bringing the media in: newspaper readership and human rights

25
Bringing the Media in: Newspaper Readership and Human Rights* Rob Clark, University of Oklahoma Cross-national studies examining human rights outcomes have seldom considered the role of the news media. This is unfortunate, as a large body of work in media stud- ies suggests that the news industry effectively educates citizens, shapes public attitudes, and stimulates political action. I juxtapose these two literatures in a cross-national con- text to examine the print media’s impact on a state’s human rights performance. First, examining micro-level evidence from the World Values Survey, I show that an individ- ual’s level of media consumption, including newspaper readership, is positively associ- ated with participation in human rights organizations. Next, I present macro-level evidence regarding the aggregate effect of a society’s newspaper readership on its human rights record. Analyzing an unbalanced dataset with a maximum of 459 obser- vations across 138 countries covering four waves during the 1980–2000 period, I use ordered probit regression to examine the relationship between a state’s newspaper read- ership and its Amnesty International rating. I find that newspaper readership exerts strong, positive effects on a state’s human rights practices net of other standard predic- tors and temporal regional controls. Moreover, the effect of readership is robust to a number of alternative specifications that address concerns with ceiling effects, measure- ment bias, influential observations, sample composition, mediation, endogeneity, and the impact of alternative forms of media consumption (i.e., the Internet and television). Introduction Previous cross-national studies have linked a state’s human rights perfor- mance to a number of domestic and international factors (Hafner-Burton and Tsutsui 2005; Poe, Tate, and Keith 1999). However, less attention has been directed toward the potential role of the news media in shaping a state’s human rights performance. In theory, the news media occupy a critical posi- tion in society. The media industry is responsible for educating citizens and disseminating important information. Journalists function as state monitors and corporate watchdogs, holding private and public institutions accountable for their actions. In addition, media outlets serve as venues that foster public dis- course where citizens can exchange ideas, form opinions, and take appropriate action. In sum, the news media represent important monitors of societal operations. Sociological Inquiry, Vol. xx, No. x, 2012, 1–25 Ó 2012 Alpha Kappa Delta DOI: 10.1111/j.1475-682X.2012.00417.x

Upload: rob-clark

Post on 30-Sep-2016

212 views

Category:

Documents


0 download

TRANSCRIPT

Rights*

Bringing the Media in: Newspaper Readership and Human

Rob Clark, University of Oklahoma

Socio

� 20

DOI:

Cross-national studies examining human rights outcomes have seldom considered

the role of the news media. This is unfortunate, as a large body of work in media stud-

ies suggests that the news industry effectively educates citizens, shapes public attitudes,

and stimulates political action. I juxtapose these two literatures in a cross-national con-

text to examine the print media’s impact on a state’s human rights performance. First,

examining micro-level evidence from the World Values Survey, I show that an individ-

ual’s level of media consumption, including newspaper readership, is positively associ-

ated with participation in human rights organizations. Next, I present macro-level

evidence regarding the aggregate effect of a society’s newspaper readership on its

human rights record. Analyzing an unbalanced dataset with a maximum of 459 obser-

vations across 138 countries covering four waves during the 1980–2000 period, I use

ordered probit regression to examine the relationship between a state’s newspaper read-

ership and its Amnesty International rating. I find that newspaper readership exerts

strong, positive effects on a state’s human rights practices net of other standard predic-

tors and temporal ⁄ regional controls. Moreover, the effect of readership is robust to a

number of alternative specifications that address concerns with ceiling effects, measure-

ment bias, influential observations, sample composition, mediation, endogeneity, and

the impact of alternative forms of media consumption (i.e., the Internet and television).

Introduction

Previous cross-national studies have linked a state’s human rights perfor-

mance to a number of domestic and international factors (Hafner-Burton and

Tsutsui 2005; Poe, Tate, and Keith 1999). However, less attention has been

directed toward the potential role of the news media in shaping a state’s

human rights performance. In theory, the news media occupy a critical posi-

tion in society. The media industry is responsible for educating citizens and

disseminating important information. Journalists function as state monitors and

corporate watchdogs, holding private and public institutions accountable for

their actions. In addition, media outlets serve as venues that foster public dis-

course where citizens can exchange ideas, form opinions, and take appropriate

action. In sum, the news media represent important monitors of societal

operations.

logical Inquiry, Vol. xx, No. x, 2012, 1–25

12 Alpha Kappa Delta

10.1111/j.1475-682X.2012.00417.x

2 ROB CLARK

The news industry may also play an active role in monitoring human

rights activities of the state. Research suggests that international media cover-

age is highly responsive to human rights conditions across the world (Ramos,

Ron, and Thoms 2007). In fact, human rights have become an increasingly

salient theme in media coverage over the past several decades. Between 1982

and 1994, Reuters World Service increased its human rights coverage by 500

percent, while the British Broadcasting Corporation’s Summary of World

Broadcasts experienced a 600 percent increase over the same period (Ron

1997). And between 1986 and 2000, Newsweek increased its human rights

coverage by 100 percent, while the Economist increased its coverage by 300

percent (Ramos, Ron, and Thoms 2007). More importantly, past research sug-

gests that media exposure has the capacity to shape attitudes and behavior

with respect to human rights. One study surveying the attitudes of African

women toward female genital cutting found that media exposure negatively

affected a woman’s attitude toward female circumcision and reduced the odds

that these women would circumcise their daughters (Boyle, McMorris, and

Gomez 2002).

In this study, I fill a notable gap in the literature by assessing the print

media’s impact on a country’s human rights performance. First, I review pre-

vious studies that demonstrate the newspaper industry’s capacity to educate

citizens, shape their attitudes, and mobilize them to become politically active.

Next, using data from the World Values Survey, I examine micro-level evi-

dence linking media consumption to human rights activities, showing that two

survey items (regularly reading a daily newspaper and actively following the

news on television, radio, or daily newspapers) are positively associated with

participation in human rights organizations net of controls. Finally, I present

analyses of macro-level data that show the positive impact of newspaper read-

ership on a country’s human rights record. I conclude by underscoring the

importance of the news media in empowering citizens to monitor and regulate

state behavior.

Media Studies

The Impact of Newspapers

A number of empirical studies have documented a range of empowering

effects that the news media has had on citizens. First, the industry is generally

regarded as a popular source for acquiring knowledge. ‘‘The most regularly

used information source, by virtually all accounts, is the mass media, espe-

cially television and the press’’ (Dalton, Beck, and Huckfeldt 1998:111). In

particular, several early studies provided initial support for the idea that

newspaper exposure positively predicts knowledge. One study found that

NEWSPAPER READERSHIP AND HUMAN RIGHTS 3

individuals living in communities with more daily newspapers were able to

identify more problems facing their state (or the country) than individuals

living in areas with fewer newspapers (Chaffee and Wilson 1977). Another

early study found that political coverage in newspapers positively predicts the

number of reasons individuals are able to give for supporting one political

candidate over another (Clarke and Fredin 1978). Since that time, other

research has shown that newspaper reading positively affects political

knowledge (Kleinnijenhuis 1991; Nicholson 2003; Pattie and Johnston 2003).

In addition, newspapers have the capacity to shape public opinion and

mobilize readers into political action. Newspaper exposure stimulates different

forms of political participation, such as attending a political meeting, writing a

letter to the editor of a newspaper, television, or magazine, contacting an

elected official, working for a political campaign, or circulating a petition

(McLeod and McDonald 1985; McLeod, Scheufele, and Moy 1999). In fact,

newspaper reading has been shown to positively affect voter turnout among

young adults (Bybee et al. 1981).

Extending this line of research to other media forms, studies have also

examined the impact of television news on viewers. However, researchers gen-

erally conclude that newspaper reading yields far greater returns than televi-

sion viewing, and there are strong theoretical reasons for believing so. First,

television consumers have less control over the pace at which information is

digested. Thus, not surprisingly, experimental research has shown that individ-

uals are able to recall news stories presented in print more easily than those

presented through television or radio (DeFleur et al. 1992). And, second, tele-

vision news provides considerably less information than newspapers. Televi-

sion broadcasts feature fewer words and ideas per news story than appear in a

front page newspaper article (McLeod, Scheufele, and Moy 1999:321; Robinson

and Davis 1990:109).

A number of empirical studies have shown that newspaper readers are

better informed than television viewers. Several studies have found that news-

paper reading positively predicts political knowledge with an effect about

twice as large as TV news viewing (Kwak 1999; McLeod, Scheufele, and

Moy 1999). In other research, reading newspapers significantly increases

knowledge and awareness of political issues, while watching TV news does

not (Robinson and Davis 1990; Scheufele, Shanahan, and Kim 2002). In addi-

tion, research suggests that newspapers better equip citizens with knowledge

of political candidates (Becker and Dunwoody 1982). Past research has shown

that reliance on newspapers for political information significantly increases a

voter’s ability to discriminate between candidates on specific issues, while

reliance on television, radio, and magazines for political information does not

(Berkowitz and Pritchard 1989).

4 ROB CLARK

Moreover, studies find that exposure to newspapers produces more active

citizens than exposure to TV news. For example, several studies find that the

effect of newspaper reading on political knowledge and participation is nota-

bly stronger than that of watching television news (McLeod and McDonald

1985; Scheufele, Shanahan, and Kim 2002). And yet another study reports that

newspaper reading positively predicts political knowledge, voting behavior,

and overall political participation, while TV news viewing has no significant

impact on any of these outcomes (Eveland and Scheufele 2000). Finally, in a

study of university students, exposure to print media was a better predictor of

political knowledge and participation (e.g., voting, demonstrating, participation

in political meetings and discussion) than exposure to TV news (Feldman and

Kawakami 1991). In sum, these studies suggest that it is important to distin-

guish between television and newspapers when considering the capacity for

the news media to educate citizens, shape public sentiment, and stimulate

political action. Thus, in this study, I investigate the mobilizing impact of the

newspaper industry. Specifically, I examine whether the print media can effec-

tively trigger social change in one of the most important sectors of govern-

ment activity at the cross-national level: human rights practices.

Generalizability

However, one reason for skepticism is that most of the above studies

were conducted within the United States, possibly leading one to draw conclu-

sions about the print media’s efficacy that may not be warranted in different

societal contexts. Interestingly, though, the positive effects of media consump-

tion on the mass public do emerge elsewhere. Comparative studies covering

European nations have linked media exposure to knowledge (Curran et al.

2009; Kleinnijenhuis 1991; Pattie and Johnston 2003). And one study conducted

in Japan shows that exposure to, and reliance on, print media for political infor-

mation positively predicts political knowledge and participation (Feldman and

Kawakami 1991). These findings have even been extended to the developing

world. For example, in Mali, exposure to television and radio news positively

predicts political knowledge (e.g., correctly naming political incumbents),

while exposure to newspaper and radio news positively predicts political par-

ticipation (e.g., attending a community meeting, demonstration, or election

rally, working for a political candidate or party, or signing a petition) (Nisbet

2008).

Analytic Strategy

To demonstrate the link between a country’s newspaper industry and its

human rights performance, I present evidence in two stages. Because

the impact of the print media theoretically operates at the micro-level, it is

NEWSPAPER READERSHIP AND HUMAN RIGHTS 5

important to first establish an empirical connection at the level of the individ-

ual (stage one) prior to addressing the primary research question regarding the

macro-level impact of newspaper readership on human rights practices (stage

two). Thus, I first examine micro-level evidence linking media consumption to

participation in human rights organizations. Prior studies have shown that

news media exposure, especially print media, leads individuals to become

more politically active. Accordingly, I examine whether individuals that read

daily newspapers and follow television or radio news are more likely to join

human rights organizations. I then shift to my primary analyses, examining

macro-level evidence linking newspaper readership at the societal level to a

country’s human rights practices. At the aggregate level, I expect the cumula-

tive effect of print media exposure to significantly improve a country’s human

rights record.

Micro-Level Evidence

If exposure to the news media has a mobilizing effect on the citizenry in

the area of human rights, we should expect to see evidence of this at the indi-

vidual level. In this section, I draw from data covering several waves of

the World Values Survey (2006) to examine whether media consumption is

Table 1Membership in Human Rights Organizations by Media Consumption

Membership in human

rights organizations

Regularly read a daily newspaper?

Wave: 1981–1984

0 = No 1.36% (79 of 5,792)

1 = Yes 2.62% (356 of 13,572)

Follow news on television, radio, or daily papers?

Wave: 1999–2004

1 = Never 1.78% (112 of 6,292)

2 = Less often 2.60% (329 of 12,664)

3 = Once or twice a week 2.66% (249 of 9,374)

4 = Several times a week 3.04% (470 of 15,444)

5 = Every day 4.16% (1,422 of 34,180)

Source: World Values Survey (2006).

6 ROB CLARK

positively associated with participation in human rights institutions. Table 1

presents two questionnaire items. Both items are behavioral questions, asking

respondents to indicate (1) whether they regularly read a daily newspaper (i.e.,

four of every six issues) (asked during the 1981–1984 survey wave), and (2)

how often they follow politics in the news on television, the radio, or daily

newspapers (asked during the 1999–2004 survey wave). I organize the

responses to these items according to whether or not these same individuals

reported belonging to a human rights organization. As Table 1 indicates, those

who regularly read newspapers were about twice as likely (2.62%) to report

being a member of a human rights organization as those who were not regular

readers (1.36%). Similarly, those who reported following the news more fre-

quently on television, radio, or newspapers were progressively more likely to

participate in a human rights organization, ranging from ‘‘never’’ following

the news (1.78% participation rate) to following the news ‘‘every day’’ (4.16%

participation rate). In sum, these results suggest that media consumption in

general, and newspaper readership in particular, is positively associated with

participation in human rights activities.

In Table 2, I examine whether these differences across levels of media

consumption are statistically significant, net of controls, using a set of probit

models. All models include a set of country dummies (results not reported to

preserve space) and are estimated with robust standard errors. Models 1–3

consist of respondents from 16 Western countries from the 1981–1984 wave

of the World Values Survey, while models 4–6 consist of respondents from a

broader set of 24 countries from the 1999–2004 wave. Because different coun-

tries participate in different survey waves, and exposure to specific survey

items may vary across both country and wave, the resulting samples for these

models are different. Prior to analyses, I standardized all predictors (mean = 0;

standard deviation = 1) so that the reported coefficients allow interested read-

ers to compare the magnitude of each variable’s effect. Thus, each cell reports

the standardized coefficient with the robust standard error in parentheses. After

estimating the focal predictors in isolation, I add controls for each respon-

dent’s gender, age, education, and income (both education and income are

ordinally coded as low, medium, and high). In addition, I include three rele-

vant attitudinal measures, including (1) an item that asks respondents to evalu-

ate their country’s level of respect for individual human rights, ranging from

‘‘a lot of respect’’ (low) to ‘‘no respect at all’’ (high), (2) an item that asks

respondents whether they think human rights policies should be left to national

governments to decide for themselves (low), decided by national governments

with coordination from the United Nations (middle), or handled exclusively by

the United Nations (high), and (3) an item that asks respondents to indicate

which geographic group they belong to most: locality, region, country,

Ta

ble

2P

rob

itM

od

els

of

Mem

ber

ship

inH

um

anR

igh

tsO

rgan

izat

ion

sa

Mo

del

1M

od

el2

Mo

del

3M

od

el4

Mo

del

5M

od

el6

Gen

der

(0=

mal

e;1

=fe

mal

e).0

07

(.0

22

).0

06

(.0

22

))

.02

7†

(.0

15

))

.02

7†

(.0

15

)A

ge

(15

–1

01

yea

rs)

.04

0†

(.0

24

).1

57

(.1

37

).0

12

(.0

16

))

.12

4(.

08

1)

Ag

e(s

qu

ared

))

.11

9(.

14

1)

.13

9†

(.0

82

)E

du

cati

on

lev

el(1

=lo

w;

2=

mid

dle

;3

=h

igh

).0

82

**

*(.

01

7)

.08

3*

**

(.0

17

)

Inco

me

lev

el(1

=lo

w;

2=

mid

dle

;3

=h

igh

).1

19

**

*(.

02

4)

.11

3*

**

(.0

25

).0

57

**

*(.

01

6)

.06

0*

**

(.0

16

)

Hu

man

rig

hts

resp

ect

(1=

resp

ect;

4=

no

resp

ect)

.00

3(.

01

7)

.00

3(.

01

7)

Hu

man

rig

hts

dec

isio

ns

(1=

cou

ntr

y;

3=

UN

))

.00

8(.

01

6)

).0

08

(.0

16

)

NEWSPAPER READERSHIP AND HUMAN RIGHTS 7

Ta

ble

2(C

on

tin

ued

)

Mo

del

1M

od

el2

Mo

del

3M

od

el4

Mo

del

5M

od

el6

Geo

gra

ph

icid

enti

ty(0

=o

ther

;1

=w

orl

d)

.06

9*

**

(.0

14

).0

69

**

*(.

01

4)

Rea

dn

ewsp

aper

reg

ula

rly

?(0

=n

o;

1=

yes

).0

81

**

(.0

27

).0

63

*(.

02

7)

.06

1*

(.0

27

)

Fo

llo

wn

ews

inm

edia

?(1

=n

ever

;5

=ev

ery

day

).1

38

**

*(.

01

7)

.11

3*

**

(.0

18

).1

13

**

*(.

01

8)

Sta

tes

16

16

16

24

24

24

Ob

serv

atio

ns

15

,96

71

5,9

67

15

,96

72

5,1

78

25

,17

82

5,1

78

Pse

ud

oR

-sq

uar

e.0

44

.05

2.0

52

.10

9.1

19

.11

9B

IC3

,64

03

,64

23

,65

18

,35

58

,33

88

,34

6

No

te:

Eac

hce

llre

po

rts

the

stan

dar

diz

edco

effi

cien

tw

ith

the

rob

ust

stan

dar

der

ror

inp

aren

thes

es.

All

mo

del

sin

clu

de

ase

to

fco

un

try

du

mm

ies

(res

ult

sn

ot

rep

ort

edto

pre

serv

esp

ace)

.aM

od

els

1–

3in

clu

de

the

foll

ow

ing

cou

ntr

ies:

Bel

giu

m,

Can

ada,

Den

mar

k,

Fra

nce

,G

erm

any

,Ic

elan

d,

Irel

and

,It

aly

,M

alta

,th

eN

eth

erla

nd

s,N

ort

her

nIr

elan

d,

No

rway

,S

pai

n,

Sw

eden

,U

nit

edK

ing

do

m,

and

Un

ited

Sta

tes;

Mo

del

s4

–6

incl

ud

eth

efo

llo

win

gco

un

trie

s:A

lban

ia,

Arg

enti

na,

Ban

gla

des

h,

Bo

snia

-Her

zeg

ov

ina,

Can

ada,

Ch

ile,

Ch

ina,

Ind

ia,

Jap

an,

Ky

rgy

zsta

n,

Mac

edo

nia

,M

exic

o,

Mo

ldo

va,

Mo

rocc

o,

Ph

ilip

pin

es,

So

uth

Afr

ica,

So

uth

Ko

rea,

Sp

ain

,S

wed

en,

Tan

zan

ia,

Ug

and

a,U

nit

edS

tate

s,V

ietn

am,

and

Zim

bab

we.

†p

<.1

;*p

<.0

5;

**p

<.0

1;

**

*p

<.0

01

(tw

o-t

aile

dte

sts)

.

8 ROB CLARK

NEWSPAPER READERSHIP AND HUMAN RIGHTS 9

continent, or the world. I dichotomize this last item into ‘‘world versus other’’

(world = 1). Models from the earlier survey wave do not include controls for

education and the attitudinal measures because of missing data.

Overall, these models indicate that media consumption is a positive pre-

dictor of participation in human rights activities. Specifically, model 1 shows

that newspaper readership positively predicts membership in a human rights

organization, and this relationship persists net of the controls introduced in

models 2 and 3 (I add a second-order polynomial for age in model 3, as the

elderly may participate less actively than the middle-aged). Similarly, model 4

shows that media consumption, more generally, is a positive predictor of

human rights activity, and this association holds net of a larger set of controls.

In sum, these models suggest that media consumption is a significant predictor

of membership in human rights organizations. Thus, the descriptive patterns

identified in Table 1 hold up in a multivariate context, as revealed by the anal-

yses presented in Table 2.

Interestingly, there is some indication that these human rights practices at

the micro-level may be linked to human rights outcomes at the societal level.

For example, among the Western countries featured in models 1–3 above,

their average level of participation in human rights organizations during the

1981–1984 period is positively associated with their human rights record (as

measured by their Amnesty International rating) during the 1980–1985 period

(r = .520). Western countries that featured a perfect Amnesty rating of 5 dur-

ing this period had an average participation rate in human rights organizations

of 2.8 percent, while those with an Amnesty rating of <5 only had an average

participation rate of 1.2 percent. In sum, this suggests that the impact of media

exposure on human rights activities at the micro-level may translate into the

news media’s broader impact on a country’s human rights record.

Macro-Level Evidence1

Dependent Variable

Amnesty Rating. Data on human rights practices come from Hafner-

Burton and Tsutsui (2005), who update existing data based on their content

analysis of human rights reports, using an inverted standards-based ordinal

Political Terror Scale. As they note, scores range from 1 to 5, with a score of

1 indicating ‘‘systematic repression,’’ 2 indicating ‘‘extensive repression,’’ 3

indicating ‘‘widespread repression,’’ 4 indicating ‘‘limited repression,’’ and 5

indicating ‘‘rare repression’’ (for a more detailed description, see Hafner-Burton

and Tsutsui 2005:1392). Separate scores, based on annual reports from Amnesty

International (AI) and the U.S. State Department (USSD) are available. The

USSD tends to rate its allies, trading partners, and foreign aid recipients more

10 ROB CLARK

favorably than AI, and rates leftist regimes relatively more harshly (Poe, Carey,

and Vazquez 2001). Therefore, I rely primarily on the AI version. However, I

follow previous studies that ‘‘blend’’ the two (e.g., Poe, Tate, and Keith 1999),

primarily using AI, and relying on the USSD when AI ratings are not available.

Despite potential bias in the U.S. State Department ratings, the AI and USSD

scores that come from Hafner-Burton and Tsutsui (2005) are highly correlated

(r = .840) and have become increasingly similar over time (Poe, Carey, and

Vazquez 2001). In most cases, there is no difference between the two scores,

and in only 4 percent of cases do the scores differ by more than one point (Poe,

Carey, and Vazquez 2001:659). Of the 459 cases in my base sample, I rely on

USSD ratings for only two observations: Gabon (wave 4) and Norway (wave 4).

Moreover, when I exclude these two cases from the analyses, the results are

quite similar (see Table 4, model 3).

Independent Variable

Newspaper Readership. I measure each country’s daily newspaper

readership using circulation rates for newspapers published at least four times

a week, calculated as average circulation (or copies printed) per 1,000 people.

Data come from the United Nations Educational, Scientific, and Cultural Orga-

nization (UNESCO) Institute for Statistics, as made available in the WorldDevelopment Indicators (International Bank for Reconstruction and Develop-

ment 2004). Newspaper circulation is generally regarded as a conservative

measure of readership because each copy is read by more than one person on

average. In the United States, readership is on average about 2.3 times larger

than daily circulation (World Association of Newspapers 2007). For example,

the Chicago Tribune, Los Angeles Times, and Washington Post average over

two readers per copy, while USA Today averages over three readers per copy,

and the New York Times averages over four readers per copy (Audit Bureau of

Circulations 2009; World Association of Newspapers 2007). These readership

levels are comparable with those achieved by major newspapers in other

advanced countries, such as Germany and Ireland (World Association of

Newspapers 2007). And when considering less developed countries, circulation

figures tend to be even more conservative estimates of readership, as news-

papers in these countries typically have higher ‘‘pass-on’’ rates (World

Association of Newspapers 2007).

Technical Controls

I estimate the effect of newspaper readership net of the following techni-

cal controls.

NEWSPAPER READERSHIP AND HUMAN RIGHTS 11

Amnesty Rating (lagged value). I include the lagged value of the

dependent variable so that I can more appropriately measure the impact

of readership on change in a state’s human rights performance.

Time Period. I also control for longitudinal change in Amnesty scores

across the sample period by including time period as a predictor in the analy-

ses (1980–1985 = 1; 1985–1990 = 2; 1990–1995 = 3; 1995–2000 = 4).

World Region. In addition, I control for cross-sectional variation in

Amnesty ratings by including world region as a predictor in the analyses. I

classify states as belonging to one of the six world regions shown in Table 3,

with Europe and the West serving as the excluded reference category.

Substantive Controls

I also include a broad set of substantive predictors, many of which have

been used in prior human rights research. First, I control for institutional and

societal features that may shape the role and impact of the news media. For

example, it is possible that the news industry may only serve as a critical and

independent monitor of societal operations when media outlets enjoy substan-

tial autonomy and ⁄ or when they are situated within democratized nations. In

authoritarian states, the government may own or control the media and attempt

to influence public opinion by deliberately distorting or excluding information

that is disseminated to the public, and ⁄ or remaining silent on (or diverting

attention away from) particular issues (Ojo 2003). Moreover, previous work

suggests that the ability for a free press to positively impact a country’s

human rights performance is restricted to those states that are fully democratic

(Whitten-Woodring 2009). In addition, the capacity for the news media to

effectively inform and mobilize the citizenry may depend on how educated

these populations are. For example, previous studies have shown that the posi-

tive effects of media consumption and newspaper reading on political knowl-

edge are greater among highly educated individuals (Kleinnijenhuis 1991;

Nisbet 2008). Consequently, it is important to control for a country’s level of

press freedom, democracy, and formal education when estimating the impact

of the newspaper industry on its human rights practices.

Press Freedom. Data on press freedom come from Whitten-Woodring

(2009) who transforms Van Belle’s (1997) ordinal measure on media freedom

by dichotomizing states into two groups: (1) those states with a ‘‘free media,’’

and (2) those states with a ‘‘government-controlled media’’ (Whitten-Woodring

2009:607). Van Belle’s measure covers the 1948–1995 period. All states des-

ignated as having ‘‘free media’’ throughout the 1980–1995 period (for which

Ta

ble

3R

egio

nal

Pat

tern

san

dL

on

git

ud

inal

Tre

nd

sin

New

spap

erR

ead

ersh

ip

19

80

–1

98

51

98

5–

19

90

19

90

–1

99

51

99

5–

20

00

Ch

ang

e(%

)

All

stat

es(N

=1

26

)9

9.8

29

9.9

49

9.6

09

7.3

8)

2.4

4

Eu

rop

e&

the

Wes

t(N

=2

3)

29

1.1

52

90

.25

27

8.0

22

54

.80

)1

2.4

8

Lat

inA

mer

ica

&th

eC

arib

bea

n(N

=2

6)

86

.40

83

.76

82

.10

81

.48

)5

.69

Cen

tral

&su

b-S

ahar

anA

fric

a(N

=3

8)

12

.22

11

.13

10

.40

11

.37

)6

.96

No

rth

Afr

ica

&th

eM

idd

leE

ast

(N=

20

)6

7.7

57

2.0

58

2.9

69

0.8

33

4.0

7

Eas

tA

sia

&th

eP

acifi

c(N

=1

9)

95

.55

98

.70

10

3.4

71

07

.47

12

.48

Eas

tern

Eu

rop

e&

Cen

tral

Asi

a(N

=2

3)

––

12

4.3

01

13

.23

)8

.91

No

tes:

Eu

rop

e&

the

Wes

t=

Au

stra

lia,

Au

stri

a,B

elg

ium

,C

anad

a,C

yp

rus,

Den

mar

k,

Fin

lan

d,

Fra

nce

,G

reec

e,Ic

elan

d,

Irel

and

,It

aly

,L

ux

emb

ou

rg,

Mal

ta,

Net

her

lan

ds,

New

Zea

lan

d,

No

rway

,P

ort

ug

al,

Sp

ain

,S

wed

en,

Sw

itze

rlan

d,

Un

ited

Kin

gd

om

,U

nit

edS

tate

s;L

atin

Am

eric

a&

the

Car

ibb

ean

=A

rgen

tin

a,B

aham

as,

Bar

bad

os,

Bo

liv

ia,

Bra

zil,

Ch

ile,

Co

lom

bia

,C

ost

aR

ica,

Cu

ba,

Do

min

ican

Rep

ub

lic,

Ecu

ado

r,E

lS

alv

ado

r,G

uat

emal

a,G

uy

ana,

Hai

ti,

Ho

nd

ura

s,Ja

mai

ca,

Mex

ico

,N

icar

agu

a,P

anam

a,P

arag

uay

,P

eru

,S

uri

nam

e,T

rin

idad

-To

bag

o,

Uru

gu

ay,

Ven

ezu

ela;

Cen

tral

&su

b-S

ahar

an

Afr

ica

=A

ng

ola

,B

enin

,B

ots

wan

a,B

urk

ina

Fas

o,

Bu

run

di,

Cam

ero

on

,C

had

,C

on

go

(DR

),C

on

go

(R),

Eq

uat

ori

alG

uin

ea,

Eth

iop

ia,

Gab

on

,G

amb

ia,

Gh

ana,

Gu

inea

-Bis

sau

,Iv

ory

Co

ast,

Ken

ya,

Les

oth

o,

Lib

eria

,M

adag

asca

r,M

alaw

i,M

ali,

Mau

riti

us,

Mo

zam

biq

ue,

Nam

ibia

,N

iger

,N

iger

ia,

Sen

egal

,S

ierr

aL

eon

e,S

om

alia

,S

ou

thA

fric

a,S

ud

an,

Sw

azil

and

,

Tan

zan

ia,

To

go

,U

gan

da,

Zam

bia

,Z

imb

abw

e;N

ort

hA

fric

a&

the

Mid

dle

Eas

t=

Afg

han

ista

n,

Alg

eria

,B

ahra

in,

Eg

yp

t,

Iran

,Ir

aq,

Isra

el,

Jord

an,

Ku

wai

t,L

eban

on

,L

iby

a,M

oro

cco

,O

man

,P

akis

tan

,Q

atar

,S

aud

iA

rab

ia,

Sy

ria,

Tu

nis

ia,

Tu

rkey

,

Un

ited

Ara

bE

mir

ates

;E

ast

Asi

a&

the

Pac

ific

=B

ang

lad

esh

,B

run

ei,

Fij

i,In

do

nes

ia,

Jap

an,

Lao

s,M

alay

sia,

Mal

div

es,

Mo

ng

oli

a,M

yan

mar

,N

epal

,N

ort

hK

ore

a,P

apu

aN

ewG

uin

ea,

Ph

ilip

pin

es,

Sin

gap

ore

,S

ou

thK

ore

a,S

riL

ank

a,T

hai

lan

d,

Vie

tnam

;E

aste

rnE

uro

pe

&C

entr

alA

sia

=A

lban

ia,

Arm

enia

,A

zerb

aija

n,

Bel

aru

s,B

osn

ia-H

erze

go

vin

a,B

ulg

aria

,C

roat

ia,

Cze

chR

epu

bli

c,E

sto

nia

,H

un

gar

y,

Ky

rgy

zsta

n,

Lat

via

,L

ith

uan

ia,

Mac

edo

nia

,M

old

ov

a,P

ola

nd

,R

uss

ia,

Ser

bia

-

Mo

nte

neg

ro,

Slo

vak

ia,

Slo

ven

ia,

Taj

ikis

tan

,U

kra

ine,

and

Uzb

ekis

tan

.

12 ROB CLARK

NEWSPAPER READERSHIP AND HUMAN RIGHTS 13

data are available) are coded as one (i.e., those states with press freedom)

throughout the sample period, while those states designated as having ‘‘gov-

ernment-controlled media’’ during any part of the 1980–1995 period are coded

as zero (i.e., those states without press freedom).2

Democratization. Democracy scores come from Marshall and Jaggers

(2005) Polity IV project that provides longitudinal political regime characteris-

tics for countries. I use their combined polity score, which is computed by

subtracting a state’s autocracy score from its democracy score. Scores range

from )10 to 10.

Tertiary Education. I measure each state’s level of formal education by

their tertiary school enrollment, which refers to the proportion of people in the

age group that officially corresponds to the tertiary level who are currently

enrolled.

In addition to the above measures, previous studies reveal a number of

other domestic forces that impact a state’s human rights record. For example,

state repression is more likely to occur within nations that suffer from under-

development, civil war, and ⁄ or population pressures (Hafner-Burton and Tsutsui

2005; Poe, Tate, and Keith 1999). Previous studies have also considered the

importance of international linkages, focusing on international participation in

treaties, organizations, and trade (Hafner-Burton and Tsutsui 2005). Thus, I

also control for a broad set of domestic and international factors that may

influence human rights practices.

GDP PC (PPP) (log). Gross domestic product per capita is based on

purchasing power parity. Data are in 1995 international dollars. An interna-

tional dollar has the same purchasing power over GDP as the U.S. dollar

has in the United States.

Civil War. Civil war is a dummy variable, referring to the presence or

absence of ‘‘intra-state war’’ in each state during each wave of the sample

period. Data come from Sarkees (2000) Correlates of War project (version

3.0).

Population Density (log). Population density refers to a country’s popu-

lation divided by its land area in square kilometers. Population estimates are

based on all residents regardless of legal status or citizenship (except for refu-

gees not permanently settled in the country of asylum) who are generally con-

sidered part of the population of their country of origin.

14 ROB CLARK

Treaty Ratifications. Following Hafner-Burton and Tsutsui (2005:

1394), I consider a state’s ratification of six human rights treaties, along with

two optional protocols that are designed to (1) enhance monitoring and

enforcement, and (2) eliminate the death penalty. In total, this measure refers

to the number of human rights instruments a state has ratified by the end of

each wave, with scores ranging from 0 to 8 across the sample period. Data

come from the Office of the United Nations High Commissioner for Human

Rights (http://www.ohchr.org).

International Organizations (log). International organizations refer to

the sum of conventional international governmental organizations and interna-

tional non-governmental organizations to which each country belongs, listed in

sections A–D of the Yearbook of International Organizations (Union of Inter-

national Associations). Data for each wave come from single years: 1980,

1985, 1990, and 1995.

Trade Openness (log). Trade openness refers to the sum of exports and

imports of goods and services, measured as a share of gross domestic product.

Analysis

Following Hafner-Burton and Tsutsui (2005), I estimate ordered probit

models of the ordinal dependent variable (Amnesty International rating) on the

independent variables. Also, adhering to Hafner-Burton and Tsutsui’s (2005)

methodology, I address potential heteroskedasticity by clustering on states.

Thus, observations are assumed to be independent across states, but not neces-

sarily within states over time. I establish time order by lagging the indepen-

dent variables one wave behind the dependent variable. I also include a

lagged-dependent variable as a predictor so that I am assessing the impact of

the independent variables on change in the Amnesty rating. I am sensitive to

potential collinearity in my models. Therefore, I report the OLS-based maxi-

mum variance inflation factor (VIF) score for each model. The maximum VIF

score across all models in Table 4 is always below 10, suggesting that collin-

earity is not a problem in these analyses.3

Sample

My base sample consists of 459 observations across 138 countries over

four waves during the 1980–2000 period. The pooled data are unbalanced

with some states contributing more observations than others. However, 114

states (83% of the total) are present across two or more waves, and 98 states

(71% of the total) are present in all four waves. Each wave represents a 6-

year period (1980–1985, 1985–1990, 1990–1995, and 1995–2000), where each

Ta

ble

4O

rder

edP

rob

itM

od

els

of

the

Am

nes

tyR

atin

g

(1)

(2)a

(3)b

(4)c

(5)d

(6)e

(7)f

(8)g

(9)h

New

spap

er

read

ersh

ip

.366**

(.105)

.375***

(.105)

.363**

(.106)

.374***

(.106)

.333**

(.106)

.354**

(.112)

.374**

(.112)

.444**

(.134)

.438**

(.147)

Am

nes

ty

rati

ng

(Lag

)

.555***

(.110)

.554***

(.110)

.553***

(.110)

.548***

(.110)

.600***

(.117)

.559***

(.131)

.412***

(.114)

.601***

(.146)

.186***

(.048)

Tim

eper

iod

.002

(.078)

.012

(.078)

.000

(.078)

).0

05

(.079)

.011

(.080)

.061

(.086)

.281**

(.093)

).0

86

(.158)

.056

(.046)

Lat

inA

mer

ica

).5

70***

(.132)

).5

27***

(.133)

).5

69***

(.132)

).5

73***

(.133)

).5

39***

(.136)

).5

62***

(.148)

).5

80***

(.128)

).6

20***

(.177)

).2

78***

(.073)

Afr

ica

).6

43**

(.193)

).6

01**

(.196)

).6

45**

(.194)

).6

48**

(.194)

).6

14**

(.207)

).5

25*

(.227)

).6

71***

(.188)

).8

41**

(.250)

).3

32**

(.097)

Mid

dle

Eas

t)

.643***

(.137)

).6

17***

(.136)

).6

42***

(.136)

).6

44***

(.136)

).6

26***

(.141)

).6

55***

(.153)

).6

16***

(.129)

).7

49**

(.224)

).3

59***

(.068)

Eas

tA

sia

).5

38***

(.124)

).5

15***

(.125)

).5

38***

(.125)

).5

39***

(.125)

).5

34***

(.129)

).5

83***

(.139)

).5

30***

(.121)

).6

72***

(.162)

).3

08***

(.065)

Eas

tern

Euro

pe

).3

75**

(.118)

).3

41**

(.117)

).3

74**

(.118)

).3

73**

(.118)

).3

07*

(.148)

).4

28*

(.171)

).4

00***

(.105)

).5

00**

(.144)

).1

92***

(.050)

Pre

ssfr

eedom

.041

(.119)

.022

(.121)

.041

(.119)

.040

(.119)

.019

(.124)

).0

45

(.130)

.063

(.117)

).1

10

(.137)

Dem

ocr

atiz

atio

n.1

21

(.104)

.133

(.104)

.123

(.104)

.124

(.104)

.108

(.107)

.188

(.122)

.107

(.101)

.275†

(.157)

.084†

(.049)

NEWSPAPER READERSHIP AND HUMAN RIGHTS 15

Ta

ble

4(C

on

tin

ued

)

(1)

(2)a

(3)b

(4)c

(5)d

(6)e

(7)f

(8)g

(9)h

rtia

ry

duca

tion

.121

(.121)

.045

(.125)

.123

(.121)

.090

(.127)

.176

(.127)

.175

(.138)

.201†

(.116)

.107

(.177)

.012

(.066)

DP

PC

PP

P)

.086

(.144)

.113

(.145)

.080

(.147)

.087

(.145)

.088

(.158)

.244

(.175)

.083

(.144)

).0

25

(.209)

vil

war

).3

43***

(.077)

).3

38***

(.077)

).3

43***

(.077)

).3

55***

(.078)

).3

43***

(.080)

).3

66***

(.089)

).3

27***

(.074)

).2

55**

(.094)

).1

96***

(.033)

pula

tion

ensi

ty

).0

96

(.084)

).0

63

(.088)

).0

94

(.085)

).0

96

(.085)

).0

81

(.088)

).0

96

(.093)

).0

39

(.084)

).1

25

(.112)

).0

34

(.042)

eaty

atifi

cati

ons

).0

27

(.086)

).0

36

(.087)

).0

29

(.086)

).0

14

(.089)

).0

33

(.087)

).0

76

(.093)

).1

01

(.092)

).1

71

(.119)

).0

49

(.047)

tern

atio

nal

rgan

izat

ions

).3

24**

(.119)

).3

13**

(.119)

).3

22**

(.119)

).3

19**

(.119)

).3

28*

(.131)

).4

21**

(.149)

).2

81*

(.110)

).3

79*

(.163)

).2

80***

(.080)

ade

pen

nes

s

.176*

(.071)

.172*

(.072)

.176*

(.071)

.184*

(.072)

.156*

(.074)

.116

(.080)

.129†

(.067)

.167†

(.100)

.127**

(.041)

Inew

s

elea

ses

).4

81***

(.078)

tern

etusa

ge

).0

83

(.106)

levis

ion

ouse

hold

s

.049

(.206)

ates

138

134

137

138

114

98

138

129

138

16 ROB CLARK

Te e

G

(

Ci

Po d

Tr r

In

o

Tr o

A

r

In Te h

St

Ta

ble

4(C

on

tin

ued

)

(1)

(2)a

(3)b

(4)c

(5)d

(6)e

(7)f

(8)g

(9)h

Obse

rvat

ions

459

443

457

458

435

392

459

224

459

Max

imum

var

iance

infl

atio

nfa

ctor

7.6

77.7

57.6

67.6

68.1

98.4

67.6

78.9

7R

2w

.048

i

Pse

udo

R-s

quar

e.1

87

.170

.186

.187

.190

.198

.203

.199

R2

b.7

17

j

BIC

2,5

99

2,5

80

2,5

95

2,5

95

2,4

87

2,2

16

2,5

62

1,2

58

R2

o.6

30

k

Note

s:E

ach

cell

report

sth

est

andar

diz

edco

effi

cien

tw

ith

the

robust

stan

dar

der

ror

inpar

enth

eses

.A

llm

odel

sin

clude

a

clust

ered

sandw

ich

esti

mat

or

toal

low

for

intr

a-st

ate

corr

elat

ion;

aE

xcl

udes

stat

esw

ith

per

fect

Am

nes

tyra

tings;

bE

xcl

udes

Am

nes

tyra

tings

bas

edon

the

US

SD

;cE

xcl

udes

outl

iers

;dE

xcl

udes

stat

esco

ntr

ibuti

ng

one

obse

rvat

ion;

eF

eatu

res

abal

ance

d

sam

ple

;f In

cludes

contr

ol

for

Am

nes

tynew

sre

leas

es;

gIn

cludes

contr

ols

for

Inte

rnet

usa

ge

and

tele

vis

ion

house

hold

s;hT

wo-

stag

ele

ast

squar

esre

gre

ssio

n,

wit

hpre

ssfr

eedom

and

GD

PP

Cse

rvin

gas

inst

rum

ents

for

new

spap

erre

ader

ship

;i R

2w

refe

rsto

R2

Wit

hin

;j R

2b

refe

rsto

R2

Bet

wee

n;

kR

2o

refe

rsto

R2

Over

all.

†p

<.1

0;

*p

<.0

5;

**p

<.0

1;

***

p<

.001

(tw

o-t

aile

dte

sts)

.

NEWSPAPER READERSHIP AND HUMAN RIGHTS 17

18 ROB CLARK

data point for each measure represents a state’s average value across the

entire wave (except where otherwise noted above). The four waves of the

dependent variable cover the 1985–2002 period (1985–1990, 1990–1995,

1995–2000, and 2000–2002), while the four waves of the lagged-dependent

variable cover the 1975–1995 period (1975–1980, 1980–1985, 1985–1990,

and 1990–1995).

Results

Regional Patterns and Longitudinal Trends

Table 3 reports newspaper readership trends and regional patterns for the

126 countries in my dataset that have circulation scores across all four waves.

In addition, I include trends for 23 Eastern European countries that report cir-

culation scores during the final two waves (see bottom row). The data reveal

considerable regional disparities that correspond broadly with socioeconomic

levels of development. In particular, Western nations feature readership rates

that are more than twice as large as any other region (254.80 during the

1995–2000 period). Latin America (81.48), the Middle East (90.83), East Asia

(107.47), and Eastern Europe (113.23) all have rates that are close to 100 by the

end of the sample period. By contrast, countries in Central and sub-Saharan

Africa (11.37) have circulation figures that are far lower than anywhere else in

the world.

In addition, Table 3 also reports longitudinal trends within each region

(see final column for overall rate of change). The first row reports the overall

trend. Readership increased slightly during the 1980s, before falling over the

last two time periods during the 1990s, contributing to an overall decline of

2.44 percent during the sample period. The 1990s decline roughly coincides

with the emergence of the Internet, particularly among advanced economies.

Accordingly, Western nations have driven the decline in newspaper circula-

tion. In this region, readership fell slightly during the 1980s, before falling

more rapidly during the 1990s, contributing to a regional decline in newspaper

circulation of 12.48 percent. By contrast, readership in Latin America

(5.69%), Africa (6.96%), and Eastern Europe (8.91%) show smaller declines.

Moreover, the Middle East (34.07%) and East Asia (12.48%) featured substan-

tial increases in newspaper circulation. Thus, while the decline of the news-

paper industry has been a salient theme in the West, this decline has not been

universal, at least during the sample period. Nevertheless, as a robustness

check, I test whether the effect of newspaper readership weakens or disappears

when controlling for Internet usage during the latter half of the sample period

(see Table 4, model 8).

NEWSPAPER READERSHIP AND HUMAN RIGHTS 19

Analyses

Table 4 presents results from the main analyses. In these models, I report

results from the ordered probit regression of the Amnesty International rating

on newspaper readership. Prior to analyses, I standardized all predictors

(mean = 0; standard deviation = 1) so that the reported coefficients allow

interested readers to compare the magnitude of each variable’s effect. Thus,

each cell reports the standardized coefficient with the robust standard error in

parentheses. Model 1 represents the base model, estimating the effect of news-

paper readership on a country’s Amnesty rating when controlling for the tech-

nical and substantive controls. The effect of newspaper readership is positive

and reaches statistical significance (p < .01). Thus, higher levels of newspaper

readership significantly improve a country’s human rights record when the

other variables are held constant. In fact, readership exerts the strongest effect

among all substantive predictors, displaying positive effects that are larger

(B = .366) than the negative effects of being in a civil war (B = ).343). The

model also reveals considerable regional disparities in human rights records,

as Western nations feature significantly higher human rights ratings than all

five non-Western regions.

In the remaining models featured in Table 4, I examine the effect of

newspaper readership under alternative specifications. In model 2, the sample

is restricted by excluding four states (16 observations) with perfect Amnesty

scores across all four waves of the sample period (Australia, Canada, the

Netherlands, and New Zealand). In model 3, the sample is restricted by drop-

ping two observations, Gabon (wave 4) and Norway (wave 4), based on the

U.S. State Department’s ratings of human rights. In model 4, the sample is

restricted by excluding one outlier (United States, wave 3) identified by the

Hadi procedure available in Stata 11.1. The procedure identifies multiple out-

liers in multivariate data using OLS (I use the p < .05 significance level for

my outlier cutoff). In model 5, I restrict the sample to those 114 states that

contribute two or more observations (N = 435). In model 6, the sample is bal-

anced, restricted to those 98 states present in all four waves of the sample per-

iod (N = 392). In each of these alternative specifications, newspaper

readership continues to positively affect a state’s Amnesty rating at a high

level of significance (p < .01 or greater).

In models 7–8, I estimate the impact of newspaper readership net of

additional controls. In model 7, I include a control that gauges the dispro-

portionate amount of attention that Amnesty International devotes to differ-

ent countries. One reason to include this control is that Amnesty’s human

rights ratings may be influenced by the amount of attention that it devotes

to each country. Thus, it may be interesting to examine the effect of

20 ROB CLARK

newspaper readership net of this possibility. Second, the control could also

be mediating the effect of newspaper readership. Previous studies have

examined factors that increase the support of human rights issues (Carpenter

2007) and domestic social movements (Bob 2001) by transnational advocacy

networks, including groups such as Amnesty International and other non-

governmental organizations. In particular, Keck and Sikkink (1998) describe

a ‘‘boomerang pattern,’’ in which local actors bypass the state and seek

allies in global civil society who will, in turn, pressure states from the out-

side. Accordingly, one possible explanation for the positive effect of news-

paper readership is that the print media may influence external actors by

providing them with information regarding domestic human rights practices.

This, in turn, may lead international advocacy groups to apply greater pres-

sure on the state. Thus, in model 7, I consider Amnesty International’s

‘‘advocacy effort’’ by controlling for the number of press releases it issues

per country during each wave of the sample period. Data come from

Amnesty International’s Country Dossiers and Publications, 1962–2009(Amnesty International 2011). The results show that controlling for

Amnesty’s news release activity does not substantially influence the positive

effect of newspapers (p < .01).

In model 8, I consider alternative forms of media consumption during the

1990s by controlling for each country’s Internet usage and television market.

Internet usage is defined as the number of people (per 100 in the population)

with access to the worldwide network. I also include a control for television

saturation, measured as the percentage of households with television sets. Data

for these measures come from the World Development Indicators (Interna-

tional Bank for Reconstruction and Development 2007, 2010) and are avail-

able during the latter half of the sample period, 1990–2000, which coincides

with the emergence of the electronic news media and the decline of the news-

paper industry among developed nations. Including these controls reduces my

sample to 224 observations across 129 states during the final two waves of the

sample period. Newspaper readership is correlated with both Internet usage

(r = .571) and television households (r = .598). However, as the maximum

VIF score indicates (8.97), collinearity is not problematic. More importantly,

as the results from model 8 show, newspaper readership remains highly signif-

icant as a positive predictor (p < .01).

Finally, it is possible that the relationship between newspaper readership

and human rights suffers from endogeneity (i.e., reverse causality). For exam-

ple, repressive human rights practices may compel governments to shut down

or hinder press activity, while reformed governments may encourage greater

press activity. Thus, in model 9, I corrected for potential endogeneity via

instrumental variables regression using two-stage least squares (2SLS)

NEWSPAPER READERSHIP AND HUMAN RIGHTS 21

estimation for panel-data models. Diagnostics reveal that the instruments

(press freedom and GDP PC) are both strong (i.e., correlated with newspaper

readership) and valid (i.e., uncorrelated with the error term). Moreover, the

results shown in model 9 reveal that newspaper readership remains a signifi-

cant, positive predictor of human rights (p < .01).

Discussion

Previous research has found that the print media can effectively educate

citizens, shape their attitudes, and mobilize them to become politically active.

In this study, I examine the print media’s impact on human rights in a cross-

national context. The findings from this study reveal that a nation’s level of

newspaper readership positively affects change in its Amnesty International

rating, net of a broad set of controls. This relationship is robust to a number

of alternative specifications, including models that consider the impact of

ceiling effects, measurement bias, influential observations, sample composi-

tion, mediation, and endogeneity. Moreover, the magnitude of the print

media’s impact is substantial, exerting large effects relative to other substan-

tive predictors and rivaling the negative impact of being in a civil war. In

addition, I do not find that the effect of newspaper readership on human rights

significantly declines in the latter half of the sample period or that the impact

of the newspaper industry was weakened by the emergence of alternative

media, such as the Internet or television.

The positive effect of newspaper readership on human rights during the

final years of the twentieth century coincides with an era that features substan-

tial critique and scrutiny directed toward the news media. The media have

been accused of bending to crass commercialism, political bias, and govern-

ment influence across a number of countries. Within this context, it is unclear

what impact (if any) the recent trend of liberalization has had on media sys-

tems. In most parts of the world, the news media have become more market-

oriented since the 1980s because of privatization and deregulation (Curran

et al. 2009). However, this has not always resulted in greater press freedom, a

more educated citizenry, or more transparent governments. Rather, the privati-

zation of media has led scholars to draw a variety of conclusions across differ-

ent countries around the world. By one account, the commercialization of

India’s television industry has provided viewers with independent media alter-

natives to state-monitored media outlets and has provided its TV journalists

with new opportunities for voicing the concerns of marginalized populations

and demanding government accountability (Rao 2008). Similarly, as a result

of Mali’s democratic transition and media liberalization during the 1990s, it

now has the greatest amount of press freedom in West Africa (Nisbet 2008).

Conversely, other research points to shortcomings in privatized media systems.

22 ROB CLARK

For example, one study comparing several Western countries finds that media

systems operating under a public service model (i.e., Finland and Denmark)

produce more knowledgeable citizens than media systems representing a pure

market model (i.e., the United States) (Curran et al. 2009). In sum, the privati-

zation of media outlets has led scholars to draw different conclusions across

different national contexts.

In addition to liberalization, the newspaper industry has had to face its

own mortality, weathering significant declines in circulation in some parts of

the world beginning in the early 1990s, which has also caused many to ques-

tion its relevance. Against this backdrop, the present study addresses an impor-

tant set of issues in media research. Can the media be rightfully construed as

public service representatives, capable of monitoring the state, while educating

and mobilizing citizens? The findings from this study answer in the affirmative

and suggest that newspapers continued to serve an important public function

during the sample period. Rather than viewing the newspaper industry as a

declining worldwide institution, it is more accurate to suggest that declines in

readership are region-specific, as many nations in the developing world (e.g.,

those in the Middle East and East Asia) experienced significant circulation

increases during the sample period. Moreover, there is evidence that newspa-

per circulation has continued to rise (or has remained stable) in most parts of

the world during the post-2000 period. Over the past 5 years, newspaper sales

worldwide increased by 9.39 percent, with circulation rates rising or remaining

stable in 75 percent of the world during this time (World Association of

Newspapers 2008). Moreover, despite the continued circulation declines in

Western countries, many of these countries have begun to distribute ‘‘free dai-

lies,’’ which now account for 23 percent of all newspaper circulation in

Europe. In fact, 70 percent of the world’s free dailies are distributed in Europe

(Bakker 2007). Between 1995 and 2007, the worldwide circulation of free dai-

lies has grown from 200,000 to over 40 million (Bakker 2007). And, over the

past 5 years, the global circulation of free dailies has increased by 173.2 per-

cent, such that they now account for 7 percent of circulation around the world

(World Association of Newspapers 2008).

Nevertheless, even with the injection of free dailies, persistent declines in

circulation in some countries during the post-2000 era suggest that the indus-

try may have to fulfill its public service role in an electronic format. To that

end, online newspaper sites around the world have grown by over 50 percent

during the past 5 years (World Association of Newspapers 2008). Future

research should investigate more closely whether newspapers will be able to

make the Internet transition and continue to monitor state activity in the same

capacity. In this study, I report the effects of newspaper readership during the

post-1990 period when controlling for a country’s Internet usage (see Table 4,

NEWSPAPER READERSHIP AND HUMAN RIGHTS 23

model 8). While this measure has limitations, Internet usage does provide us

with a preliminary glimpse of the impact that this form of media may have on

human rights practices. As the above model indicates, the effect of Internet

usage is non-significant. To the extent that these results reflect true differences

between newspapers and the Internet, one can speculate that the experience of

online news consumption may mimic that of television viewing more so

than that of newspaper reading. However, future work that examines more pre-

cise measures of online news consumption are required before definitive

conclusions can be drawn.

ENDNOTES

*I thank Amy Kroska for her generous assistance. I am also grateful to Emilie Hafner-Bur-

ton and Jenifer Whitten-Woodring for kindly sharing their data. Please direct correspondence to:

Rob Clark, Department of Sociology, University of Oklahoma, Kaufman Hall 331, 780 Van Vleet

Oval, Norman, OK 73019, USA; e-mail: [email protected] come from the World Development Indicators (International Bank for Reconstruction

and Development 2004), unless otherwise noted.2In my base sample of 138 states, 37 are coded as having press freedom throughout the sam-

ple period: Australia, Austria, Belgium, Botswana, Canada, Colombia, Costa Rica, Croatia,

Cyprus, Czech Republic, Denmark, Dominican Republic, Ecuador, Finland, France, Gambia,

Germany, India, Ireland, Israel, Italy, Jamaica, Japan, Latvia, Mauritius, Mexico, Nepal, Nether-

lands, New Zealand, Norway, Spain, Sweden, Switzerland, Trinidad-Tobago, Uganda, United

Kingdom, and United States.3In separate analyses, I also estimated a fixed-effects model that restricts attention to longitu-

dinal variation (thereby ignoring all cross-sectional differences). The results are similar to those

presented below, with newspaper readership (b = .363; p < .05) exerting a positive, significant

effect on the Amnesty rating.

REFERENCES

Amnesty International. 2011. Amnesty International’s Country Dossiers and Publications, 1962–

2009. London: IDC Publishers.

Audit Bureau of Circulations. 2009. Reader Profiles. Retrieved September 2009 (http://

www.accessabc.com).

Bakker, Piet. 2007. ‘‘Free Newspaper Readership.’’ Worldwide Readership Research Symposium.

Retrieved September 2009 (http://www.readershipsymposium.org/papers/791.pdf).

Becker, Lee and Sharon Dunwoody. 1982. ‘‘Media Use, Public Affairs Knowledge, and Voting in

a Local Election.’’ Journalism Quarterly 59:212–18, 255.

Berkowitz, Dan and David Pritchard. 1989. ‘‘Political Knowledge and Communication

Resources.’’ Journalism Quarterly 66:697–701.

Bob, Clifford. 2001. ‘‘Marketing Rebellion: Insurgent Groups, International Media, and NGO

Support.’’ International Politics 38:311–34.

24 ROB CLARK

Boyle, Elizabeth, Barbara McMorris, and Mayra Gomez. 2002. ‘‘Local Conformity to International

Norms.’’ International Sociology 17:5–33.

Bybee, Carl, Jack McLeod, William Luetscher, and Gina Garramone. 1981. ‘‘Mass

Communication and Voter Volatility.’’ Public Opinion Quarterly 45:69–90.

Carpenter, R. Charli. 2007. ‘‘Studying Issue (Non)-Adoption in Transnational Advocacy

Networks.’’ International Organization 61:643–67.

Chaffee, Steven and Donna Wilson. 1977. ‘‘Media Rich, Media Poor: Two Studies of Diversity in

Agenda-Holding.’’ Journalism Quarterly 54:466–76.

Clarke, Peter and Eric Fredin. 1978. ‘‘Newspapers, Television, and Political Reasoning.’’ Public

Opinion Quarterly 42:143–60.

Curran, James, Shanto Iyengar, Anker Lund, and Inka Salovaara-Moring. 2009. ‘‘Media System,

Public Knowledge, and Democracy: A Comparative Study.’’ European Journal of

Communication 24:5–26.

Dalton, Russell, Paul Beck, and Robert Huckfeldt. 1998. ‘‘Partisan Cues and the Media:

Information Flows in the 1992 Presidential Election.’’ American Political Science Review

92:111–26.

DeFleur, Melvin, Lucinda Davenport, Mary Cronin, and Margaret DeFleur. 1992. ‘‘Audience

Recall of News Stories Presented by Newspaper, Computer, Television, and Radio.’’

Journalism Quarterly 69:1010–22.

Eveland, William and Dietram Scheufele. 2000. ‘‘Connecting News Media Use with Gaps in

Knowledge and Participation.’’ Political Communication 17:215–37.

Feldman, Ofer and Kazuhisa Kawakami. 1991. ‘‘Media Use as Predictors of Political Behavior:

The Case of Japan.’’ Political Psychology 12:65–80.

Hafner-Burton, Emilie and Kiyoteru Tsutsui. 2005. ‘‘Human Rights in a Globalizing World: The

Paradox of Empty Promises.’’ American Journal of Sociology 110:1373–1411.

International Bank for Reconstruction and Development. 2004. World Development Indicators.

Washington, DC: International Bank for Reconstruction and Development.

——. 2007. World Development Indicators. Washington, DC: International Bank for

Reconstruction and Development.

——. 2010. World Development Indicators. Washington, DC: International Bank for

Reconstruction and Development.

Keck, Margaret and Kathryn Sikkink. 1998. Activists Beyond Borders: Advocacy Networks in

International Politics. Ithaca, NY: Cornell University Press.

Kleinnijenhuis, Jan. 1991. ‘‘Newspaper Complexity and the Knowledge Gap.’’ European Journal

of Communication 6:499–522.

Kwak, Nojin. 1999. ‘‘Revisiting the Knowledge Gap Hypothesis: Education, Motivation, and

Media Use.’’ Communication Research 26:385–413.

Marshall, Monty and Keith Jaggers. 2005. Polity IV Project: Political Regime Characteristics and

Transitions, 1800–2003. Retrieved May 2005 (http://www.cidcm.umd.edu/inscr/polity/

index.htm).

McLeod, Jack and Daniel McDonald. 1985. ‘‘Beyond Simple Exposure: Media Orientations and

Their Impact on Political Processes.’’ Communication Research 12:3–33.

McLeod, Jack, Dietram Scheufele, and Patricia Moy. 1999. ‘‘Community, Communication, and

Participation: The Role of Mass Media and Interpersonal Discussion in Local Political

Participation.’’ Political Communication 16:315–36.

Nicholson, Stephen. 2003. ‘‘The Political Environment and Ballot Proposition Awareness.’’

American Journal of Political Science 47:403–10.

Nisbet, Erik. 2008. ‘‘Media Use, Democratic Citizenship, and Communication Gaps in a

Developing Democracy.’’ International Journal of Public Opinion Research 20:454–82.

NEWSPAPER READERSHIP AND HUMAN RIGHTS 25

Ojo, Emmanuel. 2003. ‘‘The Mass Media and the Challenges of Sustainable Democratic Values in

Nigeria: Possibilities and Limitations.’’ Media, Culture, and Society 25:821–40.

Pattie, C. J. and R. J. Johnston. 2003. ‘‘Civic Literacy and Falling Electoral Turnout: The United

Kingdom 1992–1997.’’ Canadian Journal of Political Science 36:579–99.

Poe, Steven, Sabine Carey, and Tanya Vazquez. 2001. ‘‘How Are These Pictures Different?

A Quantitative Comparison of the US State Department and Amnesty International Human

Rights Reports, 1976–1995.’’ Human Rights Quarterly 23:650–77.

Poe, Steven, C. Neal Tate, and Linda Keith. 1999. ‘‘Repression of the Human Right to Personal

Integrity Revisited: A Global Cross-National Study Covering the Years 1976–1993.’’

International Studies Quarterly 43:291–313.

Ramos, Howard, James Ron, and Oskar Thoms. 2007. ‘‘Shaping the Northern Media’s Human

Rights Coverage, 1986–2000.’’ Journal of Peace Research 44:385–406.

Rao, Shakuntala. 2008. ‘‘Accountability, Democracy, and Globalization: A Study of Broadcast

Journalism in India.’’ Asian Journal of Communication 18:193–206.

Robinson, John and Dennis Davis. 1990. ‘‘Television News and the Informed Public: An

Information-Processing Approach.’’ Journal of Communication 40:106–19.

Ron, James. 1997. ‘‘Varying Methods of State Violence.’’ International Organization 51:275–300.

Sarkees, Meredith. 2000. ‘‘The Correlates of War Data on War: An Update to 1997.’’ Conflict

Management and Peace Science 18:123–44.

Scheufele, Dietram, James Shanahan, and Sei-Hill Kim. 2002. ‘‘Who Cares About Local Politics?

Media Influences on Local Political Involvement, Issue Awareness, and Attitude Strength.’’

Journalism and Mass Communication Quarterly 79:427–44.

Union of International Associations. 1980, 1985, 1990, 1995. Yearbook of International

Organizations. Munich: K.G. Saur.

Van Belle, Douglas. 1997. ‘‘Press Freedom and the Democratic Peace.’’ Journal of Peace

Research 34:405–14.

Whitten-Woodring, Jenifer. 2009. ‘‘Watchdog or Lapdog? Media Freedom, Regime Type, and

Government Respect for Human Rights.’’ International Studies Quarterly 53:595–625.

World Association of Newspapers. 2007. World Press Trends. Retrieved September 2009 (http://

www.wan-press.org/worldpresstrends/home.php).

——. 2008. World Press Trends. Retrieved September 2009 (http://www.wan-press.org/

worldpresstrends/home.php).

World Values Survey. 2006. Values Surveys 1981–2004: Integrated Questionnaire, v.20060423.

Retrieved January 2009 (http://www.worldvaluessurvey.org).