Journalism
2015, Vol. 16(3) 412­–428
© The Author(s) 2014
Reprints and permissions:
sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/1464884914521579
jou.sagepub.com
Visible and invisible countries:
News flow theory revised
Elad Segev
Tel Aviv University, Israel
Abstract
In this paper the validity and limits of the news flow theory are examined, utilizing a
large digital corpus of 35 popular news sites in 10 different languages over a threeyear
period. Three key variables were identified: GDP, foreign population and conflict
intensity, collectively accounting for more than 70% of the variance of country
prominence in the news. After offering a robust model, over-represented and underrepresented
countries are listed and divided based on their different characteristics
and news trends.
Findings show that conflicts tend to be visible only if they are in the Middle East.
Over-represented regional centers in West Europe and Asia overshadow the underrepresented
regional peripheries in East Europe and the Middle East. Finally, the US and
Africa serve as a global center and a periphery respectively. The implications of these
findings are discussed in order to invite scholars to further expand the theory and the
explanatory model accordingly.
Keywords
News flow theory, news sites, country prominence, over- and under-representation
What makes a country newsworthy? Over the years scholars have studied the relationship
between the news prominence of a country and its physical, economic, political,
social and cultural characteristics. It has been widely accepted that the international news
in different countries and languages is fundamentally different (Gerbner and Marvanyi,
1977; Kayser, 1953; Segev, 2010; Segev and Hills, 2013; Wilke, Heimprecht and Cohen,
2013), reflecting local interests and agendas. It has also been argued that the presence or
Corresponding author:
Elad Segev, Department of Communication, Tel Aviv University, P.O. Box 39040, Ramat-Aviv,
Tel Aviv 69978, Israel.
Email: eladseg@tau.ac.il
521579JOU0010.1177/1464884914521579JournalismSegev
research-article2014
Article
at University of Sussex Library on March 13, 2016jou.sagepub.comDownloaded from
Segev	 413
absence of countries in the news has further implications on decision-making (Hawkins,
2008), and on general public opinion (Wanta, Golan and Cheolhan, 2004).
While previous studies largely agree on the most important determinants of country
newsworthiness, this paper goes a step further by looking at the outliers; in other words,
countries that are over-represented or under-represented in the news when explained by
traditional factors documented by the literature. Advanced web-mining techniques are
employed in order to analyze a very large corpus of news sites from around the world,
identify and classify countries that do not fall into traditional categories, and suggest
directions to expand the news flow theory accordingly.
The current news flow theory
The different ways in which international information was reported and disseminated worldwide
had attracted academic attention already in the 1950s (International Press Institute,
1953; UNESCO, 1954). Theories attempting to explain the newsworthiness of certain issues
and actors soon emerged (Galtung and Ruge, 1965; Ostgaard, 1965). These theories further
stimulated a growing body of empirical research in order to identify the main determinants
of the international news flow (Robinson and Sparkes, 1976; Rosengren, 1970).
Today it is widely accepted that the prominence of a foreign country in the news is
attributed to three groups of variables: (a) national traits (e.g. the size and power of the
foreign country), (b) relatedness (namely, proximity to that foreign country in terms of
geography, demography, etc.) and (c) events (e.g. disasters, wars, conflict, local protest)
(Sheafer et al., 2013; Wu, 2000). While national trait variables can explain the similar
focus of news from different countries on the core countries such as the US and western
European countries, relatedness variables explain their different regional focus. Event
oriented variables can explain both the regional and global focus based on the extent of
the event and its relevance to the reporting country.
National traits
Variables of this group often measure the size of a country and its economic and political
power. Following Wallerstein’s (1974) World System Theory (WST), Chang (1998) classified
states as core, semi-peripheral, or peripheral, and found that core countries get
much greater news attention than semi-peripheral and peripheral countries. Particularly,
the economic power of a country was found to be a strong indicator of its news prominence
(Ishii, 1996; Kim and Barnett, 1996; Robinson and Sparkes, 1976; Wu, 2000,
2007). Military power is another important indicator for locating countries on that divide
(Kariel and Rosenvall, 1984; Shenhav et al., 2012), followed by population size (Charles
et al., 1979; Dupree, 1971; Rosengren, 1977).
Relatedness
The second group of variables measures the economic, political, social and cultural ties
between a reporting and a reported country. It was found, for example, that bilateral trade
is a strong predictor of the mutual newsworthiness of two countries (Charles et al., 1979;
Kariel and Rosenvall, 1984; Rosengren, 1977; Wu, 2000). Geographic proximity
at University of Sussex Library on March 13, 2016jou.sagepub.comDownloaded from
414	 Journalism 16(3)
(Dupree, 1971; Galtung and Ruge, 1965) and cultural proximity, which often refers to
ethnic similarity (Shoemaker et al., 1991), as well as immigration, travel, and shared
languages (Chang et al., 1987; Kariel and Rosenvall, 1984), were also found to be determinants
of international news coverage.
Events
The third group of variables measures the deviance of a country, such as involvement in
conflicts (Golan and Wanta, 2003) as well as the political, economic and social changes
it is experiencing (Chang et al., 1987; Shoemaker et al., 1986, 1991). The recent uprisings
in the Middle East are a good example of this, demonstrating that smaller and less
powerful countries such as Tunisia and Syria can get very high news attention worldwide,
at least for a time, due to their provision of outstanding news events. The tsunami
in Japan and the earthquake in Haiti underscore the same point.
In order to explain the news prominence of countries based on variables from some of
the groups mentioned above (hereafter: “news flow variables”), Wu (2000) studied the
international news section of newspapers from 38 countries. He used multiple regression
models for each country individually, and found that trade volume with the reporting
country and the number of international news agencies are the strongest predictors of
countries’news prominence. Additionally, when combining the data from newspapers of
all 38 countries he found that the US was by far the most prominent country, followed by
France, the UK, Russia and China. In other words, it was found that the economic power
of a country is a particularly strong predictor of its news prominence.
As a first stage, this paper systematically surveys the most significant news flow variables.
In line with the theoretical arguments and empirical findings of previous studies
(Galtung and Ruge, 1965; Ishii, 1996; Kariel and Rosenvall, 1984; Kim and Barnett,
1996; Robinson and Sparkes, 1976; Segev and Blondheim, 2013a; Wu, 2000, 2007) the
first hypothesis is that:
H1: Out of all news flow variables, the GDP of a country would have the strongest
correlation with its news prominence.
Yet, previous studies suggest other important factors. In particular, apart from national
traits, relatedness and event-oriented variables are also associated with the newsprominence
of countries. The second stage is therefore to confirm the current news flow
theory by finding the best-fit model that explains most of the news prominence of countries.
While looking at the best-fit model for each country separately, Wu (2000) found that his
news flow variables could explain between 17.2% (US news) and 36% (Iranian news) of the
variance of country prominence. In a later study, comparing online and traditional media1 in
the US alone, Wu (2007) could explain between 38% (online media) and 58.6% (traditional
media) of the variance of country prominence. While Wu’s studies examined the best-fit
model for each country separately, this study examines the prominence of a country in the
news of all other countries together. Additionally, Wu’s study does not count in his model
variables from the event group. It is therefore expected that the model in the present study
would be more comprehensive and explain more of the news prominence variance.
at University of Sussex Library on March 13, 2016jou.sagepub.comDownloaded from
Segev	 415
H2: News flow variables would explain together at least 50% of the variance of countries’
news prominence.
Once the best-fit model is found, the final and most important stage is to identify the
outliers – countries that are over-represented and under-represented. It is believed that
countries experiencing sudden incidents such as natural disasters or social protests would
get much higher news attention than expected if solely explained by news flow variables.
One of the methodological reasons for this is that the translation of these unexpected
events into measurable variables is not always possible (see Pritchard and Hughes, 2006;
and Shoemaker et al., 1986, 1991 for some examples for the operationalization of deviance
in the news). Hence, it is expected that:
H3: The news prominence of countries experiencing unexpected events would not be
explained by the news flow variables.
The outlying countries are classified based on the explanatory power of each group of
variables. This enables patterns to be found among the over- and under-represented
countries in the news, and thus also sheds light on the limitations of the current news
flow literature and the possible directions to expand and develop it further.
Methodology
News collection and analysis
The data analyzed in this study were collected from a variety of news sites in 11 different
countries (the US, the UK, Germany, France, Spain, Russia, China, Japan, Iran, Egypt
and Israel). These countries were selected on the basis of several considerations. First,
countries with a large number of online users were chosen, since they often act as cultural
and media centers for the smaller countries in their peripheries (Tunstall, 2008).
Applied to the internet, this factor must reflect the popularity of the language used in the
country (i.e. the most popular online languages such as English, Chinese, Spanish,
Japanese, French, German and Arabic were preferred).
Second, economically leading countries were chosen. As the previous studies
reviewed above suggest, a combination of economic, political and cultural factors contribute
to their news prominence. Hence, analysis of this kind should particularly include
countries with high GDPs such as Japan, China, Germany, the UK and France.
Finally, when addressing the question of news prominence, previous studies pointed
out the particularly high attention given to some Middle Eastern countries (Segev and
Blondheim, 2013b). News sites from Iran, Egypt and Israel were therefore included, as
they represent the prominent geopolitical centers of that region.
In each of the countries selected for analysis, three popular news sites were chosen for
tracking. Two of them were popular and well-established news sources, such as The New
York Times in the US, or the BBC in the UK. The third news source was the Google News
site of each country, a news aggregator of several hundreds and sometimes thousands of
popular country-specific news sources. The popularity of news sites was determined by
cross-referencing several indicators and sources, including the recent statistics provided
at University of Sussex Library on March 13, 2016jou.sagepub.comDownloaded from
416	 Journalism 16(3)
by the World Association of Newspapers (2013), the Pew Research Center (2013),
Nielsen Online (2013), and IVW (Informationsgemeinschaft zur Feststellung der
Verbreitung von Werbeträgern e.V.). The list of popular news sites was further supported
and validated by online tools such as Alexa and Google Trends.
In each news source five main topical categories were observed, including “top news,”
“world news,” “business and economy,” “technology,” and “entertainment and culture.”
These categories were chosen because they were common to all news-sites and thus enabled
a cross-national comparison. The data of each of the chosen news sites was collected
based on its RSS feeds if available or by direct web-mining of the text in each news category
every other day over a period of three years between February 1, 2009 and January
31, 2012 at 12:00 UTC. In total, 1,041,283 news items from 35 news sites2 were collected
and analyzed. It is important to note that these news items were not a random sample, but
rather the entire collection of news items that appeared in the RSS feeds of the chosen
news sites during the three-year period. Table 1 summarizes the list of countries and news
sites used for the analysis as well as the number of news items provided by each.
For each news item the date, title, content, category and source were automatically
identified and documented. The title of each news item and its content were used to
extract the countries mentioned in it. For this purpose, a database of 195 country names
in 10 different languages was built.3 Several native-speaker research assistants were
employed to translate country names into these languages. For each country name, the
research assistants were asked to provide all the common names and alternative names
(e.g. “United States,” “USA,” and so on). Then they were asked to omit all alternative
country names that might be ambiguous and therefore yield irrelevant search results.
On the basis of this list, the software could automatically identify which countries are
mentioned in each news item. A validation process was carried out by randomly choosing
100 news items and manually coding the mentioned countries. There was 78% agreement
between the human coder and the software. The difference was mainly due to the
software only counting nouns (such as “the US” or “USA”), while human coders also
include adjectives (such as “American”). This decision to strictly focus on country names
and not include nationalities was based on the fact that it would complicate the translation
process due to divergent grammatical rules in different languages. Previous studies
(Segev, 2010; Segev and Blondheim, 2013a; Segev, Sheafer and Shenhav, 2013), however,
show that focusing on country names in a high volume of news items provides a
very good proxy of the actual attention a country gives to another in its news outlets.
The main variable “News prominence” of a country was calculated based on the percentage
of news items that mentioned a country out of the total news items that mentioned
country names. For this purpose, self-reporting of country names was omitted, thus
American news items that mentioned the US or Chinese news items that mentioned China
were not included. The combined news prominence of a country by all other countries
was then tested for correlation with the various independent news flow variables listed
below in order to find the best-fit model and to identify the outlying countries.Additionally,
the monthly news prominence of the outlying countries was studied in order to examine
their trends over the three-year period and divide them into distinctive groups.
A possible caveat is that new sites within each country may vary greatly in terms of
their international scope. This could particularly be the case when comparing Google
at University of Sussex Library on March 13, 2016jou.sagepub.comDownloaded from
Segev	 417
News, an aggregator that automatically selects news items, with other news sources that
employ human editors. However, a Spearman test shows a very high correlation between
the country rankings in Google News and that of other news sites of the same county
(ranging from r = .836, p < .01 in Russia to r = .907, p < .01 in the US and France). The
country ranking correlations between news sites and Google News from different countries
were significantly lower (A Fisher z-transformation indicates significant differences
between the two with p < .01). In other words, Google News presents a significantly
similar international scope to that of other news sites from the same country.
News flow variables
A series of political, economic, social and geographic indicators was gathered based
on the three groups of variables (i.e. national traits, relatedness and event-oriented
variables) suggested by the literature surveyed above on international news flow. The
initial list included some 30 variables, but the final 14 variables that were found to be the
most significantly correlated with the news prominence of countries are reported here:
National traits.  CINC (political) – the Composite Index of National Capability developed
by Singer et al. (1972) is a comprehensive index for the general power of a country based
on its population, urban population, iron and steel production, energy consumption, military
personnel, and military expenditure. The latest CINC scores are from 2007 (http://
correlatesofwar.org).
Table 1.  Countries, news sites, and the number of news items collected for analysis.
Country News site News items Country News site News items
USA Google News 21,542 Russia Google News 22914
USA CNN 15,240 Russia Gazeta 117,125
USA The New York Times 36,269 Russia Pravda 30,989
UK Google News 36,995 Arabic Google News 22,361
UK BBC 57,922 Arabic Al Jazeera 35,181
UK Guardian 42,221 Egypt Al Ahram 3340
Israel Google News 22,536 Egypt Al Masry Alyoum 9700
Israel Ynet 24,447 Iran PressTV 2441
Israel Haaretz 8497 Iran Tabnak 71,446
Germany Google News 21,541 Iran Aftab 16,768
Germany Bild 29,300 China Google News 30,903
Germany Spiegel 29,294 China Sina 26,633
Germany Welt 4453 China People Daily 95,697
France Google News 21,304 Japan NHK 38,435
France Le Monde 26,770 Japan Yomiuri 2650
France Le Figaro 22,669 Japan Yahoo 15,518
Spain Google News 21,574  
Spain El Mundo 25,752  
Spain El Pais 30,856 Total news items 1,041,283
at University of Sussex Library on March 13, 2016jou.sagepub.comDownloaded from
418	 Journalism 16(3)
Military capability (political) – based on a qualitative assessment of the Economist
Intelligence Unit’s analysts, this variable measures the level of military sophistication
and extent of research and development in each country on a scale from 1 to 5. The
2010 scores are used in this study (http://www.visionofhumanity.org/#/page/indexes/
global-peace-index).
GDP(economic) – calculated in US dollars and based on the World Economic Outlook
Database of the International Monetary Fund. The 2010 data are used in this study (http://
www.imf.org/external/data.htm#data).
Population (social) – based on the World Population Prospects of the UN Department
of Economic and Social Affairs Population Division, the 2010 data are used in this study
(http://esa.un.org/unpd/wpp/unpp/panel_population.htm).
Area (geographic) – measures the size of a country in square kilometers based on the
Demographic Yearbook of the UN Statistics Division (http://unstats.un.org/unsd/demo-
graphic/products/dyb/dyb2008/Table03.pdf).
Relatedness4.  Bilateral trade (economic) – measures the total value of imports and exports
of a country in US dollars. The 2010 data is based on the Direction of Trade Statistics
Yearbook of the International Monetary Fund (http://www.imf.org/external/data.htm).
Foreign population (social) – the foreign-born population in a country is based on
national censuses conducted between 2000 and 2011 as reported by the UN Statistics
Division (http://data.un.org/Data.aspx?d=POP&f=tableCode%3a44).
Region (geographic) – the number of countries sharing the same region with a country
is based on the UN classification of regions (http://unstats.un.org/unsd/methods.htm).
Border (geographic) – the number of countries sharing the same border with a country
is based on the CIA World Factbook (https://www.cia.gov/library/publications/
the-world-factbook).
Event-oriented. Conflict intensity (political) – the extent of national and international
conflicts of a country on a scale from 0 to 2 between 2009 and 2010 is based on PRIO
Armed Conflict Dataset v.4-2011 (http://www.pcr.uu.se/research/ucdp/datasets/ucdp_
prio_armed_conflict_dataset).
GPI (political) – the Global Peace Index is composed by the Institute for Economics
and Peace, and measures the level of a country’s peacefulness. It is scaled from 1 to 5;
the lower the score, the more peaceful the country. The 2010 data is available from http://
www.visionofhumanity.org/#/page/indexes/global-peace-index.
Unemployment (economic) – the unemployment rate measures the percentage of the
economically active population without work. The 2010 data is based on the CIA World
Factbook (https://www.cia.gov/library/publications/the-world-factbook).
GDP change (economic) – the change in percentages between the GDP level of 2009
and that of 2010. See also the “GDP” variable.
Death/disaster (social) – the total death toll from natural disasters occurring between
2009 and 2010 based on the International Disaster Database (EM-DAT) (http://www.
emdat.be/database).
at University of Sussex Library on March 13, 2016jou.sagepub.comDownloaded from
Segev	 419
Results
Best-fit model based on the literature
Table 2 presents the Pearson correlation between the news prominence of countries and
their characteristics. In case variables were not normally distributed a logarithmic transformation
was used to allow a normal distribution of the variables.5 It shows that all variables
from the national trait group are significantly correlated with news prominence. In
other words, the size and power of a country is a particularly good predictor of its news
prominence. Of all the variables, the GDP has the strongest correlation (r = .791, p < .01)
with news prominence. A Fisher z-transformation shows that this correlation coefficient
is significantly different from that of the population variable (z = 2.79, p < .01), which
has the second highest correlation with news prominence in the national trait group.
From the relatedness group, bilateral trade (imports and exports) is a very strong predictor
of countries’ salience (r = .727, p < .01), but also the number of foreigners in a
country (r = .654, p < .01).AFisher z-transformation indicates that the difference between
the correlation coefficient of the two variables with news prominence is not significant.
Hence, the first hypothesis was partly confirmed. Indeed, the economic power of a country
is the strongest predictor of its news prominence. However, a country that has many
foreigners also has a higher chance of being mentioned in the news than countries with
few foreigners.
Finally, from the event group, the intensity of a conflict has a weak yet significant
correlation with news prominence. Similarly, the death toll from disaster is significantly
correlated with news prominence. A Fisher z-transformation indicates no significant difference
between the correlation coefficient of these two variables. In contrast, economicrelated
variables that reflect deviance or instability, such as GDPchange or unemployment
rate, are not significantly correlated with news prominence. In short, economic strength
on the one hand, and political and social instabilities on the other, have the strongest
association with news prominence.
All inner correlations between the news flow variables were studied in order to reach
the best-fit multiple regression model that explains the maximum variance of countries
news prominence. When conducting multicollinearity tests, it was found that out of all
potential variables, three variables – GDP, Foreign population and conflict intensity –
account for more than 70% of the news prominence variance (R2 = .714). This provides
strong support for the second hypothesis. Adding other variables, however, does not
increase the explanatory power of the model, and provokes a multicollinearity problem.
These three variables also have the strongest correlation with news prominence and are
taken from the three groups of variables suggested by the literature – national traits,
relatedness and events. Interestingly, they represent three different spheres of explanation
for the news prominence – economy, society and politics respectively. Table 3 presents
the correlation matrix between these variables and the results of the multiple linear
regression model. It is worth mentioning here that GDP is also strongly correlated with
Foreign population and very weakly correlated with Conflict intensity. Yet, the high tolerance
and low variance inflation factor (VIF) levels indicate that these inner correlations
do not reduce the predictive power of the model or the individual predictors.
at University of Sussex Library on March 13, 2016jou.sagepub.comDownloaded from
420	 Journalism 16(3)
The limits of the model
In order to identify the over-represented and under-represented countries in the news, Table
4 lists the outlying countries that are one standard deviation or more away from the expected
value if predicted by the model. Over-represented countries are countries that got much
higher attention in the news than would be expected by their GDP, foreign population and
involvement in conflicts. When looking carefully at the values of these variables as well as
at the monthly news trends of these countries they could be divided into four groups:
Ad hoc events.  The first group of countries includes Haiti, Iceland, Egypt, Honduras and
Madagascar. They are located in different parts of the world, they all have a relatively
low GDP and foreign population and they are not involved in major international
Table 2.  Pearson correlation of countries’ news prominence with news flow variables.
News prominence Fisher z-transformation
National traits GDP .791** z = 2.79 (p < .01)
(180)  
Population .654**  
(191)  
CINC (composite index
of national capability)
.649**  
(188)  
Area .491**  
(187)  
Military capability .470**  
(150)  
Relatedness Bilateral trade .727** z = 1.35 (n.s.)
(195)  
Foreign population .654** z = 3.35 (p < .01)
(182)  
Border .409**  
(195)  
Region .002  
(195)  
Events Conflict intensity .328** z = 0.35 (n.s.)
(191)  
Death/disaster .293** z = 1.9 (p = .05)
(153)  
GPI (global peace
index)
.080  
(147)  
GDP change .052  
(177)  
Unemployment –.030  
(142)  
Note: Numbers in parentheses indicate the total number of countries (n). The data was log-transformed
to allow a normal distribution. Variables are sorted by their correlation values to news prominence within
each group. A Fisher z-transformation shows the difference between the correlation coefficient of a variable
with that of the variable below. ** indicates p < 0.01, * indicates p < 0.05.
at University of Sussex Library on March 13, 2016jou.sagepub.comDownloaded from
Segev	 421
conflicts. What made them newsworthy was the occurrence of an unexpected major
event during the data-gathering period such as a natural disaster (in the case of Haiti and
Iceland), or a sudden uprising (in the case of Egypt, Honduras and Madagascar). This
could not be reflected by the independent variables and thus rendered these countries as
residuals with more than one standard deviation than the expected value.
In order to quantify the level of change in news attention for the over-represented countries
a surprise index was developed. It is defined as the maximum monthly news coverage
divided by the average news coverage of each country. Table 4 shows that the surprise
index for all countries in the first group is higher than five, suggesting that the news attention
for these countries was unusual during one month, namely at least five times (or in the
case of Haiti’s earthquake 13 times) more than the average. While these findings provide
an initial support for the third hypothesis, the rest of the outlying countries discussed below
fall into different categories and thus shed new light on the limits of the current model.
Ongoing political events. The second group includes countries that are involved in conflicts.
They all have medium to low levels of GDP and foreign population, yet news
around the world covers them more than would be expected. Interestingly, this group
includes: Afghanistan, Iran, Iraq, Lebanon, Jordan, Israel, Syria, Sudan and Georgia,
most of which are Middle Eastern countries. Unlike the first group of ad-hoc events,
countries with ongoing political events have a relatively low surprise index since their
involvement in conflicts is covered by the world continuously.
Ongoing economic events. The third group includes the European countries Greece and
Ireland, which have medium GDP and foreign population and no military conflicts. The
ongoing news coverage of these countries is related to their recent economic crises and
relation to the broader EU.
Table 3.  Correlation matrix and best-fit multiple linear regression of news prominence.
GDP Foreign population Conflict intensity
News prominence .791** .654** .328**
(180) (182) (191)
GDP .727** .149*
  (170) (181)
Foreign population 0.121
  (180)
R Square = .714 Durbin–Watson = 1.847
  Beta T Sig. Collinearity statistics
  Tolerance VIF
(Constant) –19.020 0.000  
GDP 2010 0.652 10.634 0.000 0.463 2.159
Foreign population 0.184 3.007 0.003 0.468 2.137
Conflict intensity 0.192 4.559 0.000 0.979 1.021
Note: Numbers in parentheses indicate the total number of countries. ** indicates p < 0.01.
VIF: Variance Inflation Factor.
at University of Sussex Library on March 13, 2016jou.sagepub.comDownloaded from
422	 Journalism 16(3)
Table 4.  List of over- and under-represented countries in the news as predicted by the best-fit
model.
Country News
prominence
(%)
Standardized
residuals
Surprise
index
GDP FP CI Group
Over-represented countries
Haiti 1.06 2.99 13.72 134 48 0 Ad hoc events
(Low GDP, FP,
No CI)
Different parts of the
world
Iceland 0.36 2.04 9.70 115 162 0
Egypt 1.28 1.52 5.68 40 78 0
Honduras 0.38 1.37 9.13 109 60 0
Madagascar 0.22 1.29 7.44 129 126 0
Afghanistan 2.84 2.40 1.76 108 102 2 Ongoing political
events (med/low
GDP, FP, high
CI) Mostly Middle
Eastern countries
Lebanon 0.96 2.07 2.04 81 80 0
Georgia 0.56 2.06 3.64 120 85 0
Israel 3.85 1.82 1.68 41 24 1
Jordan 0.54 1.75 2.09 91 113 0
Iran 3.90 1.66 2.07 26 45 1
Syria 0.66 1.52 2.35 66 104 0
Sudan 0.93 1.26 3.21 64 135 1
Iraq 2.52 1.25 1.69 62 65 2
Greece 1.58 1.35 3.24 32 22 0 Ongoing
economic events
European countries
Ireland 1.06 1.09 2.92 42 25 0
China 5.53 1.41 1.51 2 28 0 Regional center
(High GDP, FP,
no CI)
Europe and Asia
South Korea 1.97 1.37 1.75 15 98 0
UK 4.34 1.27 1.44 6 4 0
France 3.94 1.12 1.47 5 5 0
Country News
prominence
(%)
Standardized residuals GDP FP CI Group
Under-represented countries
Angola 0.07 –2.19 61 93 1 Invisible conflicts
(medium GDP,
FP, high CI)
Africa and Asia
Ethiopia 0.09 –1.50 89 90 1
Central African Republic 0.02 –1.45 159 163 1
Algeria 0.23 –1.41 49 26 1
Philippines 0.32 –1.26 45 13 1
Sri Lanka 0.24 –1.14 73 57 2
Dominican Republic 0.05 –1.78 72 35 0 Regional
peripheries
(Medium–low
GDP, FP, no CI)
Mainly East European
and Middle Eastern
peripheries
Slovakia 0.06 –1.72 60 59 0
Oman 0.07 –1.43 68 36 0
Lithuania 0.06 –1.37 84 71 0
Albania 0.04 –1.36 118 50 0
Bangladesh 0.14 –1.23 57 10 0
Slovenia 0.07 –1.08 74 108 0
Kuwait 0.09 –1.06 54 143 0
Suriname 0.01 –3.01 151 99 0 Global peripheries
(Low GDP, FP,
no CI)
Mainly Africa
Swaziland 0.01 –1.79 150 154 0
Botswana 0.02 –1.40 110 159 0
Zambia 0.04 –1.25 106 119 0
Barbados 0.02 –1.16 147 128 0
Lesotho 0.02 –1.07 155 117 0
Note: GDP and Foreign Population (FP) are represented here in their ranks out of the 200 countries rather
than by the absolute values. Conflict Intensity (CI) is based on PRIO Armed Conflict Dataset. Surprise index
is calculated as the maximum news prominence divided by the average news prominence over three years.
The list is sorted by the standardized residuals within each group of countries.
CI: Conflict Intensity, FP: Foreign Population.
at University of Sussex Library on March 13, 2016jou.sagepub.comDownloaded from
Segev	 423
Regional center.  Finally, the fourth group includes both European and Asian countries:
the UK, France, China and South Korea. These countries have relatively high levels of
GDP and foreign population, but they were not involved in major conflicts during the
data-gathering period. Still, the attention given to these countries by world news is much
higher than would be expected according to these variables.
Similarly, Table 4 lists the under-represented countries. These countries usually do
not get more than 0.1% of news attention, and can be again divided into three groups
based on the independent variables. In most cases, it would be meaningless to look at
changes in their coverage over time and compute the surprise index due to their very low
average of news coverage.
Invisible conflicts.  The first group includes countries that are involved in major conflict. Most
of them also have medium levels of GDP and foreign population, but still they were not
found to be newsworthy by news sites around the world. These countries are mainly from
Africa and Asia: Angola, Ethiopia, Central African Republic, Algeria, and the Philippines.
Regional peripheries. The second group includes countries that were not involved in
major conflicts. Their news prominence was, however, much lower than would be
expected based on their medium GDP or foreign population levels. This group includes
mainly East European and Middle Eastern peripheries: Slovakia, Slovenia, Lithuania,
Albania, Oman, Kuwait, but also the Dominican Republic and Bangladesh.
Global peripheries. Finally, the third group includes countries with very low GDP and
foreign population levels. They were also not involved in major conflicts. Still, their news
prominence was much lower than would be expected. This group includes mostly African
countries: Swaziland, Botswana, Zambia, and Lesotho, but also Suriname and Barbados.
Discussion
This paper provides a clear empirical support for the general news flow theory, but utilizing
advanced web-mining tools and a large corpus of news data over a three year
period, it indicates some of its current limitations, and identifies possible directions for
developing it a step further. Following previous studies (Ishii, 1996; Kariel and
Rosenvall, 1984; Kim and Barnett, 1996; Robinson and Sparkes, 1976; Segev and
Blondheim, 2013a; Segev and Hills, 2013; Wu, 2000, 2007), the first hypothesis was
that country’s GDP would have the strongest association with news prominence. It is
worth mentioning here that most previous studies were based on human observations
and therefore also limited in their scope. The present study confirms previous observations
in support of the first hypothesis. The economic power of a country was also the
most significant variable associated with its prominence in online news around the
world. Similarly, bilateral trade was a very prominent factor among the relatedness variables.
In line with some previous observations (Chang et al., 1987; Kariel and Rosenvall,
1984), foreign population was found to be another important factor of country news
prominence among the relatedness variables. Yet, when it comes to events, the political
and social factors (such as conflicts and death toll from disasters) rather than the economic
factors were much more significantly associated with news prominence.
at University of Sussex Library on March 13, 2016jou.sagepub.comDownloaded from
424	 Journalism 16(3)
In the search for the best-fit model, the findings show that a combination of three factors:
GDP, foreign population and conflict intensity account for more than 70% of the
variance of country news prominence. Strikingly, these variables also represent three
theoretical explanations – national traits, relatedness and events, taken from three different
spheres – economy, society and politics respectively. The model developed in this
paper outperformed the previous model suggested by Wu (2000, 2007), mainly because
it was based on a much larger data set and much longer period of observation.Additionally,
the operationalization of variables from the three different theoretical groups contributed
to locating the most powerful combination of factors. This provides an empirical support
for the second hypothesis. The remaining 30% of the news prominence variance could
be partly revealed when looking at the over- and under-represented countries and grouping
them based on their characteristics and news trends.
The third hypothesis assumed that over-represented countries would be those experiencing
unexpected events that could not be measured by the current variables. Indeed,
previous studies suggested ways to operationalize the deviance of countries (Pritchard and
Hughes, 2006; Shoemaker et al., 1986, 1991), and subsequently this study included a list
of event-oriented variables such as conflict intensity or death toll from disaster. Yet, among
the over-represented countries, there was a group of countries such as Haiti, Iceland and
Egypt, in which the present measurements could not reflect the magnitude of their deviance.
This, however, was only one group of countries found to be over-represented, providing
only a partial support for the third hypothesis. While it might be challenging to
operationalize variables that predict ad hoc events, there are still many ways to improve our
understanding of the national trait and relatedness groups of variables as suggested below.
Visible and invisible conflicts
The results suggest that conflict is indeed an important factor of the newsworthiness of
countries. Yet, there are some conflicts that are more newsworthy than others. In particular,
findings indicate that Middle Eastern conflicts are more newsworthy than African or
Asian conflicts. Over-represented countries in relation to ongoing political events and
conflicts were mainly from the Middle East. Under-represented countries in relation to
conflicts were solely from Africa and Asia.
There might be several reasons for this finding. First, the Middle East is physically
closer to the West, and particularly to the strongest and wealthiest countries in Europe,
which get more news attention anyway (see the discussion on regional centers). This
could also be related to the growing number of immigrants coming from the Middle East
to Europe. Second, all countries that were found to be over-represented in relation to
ongoing political events involve religious clashes between Muslims and Jews/Christians.
These findings provide an initial indication of the possible inclusion of religious parameters
when looking at the conflict component in relation to the news flow theory (see also
Segev and Blondheim, 2013b).
Apart from political conflicts, two other over-represented countries – Ireland and
Greece – were identified. Being an integral part of the EU it is clear why a westerndominated
international news system would give priority to ongoing economic changes
in Europe. Yet, in order to include these middle-term events, scholars should find creative
ways to operationalize them into measurable variables. Similar to ongoing political
at University of Sussex Library on March 13, 2016jou.sagepub.comDownloaded from
Segev	 425
events, in the case of ongoing economic events, variables should also interact with the
general national trait variables, since findings clearly show that middle-term events are
reported only when they are related to the economically leading countries.
Regional centers and peripheries
Findings indicate that over-represented countries with a high GDP and foreign population
but with no major military conflicts are mostly the regional centers of West Europe
and Asia. While previous studies already indicated the importance of regional centers
(Tunstall, 2008), and particularly the European ones when it comes to the news (Kim and
Barnett, 1996; Wu, 2000, 2004), the present findings indicate that over-representation
cannot be attributed solely to their economic power. Perhaps the gap between their
observed news prominence and their expected values should also be understood in the
regional historical and cultural context (particular when it comes to the UK, France and
China). There is possibly also a certain “the rich get richer” effect in news prominence.
The most prominent country in a region – for example, the leading regional economy, the
leading regional military power, or the leading regional conflict – gets more news attention
because of its being number one. These new considerations should be taken into
account in future studies attempting to operationalize and provide a broader explanation
for the news flow theory.
On the other hand, under-represented countries with a medium GDP and foreign population
but with no major conflicts are mostly regional peripheries in East Europe and the
Middle East. Countries such as Slovakia, Slovenia, Lithuania, Oman or Kuwait are
expected to have a greater news prominence based on their economies, but are forgotten,
perhaps being in the shadow of the bigger regional players. This observation was not
taken into account in the previous studies reviewed above.
Global centers and peripheries
At the global level, under-represented countries were mostly from Africa. Despite the
relatively low values of their independent variables, they were expected to generate some
level of news coverage. Yet, it is possible to say that these countries are literally invisible
to the world. While this finding is not new (see for example Golan, 2008), it certainly
requires operationalization in a new model of news flow. Perhaps, future studies should
first identify the core economic and political regional centers, and only then re-calculate
the traditional relatedness and event-oriented variables based on their economic and
political interactions with the core countries.
On the other hand, a candidate for an over-represented country in the global level
would be the US. Indeed, the data in this study confirm that the US was about one standard
deviation away from its expected value. It was, however, not included among the top
outliers. The US is a particular case since its GDP and foreign population level is the highest
by far and it is certainly involved in various international conflicts. Its news prominence
around the world reflects this trend closely, with 18.61% of all news items
mentioning the US. The US would be therefore a very good example for a world hegemon,
demonstrating that the “rich get richer” effect also works on the global level. The US gets
much higher news attention than the rest of the world not only because of its high
at University of Sussex Library on March 13, 2016jou.sagepub.comDownloaded from
426	 Journalism 16(3)
economic, social or political power, but also because of its being number one (see also
Segev and Blondheim, 2013a for a broader study of the US’s news dominance).
It is important to acknowledge, however, the boundaries of this study. First, the present
analysis is based on news sites from countries with the largest number of online
users. An ideal study should include more countries representing more regions.
Specifically, there are much fewer news flow studies in developing countries. Although
they get most of their news from international news agencies, their news flow might well
be more regional in focus and thus require further modifications in the current theory.
Second, one of the assumptions of many news flow related studies is that “external”
variables such as economic power or political events in a country can explain its news
prominence around the world. In fact, there are also various “internal” factors such as the
structure of international telecommunications, the presence of news agencies (Wu,
2000), as well as the editorial practices and traditions that affect the news prominence of
a country. Yet, the approach taken here and in many similar studies is that the external
factors also have a certain effect on internal factors, particularly when looking at a long
period of news coverage from various news sources. For example, although one outlet in
China may take a more regional approach than another, they will both find it difficult to
ignore the economic crisis in the US, which directly affects the Chinese economy as
well. It might be worth, however, examining more thoroughly this specific relationship
between external variables and those related to the editorial decision-making and the
journalistic tradition in each outlet separately.
It is out of the scope of this study to draw a comprehensive picture of the news coverage
in the entire world, or to explore the journalistic practices behind it. The main goal of this
paper was rather to develop a strong predictive model of news flow based on the existing
theory, and to unveil its limitations. To this end, the web-mining tools employed here
allowed the exploration of a much larger corpus of data and many more languages than
those conducted previously. Findings clearly indicate the visible and the invisible countries
in world news based on their relation to the local, regional and global economy, politics and
society. Most importantly, they invite scholars to expand the news flow theory accordingly.
Funding
This research received no specific grant from any funding agency in the public, commercial, or
not-for-profit sectors.
Notes
1.	 Wu studied the countries mentioned in websites of The New York Times and the CNN, comparing
them with the countries mentioned in the newspaper and the TV channel respectively.
2.	 Google News did not include special Egyptian or Iranian editions in the course of this study,
but its Arabic edition as well as Al Jazeera were used as more general and popular news
sources in the Arab world.
3.	 This database was based on the most complete list of country names available from ISO
(International Organization for Standardization). This list was translated into the following
languages: English, French, German, Spanish, Russian, Chinese (Mandarin), Japanese,
Persian, Arabic and Hebrew.
at University of Sussex Library on March 13, 2016jou.sagepub.comDownloaded from
Segev	 427
4.	 Since the dependent variable – countries’ news prominence – is an aggregation of news from
all countries, the relatedness variables are also built as an aggregation of the total relations of
a country with the outside world. For example, the bilateral trade variable represents the total
imports and exports of each country with all other countries.
5.	 The logarithmic transformation also ensured that there were no outliers to significantly divert
the correlation results. Another validation was a Spearman correlation test, which yielded
identical results.
References
Chang TK (1998) All countries not created equal to be news: World system and international communication.
Communication Research 25: 528–563.
Chang TK, Shoemaker P and Brendlinger N (1987) Determinants of international news coverage
in the US media. Communication Research 14: 396–414.
Charles J, Shore L and Todd R (1979) The New York Times coverage of equatorial and lower
Africa. Journal of Communication 29: 148–155.
Dupree JD (1971) International communication: View from ‘a window on the world’. The
International Communication Gazette 17: 224–235.
Galtung J and Ruge MH (1965) The structure of foreign news. Journal of Peace Research 2(1): 64–91.
GerbnerGandMarvanyiG(1977)Themanyworldsoftheworld’spress.JournalofCommunication
27(1): 52–66.
Golan GJ (2008) Where in the world is Africa? Predicting coverage of Africa by US television
networks. The International Communication Gazette 70: 41–57.
Golan G and Wanta W (2003) International elections on the US network news: An examination of
factors affecting newsworthiness. Gazette 65(1): 25–40.
Hawkins V (2008) Stealth Conflicts: How the World’s Worst Violence is Ignored. Hampshire,
UK: Ashgate.
International Press Institute (1953) The Flow of News. Zurich: International Press Institute.
Ishii K (1996) Is the U.S. over-reported in the Japanese press? Factors accounting for international
news in the Asahi. Gazette 57: 134–144.
Kariel HG and Rosenvall LA (1984) Factors influencing international news flow. Journalism
Quarterly 60: 434–436.
Kayser J (1953) One Week’s News: Comparative Study of 17 Major Dailies for a Seven-Day
Period. Paris: UNESCO.
Kim K and Barnett GA (1996) The determinants of international news flow: A network analysis.
Communication Research 23(3): 323–352.
Nielsen Online (2013) Today’s top headline: 111 million Americans go on the Web for daily
news. Available at: http://www.nielsen.com/us/en/newswire/2013/todays-top-headline-
111-million-americans-go-on-the-web-for.html (accessed 16 Feburary 2014).
Ostgaard E (1965) Factors influencing the flow of news. Journal of Peace Research 2(1): 39–64.
Pew Research Center (2013) The State of the News Media: An Annual Report on American
Journalism. Available at: http://stateofthemedia.org/2013/newspapers-stabilizing-but-stillthreatened/newspapers-by-the-numbers
(accessed 16 Feburary 2014).
Pritchard D and Hughes KD (2006) Patterns of deviance in crime news. Journal of Communication
47(3): 49–67.
Robinson GJ and Sparkes VM (1976) International news in the Canadian and American press:
Comparative news flow study. International Communication Gazette 22: 203–218.
Rosengren KE (1970) International news: Intra and extra media data. Acta Sociologica 8(2):
96–109.
at University of Sussex Library on March 13, 2016jou.sagepub.comDownloaded from
428	 Journalism 16(3)
Rosengren KE (1977) Four types of tables. Journal of Communication 27: 67–75.
Segev E (2010) Mapping the international: Global and local salience and news-links between countries
in popular news sites worldwide. International Journal of Internet Science 5(1): 48–71.
Segev E and Blondheim M (2013a) America’s global standing according to popular news sites
from around the world. Political Communication 30(1): 139–161.
Segev E and Blondheim M (2013b) Online news about Israel and Palestine: A cross-national comparison
of prominence and trends. Digital Journalism 1(3): 1–13.
Segev E and Hills T (2013) When news and memory come apart: A cross-national comparison of
countries’ mentions. International Communication Gazette 76(1): 67–85.
Segev E, Sheafer T and Shenhav S (2013) Is the world getting flatter? A new method for examining
structural trends in the news. Journal of the American Society for Information Science and
Technology (JASIST) 64(12): 2537–2547.
Sheafer T, Ben Nun P, and Shenhav S, et al. (2013) The conditional nature of value-based proximity
between countries: Strategic implications for mediated public diplomacy. American
Behavioral Scientist (ABS) 57(9): 1256–1276.
Shenhav SR, Rahat G and Sheafer T (2012) Testing the language–power assumption of critical
discourse analysis: The case of Israel’s legislative discourse. Canadian Journal of Political
Science 45(1): 207–222.
Shoemaker PJ, Chang TK and Brendlinger N (1986) Deviance as a predictor of newsworthiness:
Coverage of international events in the US media. Communication Yearbook 10: 348–365.
Shoemaker PJ, Danielian LH and Brendlinger N (1991) Deviant acts, risky business and US interests:
The newsworthiness of world events. Journalism Quarterly 68: 781–795.
Singer JD, Bremer S and Stuckey J (1972) Capability distribution, uncertainty, and major power
war, 1820–1965. In: Russett B (ed.) Peace, War, and Numbers. Beverly Hills: Sage, pp. 19–48.
Tunstall J (2008) The Media Were American: US Mass Media in Decline. Oxford: Oxford
University Press.
UNESCO (1954) How Nations See Each Other? Paris: UNESCO Publications.
Wallerstein I (1974) The Modern World-System. New York: Academic Press.
Wanta W, Golan G and Cheolhan L (2004) Agenda setting and international news: Media influence
on public perceptions of foreign nations. Journalism and Mass Communication 81(2): 364–377.
Wilke J, Heimprecht C and Cohen A (2012) The geography of foreign news on television: A
comparative study of 17 countries. International Communication Gazette 74(4): 301–322.
World Association of Newspapers and News Publishers (2013). WPT Database. Available at:
http://www.wptdatabase.org (accessed 16 Feburary 2014).
Wu DH (2000) Systematic determinants of international news coverage. Journal of Communication
50(2): 110–130.
Wu DH (2004) The world’s windows to the world: An overview of 44 nations’ international
news coverage. In: Paterson C and Sreberny A (eds) International News in the Twenty-First
Century. UK: John Libbey Publishing, pp. 95–108.
Wu DH (2007) A brave new world for international news? Exploring the determinants of the coverage
of foreign news on US Websites. International Communication Gazette 69(6): 539–552.
Author biography
Elad Segev is lecturer of new media in the Department of Communication at Tel Aviv University. His
research focuses on international news, search engines, public understanding of science, and new
applications and methodologies to utilize data mining for social and communication research. His
studies are published in books and journals in the field of communication and new media such as the
Journal of Computer-Mediated Communication, Political Communication, The International Journal
of Press/Politics, and The Journal of the American Society for Information Science and Technology.
at University of Sussex Library on March 13, 2016jou.sagepub.comDownloaded from