A Little Bird Told Me, So I Didn't Believe It: Twitter, …...Twitter, which both media outlets and...

This article was downloaded by: [University Town Library of Shenzhen]On: 11 July 2012, At: 06:02Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Communication QuarterlyPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/rcqu20

A Little Bird Told Me, So I Didn't BelieveIt: Twitter, Credibility, and IssuePerceptionsMike Schmierbach & Anne Oeldorf-Hirsch

Version of record first published: 28 Jun 2012

To cite this article: Mike Schmierbach & Anne Oeldorf-Hirsch (2012): A Little Bird Told Me, So I Didn'tBelieve It: Twitter, Credibility, and Issue Perceptions, Communication Quarterly, 60:3, 317-337

To link to this article: http://dx.doi.org/10.1080/01463373.2012.688723

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions

This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representationthat the contents will be complete or accurate or up to date. The accuracy of anyinstructions, formulae, and drug doses should be independently verified with primarysources. The publisher shall not be liable for any loss, actions, claims, proceedings,demand, or costs or damages whatsoever or howsoever caused arising directly orindirectly in connection with or arising out of the use of this material.

http://www.tandfonline.com/loi/rcqu20

http://dx.doi.org/10.1080/01463373.2012.688723

http://www.tandfonline.com/page/terms-and-conditions

A Little Bird Told Me, So I Didn’tBelieve It: Twitter, Credibility,and Issue PerceptionsMike Schmierbach & Anne Oeldorf-Hirsch

This article investigates how media use of the microblogging tool Twitter affects percep-

tions of the issue covered and the credibility of the information. In contrast to prior stu-

dies showing that ordinary blogs are often judged credible, especially by their users, data

from 2 experiments show that Twitter is considered less credible than various forms

of stories posted on a newspaper Web site, and fails to convey importance as well as a

newspaper or blog.

Keywords: Credibility; Information Seeking; Twitter

In the last decade, individuals have increasingly turned to the Internet for infor-

mation, including the news. From 1994 to 2008 reading the news online (at least 3

days per week) increased from 2% to 37% (Pew Research Center, 2008). In turn,

news organizations have struggled to adjust both to competition from new outlets

and to distribution through new technologies. Individuals get information directly

from organization Web sites, but also indirectly through individual or third-party

blogs that re-post news. Somewhere between these lies the microblogging tool

Twitter, which both media outlets and others can use to provide quick tweets alerting

followers to the day’s headlines and directing them to full stories.

In this context, judgments about credibility are both critical and complex. Faced

with a diversity of news sources and stories, individuals must decide how much they

trust information and whether to act on it. Yet, new technologies can mask or com-

plicate the underlying source of information. Consider, for example, a news story that

Mike Schmierbach (Ph.D., University of Wisconsin, 2004) is an assistant professor in the College of Commu-

nications at Pennsylvania State University. Anne Oeldorf-Hirsch (Ph.D., Pennsylvania State University, 2011) is

a doctoral candidate in the College of Communications at Pennsylvania State University. Correspondence: Mike

Schmierbach, College of Communications, Pennsylvania State University, 217 Carnegie Bldg., University Park,

PA 16802; E-mail: [email protected]

Communication Quarterly

Vol. 60, No. 3, July–August 2012, pp. 317–337

ISSN 0146-3373 print/1746-4102 online # 2012 Eastern Communication Association

DOI: 10.1080/01463373.2012.688723

Dow

nloa

ded

by [

Uni

vers

ity T

own

Lib

rary

of

Shen

zhen

] at

06:

02 1

1 Ju

ly 2

012

initially appears in The New York Times. A person might find the story on the

organization’s Web site, find a link from a prominent blog, be sent an e-mail recom-

mending the story or the blog post, or come upon a tweet referencing the story. In

such circumstances, opinions about the initial news outlet may become intertwined

with opinions about blogs or Twitter, as well as the individuals making the

recommendation.

In this study, we consider how these complications may affect judgments about

the credibility of information posted to Twitter, as well as further interest in the story

topic. By comparing evaluations of Twitter posts to evaluations of the original story

in different formats and locations, we explore how a specific microblogging tech-

nology influences perceptions of a news story.

Blogs, Microblogs, and News

Before considering the specific microblogging tool of Twitter, we first consider the key

antecedent technology. Blogs have been defined as personal Web sites that are regu-

larly updated and contain archived dated entries in reverse chronological order, which

contain primarily text, but may also contain photos or other multimedia, and usually

allow for audience comments (Herring, Scheidt, Bonus, & Wright, 2004; Herring,

Scheidt, Wright, & Bonus, 2005; Nardi, Schiano, & Gumbrecht, 2004). Technorati,

the first blog search engine and leading authority on blogging, has indexed 133 million

blogs in 88 languages in at least 66 countries (see ‘‘State of the Blogosphere 2008’’ at

http://technorati.com/blogging/feature/state-of-the-blogosphere-2008/). As of 2009, a

reported 77% of active Internet users read blogs regularly, totaling up to 346 million

blog readers (see ‘‘State of the Blogosphere 2009’’ at http://technorati.com/blogging/

feature/state-of-the-blogosphere-2009/).

Only a small portion of these blogs are news focused; the most popular topic for

blogs continues to be the author’s personal life, and few serve the ‘‘filtering’’ and

link-sharing role often associated with blogs in the political communication domain

(Wei, 2009). Similarly, most blogs have no professional aspirations, and the majority

(72%) of bloggers consider themselves ‘‘hobbyists,’’ writing for personal satisfaction

(see ‘‘State of the Blogosphere 2009’’ at http://technorati.com/blogging/feature/

state-of-the-blogosphere-2009/). Readers have differing motivations, particularly

information seeking and media checking, convenience, personal fulfillment, political

surveillance, social surveillance, and expression and affiliation (Kaye, 2005). These

motivations may explain why even personal blogs often have informational elements

and why many of the most widely read blogs are more explicitly informative or news

focused.

Nonetheless, blogs are not a primary channel for media sources to spread hard

news, no doubt in part because they rely on the blog owner to pass on stories and,

thus, provide context that the press cannot control. Instead, many news organizations

have turned to microblogging and social media to share content. Microblogging is a

hybrid of blogging and instant messaging in which updates form a feed, similar to a

318 M. Schmierbach & A. Oeldorf-Hirsch

Dow

nloa

ded

by [

Uni

vers

ity T

own

Lib

rary

of

Shen

zhen

] at

06:

02 1

1 Ju

ly 2

012

blog, and users can quickly reply to or re-post others’ updates. The most popular ser-

vice is Twitter, a site that allows users 140 characters to send tweets to their followers.

Users can enter text and shortened URLs that lead to pictures or other sites using the

Twitter homepage, their mobile phone, or one of hundreds of microblogging applica-

tions that show up on their desktop, their browser, or other Web sites (for a list of

more than 260 applications that allow Twitter monitoring, see Goldstein, 2011).

Other users can then choose to ‘‘follow’’ this user, meaning that they receive all of

that person’s updates in an ongoing feed. Alternatively, they can also visit the Twitter

site or another page where the individual feed can be displayed.

As of 2011, Twitter attracts at least 190 million users per month (Schonefld, 2010),

who post 150 million tweets per day (Siegler, 2011). Although daily chatter is com-

mon, news reporting and information sharing are other popular uses of this and

similar microblogging platforms (Java, Song, Finin, & Tseng, 2007). Twitter’s identity

as a news source is rapidly becoming more prominent, and the service has been

recognized as a useful news and current events tool (Kwak, Lee, Park, & Moon,

2010; Phelan, McCarthy, & Smyth, 2009; Sakaki, Okazaki, & Matsuo, 2010). In fact,

data collected and analyzed from millions of tweets on thousands of trending topics

found that over 85% of trending topics on the site are headline news topics or

persistent news topics (Kwak et al., 2010).

Considerable attention has been paid to traditional blogs and news use, and a

growing amount of work considers Twitter as well. It is beyond the scope of this arti-

cle to consider the full range of such research, but several studies highlight the com-

parisons offered between Twitter and more traditional journalism outlets. Much of

this research focuses on the use of Twitter by so-called citizen journalists, demon-

strating how the technology allows ordinary people the opportunity to shape and

comment on the mainstream media agenda (Hermida, 2010), although some ques-

tion its utility in the face of low levels of adoption (Murthy, 2011). Since its creation,

Twitter emerged as a user-generated source of real-time news, often breaking news

before any other media source, as in the Iran protests in 2009 (Grossman, 2009).

Ampofo, Anstead, and O’Loughlin (2011) argued that debate viewers used Twitter

to directly comment on and shape the dialogue about political discourse.

At the same time, research suggests that journalists are also increasingly enthusi-

astic about Twitter (Arceneaux & Weiss, 2010) and are willing to make use of the

technology in promotion (Butcher, 2009; Oberholtzer, 2011). Ahmad (2010) argued

that Twitter would function as a useful promotional tool for journalists, but the

approach was driven by critical theory rather than empirical evidence. A content

analysis by Muralidharan, Rasmussen, Patterson, and Shin (2011) concluded that

media outlets do use Twitter and other social media to disseminate content, but often

fail to take advantage of the potential for interaction with the audience. Perhaps the

most notable studies in this area were conducted by C. F. Greer and Ferguson (2011;

see also Ferguson & Greer, 2011), which evaluate the use of Twitter by television and

radio stations, respectively. For example, in their study of local television use of the

technology, these authors found that, by far, the most common use was in passing on

news, rather than other promotional techniques. Although such studies can catalog

Communication Quarterly 319

Dow

nloa

ded

by [

Uni

vers

ity T

own

Lib

rary

of

Shen

zhen

] at

06:

02 1

1 Ju

ly 2

012

how media outlets have employed Twitter, they tell us little about how audiences

respond. As Chen (2011) pointed out, social gratifications are a key motivator for

Twitter use, and one-directional news distribution may be unlikely to serve such

motivations. Thus, questions about the value of Twitter for journalists remain, and

at present the most common journalistic use of Twitter appears to be as a means

to promote or disseminate content rather than engage in social conversations, sug-

gesting traditional news outlets may not be realizing the full potential of social media.

Finally, a study by Schultz, Utz, and Goritz (2011) presented a design most similar

to the approach taken in this study, comparing the effects of different message stra-

tegies on audiences. In this study, participants were presented with crisis communi-

cation from a company through a newspaper, a blog or Twitter. Although the results

were varied, in general individuals receiving a message via Twitter were less likely to

share it with others and had a lower impression of the company. Thus, this research

at least hints that the promise of Twitter might also hold some peril. We consider this

in the context of news media and the effects Twitter could have on the perceived

credibility of the outlet and its messages.

Credibility Assessments

Credibility is generally defined as the extent to which individuals find information or

its source believable, accurate, and trustworthy (Flanagin & Metzger, 2000). Much of

the prior research exploring the perceived credibility of Internet-based information

has shown that online information is considered as or more credible than offline con-

tent. For example, Johnson, Kaye, Bichard, and Wong (2008) found that some online

news sources were rated more credible than their offline counterparts. Kiousis (2001)

found that online news was rated more credible than television, although less credible

than newspapers. Flanagin and Metzger (2000) reached a similar conclusion. Thus, it

appears that online information can be credible. As an Internet-based tool, Twitter

may be evaluated similarly to other online information sources. In particular, as a

type of microblogging, Twitter may be perceived in a manner consistent with other

blogs. Research on blog credibility provides a mixed picture. A series of studies by

Johnson and colleagues (Johnson & Kaye, 2004, 2009; Johnson et al., 2008) indicated

that blogs were widely considered credible, even by comparison to more traditional

news sources. However, these judgments were largely held by regular blog users;

non-users had lower opinions of blog credibility. In addition, other research suggests

that traditional online news sources sometimes rank higher than other online sources

in terms of credibility (Flanagin & Metzger, 2007; Melican & Dixon, 2008).

One complicating factor for both Twitter and traditional blogs is that the source of

information can be unclear. As Sundar (1999) pointed out, whereas traditional mea-

sures of credibility are mostly applied to content producers, in an online environ-

ment people assess individual pieces of information on the basis of various cues

about credibility. These include not only the original source of information but also

what Sundar and Nass (2001) called the selecting source—that is, the individual or


Dow

nloa

ded

by [

Uni

vers

ity T

own

Lib

rary

of

Shen

zhen

] at

06:

02 1

1 Ju

ly 2

012

organization responsible for suggesting a story to a reader but not for gathering the

information initially. Their study found more favorable evaluations for information

selected by other users, rather than expert editors or oneself. This could help explain

the favorable ratings blogs receive in some studies. In a study of health messages, Hu

and Sundar (2010) considered both the originating source (e.g., a doctor or layper-

son) and the selecting source (e.g., a specific Web site, blog, or personal page). Their

study found no direct effect of either. Given that other research (Eastin, 2001) has

found that health information from experts is seen as more credible, it could be that

the selecting aspect of the process acted to partly mask normal source effects. How-

ever, Sundar, Knobloch-Westerwick, and Hastall (2007) found that the originating

source still influenced credibility when individuals were presented with information

on a news-aggregating site.

Compared with blogs and news aggregation services, Twitter represents a more

complex, ambiguous case. A traditional blog that posts links to news content clearly

has both an originating source and a selecting source. However, when news outlets

use Twitter to promote their content, these lines blur. For example, the official news

feed of The New York Times does not qualify as a ‘‘selecting source’’ as technically

defined in prior research because no third party is making judgments about story

appropriateness or importance. Yet, it could be that individuals will nonetheless

allow their feelings about Twitter to influence their assessment of the credibility of

the information source, although Twitter, as an entity, did not actually select the

tweeted content. If so, this would suggest that individuals do not necessarily make

credibility judgments through a careful consideration of sources and message content

but, rather, allow superficial cues to sway their judgments. If so, these cues may also

carry over to judgments about message credibility, if readers do not attend carefully to

content features in forming such judgments. On the other hand, some individuals

appear to distinguish between message and source credibility (J. D. Greer, 2003;

Kiousis, 2006). Kiousis measured message and source credibility of online news sep-

arately and found that the use of a Web site’s multimedia features had a significant,

positive influence on the perceived credibility of the source, but not the message. This

illustrates that even when a story’s credibility is seemingly unaffected by the medium,

perceptions of the source may be influenced. In a study on advertising’s effects on an

online story’s credibility, J. D. Greer found that, although message and source credi-

bility perceptions are correlated, the extent to which they are related differs by story

context. Thus, Twitter could affect judgments about source credibility to the extent

that it is perceived as a selecting or even originating source. In turn, these evaluations

might carry over to feelings about the message, but the literature offers reasons for

cautious and distinct assessment of both variables.

Hypotheses

Predicting the effects of Twitter requires combining seemingly disparate information

from several lines of research on credibility. Some of this work shows Internet sources


Dow

nloa

ded

by [

Uni

vers

ity T

own

Lib

rary

of

Shen

zhen

] at

06:

02 1

1 Ju

ly 2

012

can be seen as quite credible. Even blogs, which are similar to Twitter in important

ways, are often rated as trustworthy, although primarily by users or those otherwise

disinclined to solely rely on mainstream media (Choi, Watt, & Lynch, 2006). How-

ever, blogs present ‘‘user-selected’’ content whereas Twitter may not, at least in the

case of a feed from the media outlet itself. The literature also shows that people

may evaluate the message and the source in distinct ways, and that the originating

source (in this study, The New York Times) may be less influential than opinions

about the selecting source.

Thus, evaluations of Twitter specifically may be important in shaping perceptions

of both source and message; and, in this regard, Twitter may have factors working

against it. As noted earlier, Shultz et al. (2011) found individuals exposed to corpor-

ate crisis communication via Twitter had somewhat more negative impressions of the

company, suggesting that people may not respond as favorably to communication via

that medium. A study by Dutta-Bergman (2004) found that jargon-dependent or

incomplete health messages were seen as less credible. Given the brevity of Twitter

messages even compared with traditional blogs, this might suggest the typical reader

will respond negatively to typical tweets compared with complete news stories.

Twitter credibility may also suffer because of perceptions of its use by and for celeb-

rities, given their strong presence and ‘‘most-followed’’ status on the site (Hargittai &

Litt, 2011; Marwick & Boyd, 2011; Wu, Hofman, Mason, & Watts, 2011). To evaluate

the specific effects of ‘‘tweeting’’ on credibility judgments, we compare responses to a

story located on a newspaper Web site to judgments about a tweet from that same

newspaper advertising the story. Although the true source remains constant, it

may be that individuals will perceive Twitter as a kind of quasi-selecting source,

affecting credibility judgments. The message format could also influence these judg-

ments. Because Twitter provides fewer markers of being a traditional news source and

gives only a brief synopsis of a story, we expect that, unlike traditional blogs, Twitter

posts will not be viewed favorably, either in terms of the message itself or the overall

source credibility, as reflected in the following hypotheses:

H1a: Individuals will perceive the message as less credible when presented as aTwitter post by a news organization than as a story posted on the organization’sWeb site.

H1b: Individuals will perceive the source as less credible when presented as aTwitter post by a news organization than as a story posted on the organization’sWeb site.

By itself, credibility is a noteworthy outcome, and it is the primary focus of this

study. Credibility is often cited as a factor linked to declining media use (Johnson

et al., 2008), and it is a rich variable worthy of investigation. However, declines in

credibility may also undercut the ability of media to shape issue perceptions even

among those who continue to consume news (Wanta & Hu, 1994). If the Twitter

posting reduces credibility, it could also reduce perceived importance. In addition,

by including less information, the Twitter post and the shorter story would also


Dow

nloa

ded

by [

Uni

vers

ity T

own

Lib

rary

of

Shen

zhen

] at

06:

02 1

1 Ju

ly 2

012

rob individuals of the context needed to determine that a story was important. Story

length could also be a heuristic showing the importance of the issue—as demon-

strated in Graber’s (1988) study showing that individuals use length as one factor

in deciding which news stories matter. Overall, this leads to the following hypothesis:

H2: Individuals will perceive content as more important when presented as a longerstory than either as a short story or a Twitter post by a news organization.

Ultimately, one of the reasons news organizations use Twitter is to try to increase

readership. The hope, presumably, is that potential readers will be intrigued by the

brief headline and click the accompanying link to learn more, pushing up the number

of visitors to the organization’s site. Certainly, this seems like a logical prospect. By

limiting information, such brief tweets might increase uncertainty. Uncertainty

Reduction Theory (Berger & Calabrese, 1975; Sept, Hildebrand, & Wexler, 1992) pro-

poses that individuals will seek to resolve the anxiety and discomfort caused by this

uncertainty by seeking information, and some research on media suggests at least

limited circumstances in which this might occur (Boyle et al., 2004; Kubey & Peluso,

1990). On the other hand, if credibility or importance is seen as lower for the Twitter

condition, this would probably make people less interested in pursuing the story

further, particularly through the same news source. Ultimately, the current literature

simply is not sufficient to make a clear prediction, but potential changes in infor-

mation seeking as a result of Twitter exposure are important, as we explore in the

following research question:

RQ1: How will intended information seeking after exposure to a Twitter post differcompared with exposure to a version of the story on the news organizationWeb site?

Study 1

Method

Participants were recruited by students at a large, Northeastern U.S. university. Each

student was required to recruit at least 15 volunteers. These individuals were con-

tacted by e-mail and directed to an online consent form. Consenting participants

were randomly assigned to one of three story conditions. After dropping a small

number of participants who incorrectly answered a question about the story topic,

the total N was 225.

Most of these participants were current students (71.1%), primarily at the same

school as those recruiting (54.7%). The majority of the remaining participants had

completed college (20.9%). Participants were predominantly women (63.6%) and

White (88.9%). The participants were generally young (M¼ 25.36, SD¼ 11.02),

but ages ranged from 18 to 74. Only 28.4% of participants used Twitter. Although

this is a low rate, it is substantially higher than overall reported usage; according

to research by the Pew Internet and American Life Project (Smith & Rainie, 2010),


Dow

nloa

ded

by [

Uni

vers

ity T

own

Lib

rary

of

Shen

zhen

] at

06:

02 1

1 Ju

ly 2

012

just 8% of Internet users and 14% of those under 30 reported using Twitter when

surveyed in November 2010, and rates were most likely lower during the time of data

collection in the Fall of 2009.

As noted, participants viewed one of three versions of a news story adapted from

The New York Times. Participants were presented with a screen-captured image of the

story, which described a speech by President Obama on health care. The long form of

the story provided the first four paragraphs of this story as they appeared on the

‘‘Money & Policy’’ section of the newspaper’s Web site. All ads were removed, as

was material indicating the date. A ‘‘next page’’ box appeared at the bottom of the

story, which had the headline, ‘‘Seeking to Woo the Insured, Obama Cites Risk of

Losing Coverage.’’ In contrast to this version of the story, the short version included

just the first paragraph, and the box at the bottom of the story said ‘‘read story.’’

Otherwise, the format matched. Finally, the Twitter version showed the ‘‘nytimes’’

Twitter feed page on Twitter’s site, and listed the tweet for this story: ‘‘Obama Keeps

Up Health Care Push, Citing Uninsured,’’ along with a bit.ly link and the note that

the tweet was made recently. Although hosted by Twitter, this page clearly indicates it

is associated with The New York Times. For all three versions, the story or tweet was

the actual text posted by the newspaper, with no alterations other than shortening the

material. For all versions, other tweets, headlines, and stories were deleted.

For all three versions, participants were told the image was a screen capture of a

recent news item that appeared online, and that they should look at it before clicking

to proceed. In total, 52 individuals read the Twitter version, 77 read the short version,

and 96 saw the long version. Participants were randomly assigned to conditions.

Although true randomization means small differences in cell sizes could occur by

chance, the inequality in numbers between conditions suggests higher numbers of

participants in the short-story and Twitter versions dropped out prior to submitting

their responses. Because they did not complete the study, we cannot be sure how they

may have differed, but informal feedback from some participants suggests a portion

of those in the Twitter and short-story versions may have believed the displayed story

was incomplete or broken. If anything, this would suggest our results may underesti-

mate negative feelings toward this condition, but we cannot be certain, and potential

inequalities between conditions are a concern. To help address this, several control

variables are included in analysis, including age, student status, prior use of Twitter,

and gender, as described earlier.

Dependent Variables

Source credibility. Participants were asked to indicate how credible they found the

source of the information. ‘‘Source’’ was used as an intentionally ambiguous label,

allowing participants to perceive source in their own terms, rather than forcing them

to evaluate either The New York Times by itself or explicitly prompting an assessment

that included Twitter. This also ensured equivalent wording across conditions. An

index was formed based on two questions, which were presented on a 7-point scale

ranging from 1 (strongly disagree) to 7 (strongly agree) (M¼ 5.30, SD¼ 1.42;


Dow

nloa

ded

by [

Uni

vers

ity T

own

Lib

rary

of

Shen

zhen

] at

06:

02 1

1 Ju

ly 2

012

Cronbach’s a¼ .91). Specifically, participants were asked how much they agreed that

they ‘‘trust the source of this information’’ and ‘‘believe this source is credible.’’

Message credibility. Participants were also asked to indicate how credible they

found the specific message they were presented with, using an index based on two

questions presented on the same 7-point scale (M¼ 5.04, SD¼ 1.43; a¼ .93). Specifi-

cally, participants were asked how much they agreed that they ‘‘believe the content of

this text is true’’ and ‘‘trust the specific information in this text.’’

Issue importance. Participants were asked how important they found health care

reform, which likely indicates both involvement in the topic and support for reform.

An index was formed based on six questions, all of which were presented on the

7-point agreement scale (M¼ 5.39, SD¼ 1.35; a¼ .93). Specifically, participants were

asked how much they agree that health care reform and (separately) the number of

uninsured is important to them personally, that it is important to the nation as a

whole, and that it will be important in 10 years.

Information-seeking intent. Participants were asked how likely they were to pursue

this issue further. An index was formed based on nine questions, all of which were pre-

sented on a 7-point scale ranging from 1 (not likely) to 7 (very likely) (M¼ 4.38,

SD¼ 1.34; a¼ .89). Participants indicated how likely they were to take the following

actions regarding ‘‘the issue of health care and the number of uninsured’’:

Click the link found in the story, read additional online stories, search for infor-mation, read additional stories in a print newspaper, watch network television newscoverage, watch cable news coverage, read blogs, talk with friends, and talk withfamily.

Information satisfaction. Finally, participants were asked how ‘‘satisfied [they

were] with the information found in the text.’’ This was measured with a single ques-

tion, as described earlier, using a 7-point agreement scale (M¼ 3.64, SD¼ 1.64).

Study 1 Results

As noted, analysis included four control variables: age, prior use of Twitter, gender,

and student status. All hypotheses were initially tested using an analysis of covari-

ance (ANCOVA), with these control variables as covariates and story condition as

the independent variable. H1a predicted that people would see stories posted on

The New York Times Web site as more credible than those posted via Twitter.

The analysis shows no significant effect of story condition on message credibility,

F(2, 184)¼ 1.13, ns. As shown in Table 1, perceived message credibility, although

somewhat lower for those in the Twitter condition, did not significantly vary

between groups. Thus, H1a was not supported. H1b predicted that people would

see stories posted on The New York Times Web site as having a more credible

source than those posted via Twitter, although they also come from The New York

Times. This finding was supported. The analysis showed an overall significant effect


Dow

nloa

ded

by [

Uni

vers

ity T

own

Lib

rary

of

Shen

zhen

] at

06:

02 1

1 Ju

ly 2

012

of story condition on source credibility, F(2, 184)¼ 4.90, p< .01 (g2p ¼ .05). As

shown in Table 1, perceived source credibility was significantly lower in the Twitter

condition than in either the short- or long-story conditions, which did not signifi-

cantly differ from one another.

H2 predicted that individuals would see the issue as more important if presented

with a long story. This hypothesis was not supported, as the analyses found no sig-

nificant difference between story conditions, F(2, 187)¼ 0.24, ns. Finally, RQ1 asked

how story length and location would affect information-seeking intention. The analy-

ses show no evidence that length or location influenced intended information seek-

ing, F(2, 186)¼ 0.31, ns. The literature suggested potentially competing mechanisms

by which length and location would affect information seeking. One possible mech-

anism, importance, was not influenced. Additional analysis confirms that condition

had an effect on the alternative mechanism of information satisfaction, F(2,

184)¼ 6.54, p< .01 (g2p ¼ .07). However, as Table 1 shows, the levels of satisfaction

did not match expectations. Whereas those who saw the short story were noticeably

less satisfied than those reading the long story, those who saw only Twitter were as

satisfied as those reading the long story. Stories posted on Twitter are seen as less

credible, but fail to evoke curiosity about the issue.

To further explore the results, a regression analysis was run using issue impor-

tance, credibility, information satisfaction, and the control variables to predict infor-

mation seeking. Perceived importance was a positive predictor of intent to seek

future information (b¼ .36, p< .001), whereas neither source (b¼� .46, ns) nor

message credibility (b¼ .17, ns) were significant predictors. Information satisfaction

was unrelated to information seeking (b¼ .01, ns). It is possible the null findings for

credibility are a reflection of colinearity, as the two measures are strongly correlated

with one another (r¼ .85, p< .001). Thus, we evaluated the zero-order correlation of

both measures of credibility to information seeking, and found that both source

credibility (r¼ .38, p< .001) and message credibility (r¼ .38, p< .001) were signifi-

cantly related to information-seeking intention. Thus, it may be that story location

and length had an indirect effect on information seeking by way of credibility judg-

ments, but the evidence for this is weak.

Table 1 Credibility, Importance, Information-Seeking, and Information Satisfaction

Scores by Post Length and Location (Study 1)

Dependent variable Twitter Short story Long story

Message credibility 4.77a 5.11a 5.18a

Source credibility 4.76a 5.47b 5.56b

Importance 5.43a 5.36a 5.51a

Information seeking 4.41a 4.38a 4.35a

Information satisfaction 3.92a 3.04b 3.92a

Note. Scores shown are estimated means with age, gender, Twitter use, and student status as covariates. Means

not sharing a subscript significantly differ on pairwise comparisons (p< .05).


Dow

nloa

ded

by [

Uni

vers

ity T

own

Lib

rary

of

Shen

zhen

] at

06:

02 1

1 Ju

ly 2

012

One common explanation for differences in credibility in the literature stems from

differences in use. Thus, we explored whether individuals who reported using Twitter

had different credibility perceptions between conditions, relative to non-users. To do

so, we replicated the ANCOVA analysis reported earlier, but treated Twitter use as an

additional moderating factor, rather than a covariate. The results gave no indication

of a moderating role for Twitter use when considering either source credibility, F(2,

182)¼ 0.17, ns; or message credibility, F(2, 181)¼ 0.67, ns. Unfortunately, there are

too few individuals in the sample who used Twitter to further probe for an influence

based on differing levels or types of use.

Overall, then, the results suggest that Twitter is seen as a less credible source that

may or may not present a less credible message, but that this might not affect opi-

nions about the issue. The similar credibility for short and long stories suggests that

level of information does not account for Twitter’s lower credibility. To further

explore how the ‘‘location’’ in which a story appears relates to credibility, we carried

out a second study holding the length and level of information more constant. In this

study, we added a third potential ‘‘source,’’ a private blog page posting a story link.

This allows us to more fully explore the outlets through which individuals might

gain access to the same news: the outlet’s Web site, the official Twitter feed, and

an unofficial blog reference. Given the unclear findings for information satisfaction

and the removal of variance in level of information provided, we omit that variable

in Study 2.

Study 2

Method

As in Study 1, participants were recruited by students at the same university, sent an

e-mail directing them to the online study, and randomly assigned to one of three

conditions: Twitter, newspaper Web site, or blog. Participants were more explicitly

told that they were viewing a screen capture that might include only brief text, based

on the belief that many of the participants who prematurely quit in Study 1 did so

because they did not perceive the Twitter or short versions as a complete ‘‘story’’

and, thus, were reluctant to answer questions about the presented content. This

appears to have been effective, as numbers across conditions were more equal. No

participants had to be removed for incorrectly identifying the topic. The total N

was 435. Similar to the first study, participants were primarily current students

(69.1%), women (62.1%), White (87.6%), and young (ages ranged from 18–68;

M¼ 26.31, SD¼ 11.05). As with Study 1, a minority of participants (21.1%) used

Twitter, but this rate exceeded estimates for overall population use. For this study,

data were collected in the Spring of 2010.

Participants viewed the story description on The New York Times Web site, The

New York Times’ Twitter feed on Twitter’s site, or on an individual’s blog created

specifically for this study. In all conditions, the same story was used. This story

appeared in the ‘‘Energy & Environment’’ section of The New York Times, with the


Dow

nloa

ded

by [

Uni

vers

ity T

own

Lib

rary

of

Shen

zhen

] at

06:

02 1

1 Ju

ly 2

012

headline ‘‘Wind Power Grows 39% for the Year.’’ To minimize the potential con-

found of level of information, The New York Times version appeared on a page listing

a number of different stories, giving only the headline and a one-sentence summary,

with the byline of the author. The other stories listed were blurred. The Twitter ver-

sion used the actual ‘‘nytimes’’ feed page, as in Study 1, with the original tweet visible

(reading ‘‘Wind Power Grows 39% for the Year’’), and followed by a bit.ly link and a

note that it was posted recently. As with the version from the newspaper Web site,

other stories were listed above and below this tweet, but they were blurred. Finally,

the blog version appeared on a blog created just for this study using Google’s Blogger

software. The blog was created under a fictitious username with the title ‘‘A day in the

life . . .,’’ and some fictitious entries were created. The New York Times story was listed

in a blog posting with the headline ‘‘NYTimes: Wind Power Grows 39% for the

Year,’’ text saying ‘‘Check out this New York Times article on wind power,’’ the same

image of windmills that appeared on the original New York TimesWeb site, and a link

reading ‘‘New York Times article.’’ Other entries above and below this post were

blurred. In total, 158 participants viewed The New York Times version, 131 viewed

the Twitter version, and 146 viewed the blog version.

Dependent Variables

Message credibility. Participants were asked to evaluate the credibility of the mess-

age using the same two-item index as in Study 1 (M¼ 4.22, SD¼ 1.39; Cronbach’s

a¼ .86).

Source credibility. Participants were also asked to evaluate the credibility of the

message source, again using the same two-item index as in Study 1 (M¼ 4.36,

SD¼ 1.55; a¼ .93).

Issue importance. Because analysis in the prior study showed little difference

between personal and general importance, as well as immediate and future impor-

tance, in this study we relied on a single question to measure issue importance. Part-

icipants were asked how much they agreed with the statement, ‘‘This is an important

issue,’’ using a 7-point agreement scale (M¼ 4.96, SD¼ 1.41).

Information-seeking intent. Participants were asked how likely they were to pursue

this issue further through a variety of venues. An index was formed based on 11 ques-

tions, all of which were presented on a 7-point scale ranging from 1 (not likely) to 7

(very likely) (M¼ 3.61, SD¼ 1.33; a¼ .92). Participants were asked about the same

venues as Study 1, and because of the added blog condition, participants also indi-

cated how likely they were to read comments about this issue and read blogs discuss-

ing this issue.

Results

Initial analyses were ANCOVA controlling for Twitter use, age, gender, and student

status, with story condition as the independent variable. In Study 1, H1a was not


Dow

nloa

ded

by [

Uni

vers

ity T

own

Lib

rary

of

Shen

zhen

] at

06:

02 1

1 Ju

ly 2

012

supported, although the results were directionally consistent with expectations. Here,

however, the results show a clear effect of story location on message credibility judg-

ments, F(2, 385)¼ 10.56, p< .001 (g2p ¼ .05). As shown in Table 2, the story appear-

ing directly on The New York Times Web page was seen as significantly more credible

than stories linked from a blog or Twitter. The latter were not significantly different

from one another, although the blog actually scored slightly higher on the scale,

despite not having any official affiliation with The New York Times.

Data for H1b and judgments about source credibility are consistent with the pre-

vious result and the findings from Study 1. Story location had a significant effect on

source credibility judgments, F(2, 385)¼ 10.37, p< .001 (g2p ¼ .05). As Table 2 shows,

the findings are consistent with the results for message credibility, with scores for The

New York Times significantly higher than the other outlets and the blog actually

scoring slightly, albeit not significantly, higher than Twitter.

Despite the differences in credibility, Study 1 found no support for H2, involving

issue importance and length. These analyses can only partly address this hypothesis,

as only one story length was included. We can still evaluate whether the posting

location for the story mattered, and these results do show an effect of story location

on perceived importance, F(2, 383)¼ 6.68, p< .01 (g2p ¼ .03). As shown in Table 2,

the results are consistent with the idea that Twitter posts would make the story seem

less important. The New York Times story was associated with greater issue impor-

tance than the Twitter post, but the blog post actually led to the greatest perceived

importance (albeit not significantly above the score for The New York Times).

Finally, the findings show no significant effect of story location on information-

seeking intention, F(2, 395)¼ 0.11, ns. We carried out additional analyses to further

explore this result. A regression model predicting information seeking with credibility

and issue importance, along with the control variables, showed that message credi-

bility (b¼ .36, p< .05) and issue importance (b¼ .28, p< .001) were both signifi-

cant, positive predictors of information seeking. Once again, the results may be

somewhat complicated by the strong correlation between the two measures of credi-

bility (r¼ .86, p< .001), and zero-order correlation tests showed a significant, posi-

tive correlation of both message credibility (r¼ .26, p< .001) and source credibility

Table 2 Credibility, Importance, and Information-Seeking Scores by

Post Location (Study 2)

Dependent variable Twitter New York Times Blog

Message credibility 3.90a 4.64b 4.14a

Source credibility 4.02a 4.82b 4.25a

Importance 4.63a 5.03b 5.27b

Information seeking 3.65a 3.59a 3.65a

Note. Scores shown are estimated means with age, gender, Twitter use, and student status

as covariates. Means not sharing a subscript significantly differ on pairwise comparisons

(p< .05).


Dow

nloa

ded

by [

Uni

vers

ity T

own

Lib

rary

of

Shen

zhen

] at

06:

02 1

1 Ju

ly 2

012

(r¼ .19, p< .001) with information seeking. Thus, similar to Study 1, it appears that

story location may affect information-seeking intent indirectly—in this case, by influ-

encing both credibility and importance judgments.

As in Study 1, we also carried out further analyses to consider why participants

viewed Twitter as less credible, introducing Twitter use as a moderating factor

instead of as a covariate. As in the first study, there was no significant interaction

of Twitter use and story location in affecting either source credibility, F(2,

383)¼ 0.40, ns; or message credibility, F(2, 383)¼ 0.41, ns. Similarly, there was also

no significant interaction involving these variables in predicting issue importance,

F(2, 381)¼ 0.45, ns. In Study 1, we were unable to probe this further due to the

small number of Twitter users. Thanks to the larger N for this study, we were able

to carry out exploratory analyses to more fully address the question of how Twitter

use might affect credibility judgments. We considered the differences in credibility

between the Twitter and newspaper conditions described earlier, focusing specifi-

cally on the 54 individuals in one of those conditions who indicated they used Twit-

ter. Obviously, this gives us limited power, and we note results even approaching

significance for that reason. We asked individuals who used Twitter to indicate

(on a 7-point Likert scale) how much they agreed that ‘‘Twitter is a great way to

get information’’ (M¼ 4.03, SD¼ 1.90), and ‘‘I follow a number of other users

on Twitter’’ (M¼ 4.90, SD¼ 1.92).

We then carried out a series of regression analyses that included the control vari-

ables of age, gender, student status, a dummy-coded variable indicating whether indi-

viduals were in the Twitter or The New York Times conditions, one of the Twitter

use=attitude measures, and an interaction term reflecting the product of that measure

and the condition variable.1 The primary point of these analyses is that last term; a

significant interaction would suggest that, although Twitter use itself did not moder-

ate the findings, the specific amount or nature of use might. Overall, the results fall

shy of significance, although the patterns are consistent with one another and with

what one might expect if positive attitudes toward Twitter could enhance credibility.

Indeed, if the analyses were run including the blog condition as well, many of these

coefficients would be significant, but interpretation would be more suspect. Both

interaction terms involving the measure ‘‘Twitter is a great way to get information’’

approach traditional significance levels. For source credibility, b¼ .38, p¼ .08; and

for message credibility, b¼ .33, p¼ .10. In both cases, then, the result suggests for

the participants in this study, at least, the gap between credibility in the Twitter

and The New York Times conditions—which was negative, on average, reflecting

lower credibility for Twitter—grew smaller as positive perceptions of Twitter

increased. Yet, as the results show, the difference in credibility is so low at the lowest

level of perceptions (the coefficient for the Twitter condition reflects the difference

when perception scores equal 0) that even at the highest levels of perception, credi-

bility between the two sources is essentially equal. It does not appear that even those

most favorably disposed toward Twitter ever see it as notably more credible than

the newspaper’s own Web site. Results showed a similar pattern for the variable mea-

suring whether individuals follow others on Twitter, but were even further from


Dow

nloa

ded

by [

Uni

vers

ity T

own

Lib

rary

of

Shen

zhen

] at

06:

02 1

1 Ju

ly 2

012

significance. For source credibility, the interaction of following and condition had

b¼ .25, p¼ .19; and for message credibility had b¼ .27, p¼ .13.

In summary, it may be that in a study specifically of Twitter users, with greater

power to detect relationships, active users of Twitter and those with especially favor-

able impressions of the medium would not display the same skeptical outlook found

in the results as a whole. However, the data here are not strong enough to conclus-

ively support that, and they show that when expressed in terms of mere use, those

who are ‘‘on’’ Twitter do not show significantly more favorable responses to the

medium than the majority of people online, who are non-users.

Discussion

Although research has considered differences in credibility among online information

sources, it has generally not explored how the same information, from the same

source, but distributed through different channels, would be perceived differently.

In focusing on the use of Twitter by a news organization, we were able to consistently

demonstrate lowered source credibility, although The New York Times is identified as

the source of the information, and the Twitter post actually appears on the ‘‘official’’

New York Times feed page. We found a similar pattern for message credibility,

although only in Study 2 was the difference significant. Some caution is necessary,

as the net effect size in these results was fairly small and the sample was limited.

The study offers good reason to question the credibility of media messages distribu-

ted by Twitter, but it would be premature to suggest that content shared this way

suffers from a major credibility problem.

Nonetheless, this result provides a notable contrast to studies that show blogs have

similar credibility to mainstream news. Whereas the ‘‘filtering’’ performed by a blog

may suggest the information is worthwhile, Twitter apparently lacks this effect. Per-

haps this stems from media coverage of Twitter linking it to celebrities and shallow

posts from ‘‘average’’ Americans. Exploring exact perceptions of Twitter and their

role in credibility judgments would be a worthy future direction. In this study, we

did not want to prompt people to necessarily view the ‘‘source’’ as either Twitter

or The New York Times, as the ambiguity of the source is a meaningful feature of

media use of Twitter. By contrast, blogs more explicitly indicate both an originating

and selecting source; a future study of a Twitter feed that serves a selecting function

would be a useful comparison. Notably, however, in this study even the blog post was

seen as less credible; and, although scores were higher than those of Twitter, the dif-

ference was not significant. Of course, whereas most widely read blogs would have a

recognizable ‘‘brand’’ themselves, as well as ongoing discussion and further analysis,

in this study we relied on a simple, direct link to the story—an expression of ‘‘I

noticed this,’’ similar to what happens with Twitter. Thus, the blog created for this

study probably scored worse in terms of credibility than a typical, widely known blog

might, particularly among Internet-savvy individuals. Yet, the blog did not perform

worse than Twitter, despite not having a recognizable brand or being presented as the


Dow

nloa

ded

by [

Uni

vers

ity T

own

Lib

rary

of

Shen

zhen

] at

06:

02 1

1 Ju

ly 2

012

official feed of a major newspaper, suggesting that the results probably underestimate

the degree to which Twitter posts may be viewed more negatively than blog links.

Meanwhile, these data do not fully illuminate the reasons for the differences. Prior

studies have suggested that individuals who are regular consumers of a specific news

source will see it as more credible. Yet, our analysis did not show a moderating effect

of mere Twitter use. However, it could be that many of those saying they ‘‘used’’

Twitter did so only rarely and did not specifically follow news organizations. Further

analyses in Study 2 were limited by the small number of Twitter users, but did offer

hints that people with more favorable impressions or greater use of Twitter saw smal-

ler credibility differences between sources, although not to the degree that they actu-

ally preferred Twitter. Further analysis in a study with greater numbers of Twitter

users or with users of the specific feed being considered would be valuable.

This raises a broader question regarding the implications of these results in natural

settings. On the one hand, it is tempting to dismiss the findings because a minority of

participants made any use of Twitter. Yet, our study actually showed higher levels of

use than those reported on representative surveys of Internet users. It is an open

question whether ‘‘non-users’’ would actually be affected by Twitter content. For

the most part, Twitter is a ‘‘subscription’’ driven service, with individuals receiving

only those feeds they have deliberately chosen. However, some inadvertent exposure

is possible. Google has at least intermittently provided Twitter results among search

outcomes. Many blogs and other Web sites display a scrolling list of recent tweets that

might be available even to those who do not see themselves as Twitter users. The New

York Times recently moved to a subscription wall model that allows individuals to

read a limited number of stories directly from the front page, but providing access

to additional content when it is linked, as through tweets or blogs. This might

prompt more people to initially encounter stories in those ways, even if they are

not primarily users of those media. Moreover, we do not know whether greater

use of Twitter is causing more positive credibility judgments. It could be that those

with the greatest trust in Twitter gravitated toward it first, and later users will remain

skeptical. Ultimately, the majority of Twitter users may differ from those represented

in this study, but the relatively rare case of individuals first encountering a media out-

let via Twitter, as reflected here, may be more meaningful. These individuals, after all,

are the ones who will decide whether to subscribe to a feed or otherwise pursue

further content. Nonetheless, future research should evaluate the sampling strategy

carefully. The study used here relied on a student-heavy population, which actually

helped boost the number of Twitter users, but compromises the ability to generalize

to the population as a whole. A more representative sample would have fewer users

but greater external validity, whereas a more deliberate sample of users would allow

further probing of the effects of usage and perceptions of Twitter.

The study also explores the potential consequences of credibility judgments. We

argued that Twitter should lead to stories seeming less important due to both low

credibility and brevity. In Study 1, we did not find support for this; issue importance

did not vary between conditions. In Study 2, however, the Twitter version was seen as

less important. It may be that in Study 1, the issue was already too widely known and


Dow

nloa

ded

by [

Uni

vers

ity T

own

Lib

rary

of

Shen

zhen

] at

06:

02 1

1 Ju

ly 2

012

discussed for a single manipulation to meaningfully affect judgments, making Study 2

a better test. If so, then the decreased credibility of Twitter-posted content has other

meaningful effects. Not only are people more skeptical of the information, but they

are also concluding that the underlying issues matter less. Notably, although the blog

posting also suffered from credibility concerns, it was seen as more important than

the Twitter post, suggesting that the added importance cue of having someone select

this story from among many may have influenced perceived importance. This under-

scores the ambiguity of Twitter as a source. Although clearly perceived as distinct

from content directly obtained from The New York Times, materials posted on The

New York Times’ Twitter feed do not appear to benefit from the positive influence

of coming from a ‘‘selecting’’ source.

Although credibility and importance are significant predictors of information-

seeking intent, in both studies the manipulations failed to influence this third out-

come variable. There may be some offsetting factor at play. Perhaps, as speculated,

the tweet format created curiosity about the issue, but the effects on credibility

and importance canceled this out; or, perhaps because participants were primarily

college students, their willingness to pursue further news is too low, regardless of

stimulus. That said, there are hints that the manipulation may have an indirect effect,

as some aspects of credibility were correlated with information-seeking intent, and

importance was a consistent, strong predictor of such intent. Perhaps repeated

exposure to content on a media Web site would amplify information seeking relative

to monitoring the same site via Twitter. Another consideration is the measurement of

intended information-seeking behavior. Any such self-report data are speculative,

and it is socially desirable to overestimate likely news use. Because of the software

used to gather data, it was impossible to allow participants to actually click on the

link in the various stories, but a design where this was possible would offer a useful

measure.

Overall, however, the findings here offer important theoretical and practical

insights. In contrast to many studies of online credibility, this study shows that even

somewhat regular users of Twitter do not see it as providing more credible infor-

mation, and the population as a whole is unusually skeptical of Twitter relative to

other means of distribution. Twitter seems to elicit a negative reaction from many.

The exact mechanisms are unclear, and we cannot say if this would carry over to

other microblogging tools that may be under development. On the surface, however,

it is noteworthy simply because unlike traditional blogs, Twitter here is not serving as

a selecting source—the decision to share this story was made by the originating

source, The New York Times. Yet, participants still viewed the content on Twitter dif-

ferently, suggesting that the Twitter ‘‘brand’’ has some independent cue. Perhaps the

positive responses to other selecting sources are also due to cues, and not to reasoned

evaluations about the benefits of custom-selected material. At an applied level, this

study suggests the need for caution in the use of Twitter as a way to distribute news.

Despite the official New York Times ‘‘stamp,’’ these stories were still viewed in a more

negative light when posted to Twitter. Participants even saw the larger news organi-

zation as less credible. It would be premature to say that using Twitter is necessarily


Dow

nloa

ded

by [

Uni

vers

ity T

own

Lib

rary

of

Shen

zhen

] at

06:

02 1

1 Ju

ly 2

012

hurting The New York Times, but absent evidence showing a positive effect, skepti-

cism seems warranted. The recent move by The New York Times to promote subscrip-

tions appears mixed, from this perspective; however, because it came after this study

was conducted, we cannot offer direct empirical insights on this point. The news-

paper may be cultivating a more engaged, trusting set of core users, but it may also

be increasing the chance that non-subscribers will come away with a negative

impression of content if they encounter it through Twitter or some other means.

At the very least, news organizations and others (such as corporations) should

pay attention to continued research in this area and not reflexively embrace all

new distribution technologies as equally beneficial.

Note

[1] To save space, we present the full results for these analyses in this note, and we describe only

a few critical findings. For source credibility, there are two models. The first includes the

measure where participants indicated Twitter was a great way to get information, along with

the relevant interaction term. Unstandardized coefficients and standard errors, in parenth-

eses, in this model were as follows: age, �0.06 (0.03); gender �0.07 (0.38); student status,

0.60 (0.63); Twitter is a great source, 0.04 (0.15); Twitter condition, �1.85 (1.04); great

Source�Condition, 0.37 (0.21); and total R2¼ .26. The second model includes the measure

where participants indicated following a number of other users on Twitter, along with the

relevant interaction term. The coefficients and standard errors, in parentheses, for this model

were as follows: age, �0.06 (0.03); gender �0.07 (0.38); student status, 0.60 (0.63); follow

others, 0.04 (0.15); Twitter condition, �1.85 (1.04); Follow�Condition, 0.37 (0.21); and

total R2¼ .26. For message credibility, there were also two models. For the model with

‘‘Twitter is a great way to get information,’’ the coefficients and standard errors, in parenth-

eses, were as follows: age, �0.03 (0.02); gender �0.05 (0.36); student status, �0.15 (0.60);

Twitter is a great source, �0.04 (0.14); Twitter condition, �1.84 (0.82); great Sour-

ce�Condition, 0.33 (0.20); and total R2¼ .16. For the model with ‘‘I follow a number

of other users,’’ the coefficients and standard errors, in parentheses, were as follows: age,

�0.05 (0.03); gender, �0.29 (0.38); student status, 0.59 (0.61); follow others, 0.16 (0.13);

Twitter condition, �1.69 (0.99); Follow�Condition, 0.25 (0.18); and total R2¼ .28.

References

Ahmad, A. N. (2010). Is Twitter a useful tool for journalists?. Journal of Media Practice, 11, 145–155.

doi: 10.1386=jmpr.11.2.145_1

Ampofo, L., Anstead, N., & O’Loughlin, B. (2011). Trust, confidence, and credibility. Information,

Communication & Society, 14, 850–871. doi: 10.1080=1369118X.2011.587882Arceneaux, N., & Weiss, A. (2010). Seems stupid until you try it: Press coverage of Twitter, 2006–9.

New Media & Society, 12, 1262–1279. doi: 10.1177=1461444809360773Berger, C. R., & Calabrese, R. J. (1975). Some explorations in initial interaction and beyond:

Toward a developmental theory of interpersonal communication. Human Communication

Research, 1, 99–112. doi: 10.1111=j.1468–2958.1975.tb00258.xBoyle, M. P., Schmierbach, M., Armstrong, C. L., McLeod, D. M., Shah, D. V., & Pan, Z. (2004).

Information seeking and emotional reactions to the September 11 terrorist attacks. Journal-

ism & Mass Communication Quarterly, 81, 155–167.


Dow

nloa

ded

by [

Uni

vers

ity T

own

Lib

rary

of

Shen

zhen

] at

06:

02 1

1 Ju

ly 2

012

Butcher, M. (2009, March 5). Sky News realises news breaks first on Twitter, not TV—Creates a

Twitter correspondent. TechCrunch. Retrieved from http://techcrunch.com/2009/03/05/

sky-news-realises-news-breaks-first-on-twitter-not-tv-creates-a-twitter-correspondent/.

Chen, G. M. (2011). Tweet this: A uses and gratifications perspective on how active Twitter use

gratifies a need to connect with others. Computers in Human Behavior, 27, 755–762. doi:

10.1016=j.chb.2010.10.023Choi, J. H., Watt, J. H., & Lynch, M. (2006). Perceptions of news credibility about the war in Iraq:

Why war opponents perceived the Internet as the most credible medium. Journal of

Computer-Mediated Communication, 12, 209–229. doi: 10.1111=j.1083–6101.2006.00322.xDutta-Bergman, M. J. (2004). The impact of completeness and Web use motivation on the

credibility of e-health information. Journal of Communication, 54, 253–269. doi: 10.1111=j.1460–2466.2004.tb02627.x

Eastin, M. S. (2001). Credibility assessments of online health information: The effects of source

expertise and knowledge of content. Journal of Computer-Mediated Communication, 6.

Retrieved from http://jcmc.indiana.edu/vol6/issue4/eastin.html.

Ferguson, D. A., & Greer, C. F. (2011). Local radio and microblogging: How radio stations in

the U.S. are using Twitter. Journal of Radio & Audio Media, 18, 33–46. doi: 10.1080=19376529.2011.558867

Flanagin, A. J., & Metzger, M. J. (2000). Perceptions of Internet information credibility. Journalism

& Mass Communication Quarterly, 77, 515–540.

Flanagin, A. J., & Metzger, M. J. (2007). The role of site features, user attributes, and information

verification behaviors on the perceived credibility of Web-based information. New Media &

Society, 9, 319–342. doi: 10.1177=1461444807075015Goldstein, E. (2011, May). The most complete Twitter application list available—2011 edition.

Social Media Today. Retrieved from http://socialmediatoday.com/eric-goldstein/295232/

most-complete-twitter-application-list-available-2011-edition

Graber, D. A. (1988). Processing the news: How people tame the information tide. White Plains, NY:

Longman.

Greer, C. F., & Ferguson, D. A. (2011). Using Twitter for promotion and branding: A content

analysis of local television Twitter sites. Journal of Broadcasting & Electronic Media, 55,

198–214. doi: 10.1090=08838151.2011.570824Greer, J. D. (2003). Evaluating the credibility of online information: A test of source and advertising

influence. Mass Communication & Society, 6, 11–28. doi: 10.1207=S15327825MCS0601_3

Grossman, L. (2009, June 17). Iran protests: Twitter, the medium of the movement. TIME.

Retrieved from http://www.time.com/time/world/article/0,8599,1905125,00.html.

Hargittai, E., & Litt, E. (2011). The tweet smell of celebrity success: Explaining variation in Twitter

adoption among a diverse group of young adults. New Media & Society, 13, 824–842. doi:

10.1177=1461444811405805Hermida, A. (2010). Twittering the news. Journalism Practice, 4, 297–308. doi: 10.1080=

17512781003640703

Herring, S. C., Scheidt, L. A., Bonus, S., & Wright, E. (2004). Bridging the gap: A genre analysis of

Weblogs. In Proceedings of the 37th annual Hawaii international conference on System Sciences,

2004 (pp. 101–111). Honolulu, HI: IEEE. doi:10.1109=HICSS.2004.1265271

Herring, S. C., Scheidt, L. A., Wright, E., & Bonus, S. (2005). Weblogs as a bridging genre. Infor-

mation Technology & People, 18, 142–171. doi: 10.1108=09593840510601513Hu, Y., & Sundar, S. S. (2010). Effects of online health sources on credibility and behavioral inten-

tions. Communication Research, 37, 105–132. doi: 10.1177=0093650209351512Java, A., Song, X., Finin, T., & Tseng, B. (2007). Why we Twitter: Understanding microblogging

usage and communities. In Proceedings of the 9th WebKDD and 1st SNA–KDD 2007 workshop

on Web Mining and Social Network Analysis—WebKDD=SNA–KDD ‘07 (pp. 56–65). New

York, NY: ACM Press. doi:10.1145=1348549.1348556


Dow

nloa

ded

by [

Uni

vers

ity T

own

Lib

rary

of

Shen

zhen

] at

06:

02 1

1 Ju

ly 2

012

Johnson, T. J., & Kaye, B. K. (2004). Wag the blog: How reliance on traditional media and the

Internet influence credibility perceptions of Weblogs among blog users. Journalism & Mass

Communication Quarterly, 81, 622–642.

Johnson, T. J., & Kaye, B. K. (2009). In blog we trust? Deciphering credibility of components of

the Internet among politically interested Internet users. Computers in Human Behavior, 25,

175–182. doi: 10.1016=j.chb.2008.08.004Johnson, T. J., Kaye, B. K., Bichard, S. L., & Wong, W. J. (2008). Every blog has its day: Politically-

interested Internet users’ perceptions of blog credibility. Journal of Computer-Mediated Com-

munication, 13, 100–122. doi: 10.1111=j.1083–6101.2007.00388.x.Kaye, B. K. (2005). It’s a blog, blog, blog, blog world. Atlantic Journal of Communication, 13, 73–95.

doi: 10.1207=s15456889ajc1302_2Kiousis, S. (2001). Public trust or mistrust? Perceptions of media credibility in the Information Age.

Mass Communication and Society, 4, 381–403. doi: 10.1207=S15327825MCS0404

Kiousis, S. (2006). Exploring the impact of modality on perceptions of credibility for online news

stories. Journalism Studies, 7, 348–359. doi: 10.1080=14616700500533668Kubey, R. W., & Peluso, T. (1990). Emotional response as a cause of interpersonal news diffusion:

The case of the space shuttle tragedy. Journal of Broadcasting & Electronic Media, 34, 69–76.

doi: 10.1080=08838159009386726Kwak, H., Lee, C., Park, H., & Moon, S. (2010). What is Twitter, a social network or a news media?

In Proceedings of the 19th international conference on World Wide Web—WWW ‘10 (pp.

591–600). New York, NY: ACM Press. doi:10.1145=1772690.1772751Marwick, A., & Boyd, D. (2011). To see and be seen: Celebrity practice on Twitter. Convergence:

The Journal of Research Into New Media Technologies, 17, 139–158. doi: 10.1177=1354856510394539

Melican, D. B., & Dixon, T. L. (2008). News on the net: Credibility, selective exposure, and racial

prejudice. Communication Research, 35, 151–168. doi: 10.1177=0093650207313157Muralidharan, S., Rasmussen, L., Patterson, D., & Shin, J.-H. (2011). Hope for Haiti: An analysis of

Facebook and Twitter usage during the earthquake relief efforts. Public Relations Review, 27,

175–177. doi: 10.1016=j.pubrev.2011.01.010Murthy, D. (2011). Twitter: Microphone for the masses?. Media, Culture & Society, 33, 779–789.

doi: 10.1177=0163443711404744Nardi, B. A., Schiano, D. J., & Gumbrecht, M. (2004). Blogging as social activity, or, would you let

900 million people read your diary? In Proceedings of the 2004 ACM conference on

Computer-Supported Cooperative Work—CSCW ‘04 (p. 222). New York, NY: ACM Press.

doi:10.1145=1031607.1031643Oberholtzer, J. (2011, April 12). The top 25 newspapers on Twitter. Forbes. Retrieved from http://

www.forbes.com/sites/jasonoberholtzer/2011/04/12/the-top-25-newspapers-on-twitter/.

Pew Research Center for the People, & the Press. (2008). Key news audiences now blend online and

traditional sources. Washington, DC: Author. Retrieved from http://people-press.org/report/

444/news.media

Phelan, O., McCarthy, K., & Smyth, B. (2009). Using Twitter to recommend real-time topical news.

In Proceedings of the 3rd ACM conference on Recommender Systems—RecSys ‘09 (pp.

385–388). New York, NY: ACM Press. doi:10.1145=1639714.1639794Sakaki, T., Okazaki, M., & Matsuo, Y. (2010). Earthquake shakes Twitter users: Real-time event

detection by social sensors. In Proceedings of the 19th international conference on World Wide

Web—WWW ‘10 (p. 851). New York, NY: ACM Press. doi:10.1145=1772690.1772777Schonfeld, E. (2010, June 8). Costolo: Twitter now has 190 million users tweeting 65 million times a

day. TechCrunch. Retrieved from http://techcrunch.com/2010/06/08/twitter-190-million-users/

Schultz, F., Utz, S., & Goritz, A. (2011). Is the medium the message? Perceptions of and reactions to

crisis communication via twitter, blogs and traditional media. Public Relations Review, 37,

20–27. doi: 10.1016=j.pubrev.2010.12.001


Dow

nloa

ded

by [

Uni

vers

ity T

own

Lib

rary

of

Shen

zhen

] at

06:

02 1

1 Ju

ly 2

012

Sept, R., Hildebrand, K., & Wexler, M. N. (1992). An evaluation of Uncertainty Reduction Theory.

Free Inquiry in Creative Sociology, 20, 97–108.

Siegler, M. G. (2011, April 6). Twitter tweets some big Q1 stats; 155 million tweets a day now. Tech-

Crunch. Retrieved from http://techcrunch.com/2011/04/06/twitter-q1-stats/

Smith, A., & Rainie, L. (2010, December 9). 8% of Americans use Twitter. Pew Internet. Retrieved

from http://pewinternet.org/Reports/2010/Twitter-Update-2010/Findings.aspx

Sundar, S. S. (1999). Exploring receivers’ criteria for perceptions of print and online news. Journal-

ism & Mass Communication Quarterly, 76, 373–386.

Sundar, S. S., Knobloch-Westerwick, S., & Hastall, M. R. (2007). News cues: Information scent and

cognitive heuristics. Journal of the American Society for Information Science and Technology,

58, 366–378. doi: 10.1002=asi.20511Sundar, S. S., & Nass, C. (2001). Conceptualizing sources in online news. Journal of Communi-

cation, 51, 52–72. doi: 10.1111=j.1460–2466.2001.tb02872.xWanta, W., & Hu, Y. (1994). The effects of credibility, reliance, and exposure on media agenda

setting: A path analysis model. Journalism Quarterly, 71, 90–98.

Wei, L. (2009). Filter blogs vs. personal journals: Understanding the knowledge production gap on

the Internet. Journal of Computer-Mediated Communication, 14, 532–558. doi: 10.1111=j.1083–6101.2009.01452.x

Wu, S., Hofman, J. M., Mason, W. A., & Watts, D. J. (2011). Who says what to whom on Twitter.

In Proceedings of the 20th international conference on World Wide Web—WWW ‘11

(pp. 705–714). New York, NY: ACM Press. doi:10.1145=1963405.1963504


Dow

nloa

ded

by [

Uni

vers

ity T

own

Lib

rary

of

Shen

zhen

] at

06:

02 1

1 Ju

ly 2

012

A Little Bird Told Me, So I Didn't Believe It: Twitter, …...Twitter, which both media outlets and...

Documents

Transcript of A Little Bird Told Me, So I Didn't Believe It: Twitter, …...Twitter, which both media outlets and...