d T

us k

M ah

at hi

r M

oh a m

a d E

dg ar

s R

in k e

v i cs

E m

i r Ta

m im

bi n

H am

ad E

m m

an ue

l M ac

ro n

E rn

es to

A ra

új o

E vo

M or

al es

G iu

se pp

e C

on te

H ill

ar y

C lin

t o n

Im ra

n K

h a n

Iv án

D uq

ue Ja

ir B

ol so

n a ro

Ja va

d Za

rif J o

k o W

id o d

o J o

r g e

A rr

ea za

M on

ts er

ra t

Ju an

O rla

nd o

H er

ná nd

ez Ju

st in

Tr ud

ea u

K ha

lid bi

n A

hm ed

K in

g A

bd ul

la h

II K

in g

S a l

m an

bi n

A bd

u l az

i z A

l S au

d K

o n o

Ta ro

Le e

N ak

- y eo

n M

ar ce

lo E

br ar

d M

ar tín

V iz

ca rr

a M

au ri c

io M

ac ri

M ev

lü t Ç

av u ş

oğ lu

M ig

u e lD

ía z-

C an

e l M

ik e

P om

pe o

M oh

am m

e d bi

n Z a

y e d

A l N

ah ya

n M

oo n

J a e-

in M

uh am

m ad

u B

u h ar

i N

ar e n

dr a

M od

i N

ic ol

ás M

ad ur

o P

au lK

ag am

e P

e d ro

S án

ch ez

P et

ro P

or o s

he nk

o P

op e

Fr an

c i s

R an

ia A

lA bd

ul la

h R

ec ep

Ta yy

ip E

rd oğ

an R

ic ar

do R

os se

lló S

aa d

H ar

iri S

eb as

tia n

K ur

z S

eb as

tiá n

P iñ

er a

S h a

h M

ah m

oo d

Q ur

es hi

S h e

i k h

M oh

am m

ed A

l M a k

to um

S hi

nz o

A be

Te od

or o

Lo cs

in Th

er es

a M

ay U

hu ru

K e n

ya tta

V la

di m

ir P

ut in

Twitter polarity Press articles polarity

Figure 2. Compound sentiment (Panel A) and polarity (Panel B) per leader for Twitter and press articles.

Peres et al. 55

specific notation (i.e., “b’,” “RT,” “b’RT”), although

we used them for the sentiment analysis. We also con-

ducted stemming (using NLTK Porter Stemmer) to exo-

genously force words with similar meanings, such as

“great,” “greatly,” “greatest,” and “greater,” to be

treated as a bundle. We filtered out tokens that appear

in less than 15 documents or in more than 50% of the documents (a token is a word or a short word combina-

tion). From the tokens left, we kept the 100,000 most

common speech tokens. The algorithm yielded its best

results for 15 topics (we manually looked through

results of different numbers of subjects to finalize the

number), and the different topics were manually named.

Then the chosen LDA model was applied on the Twitter

data and the LexisNexis data, with each document

being all the tweets or all the reports (respectively)

connected to each single leader. In this way, we

obtained the distribution of topics discussed in by each

political leader on Twitter, and in connection with the

leader in the press.

Results

Universality and Media Compatibility in Sentiment

Figure 2 illustrates the values of the sentiment/polarity mea-

sures per leader for Twitter and for press articles. In general,

the sentiment/polarity of leaders is neutral to positive, with the

Pakistani leader Shah Mahmood Qureshi being the most posi-

tive (.63, .231, .667, and .127 for Twitter sentiment, Twitter

polarity, press sentiment, and press polarity, respectively), and

politicians such as the British leader Theresa May and Abdul-

lah bin Zayed Al Nahyan from the United Arab Emirates being

at the bottom for positivity (note the relatively negative press

coverage of the latter). The average sentiment (using the com-

pound sentiment measure described previously) is .279 (SD ¼ .134) for Twitter and .272 (SD ¼ .196) for the press articles, meaning that they are not significantly different from each

other (p ¼ .62). The average polarity of the leaders in our data set is .13 (SD ¼ .054) for Twitter and .064 (SD ¼ .03) for the press articles. These numbers indicate that Twitter is signifi-

cantly more positive than the press articles, but the difference is

small and is not robust across different sentiment measurement

methods.

We further tested to what extent the sentiment of the polit-

ical leaders’ communication depends on their country and per-

sonal characteristics. Table 2 describes the results of four

regression models, where the dependent variables are the senti-

ment and polarity for Twitter and press articles and the depen-

dent variables are the country characteristics. None of the

variables were found to be significant. That is, the sentiment

and the polarity of communication do not depend on the lead-

ers’ individual country characteristics. We repeated this anal-

ysis with the leaders’ personal characteristics, as well as with

other country characteristics, such as level of happiness,

freedom of press, and social inequality. None of these variables

were found to be significant.

The results show a clear universality across leaders and also

similarity across communication types. The major political

leaders in our data set are similar in the sentiment of their

communication, and their sentiment, measured in terms of both

polarity and compound sentiment, does not show a clear depen-

dence on the country characteristics. This universality is also

expressed in the similarity of sentiment of Twitter and press

articles. The traditional notion of self-presentation as portray-

ing the individual in a positive manner and representation in

journalism as being more objective and therefore less posi-

tively biased has faint support in our data. Sentiment is gener-

ally neutral to positive, and Twitter is slightly more positive in

polarity, but this difference is not robust across the other senti-

ment measure we used.

Universality and Media Compatibility in Communication Topics

Tables 3 through 5 present the results of the content analysis.

The LDA procedure identified 15 topics that span the leaders’

communication: (1) the Arab world, (2) condolences, (3) con-

gratulations, (4) corruption, (5) diplomacy, (6) economy,

(7) education, (8) elections, (9) fighting terrorism, (10) global

relationships, (11) humanitarian crisis, (12) immigration,

(13) national holiday, (14) official thank-you messages, and

(15) parliamentary issues. Table 3 describes, for each topic,

the 15 words with the highest likelihood. Note that LDA clas-

sifies all the words on all the topics, providing for each word

the likelihood that it belongs to the topic. Table 4 presents the

average weights of the topic mixture over all the leaders, for

Twitter and for the press articles, and compares them using a

series of t-tests. The table indicates that diplomacy (i.e., words

Table 2. Regression of the Average Sentiment and Polarity of a Leader on Country Characteristics for Twitter and Press Articles.

Twitter Sentiment

Twitter Polarity

Press Articles

Sentiment

Press Articles Polarity

logGDP .045 �.014 �.014 .006 Continent

Asia .116 .073 .073 .002 Europe .035 .165 .165 �.019 Central America .144 �.059 �.059 �.016 South America .118 �.001 �.001 �.019 North America .098 .122 .122 �.010

Democracy .000 �.001 �.001 .000 Hofstede Dimensions

Power distance .066 �.039 �.039 .018 Individualism �.002 .001 .001 .000 Masculinity .000 .001 .001 .000 Uncertainty avoidance �.002 .003 .003 .000 Long-term orientation .000 .003 .003 .000 Indulgence �.001 .000 .000 .000

56 Journal of International Marketing 28(1)

T a b

le 3 .

L D

A R

e su

lt s:

T h e

T o p

1 5

W o rd

s w

it h

th e

H ig

h e st

L ik

e lih

o o d

fo r

th e

1 5

T o p ic

s.

T h e

A ra

b w

o rl

d go

d b in

m o h am

za i

b le

ss k in

g m

e rc

i jo

rd an

m ar

ty r

al ca

m p

ar ab

p u ti n

h o li

ra m

ad an

.0 3 6 2

.0 3 1 2

.0 3 0 3

.0 2 5 9

.0 1 8 3

.0 1 1 1

.0 1 1 0

.0 1 0 4

.0 1 0 3

.0 1 0 1

.0 1 0 0

.0 0 9 8

.0 0 9 8

.0 0 9 7

.0 0 9 3

C o n d o le

n ce

s fa

m ili

p e o p l

vi ct

im am

p at

ta ck

to d ai

d e at

h liv

e fr

ie n d

co n d o l

co n d e m

n lo

se af

fe ct

ye ar

th o u gh

t .0

3 9 5

.0 1 7 1

.0 1 5 5

.0 1 3 5

.0 1 3 2

.0 1 2 8

.0 1 0 5

.0 1 0 4

.0 1 0 2

.0 0 9 6

.0 0 9 2

.0 0 9 2

.0 0 9 1

.0 0 7 8

.0 0 7 2

C o n gr

at u la

ti o n s

co n gr

at u l

gr e at

h ap

p i

p e o p l

w o rl

d lo

ve w

is h

d ai

co u n tr

i w

o m

e n

n at

io n

b e st

p ro

u d

p e ac

ye ar

.0 2 6 2

.0 1 9 2

.0 1 8 9

.0 1 8 6

.0 1 7 7

.0 1 6 7

.0 1 4 2

.0 1 4 0

.0 1 2 7

.0 1 1 5

.0 1 1 2

.0 1 0 7

.0 1 0 6

.0 1 0 6

.0 1 0 1

C o rr

u p ti o n

go ve

rn u k ra

in w

o rk

p u b lic

am p

fo rc

to d ai

re fo

rm p o w

e r

n at

io n

im p le

m e n t

n e e d

p e o p l

ta x

co rr

u p t

.0 2 8 7

.0 1 7 3

.0 1 3 8

.0 1 3 1

.0 0 9 8

.0 0 8 9

.0 0 8 7

.0 0 7 7

.0 0 7 5

.0 0 7 4

.0 0 6 9

.0 0 6 9

.0 0 6 9

.0 0 6 8

.0 0 6 6

D ip

lo m

ac y

m e e t

m in

is t

p re

si d

p ri

m e

fo re

ig n

d is

cu ss

co n fe

r co

u n ci

l au

st ri

a is

su vi

si t

re la

t re

ce iv

co o p e r

to d ai

.0 5 7 7

.0 5 4 4

.0 4 2 7

.0 2 8 6

.0 2 3 0

.0 1 8 7

.0 1 4 0

.0 1 3 6

.0 1 3 2

.0 1 3 0

.0 1 2 4

.0 1 2 0

.0 1 1 5

.0 1 0 5

.0 1 0 3

E co

n o m

y d e ve

lo p

in ve

st p ro

je ct

jo b

e d u c

w o rk

cr e at

im p ro

v n e w

ye ar

m ill

io n

e co

n o m

i p la

n co

u n tr

i o p p o rt

u n

.0 1 8 4

.0 1 5 8

.0 1 4 8

.0 1 3 7

.0 1 1 8

.0 1 1 7

.0 1 1 5

.0 1 1 2

.0 1 0 9

.0 1 0 7

.0 1 0 6

.0 1 0 1

.0 0 9 6

.0 0 9 5

.0 0 9 4

E d u ca

ti o n

sc h o o l

gt lt

re ad

n e w

vo te

st u d e n t

co n te

n t

o p e n

d o o r

p h o to

am p

ta le

n t

ye ar

p ti

.0 1 9 6

.0 1 9 2

.0 1 6 0

.0 1 4 4

.0 1 4 2

.0 1 3 5

.0 1 2 7

.0 1 2 2

.0 1 1 5

.0 0 8 0

.0 0 7 6

.0 0 7 4

.0 0 7 0

.0 0 6 7

.0 0 6 5

E le

ct io

n s

e le

ct liv

e p ar

ti vo

te w

at ch

ca n d id

to d ai

p e o p l

n at

io n

co m

e lik

e co

n gr

e ss

ca m

p ai

gn sp

e e ch

ti m

e .0

3 2 8

.0 2 3 1

.0 1 7 6

.0 1 5 9

.0 1 5 9

.0 1 4 3

.0 1 3 0

.0 1 0 8

.0 0 9 1

.0 0 8 8

.0 0 8 3

.0 0 8 2

.0 0 8 1

.0 0 7 9

.0 0 7 8

F ig

h ti n g

te rr

o ri

sm fi gh

t te

rr o r

k n o w

ir an

w an

t p e o p l

ac t

p ai

re gi

m w

ar ti m

e th

at at

ta ck

te rr

o ri

st w

o rl

d .0

1 2 8

.0 1 1 3

.0 1 0 5

.0 0 9 9

.0 0 9 8

.0 0 8 6

.0 0 8 5

.0 0 8 3

.0 0 7 9

.0 0 7 9

.0 0 7 2

.0 0 6 8

.0 0 6 6

.0 0 6 1

.0 0 6 1

G lo

b al

re la

ti o n sh

ip s

go o d

ch ile

e u

ch in

a co

u n tr

i m

e e t

am p

e u ro

p to

d ai

w o rl

d co

o p e r

p re

si d

so u th

tr ad

e k o re

a .0

2 5 5

.0 2 0 1

.0 1 9 8

.0 1 8 5

.0 1 6 5

.0 1 5 8

.0 1 5 6

.0 1 4 8

.0 1 3 7

.0 1 3 4

.0 1 3 4

.0 1 2 7

.0 1 1 6

.0 1 1 3

.0 1 1 2

H u m

an it ar

ia n

cr is

is ri

gh t

le b an

o n

ve n e zu

e la

se cu

r p ro

te ct

st at

e p e o p l

la w

co u n tr

i h u m

an p e ac

su p p o rt

d e fe

n d

is ra

e l

am e ri

ca n

.0 3 1 4

.0 1 9 7

.0 1 6 4

.0 1 4 3

.0 1 4 3

.0 1 4 3

.0 1 3 9

.0 1 3 6

.0 1 2 5

.0 1 1 8

.0 1 0 8

.0 0 9 3

.0 0 9 1

.0 0 8 9

.0 0 8 7

Im m

ig ra

ti o n

h ar

ir i

b re

x it

n e w

ce n te

r to

d ai

m e e t

p ro

d u ct

m ig

ra t

te ch

n o lo

g in

d u st

ri in

n o v

o p e n

n at

io n

in au

gu r

p re

si d

.0 4 2 3

.0 1 9 5

.0 1 6 8

.0 1 3 3

.0 1 2 3

.0 1 1 7

.0 1 1 0

.0 1 0 6

.0 1 0 4

.0 1 0 1

.0 0 8 8

.0 0 8 8

.0 0 8 1

.0 0 8 0

.0 0 7 8

N at

io n al

h o lid

ay ye

ar ch

an g

d ai

to d ai

th an

n iv

e rs

ar i

in d e p e n d

co m

m it

d e m

o cr

ac i

p e o p l

ag o

h is

to ri

m o ve

m e n t

n at

io n

fr e e d o m

.0 2 3 4

.0 2 2 7

.0 1 9 9

.0 1 7 7

.0 1 3 6

.0 1 3 5

.0 1 1 9

.0 1 1 3

.0 1 0 6

.0 0 9 5

.0 0 8 5

.0 0 8 4

.0 0 7 2

.0 0 7 0

.0 0 6 6

O ff ic

ia l th

an k -y

o u

th an

k w

o rk

su p p o rt

fo rw

ar d

lo o k

to d ai

p e o p l

p ro

b le

m co

u n tr

i co

n ti n u

n at

o w

e lc

o m

co m

m it

ye ar

d ai

.1 0 8 6

.0 3 9 2

.0 1 4 2

.0 1 3 1

.0 1 2 3

.0 1 1 4

.0 1 1 0

.0 0 9 6

.0 0 9 4

.0 0 9 3

.0 0 8 6

.0 0 8 3

.0 0 7 9

.0 0 7 8

.0 0 7 6

P ar

lia m

e n ta

ry am

p p m

in te

rv ie

w co

n se

rv tu

rk e i

se n at

e u

p re

si d

tr u m

p u k

p e n si

o n

n e go

ti ti m

e m

o o n

sa i

.0 4 7 4

.0 3 0 2

.0 1 4 0

.0 1 2 8

.0 1 2 3

.0 1 1 8

.0 1 1 6

.0 1 1 5

.0 1 1 3

.0 1 1 2

.0 1 1 1

.0 1 0 4

.0 1 0 2

.0 1 0 1

.0 0 9 8

N o te

s: A

ll w

o rd

s w

e re

co n ve

rt e d

in to

lo w

e rc

as e

le tt

e rs

.W o rd

s m

ig h t b e

tr u n ca

te d

b e ca

u se

o f th

e st

e m

m in

g p ro

ce ss

(e .g

., “m

ig ra

t” st

an d s

fo r

“m ig

ra ti o n ,”

“i m

m ig

ra n ts

,” an

d o th

e r

w o rd

s) .T

h e

n u m

b e rs

ar e

th e

lik e lih

o o d

th at

th e

w o rd

b e lo

n gs

to th

e to

p ic

. *p <

.0 5 . In

th is

ca se

, n o

va ri

ab le

w as

fo u n d

si gn

if ic

an t.

57

related to meetings, conferences, discussions, etc.) is the most

popular topic for both communication outlets. Press articles

focus significantly more than Twitter on corruption, diplo-

macy, elections, and parliamentary issues, whereas Twitter,

relative to the press articles, deals more with personal mes-

sages, such as condolences, congratulations, and thank-you

messages, as well as economy. Many of the topics, such as the

Arab world, education, fighting terrorism, and global relation-

ships, receive similar emphasis both on Twitter and in the

press.

To what extent do communication topics depend on the

country characteristics? We tested, similarly to the sentiment

analysis, how the topic mixture depends on the country char-

acteristics. For each topic, we ran a regression model over the

61 leaders, in which the dependent variable is the weight of the

topic in the leader’s communication (for either Twitter or the

press articles) and the independent variables are the individual

and country characteristics. The results are presented in Table 5

(Panel A for Twitter and Panel B for press articles). The results

indicate that overall, the topic mixture does not seem to

strongly depend on the individual and country characteristics.

On Twitter, immigration is a prevalent topic for democracies

and for countries with high GDP. Interestingly, immigration

has more presence in the tweets of African leaders relative to

leaders from other continents. Immigration is also discussed in

countries with lower uncertainty avoidance, shorter-term orien-

tation, and lower indulgence. The Arab world is discussed less

in democracies (probably because it is discussed by Arab lead-

ers, who are mostly not democratic rulers). Being a democracy

is positively correlated with a country’s leader’s tweets on

corruption and education, and less with tweets on diplomacy.

For press articles, results are similar (although fewer variables

are significant). In addition to these variables, we tried a large

set of models with different variables, including individual

characteristics of the leaders and country characteristics, such

as level of happiness, freedom of press, and social inequality,

which did not lead to any results of interest.

What additional insights can we gain from the differences

between the topics discussed on Twitter and in the press

articles? As discussed, Table 4 indicates that although many

of the topics are discussed on average equally in both media,

some topics (e.g., congratulations, condolences, thank-you

messages) are more typical on Twitter, whereas topics such

as diplomacy and parliamentary issues are discussed more in

press articles. We wanted to analyze the differences between

the outlets at the individual leader’s level (in addition to the

aggregate comparison presented previously) and ask to what

extent a leader’s communication on Twitter is similar to or

different from the leader’s press coverage. To do so, we calcu-

lated the cosine similarity of the two topic mixture distribution

vectors for each leader. Cosine similarity treats the two topic

mixtures of a leader’s media as two vectors in a 15-dimensional

space and calculates their scalar (dot) product. Both the press

articles and Twitter topic distribution vectors of each leader

were normalized so that the sum of squares is 1 before the dot

product of the two vectors was calculated, so the similarity is

between 0 (lowest possible similarity; vectors are orthogonal)

and 1 (highest possible similarity). The cosine similarity values

of the leaders on our list range from .14 (Emir Tamim bin

Hamad) to .29 (Pope Francis), with an average of .22 (SD ¼ .03). Although these numbers may seem low, they represent

relatively high similarity, given the different structure and typ-

ical grammatical nature of the texts in these outlets. Hence, for

political leaders, the differences between Twitter and press

articles are no different than the typical difference between the

contents of these two channels, given their fundamentally dif-

ferent characteristics. In other words, we see no evidence that

differences in topic mixture between outlets are a result of an

outlet-specialized communication strategy.

We further checked whether the level of similarity depends

on the country characteristics. We ran a regression model in

which the dependent variable was the cosine similarity score

and the independent variables were the country characteristics

used in the previous regressions. The results, described in

Table 6, indicate that the higher the power distance and indi-

vidualism are, the higher is the topic similarity between Twitter

and press articles. Although the coefficients are significant, it is

hard to interpret them as representing a theoretically based,

fundamental mechanism.

Similarly to the sentiment analysis, the content analysis

implies universality across leaders in communication topics.

These results, grounded in the theoretical perspective of

the “mediatized communication as a meta process of

globalization” (Krotz 2007), imply that despite the distinct

differences between cultures, leaders generate a communica-

tion mix that is weakly dependent on the country characteris-

tics. In branding terminology, although political leaders are

brands that are strongly identified with their country, their

communication style does not reflect that. This is the case both

for their own generated content on Twitter and for the press

Table 4. The Average Topic Distribution over Leaders, for Twitter and Press Articles.

Topic p-Value of t-Test

Twitter Average

Press Articles (LexisNexis) Average

The Arab world .7951 .0736 .0790 Condolences .0000* .0624 .0372 Congratulations .0000* .1031 .0262 Corruption .0050* .0768 .0981 Diplomacy .0004* .1005 .1545 Economy .0345* .0816 .0615 Education .0686 .0491 .0595 Elections .0125* .0617 .0792 Fighting terrorism .2328 .0698 .0625 Global relationships .5576 .0681 .0629 Humanitarian crisis .8009 .0719 .0740 Immigration .4044 .0465 .0503 National holiday .3443 .0596 .0555 Official thank-you .0000* .0528 .0320 Parliamentary .0021* .0483 .0752

Notes: Each t-test compares the distance between the two average values, over all the leaders (N ¼ 61). *p < .05.

58 Journal of International Marketing 28(1)

T a b

le 5 .

R e gr

e ss

io n s

o f th

e T

o p ic

M ix

tu re

o f a

L e ad

e r

o n

C o u n tr

y C

h ar

ac te

ri st

ic s

fo r

T w

it te

r an

d P re

ss A

rt ic

le s.

In d e p e n d e n t

V ar

ia b le

lo gG

D P

C o n ti n e n t

H o fs

te d e

D im

e n si

o n s

A si

a E u ro

p e

C e n tr

al A

m e ri

ca S o u th

A m

e ri

ca N

o rt

h A

m e ri

ca D

e m

o cr

ac y

P o w

e r

D is

ta n ce

In d iv

id u al

is m

M as

cu lin

it y

U n ce

rt ai

n ty

A vo

id an

ce L o n g-

T e rm

O ri

e n ta

ti o n

In d u lg

e n ce

A :

R e g r e ss

io n

o f

T w

it te

r T

o p

ic M

ix tu

r e

o f

a L

e a d

e r

o n

C o

u n

tr y

C h

a r a c te

r is

ti c s

T h e

A ra

b w

o rl

d .0

1 6 0

.0 0 0 0 �

.0 1 9 9 �

.0 3 9 0 �

.0 7 1 8

.0 1 3 5

� .1

4 4 1 *

.0 0 0 4

� .0

0 0 7

.0 0 0 6

.0 0 1 4

� .0

0 1 3

� .0

0 0 5

C o n d o le

n ce

s �

.0 0 2 6

.0 8 2 4 *

.0 4 4 5

.0 2 3 0

.0 2 6 6

.0 5 3 6

.0 5 0 7 *

� .0

0 0 4

.0 0 0 0

� .0

0 0 2

.0 0 0 4

� .0

0 0 4

.0 0 0 3

C o n gr

at u la

ti o n s

� .0

0 7 0

.1 2 0 6 *

.0 8 2 2

.1 0 1 0

.0 2 8 8

.0 3 7 8

.0 2 5 5

� .0

0 0 1

.0 0 0 7

� .0

0 1 0

.0 0 1 3

� .0

0 1 2

.0 0 1 1

C o rr

u p ti o n

� .0

0 1 9

.0 2 7 9

.0 1 4 3

.0 6 9 3

.0 2 0 5

.0 0 5 0

.0 6 6 9 *

.0 0 0 4

.0 0 0 1

� .0

0 0 2

.0 0 0 1

� .0

0 0 2

� .0

0 0 1

D ip

lo m

ac y

.0 2 6 5 �

.1 4 5 8 �

.0 2 7 6 �

.1 1 3 3 �

.0 4 1 3 �

.1 2 5 6

� .1

1 2 1 *

.0 0 1 3

� .0

0 0 4

.0 0 2 2 *

� .0

0 1 4

.0 0 0 8

� .0

0 1 6

E co

n o m

y .0

0 6 6 �

.0 0 5 6 �

.0 2 6 0

.1 0 1 4 �

.0 0 1 1 �

.0 1 0 2

.0 5 1 8

.0 0 0 0

� .0

0 0 3

� .0

0 1 2

.0 0 0 6

� .0

0 0 2

.0 0 0 7

E d u ca

ti o n

.0 0 4 4

.0 3 5 8 �

.0 2 3 8 �

.0 0 3 8 �

.0 2 2 5

.0 1 7 9

.0 5 0 2 *

� .0

0 0 3

� .0

0 0 5

� .0

0 0 4

.0 0 0 2

.0 0 0 0

.0 0 0 1

E le

ct io

n s

� .0

0 2 8 �

.0 2 5 9 �

.0 6 6 1 �

.0 1 5 9 �

.0 3 1 8 �

.0 0 7 1

.0 0 7 2

.0 0 0 4

.0 0 0 4

.0 0 0 7

.0 0 0 0

.0 0 1 *

.0 0 0 0

F ig

h ti n g

te rr

o ri

sm �

.0 0 2 3

.0 5 6 4

.0 5 2 0

.0 0 5 1

.0 2 0 9

.0 5 6 6

.0 1 4 0

.0 0 0 3

.0 0 0 2

� .0

0 0 3

.0 0 0 6

� .0

0 0 7

.0 0 0 2

G lo

b al

re la

ti o n sh

ip s

.0 0 4 4 �

.0 1 0 7

.0 1 7 6 �

.0 3 6 7

.0 2 4 4 �

.0 3 4 6

� .0

0 9 1

� .0

0 0 2

.0 0 0 8

.0 0 0 0

� .0

0 0 4

.0 0 0 7 *

.0 0 0 2

H u m

an it ar

ia n

cr is

is .0

0 8 3

.0 0 3 9

.0 5 0 0 �

.0 5 2 3

.0 4 6 0

.0 2 9 9

.0 0 8 7

.0 0 0 3

� .0

0 0 4

.0 0 0 2

� .0

0 0 3

� .0

0 0 8

.0 0 0 1

Im m

ig ra

ti o n

.0 2 6 4 * �

.0 8 8 6 * �

.1 0 2 6 * �

.1 1 2 4 * �

.0 7 4 7 * �

.1 3 5 5 *

.0 4 0 1 *

.0 0 0 0

� .0

0 0 5

� .0

0 0 2

� .0

0 0 7 *

� .0

0 0 6 *

� .0

0 0 9 *

N at

io n al

h o lid

ay �

.0 0 1 5

.0 1 0 3

.0 0 5 7

.0 2 0 3

.0 3 7 5 * �

.0 0 1 0

.0 3 0 1 *

.0 0 0 2

.0 0 0 3

� .0

0 0 2

.0 0 0 0

.0 0 0 2

.0 0 0 3

O ff ic

ia l th

an k -y

o u

.0 0 2 9

.0 0 9 6

.0 1 2 8

.0 2 2 3

.0 1 4 7

.0 1 0 7

.0 1 6 7

.0 0 0 2

.0 0 0 4

� .0

0 0 1

� .0

0 0 3

.0 0 0 1

� .0

0 0 2

P ar

lia m

e n ta

ry .0

0 9 1

.0 6 4 2

.0 4 5 1

.0 2 1 5

.0 3 2 3

.0 8 6 9

.0 4 0 6

� .0

0 0 3

� .0

0 0 6

.0 0 0 1

� .0

0 0 4

� .0

0 0 6

� .0

0 0 7

B :

R e g r e ss

io n

o f

P r e ss

A r ti

c le

s (L

e x is

N e x is

) T

o p

ic M

ix tu

r e

o f

a L

e a d

e r

o n

C o

u n

tr y

C h

a r a c te

r is

ti c s.

T h e

A ra

b w

o rl

d .0

1 3 1

.0 1 0 3 �

.0 1 8 9 �

.0 0 7 5 �

.0 4 0 6

.0 2 2 1

� .1

2 2 2 *

.0 0 0 6

� .0

0 0 3

.0 0 0 3

.0 0 1 0

� .0

0 0 9

� .0

0 0 6

C o n d o le

n ce

s .0

0 1 0

.0 4 0 5

.0 1 6 1

.0 0 5 7

.0 1 2 4 �

.0 0 7 8

.0 3 0 3 *

� .0

0 0 2

.0 0 0 1

� .0

0 0 6

� .0

0 0 2

� .0

0 0 1

.0 0 0 9

C o n gr

at u la

ti o n s

.0 0 1 5

.0 3 0 3

.0 1 1 6

.0 0 1 4

.0 0 5 4

.0 1 0 4

.0 1 3 2

� .0

0 0 2

.0 0 0 0

� .0

0 0 1

.0 0 0 1

� .0

0 0 2

.0 0 0 1

C o rr

u p ti o n

.0 0 0 9 �

.0 1 5 3

.0 0 8 4

.0 5 8 3

.0 3 8 2 �

.0 0 4 9

.0 2 6 5

.0 0 1 1 *

.0 0 0 3

.0 0 0 1

� .0

0 0 2

.0 0 0 1

� .0

0 0 4

D ip

lo m

ac y

� .0

2 7 6

.1 9 0 7

.1 6 7 1 �

.0 1 0 8 �

.0 0 0 3

.0 6 0 4

.0 6 0 9

.0 0 0 6

.0 0 1 1

.0 0 0 1

.0 0 1 5

� .0

0 0 9

.0 0 1 8

E co

n o m

y .0

0 5 3 �

.0 2 8 2 �

.0 1 6 4

.2 0 6 4 *

.0 2 6 8 �

.0 1 4 7

� .0

1 7 8

.0 0 0 4

.0 0 0 2

.0 0 0 6

.0 0 0 0

� .0

0 0 1

� .0

0 0 9

E d u ca

ti o n

.0 1 1 7

.0 1 3 2 �

.0 0 8 0 �

.0 1 4 2 �

.0 0 6 7

.0 2 4 2

.0 4 4 4 *

.0 0 0 0

� .0

0 0 5

.0 0 0 1

� .0

0 0 1

� .0

0 0 7

� .0

0 0 8

E le

ct io

n s

.0 1 8 2 �

.0 3 8 6 �

.0 3 2 1 �

.1 0 6 4 * �

.0 3 0 5 �

.0 5 9 4

.0 5 8 4 *

� .0

0 0 1

� .0

0 0 7

� .0

0 0 3

� .0

0 0 8

.0 0 0 0

.0 0 0 0

F ig

h ti n g

te rr

o ri

sm .0

0 0 7

.0 2 6 0

.0 1 7 6

.0 0 2 7 �

.0 0 6 5

.0 0 0 2

� .0

0 5 6

.0 0 0 0

.0 0 0 2

� .0

0 0 5

.0 0 0 6

� .0

0 0 1

.0 0 0 6

G lo

b al

re la

ti o n sh

ip s

.0 0 7 9 �

.0 2 8 8 �

.0 2 8 1 �

.0 0 8 4

.0 1 6 4 �

.0 4 7 6

� .0

2 2 9

� .0

0 0 5

.0 0 0 4

.0 0 0 3

� .0

0 0 3

.0 0 1 3 *

.0 0 0 0

H u m

an it ar

ia n

cr is

is .0

1 2 6

.0 0 1 5

.0 1 4 1 �

.0 6 2 0

.0 3 2 6

.0 2 5 5

.0 2 0 1

� .0

0 0 2

� .0

0 0 7

.0 0 0 2

� .0

0 0 3

� .0

0 0 6

.0 0 0 0

Im m

ig ra

ti o n

.0 1 7 5 * �

.0 5 5 2 �

.0 5 4 0 �

.0 3 3 9 �

.0 4 2 3 �

.0 9 6 6 *

.0 2 4 4

.0 0 0 1

.0 0 0 1

� .0

0 0 1

� .0

0 0 3

� .0

0 0 7 *

� .0

0 0 7

N at

io n al

h o lid

ay .0

0 5 8 �

.0 2 0 0 �

.0 0 7 1 �

.0 1 0 0

.0 3 3 2 �

.0 2 3 4

.0 1 0 4

.0 0 0 3

.0 0 0 2

.0 0 0 0

� .0

0 0 4

.0 0 0 2

� .0

0 0 1

O ff ic

ia l th

an k -y

o u

.0 0 6 4 * �

.0 1 0 3 �

.0 0 2 1 �

.0 0 2 3 �

.0 0 6 3 �

.0 1 8 6

.0 0 6 2

.0 0 0 1

.0 0 0 0

.0 0 0 0

� .0

0 0 3

� .0

0 0 1

� .0

0 0 2

P ar

lia m

e n ta

ry .0

1 2 0

.0 2 4 4

� .0

0 2 1

.0 1 3 4

.1 2 6 0

.0 5 8 9

� .0

0 0 4

� .0

0 0 8

� .0

0 0 1

� .0

0 0 1

� .0

0 0 6

� .0

0 0 8

*p <

.0 5 .

59

articles, which are supposed to represent the spectrum of inter-

ests of the local readers.

The similarity exists not only among leaders, but also

between Twitter, which is a presentational social media chan-

nel, and press articles, which are a form of representational

media. The differences in sentiment between Twitter and press

articles are small. Although the two outlets differ in some

topics (Table 3), their topic mixtures have many similarities,

and the differences seem to result from the inherent nature of

these outlets, not from channel-specific communication strate-

gies. In this sense, our results are in line with previous findings

showing that journalists rely on social media as a source of

information while constructing their news articles (Shapiro and

Hemphill 2017).

Universality and Media Compatibility in “I,” “We,” “You,” and “They” Messages

As an additional analysis, we tested another linguistic charac-

teristic of the leaders’ communications: some language models

(Packard, Moore, and McFerran 2018) emphasized the use of

“I” or “we” versus “you” and “they” as a possible measure of

the amount of self-focus versus orientation toward others in

one’s communication.

We used the content analysis algorithm to count the pro-

portion of first-person, second-person, and third-person words

for Twitter and for the press articles. Words with similar

meanings were grouped together (e.g., the words “I,” “me,”

and “myself” were all counted as “I”). Figure 3 presents the

percentage of these pronouns. For both outlets, “we” mes-

sages are predominant, followed by “I” messages. While

“they” messages are more prevalent than “you” messages in

press articles, the opposite is true for Twitter. This finding is

understandable because of the more descriptive nature of

press items. Note that we did not count the use of “he” or

“she,” as these words are commonly used in press articles to

describe what the leader did and therefore are less indicative

of the leader’s communication style.

Pronoun usage is related to sentiment. Table 7 presents the

results of two regression models, where the independent variable

is the polarity for Twitter and press articles and the dependent

variables are the percentage of each pronoun type. We can see

that for both communication outlets, more positive sentiment is

associated with higher usage of “we” messages and lower usage

of “they” messages. We obtain similar results for the sentiment

measure. Note that both sentiment algorithms we use do not

consider pronoun usage as input for determining the sentiment,

so these two dimensions are not forced to be correlated.

We also tested the percentage of the four types of pronouns

versus the country characteristics. The results are presented in

Table 8. Similar to the results on sentiment and topic, the use of

“I,” “we,” “they,” or “you” messages seems to be independent

of country characteristics. The only significance we found is

that when tweeting, leaders from Asia, Europe, and Central

America use “we” messages more often than leaders from

Africa do. Also, leaders from countries with shorter-term orien-

tation were found to use first-person pronouns (“I” and “we”)

on Twitter more often than leaders from countries with longer-

term orientation. The rightmost two columns of Table 8

describe regression models in which the dependent variables

are the differences between the number of “I” or “we” mes-

sages and the number of “they” or “you” messages. This anal-

ysis indicates the use of first-person relative to second- and

third-person pronouns (the left columns relate to the overall

proportion of each type, irrespective of the others). In line with

the finding described earlier, relative to African leaders, leaders

from Asia, Europe, and Central America use first-person mes-

sages more than second- and third-person messages when

tweeting. Also, leaders from countries with shorter-term orien-

tation were found to use first-person pronouns more often than

second- and third-person pronouns on Twitter.

Mapping on the Universality–Compatibility Matrix

We combined the two dimensions to classify the leaders in our

sample on the universality–compatibility matrix. The matrix is

conceptual, and therefore, we had several options to operationa-

lize it. We chose the horizontal axis to represent the extent to

which the topic distribution of the focal leader is similar to the

average topic distribution across leaders. Our measure of simi-

larity is the cosine similarity (of the normalized vectors) between

the topic distribution vector of the focal leader’s tweets and the

average distribution vector of all leaders on Twitter. Recall that

cosine similarity ranges from 0 to 1, with 1 indicating identical

vectors. The vertical axis, representing media compatibility,

indicates the difference in average sentiment between the lead-

er’s Twitter account and the leader’s press articles. Since in our

sample the Twitter sentiment was always higher than the senti-

ment of the news articles, low compatibility means more nega-

tive press coverage, and high compatibility means that the press

articles are similar in sentiment to the Twitter account. Note that

we could have chosen this axis to be the topic similarity or a

Table 6. Average Topic Distribution over Leaders, for Twitter and Press Articles.

Coef. p-Value

GDP 4.86E-07 .882 Continent

Asia .1955 .133 Europe .1511 .271 Central America .1156 .406 South America .1664 .203 North America .0389 .835

Democracy .0434 .625 Hofstede Dimensions

Power distance .0035* .034 Individualism .0033* .048 Masculinity .0011 .533 Uncertainty avoidance .0016 .311 Long-term orientation .0002 .922 Indulgence .0017 .348

*p < .05.

60 Journal of International Marketing 28(1)

combined similarity, but sentiment seems to be the most appro-

priate for testing press bias.

Figure 4 presents the mapping of the 54 leaders who had

both press articles and Twitter data in our sample. We defined

the intersection between the four quadrants arbitrarily as the

middle point of the value range for each axis. Following the

classification in Figure 1, we see that the largest group is the

Global Icons (26 leaders), who have communication topics

similar to the average and have high media compatibility

(i.e., low differences in sentiment), such as the Russian Dmitry

Medvedev, Donald Trump from the United States, and Emma-

nuel Macron from France. The second largest group is the

Narrowband Influencers (13 leaders), with high media compat-

ibility but a more peculiar topic distribution. This group

includes Pope Francis, King Salman of Saudi Arabia, and

Mohammed bin Zayed from the United Arab Emirates. The

A: Twitter Pronoun Usage

B: Press Articles Pronoun Usage

0

.01

.02

.03

.04

.05

.06

.07

A b

d e l F

a tt a h a

l- S

i s i

A b

d u lla

h b

in Z

a ye

d A

l N a h ya

n A

d e l A

lju b e ir

A le

ks a n d a r

V u ci

c A

n d ré

s M

a n

u e l L

ó p e z

O b ra

d o r

S te

ff e n S

e ib

e rt

A ri f

A lv

i B

e n ja

m in

N e ta

n ya

h u

D a n ilo

M e d in

a D

m itr

y M

e d ve

d e v

D o n a ld

T ru

m p

D o n a ld

T u sk

M a h a th

ir M

o h a m

a d

E d

g a rs

R in

ke vi

cs E

m ir T

a m

im b

in H

a m

a d

E m

m a n u e l M

a cr

o n

E rn

e st

o A

ra ú jo

E vo

M o ra

le s

G iu

se p p e C

o n te

H ill

a ry

C lin

to n

I m ra

n K

h a n

I v á n D

u q u e

Ja ir M

e ss

ia s

B o

ls o n a ro

Ja va

d Z

a ri f

J o ko

W i d

o d o

Jo rg

e A

rr e a za

M o n ts

e rr

a t

Ju a

n O

rl a n d o H

e rn

á n

d e z

Ju st

i n T

ru d e a u

K h

a lid

b in

A h m

e d

K in

g A

b d u lla

h I

I S

a lm

a n b

in A

b d u l a

zi z

A l S

a u

d K

o n o T

a r o

L e e N

a k-

ye o n

M a rc

e l o

E b ra

rd M

a rt

ín V

iz ca

rr a

M a u ri ci

o M

a cr

i M

e vl

ü t Ç

a vu

şo ğ lu

M ig

u e l D

ía z-

C a n e l

M ik

e P

o m

p e

o M

o h a m

m e d b

in Z

a ye

d A

l N a h ya

n M

o o n J

a e -i n

M u h a m

m a d u B

u h a ri

N a re

n d ra

M o d i

N ic

o lá

s M

a d u r o

P a

u l K

a g a m

e P

e d ro

S á n ch

e z

P e

t r o P

o ro

sh e n ko

P o

p e F

ra n ci

s R

a n ia

A l A

b d u lla

h R

e ce

p T

a yy

ip E

rd o ğ a n

R ic

a rd

o R

o ss

e lló

S a

a d H

a ri ri

S e

b a st

i a n

K u rz

S e

b a st

i á n

P iñ

e ra

S h

a h M

a h m

o o d Q

u re

sh i

S h

e ik

h M

o h a m

m e d A

l M a kt

o u m

S h

in zo

A b e

T e o d o ro

L o cs

i n T

h e re

s a M

a y

U h u ru

K e n ya

tt a

V la

d im

ir P

u tin

I We They You

.0

.2

.4

.6

.8

1.0

1.2

A b

d e l F

a tt a h a

l- S

is i

A b

d u lla

h b

in Z

a ye

d A

l N a h ya

n A

d e l A

lju b e ir

A le

ks a n d a r

V u ci

c A

n d ré

s M

a n

u e l L

ó p e z

O b ra

d o r

S te

ff e n S

e ib

e rt

A ri f

A lv

i B

e n ja

m in

N e ta

n ya

h u

D a n ilo

M e d in

a D

m itr

y M

e d v e

d e v

D o n a ld

T ru

m p

D o n a l d

T u sk

M a h a t h

i r M

o h a m

a d

E d

g a rs

R in

ke vi

cs E

m ir T

a m

im b

in H

a m

a d

E m

m a n u e l M

a cr

o n

E rn

e s t

o A

ra ú jo

E v o

M o ra

le s

G iu

s e p p e C

o n t e

H ill

a ry

C lin

t o n

Im ra

n K

h a n

I v á n D

u q u e

Ja ir M

e ss

ia s

B o

ls o n a r o

Ja va

d Z

a ri f

Jo ko

W id

o d o

Jo r g

e A

rr e a za

M o n ts

e rr

a t

Ju a

n O

rl a n d o H

e rn

á n

d e z

Ju st

i n T

ru d e a u

K h

a lid

b in

A h m

e d

K in

g A

b d u lla

h I

I S

a lm

a n b

in A

b d u la

zi z

A l S

a u d

K o

n o T

a ro

L e e N

a k-

y e o n

M a rc

e lo

E b r a

rd M

a rt

ín V

iz c a

rr a

M a u r i ci

o M

a cr

i M

e v l

ü t Ç

a vu

şo ğ lu

M i g

u e l D

í a z-

C a n e l

M i k

e P

o m

p e

o M

o h a m

m e d b

i n Z

a ye

d A

l N a h ya

n M

o o n J

a e -i n

M u h a m

m a d u B

u h a ri

N a re

n d ra

M o d i

N ic

o lá

s M

a d u ro

P a

u l K

a g a m

e P

e d ro

S á n ch

e z

P e

tr o P

o ro

sh e n ko

P o

p e F

ra n ci

s R

a n ia

A l A

b d u lla

h R

e ce

p T

a yy

ip E

rd o ğ a n

R ic

a rd

o R

o s s

e lló

S a

a d H

a ri ri

S e

b a st

ia n

K u rz

S e

b a s t

iá n

P iñ

e ra

S h

a h M

a h m

o o d Q

u re

sh i

S h

e ik

h M

o h a m

m e d A

l M a kt

o u m

S h

i n zo

A b e

T e o d o ro

L o cs

in T

h e re

sa M

a y

U h u ru

K e n ya

tt a

V la

d im

ir P

u tin

I We They You

Figure 3. Pronoun usage frequency per leader for Twitter (Panel A) and press articles (Panel B).

Peres et al. 61

Cosmopolitan Antagonists (nine leaders), with generic topic

distribution and lower media compatibility, included Benjamin

Netanyahu from Israel and the Canadian leader Justin Trudeau.

Finally, the Peculiar Notorious are the least frequent in our

sample, with only six leaders, most of them near the borders

of the quadrant. Their salient member is King Abdullah II from

Jordan. To examine whether the four quadrants represent any

profiling with respect to country characteristics, we ran a set of

ANOVA tests on all the country characteristics used in the

Table 7. Average Leader Polarity for Twitter and Press Articles Versus Usage of Pronouns.

Twitter Polarity Press Articles Polarity

Intercept .0748* .0715* % of “I” 1.6114 �9.2876 % of “We” 1.759* 4.0929* % of “They” �9.8802* �7.5499* % of “You” 2.7287 18.9278

*p < .05.

Table 8. Average Leader Usage of Pronouns Versus Country Characteristics.

“I,” Twitter

“We,” Twitter

“They,” Twitter

“You,” Twitter

“I,” Press

Articles

“We,” Press

Articles

“They,” Press

Articles

“You,” Press

Articles

(“I” þ “We”) � (“They” þ “You”),

Twitter

(“I” þ “We”) � (“They” þ “You”),

Press Articles

logGDP �.00005 .00033 �.00006 .00040 .00029 .00032 .00055 .00029 �.00006 �.00022 Continent

Asia .01186 .0241* .00261 .00369 �.00181 �.00122 �.00072 �.00051 .0002* �.00001 Europe .00750 .0268* .00199 �.00008 �.00036 .00058 �.00052 �.00062 �.0001* .00007 Central America .01525 .0276* .00093 .00452 �.00002 .00214 .00007 .00000 �.0004* .00002 South America .00494 .01031 .00153 .00037 .00063 .00251 �.00107 �.00024 �.00012 .00002 North America .00533 .01261 .00176 .00309 �.00044 .00035 �.00046 �.00014 .00019 �.00006

Democracy .00586 .0141* .00157 .00081 �.00098 �.00019 �.00047 �.00038 �.0002* .00004 Hofstede Dimensions

Power distance �.00002 �.00016 .00000 �.00003 .00002 .00002 .00000 .00000 .01759 �.00032 Individualism .00006 �.00008 .00002 .00007 .00000 �.00001 �.00003 �.00001 .02966 �.00180 Masculinity �.00003 �.00015 �.00001 �.00004 .00003 .00003 �.00001 .00001 .03742 .00205 Uncertainty avoidance .00005 .00019 �.00002 .00003 .00000 �.00001 .00001 .00000 .03239 .00136 Long-term orientation �.0001* �.0004* �.00003 �.00004 .00001 .00001 �.00001 .00000 .0131* .00050 Indulgence .00002 .00018 .00001 .00001 �.00004 �.00004 �.00001 �.00001 .01335 .00444

*p < .05.

A. Aljubeir

A. Vucic

S. Seibert

A. Alvi

B. Netanyahu

D. Medina

D. Medvedev

D. Trump

D.Tusk

M. Mohamad

E. Rinkevics

T. bin Hamad

E. Macron

E. Araújo E. Morales G. Conte

H. Clinton I. Khan

I. Duque J. Bolsonaro

J. Zarif J. Widodo

J. Montserrat

J. Hernández

J. Trudeau

K. bin Ahmed

Abdullah II

M. bin Salman

K. Taro

L. Nak-yeon

M. Vizcarra

M. MacriM. Díaz-CanelM. Pompeo

M. bin Zayed Moon Jae-in

M. Buhari

N. Modi

N. Maduro

P. Kagame

P. Sanchez

P. Poroshenko

Pope Francis

R. Al Abdullah

R. Erdoğan

R. Rosselló S. Hariri

S. Kurz

S. Piñera

M. Qureshi

M. Al Maktoum

S. Abe

T. Locsin

.00

.05

.10

.15

.20

.25

.50 .55 .60 .65 .70 .75 .80 .85 .90 .95 1.00

M ed

ia C

om pa

tib ili

ty :

itt er

a nd

P re

ss

( ig

h →

Universality: Topic Similarity to Other Political Leaders (Particular → Universal)

Cosmopolitan Antagonist

Narrowband Influencer

Global Icon

Peculiar Notorious

Figure 4. Mapping of the political leaders on the universality–compatibility matrix.

62 Journal of International Marketing 28(1)

previous regressions and did not see a significant difference.

We also tried to cluster the data using k-means clustering and

did not obtain meaningful insights. The results confirm the

pattern we observed in the previous analysis of the leaders’

communication: despite being strong national brands, their

topic distribution shows a generic, universal pattern, and they

have a relatively high level of media compatibility. However,

some of them talk about topics that deviate from the general

distribution, and some enjoy press coverage that reflects their

presentational media (which, interestingly, does not show

meaningful dependence on the political system of the country).

Discussion and Conclusions

The focus of this article is to study universality and media

compatibility in the communication of political leaders. One

would expect that leading politicians, being salient icons of

their countries, will adopt a country-specific communication

persona, and hence their communication will demonstrate

country-related characteristics. On the other hand, these leaders

operate in a mediatized environment. This environment is con-

sidered a driver of globalization and suggests the value of more

homogenous, universal communication.

Heavy mediatization also affects the relationship between

presentational and representational communication outlets,

which operate simultaneously and influence each other. While

variety suggests large differences between outlets, the mutual

influence might lead to high media compatibility.

Our study is the first, to the best of our knowledge, to study

these two dimensions systematically and quantitatively, using

large-scale data. We collected the communication data of 61

global world leaders from their Twitter accounts and press

coverage, and used language processing tools to study the sen-

timent, topic mixture, and use of pronouns.

Our results show a high level of universality and media

compatibility across political leaders. Leaders are similar to

each other in their sentiment, topic mixture, and use of mes-

sages. Overall, the sentiment is positive, and the discussed

topics are mainly diplomacy, economy, corruption, and the

Arab world. Of the pronouns, “we” messages are used the most,

followed by “I” messages. The results are robust across com-

munication channels, with Twitter being slightly more positive

than the press articles. Press articles focus significantly more

than Twitter on corruption, diplomacy, elections, and parlia-

mentary issues, whereas Twitter, relative to the press articles,

deals more with personal messages, such as condolences, con-

gratulations, and thank-you messages, as well as economy.

With a few exceptions, we saw almost no significant depen-

dence on country characteristics. Relative to Africa, leaders from

Asia, Europe, and Central America use first-person messages

more than second- and third-person messages when tweeting.

Also, leaders from countries with shorter-term orientation were

found to use first-person pronouns, relative to second- or third-

person pronouns, more often on Twitter, but the effects are small.

Classifying the leaders in the universality–compatibility

space, we saw that most of them fall in the universal–high

compatibility quadrant.

This research is not free of limitations. First, our selection of

leaders is biased toward leaders with high visibility in the global

arena. Local politicians, who deal less with international affairs,

might demonstrate more country-specific communication pat-

terns. Second, our choice of outlets is limited to press articles

and Twitter. However, political leaders are active on other social

platforms (e.g., Facebook, Instagram) and other representational

media, such as television, which were not studied here and could

demonstrate different patterns. Third, universality in communi-

cation can be measured in additional ways, such as by the pro-

portion of communication in English, The number of

international press conferences initiated by the leader, and the

leader’s choice of clothing. Fourth, at the technical level, lan-

guage processing, although continuously improving, does not

yet deal reliably with multilingual data sets, nonstandard lan-

guage, grammatical complexities, and meaning extraction.

Our results add to the understanding of globalization in

brands that are typically strongly associated with their country

of origin. The universalism we observe might add to the body

of results in brand research, showing the advantages of global

positioning (Diamantopoulos et al. 2019; Guo 2013; Nijssen

and Douglas 2011). Methodologically, we show how these

questions can be addressed quantitatively and systematically,

for a large number of countries, using publicly available data.

While these approaches are commonly used in marketing, our

paper suggests that they could be implemented to gain insights

in communication research and political studies.

Is the universalism we observed good or bad for the citizens

of the world? As stated in the introduction, the citizens of the

world’s nations are increasingly facing challenges of a global

nature, such as immigration, environmental issues, and the future

of the European Union and other regional treaties. Coping with

these challenges requires world leaders to create communication

infrastructures that go beyond national boundaries. Our results

provide an optimistic view that such infrastructures might exist.

Regarding media compatibility, the high compatibility level

observed between Twitter and the press articles is a reminder of

the costs that come with the mediatization that enables this

universality. Our results indicate that the continuous spillover

of content between presentational channels (e.g., social media)

and representational channels (e.g., press, television) makes

them similar to each other and raises concerns about the actual

variety of information sources available to citizens and voters,

the objectivity of these sources, and their reliability.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to

the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for

the research, authorship, and/or publication of this article: This article

Peres et al. 63

was supported by the Israeli National Foundation and the KMart

foundation.

References

Anderson, Chris (2013), “The Shared Genius of Elon Musk and Steve

Jobs,” Fortune (November 21), https://fortune.com/2013/11/21/

the-shared-genius-of-elon-musk-and-steve-jobs/.

Bennett, Lance W. and Shanto Iyengar (2008), “A New Era of Min-

imal Effects? The Changing Foundations of Political Commu-

nication,” Journal of Communication, 58 (4), 707–31.

Bos, Linda, Wouter Van der Brug, and Claes De Vreese (2011), “How

the Media Shape Perceptions of Right Wing Populist Leaders,”

Political Communication, 28 (2), 182–206.

Bovet, Alexandre and Hernán A. Makse (2019), “Influence of Fake

News in Twitter During the 2016 US Presidential Election,”

Nature Communications, 10 (1), 7–20.

Bovet, Alexandre, Flaviano Morone, and Hernán A. Makse (2018),

“Validation of Twitter Opinion Trends with National Polling

Aggregates: Hillary Clinton vs Donald Trump,” Scientific Reports,

8 (1), 8673–88.

Brants, Kees, Claes de Vreese, Judith Möller, and Philip van Praag

(2010), “The Real Spiral of Cynicism? Symbiosis and Mistrust

Between Politicians and Journalists,” International Journal of

Press/Politics, 15 (1), 25–40.

Büschken, Joachim and Greg M. Allenby (2016), “Sentence-Based

Text Analysis for Customer Reviews,” Marketing Science, 35

(6), 953–75.

Cayla, Julien and Eric J. Arnould (2008), “A Cultural Approach to

Branding in the Global Marketplace,” Journal of International

Marketing, 16 (4), 86–112.

Chadwick, Andrew (2006), Internet Politics: States, Citizens, and

New Communications Technologies. New York: Oxford Univer-

sity Press.

Chadwick, Simon and Nick Burton (2008), “From Beckham to

Ronaldo—Assessing the Nature of Football Player Brands,” Jour-

nal of Sponsorship, 1 (4), 307–17.

Couldry, Nick and Andreas Hepp (2017), “The Continuing Lure of the

Mediated Centre in Times of Deep Mediatization: Media Events and

Its Enduring Legacy,” Media, Culture & Society, 40 (1), 114–17.

Crystal, David (2006), Language and the Internet. Cambridge, UK:

Cambridge University Press.

Dahlgren, Peter (2009), Media and Political Engagement. Cambridge,

UK: Cambridge University Press.

D’Alessio, Dave and Mike Allen (2000), “Media Bias in Presidential

Elections: A Meta-Analysis,” Journal of Communication, 50 (4),

133–56.

Diamantopoulos, Adamantios, Vasileios Davvetas, Fabian Bartsch,

Timo Mandler, Maja Arslanagic-Kalajdzic, and Martin Eisend

(2019), “On the Interplay Between Consumer Dispositions and

Perceived Brand Globalness: Alternative Theoretical Perspectives

and Empirical Assessment,” Journal of International Marketing,

27 (4), 39–57.

Draper, Robert (2018), “The Man Behind the President’s Tweets,” The

New York Times Magazine (April 16), https://www.nytimes.com/

2018/04/16/magazine/dan-scavino-the-secretary-of-offense.html.

Enli, Gunn Sara and Eli Skogerbø (2013), “Personalized Campaigns in

Party-Centred Politics: Twitter and Facebook as Arenas for Polit-

ical Communication,” Information, Communication & Society, 16

(5), 757–74.

French, Alan and Gareth Smith (2010), “Measuring Political Brand

Equity: A Consumer Oriented Approach,” European Journal of

Marketing, 44 (3/4), 460–77.

Geis, Michael L. (1987), “The Problem of Bias in Political Journal-

ism,” in The Language of Politics. New York: Springer, 58–77.

Golbeck, Jennifer, Justin M. Grimes, and Anthony Rogers (2010),

“Twitter Use by the U.S. Congress,” Journal of the American

Society for Information Science and Technology, 61 (8), 1612–21.

Guo, Xiaoling (2013), “Living in a Global World: Influence of Con-

sumer Global Orientation on Attitudes Toward Global Brands from

Developed Versus Emerging Countries,” Journal of International

Marketing, 21 (1), 1–22.

Gürhan-Canli, Zeynep, Gülen Sarıal-Abi, and Ceren Hayran (2018),

“Consumers and Brands Across the Globe: Research Synthesis and

New Directions,” Journal of International Marketing, 26 (1), 96–117.

Guzman, Francisco and Vicenta Sierra (2009), “A Political Candi-

date’s Brand Image Scale: Are Political Candidates Brands?”

Brand Management, 17 (3), 207–17.

Henneberg, Stephan C. and Nicholas J. O’Shaughnessy (2007),

“Theory and Concept Development in Political Marketing: Issues

and an Agenda,” Journal of Political Marketing, 6 (2/3), 5–31.

Hepp, Andreas, Andreas Breiter, and Uwe Hasebrink (2018), Commu-

nicative Figurations: Transforming Communications in Times of

Deep Mediatization. London: Palgrave Macmillan.

Hepp, Andread, Stig Hjarvard, and Knut Lundby (2015),

“Mediatization: Theorizing the Interplay Between Media, Culture

and Society,” Media, Culture & Society, 37 (2), 314–24.

Herz, Marc and Adamantios Diamantopoulos (2017), “I Use It but

Will Tell You That I Don’t: Consumers’ Country-of-Origin Cue

Usage Denial,” Journal of International Marketing, 25 (2), 52–71.

Hjarvard, Stig (2013), The Mediatization of Culture and Society. Lon-

don: Routledge.

Hockett, Jeremy (2005), “Brand ‘W’ and the Marketing of an Amer-

ican President: Or, Logos as Logos,” Westminster Papers in Com-

munication and Culture, 2 (2), 72–95.

Hofstede, Geert (2011), “Dimensionalizing Cultures: The Hofstede

Model in Context,” Psychology and Culture, 2 (1), article 8.

Howard, Philip. N. and Malcolm R. Parks (2012), “Social Media and

Political Change: Capacity, Constraint, and Consequence,” Jour-

nal of Communication, 62 (2), 359–62.

Kock, Florian, Alexander Josiassen, and George Assaf (2019),

“Toward a Universal Account of Country-Induced Predispositions:

Integrative Framework and Measurement of Country-of-Origin

Images and Country Emotions,” Journal of International Market-

ing, 27 (3), 43–59.

Krippendorff, Klaus (2004),”Reliability in Content Analysis: Some

Common Misconceptions and Recommendations,” Human Com-

munication Research, 30 (3), 411–33.

Krotz, Friedrich (2007), “The Meta-Process of Mediatization as a

Conceptual Frame,” Global Media and Communication, 3 (3),

256–60.

64 Journal of International Marketing 28(1)

Krotz, Friedrich and Andreas Hepp (2013), “A Concretization of Med-

iatization: How Mediatization Works and Why ‘Mediatized Worlds’

Are a Helpful Concept for Empirical Mediatization Research,” Eur-

opean Journal for the Philosophy of Communication, 3 (2), 119–34.

Marshall, David P. (2015), “Monitoring Persona: Mediatized Identity and

the Edited Public Self,” Journal of Literary Studies, 28 (1), 115–33.

Nijssen, Edwin J. and Susan P. Douglas (2011), “Consumer World-

Mindedness and Attitudes Toward Product Positioning in Adver-

tising: An Examination of Global Versus Foreign Versus Local

Positioning,” Journal of International Marketing, 19 (3), 113–33.

Özsomer, Ayşegül (2012), “The Interplay Between Global and Local

Brands: A Closer Look at Perceived Brand Globalness and Local

Iconness,” Journal of International Marketing, 20 (2), 72–95.

Packard, Grant, Sarah G. Moore, and Brent McFerran (2018), “(I’m)

Happy to Help (You): The Impact of Personal Pronoun Use in

Customer–Firm Interactions,” Journal of Marketing Research, 55

(4), 541–55.

Pang, Bo and Lillian Lee (2004), “A Sentimental Education: Senti-

ment Analysis Using Subjectivity Summarization Based on Mini-

mum Cuts,” in Proceedings of the 42nd Annual Meeting of the

Association for Computational Linguistics. Stroudsburg, PA:

Association for Computational Linguistics, 271–78.

Phipps, Marcus, Jan Brace-Govan, and Colin Jevons (2010), “The

Duality of Political Brand Equity,” European Journal of Market-

ing, 44 (3/4), 496–514.

Shang, Shari S.C., Ya-Ling Wu, and Eldon Y. Li (2017), “Field

Effects of Social Media Platforms on Information-Sharing Con-

tinuance: Do Reach and Richness Matter?” Information & Man-

agement, 54 (2), 241–55.

Shapiro, Matthew A. and Libby Hemphill (2017), “Politicians and the

Policy Agenda: Does Use of Twitter by the US Congress Direct

New York Times Content?” Policy & Internet, 9 (1), 109–32.

Smith, Gareth (2009), “Conceptualizing and Testing Brand Personal-

ity in British Politics,” Journal of Political Marketing, 8 (3),

209–32.

Steenkamp, Jan-Benedict E.M. (2019), “Global Versus Local Con-

sumer Culture: Theory, Measurement, and Future Research

Directions,” Journal of International Marketing, 27 (1), 1–19.

Strömbäck, Jesper and Peter Van Aelst (2013), “Why Political

Parties Adapt to the Media: Exploring the Fourth Dimension

of Mediatization,” International Communication Gazette, 75 (4),

341–58.

Tirunillai, Seshadri and Gerard J. Tellis (2014), “Mining Marketing

Meaning from Chatter: Strategic Brand Analysis of Big Data Using

Latent Dirichlet Allocation,” Journal of Marketing Research, 51

(4), 463–79.

Ulrich, Dave and Norm Smallwood (2007), “Building a Leadership

Brand,” Harvard Business Review, 85 (7/8), 92.

Van Aelst, Peter, Bart Maddens, Jo Noppe, and Stefaan Fiers (2008),

“Politicians in the News: Media or Party Logic? Media Attention

and Electoral Success in the Belgian Election Campaign of 2003,”

European Journal of Communication, 23 (2), 193–210.

Wattal, Sunil, David Schuf, Munir Mandviwalla, and Christine B.

Williams (2010), “Web 2.0 and Politics: The 2008 U.S. Presiden-

tial Election and an E-Politics Research Agenda,” MIS Quarterly,

34 (4), 669–88.

Zhao, YanYan, Bing Qin, and Ting Liu (2010), “Sentiment Analysis,”

Journal of Software, 21 (8), 1834–48.

Peres et al. 65

Copyright of Journal of International Marketing is the property of American Marketing Association and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.