Determinants of Internet News Use: A Structural Equation Model Approach

image_pdfimage_print

By Jin Yang and Padmini Patwardhan

[WJMCR 8:1 December 2004]

Sections:

Literature Review|Method|Findings|Discussion


By Jin Yang and Padmini Patwardhan

Abstract

The study examines how demographics, perceived credibility of Internet news and Internet dependency relations directly influence Internet current affairs news use and how perceived credibility of Internet news and Internet dependency relations mediate the influence of demographics on Internet current affairs news use. The study finds that both the perceived credibility of Internet news and Internet dependency relations affect Internet current affairs news positively. Age and education have no direct impact on Internet current affairs news, but their indirect impact is all negative. In other words, younger people and less educated people access Internet current affairs news more frequently because their relationship with the Internet appears stronger. Moreover, the study finds a positive correlation between Internet credibility and Internet dependency relations suggesting that when people hold a more trusting attitude toward Internet news, they would depend more on the Internet for various life goals achievement; when they depend on the Internet more for various life goal achievements, they would rate the Internet news more credible.

Introduction

Through its vast repository of information resources, the Internet is becoming an essential part of American politics. While 52 percent of Internet users said they went online to get news or information about politics, nationwide surveys suggested that online political news consumers grew dramatically from 18 percent of the U.S. population in 2000 to 29 percent in 2004.1 In 2000, about 69 percent of Americans claimed to find reliable and up-to-date news online,2 attesting to a growing confidence in the Internet�s informational efficacy, and the number continues to increase every year. This perceived and potential importance of the Internet in political participation is the basis for our study�s investigation of factors that may exert an influence, whether direct or indirect, on Internet current affairs news consumption. Our study brings together key antecedent variables from the literature (perceived credibility, Internet dependency and social locus) in pursuit of this exploration. Its primary contribution being the proposition and testing of a structural model of Internet news consumption.

Approaches to exploring mass media usage and antecedents are grounded in a variety of conceptual frameworks. For example, extensive scholarship on media credibility over the last decade has empirically established the close relationship between perceived credibility of a medium and its significant usage. Such a view of mass media usage is rooted in a basic sociological principle: behaviors are influenced by attitudes. If people do not trust a medium, they are less likely to pay attention to its contents.3 Another line of research that attempts to explain media usage derives from media system dependency (MSD) theory.4 Exploratory research on MSD relations in the Internet context has found Internet dependency relations to be a significant predictor of several online behaviors, including online news use.5 A third approach adopts a social locus perspective, wherein various demographic factors are seen as important predictors of media use whenever a new medium enters the news mainstream.

Our study breaks new ground to examine these factors in an integrative way for the following reasons. First, the research studies cited above are conducted in an independent manner, and are not contextualized in one theoretical domain. For instance, we know credibility plays a significant role in news consumption, but we do not know how credibility works together with dependency to exert an impact on news use. Second, researchers predominantly examine direct effects of variables influencing mass media use, seldom reviewing their indirect impact. We take a more holistic approach in our study, specifically exploring the synchronous influence of perceived Internet news credibility, Internet dependency relations and demographics on Internet current affairs news use. We also examine the indirect impact of demographics through the mediation of Internet news credibility and Internet dependency relations on Internet current affairs news use. Integrating these attitudinal, relational, and demographic perspectives of individuals and taking into account both direct and indirect impact through constructing and testing a model of Internet current affairs news use, our study examines whether the three well-established mass media use perspectives work together to produce significant Internet current affairs news consumption.

Literature Review

Internet News Consumption

A key factor that contributes to the exponential growth of the Internet is its information resources. Researchers have identified five motivations that predict hours spent on the Web, and seeking information is one of them.The three main features of the Internet that drive people to online news resources are free content, convenient updates, and search and interactive opportunity. According to another study, the informational gratification obtained from the Internet is one of the driving forces behind its popularity.7 Industry reports also find informational Internet use widely prevalent. Nielsen/Netratings reports that U.S. Internet penetration reached nearly 75 percent in 2003, i.e. three quarters of Americans have access to the Internet at home.8 Getting news is one of the many things a majority of people do online (72 percent) based on a Pew Internet and American Life May-June 2004 survey.9 In addition, seventy-one percent of webmasters of news organizations have reported significant access to their web sites, with 79 percent indicating that their sites had more than two thousand �hits� per week in 1996 (Davis, 1999).10 The sites of major national and local news organizations continue to be the leading sources of online campaign information.11 Roughly half of online election news consumers (52 percent) said they went most often to the sites of major national media organizations such as CNN and the New York Times, while 18 percent report visiting websites of local news organizations.12 With Internet based news consumption firmly established at an individual level through both academic and industry research sources, our next task was to examine factors that directly or indirectly affected this consumption.

Demographics and Internet News

Traditionally, demographic variables have always been examined as the initial predictors of mass media use. For example, the decline of newspaper readership in the 1970s and 1980s generated a large amount of research on profiling national newspaper readers versus nonreaders.13 The findings indicated that demographics accounted for a significant amount of newspaper readership. Investigations into online news behavior from a demographic perspective also revealed that people reading news online were younger, better educated and more affluent.14 In a February 2004 survey by Pew Internet & American Life, 22 percent of Americans aged 65 or older reported having access to the Internet. By contrast, 58 percent of Americans aged 50 to 64, 75 percent of 30-49 year-olds, and 77 percent of 18-29 year-olds currently go online.15 In a 2002 national study on the ever shifting Internet population16 researchers found that a digital divide persisted on the demographic side: younger American were more wired than older Americans, and well-educated Americans were more wired than those who only completed high school. The fact that young people and well-educated people were more wired suggests that they may get more chances to access Internet news, and therefore actual exposure to Internet news would be more extensive in comparison to older users. In the light of the findings and logic, this study proposes the following hypotheses.

H1a: Age has a negative effect on Internet current affairs news use.

H1b: Education has a positive effect on Internet current affairs news use.

Perceived Credibility and Internet News For more than five decades, credibility research17 has suggested a strong correlation between perceived credibility of a medium and usage of the medium.18 If people do not trust the medium, they are less likely to pay attention to its contents.19 In the same vein, researchers found that the preferred medium is generally judged as the more credible medium.20 As the Internet wins more people who use it as a primary new source of information, the credibility issue presents itself in a new setting. A study surveyed politically interested Web users and found that they relied more on Internet sources for political information and judged them as more credible than traditional sources.21Schweiger conducted another study on Internet credibility among German web-users and non-users comparing Internet credibility with that of television and newspapers. He found positive perceptions of Internet credibility even though it was not as high as that of television and newspapers.22 In another 1999 U.S. industry study by the New York-based ScreamingMedia, 63 percent said the Internet provided them the most in-depth information, 57 percent said the Internet had �the most up-to-date information� and 51 percent said the Internet had �the most accurate information.�23 In a March 2002 national U. S. survey, the Pew Internet Project found that while 62 percent of Internet users had gone online in search of health information, 73 percent turned away from the web site information because of the low credibility perception due to the site�s commercialization, lack of information source identification and other factors.24Based on these observations the following hypothesis was proposed linking perceived Internet credibility with Internet news use.

H2: Perceived credibility of the Internet has a positive effect on Internet current affairs news.

Media Dependency Relations and Online News Usage

Media system dependency (MSD) theory regards the dependency relationship of individuals with the media system as an important driver of media use. The dependency relation derives from the media�s power or control over information resources and individuals, and groups and organizations’ reliance on these information resources to attain their goals.25 At the individual level, the goals-resource dependency relations are defined as:

…the extent to which attainment of an individual�s goals is contingent upon access to the information resources of the media system, relative to the extent to which attainment of media system goals is contingent upon the resources controlled by individuals.26

This relational aspect of MSD distinguishes it from other commonly used terms of media dependency or media reliance. Specifically, its conceptual definition features a relationship, and how strong or weak the relationship is with the mass media is gauged by the extent of �perceived helpfulness� of media in attainment of three sets of personal goals: understanding, orientation and play.27 These are further conceptualized in individual and social dimensions: understanding as self-understanding and social understanding, orientation as action orientation and interaction orientation and play as solitary play and social play.28 Therefore, in essence, the dependency relationship is still an attitudinal variable because it is based on respondents� evaluation of the medium�s performance out of experience. The six sets of goals range widely from cognitive need to behavioral performance. Social understanding and self-understanding implicate the knowledge and awareness of living conditions and status29 with social understanding featuring the knowledge of the social environment and perception of individuals� role in that environment and self-understanding relating to the development of self-concept including self-esteem, beliefs, values and attitudes.30

Action and interaction orientation are concerned with formulating detailed strategies to act and interact with others.31 Play dependency implies fantasy, escape, as also sheer enjoyment of recreational activities,32 with solitary play referring to enjoyment or diversion by oneself, and social play representing enjoyment or diversion in the company of others.33

The achievement of the six goals via the dependency on a certain mass medium is the so-called �dependency relationship.� A person�s dependency relation with the medium is captured quantitatively by measuring how effectively a certain medium facilitates the accomplishment of the six goals. Researchers have empirically demonstrated the usefulness of media system dependency relations in explaining media usage. For instance, a newspaper readership study found that dependency relations for social and self understanding explained a considerable amount of variance in newspaper readership beyond the variance explained by demographic variables.34 Another study on television dependency relation found positive correlations ranging from .22 to .35 between television dependency and television exposure and noted that a positive relationship should be expected between media dependency and media exposure.35

In the online environment, a study of Internet dependency relations examined intensity of Internet dependency relations (IDR) as defined by MSD theory and concluded that a positive but restrained dependency relationship between individuals and the Internet is in place.36 A follow up study by the researchers also found that IDR was a significant predictor of online shopping and online news reading.37 Indeed, many researchers have suggested that media dependency relations may be good predictors in examining media behaviors.38

This study thus proposes that:

H3: Internet dependency relations will have a positive effect on Internet current affairs news use.

The above three sets of hypotheses establish the direct impact of the three proposed antecedent variables on Internet current affairs news consumption. For the indirect impact, i.e. to show how perceived credibility of Internet news and dependency relations mediates the influence of demographics on Internet current affairs news consumption, we need to first show the mechanism that connects attitudinal variable and relational variable with Internet current affairs news use.39Then we need to establish the connections between demographics and attitudinal and relational variables. The mechanism that connects attitudinal and relational variables with the Internet current affairs news use has already been developed (H1a, H1b, H2, H3). Therefore the following sections establish the connection between demographics and attitudinal and relational variables.

Demographics and Perceived Credibility of Online News

Attitudinal variables are very important in the study of news consumption because psychological or attitudinal differences either represent �causes� of differences in reading habits, i.e. attitude influences news consumption directly, or the secondary effects of demographic differences i.e. attitude influence news consumption indirectly via demographic differences.40 A study on network news found a significant (though weak) relation between standard demographic variables (such as age, gender and education) and the attitude toward network news.41 Most important of all, research on the Internet confirmed that those who were older, male, and of high socioeconomic status were more critical of the Internet and scored lower on the credibility ratings of the Internet.42 Thus the following hypotheses related to the influence of demographics on perceived credibility of Internet news were proposed.

H4a: Age has a negative effect on the perceived credibility of Internet news.

H4b: Education has a negative effect on the perceived credibility of Internet news.

Demographics and Media Dependency Relations

Empirical findings concerning the connection between demographics and media system dependency relations show mixed results. Ball-Rokeach pointed out the existence of association between an individual�s social location and media dependencies (p.504).43 A study on television shopping located a negative linkage between education and television shopping dependency, but no linkage between age and television shopping dependency, and no linkage between income and television shopping dependency.44 A study on the digital divide suggests older people pursue a narrower range of life goals and online activities than younger people, implying older people are less dependent on the Internet in general than younger people.45 Education wise, a Pew Internet and American Life Project survey in 2000 found that the percentage of education groups online was as follows: 28 percent high school or less; 62 percent some college, 76 percent college degree or more.46 A 2004 Pew Internet study found that educated people were more likely to access the Internet from places other than work and home and more likely to become avid Internet users.47 Two things emerge from this discussion: that younger users are overall more dependent on the Internet, and higher levels of education increase Internet dependency. The following hypotheses relating age and education with Internet dependency relations were thus proposed.

H5a: Age has a negative effect on Internet dependency relations.

H5b: Education has a positive effect on Internet dependency relations.

Perceived Credibility and Internet Dependency Relations

Is a connection between perceived credibility of a medium and dependency relations with a medium tenable? While this is a legitimate question to ask in mass communication research, there is not much literature to turn to for empirical support. One may argue that in the Internet environment, an individual who perceives Internet news sources as reliable may tend to develop a closer relationship with the Internet. In turn, stronger Internet dependency relations should lead to a higher rating for the Internet on the credibility scale. The mutual beneficial connection between IDR and perceived credibility of the Internet seems sound and logical. Perceived credibility and media dependency relations have been independently explored in the context of news usage, but no studies have empirically tested the relation..

On the other hand, it is also possible to argue for an alternative model in which perceived credibility and media dependency relations influence news use as independent or unrelated variables. Despite the attitude individuals hold toward the Internet news (either credible or not in terms of information provided online), individuals may use the Internet anyway for non-information purpose such as for fun, entertainment and communication. As a result, there may not be any link between credibility and IDR.

Based on this concern the study constructs two models, each integrating all the proposed hypotheses: one correlating perceived credibility and Internet dependency relations (Figure 1), and the other without the correlation (Figure 2). Depending on the model index comparison, a decision will be made as to which model provides a better fit to the data. The following question explores the goodness of fit of the two models.

RQ1: Which proposed model better explains Internet current affair news usage: the one with the correlation between perceived credibility of Internet news and IDR or the one without the correlation?

FIGURE 1. Path model of perceived credibility, Internet dependency relations and demographics on Internet current affairs news use with the correlation between perceived credibility and Internet dependency relations. 

FIGURE 2. Path model of perceived credibility, Internet dependency relations and demographics on Internet current affairs news use without the correlation between perceived credibility and Internet dependency relations. 

Conceptual and Operational Definitions

Internet current affair news use was conceptualized as the extent to which respondents were exposed to Internet current affair news by browsing news oriented web sites including those of well-established news organizations such as big names like the New York TimesWall-Street JournalUSA Today, CNN, Fox News, ABC and NBC etc, web-only publications such as The Onion, the Salon and the Slate, and news services of Internet portal companies such as Yahoo! MSNBC. The use was operationalized as the frequency of checking Internet current affairs news on a weekly basis: never, less than once a week, once a week, a few times a week and at least once a day for three kinds of news: international news, national news and political news.48

Perceived credibility of Internet news was defined as a global evaluation of the objectivity of Internet news. The perceived credibility in this study does not focus on specific news stories or news sources but on the news channel as a message carrier as a whole. In other words, the perceived credibility of Internet news is the extent to which respondents trust Internet news as a whole. It was operationalized as a three-item index featuring accuracy, believability, and fairness on a semantic differential scale from very believable to not at all believable. The scale was borrowed from a study comparing Internet and traditional sources credibility with reliability adequately established.49

Internet dependency relations (IDR) was conceptualized as the extent to which users depended on the Internet to meet six goals as defined by MSD theory: social understanding, self understanding, action orientation, interaction orientation, social play and solitary play. It was operationalized as the respondents� mean score on the eighteen item MSD scale (each of the six goals was measured by three items) developed by media system dependency relations researchers.50

Method

Using a cross sectional email survey, the study was conducted in a large mid-western university community in the United States. The university population is hardly representative of the U.S. population, but the study�s exploratory nature in model testing may justify sampling from this highly wired community with ubiquitous access to the Internet.

Sampling

Respondents were selected using multi-stage stratified random sampling. The university population was stratified into two groups�students and non-students�to ensure age and education variability. For the student population, 500 email addresses were randomly selected from 11,850 students. The non-student population was stratified into five groups: faculty, staff, faculty and staff, civil service and retired members and a 10 percent sample (n = 563) was drawn randomly from each group. Together the total sample size for this study was 1063 respondents (500 students, 144 faculty, 122 staff, 27 faculty and staff, 219 civil service, 51 retired members).

Measurement

Internet usage for current affairs news was measured using three questions: one for international news, one for political news, and one for national news. The Internet dependency relations scale was borrowed from the previously validated 18-item individual media dependency scale developed by Grant, Guthrie and Ball-Rokeach.51 Respondents were asked to indicate how helpful the Internet was for the eighteen statements on a scale from not at all helpful, not very helpful, neutral, somewhat helpful, and extremely helpful. A three-item scale was adopted from Johnson and Kaye52 to measure perceived credibility of Internet news. Age was measured by asking the year of birth, education by choosing from high school diploma, some college, bachelor�s degree, some graduate school, master’s degree and doctoral degree. Table 1 presents a brief definition of all the manifest variables.

Table 1. Measurements of Latent Variables
Latent VariableObserved VariableVariable Abbreviation
Internet Current Affairs News UseExposure to political web news
Exposure to international web news
Exposure to national web news
Political
International
National
Perceived Credibility Internet NewsPerceived fairness of the web news
Perceived accuracy of the web news
Perceived believability of the web news
Fair
Accurate
believable
Internet Dependency RelationUsefulness of the Internet in fulfilling understanding goals
Usefulness of the Internet in fulfilling orientation goals
Usefulness of the Internet in fulfilling play goals
Understanding
Orientation
play
AgePhysical ageage
EducationFrom high school to doctoral degree on a six-point scaleedu

Survey Administration

The survey was distributed via email to respondents� email accounts at the beginning of 2002. An e-mail survey offers an efficient way of collecting data within a short period of time and with limited expenses because it eliminates postage, printing, and interviewer costs.53 The cover letter and the questionnaire were sent out as inline text. Respondents were instructed on how to answer the email survey. The second mailing was delivered one week later and the data were collected within the two-week period. Since there was no dramatic hard news event happening during the two-week period, it was assumed that the respondents� online news behavior was as typical as any other two-week period.

Reliability and Validity

A pilot study on 60 undergraduate students to pretest the Internet dependency relations scale on the overall 18 items indicated a very high Cronbach�s alpha (.93). For social understanding dependency relations, Cronbach�s alpha was .74; self-understanding .75; action orientation .55; interaction orientation .64; social play .77; and solitary play .89. Inter-item reliability for perceived credibility of Internet news was also satisfactory (.78).

A preliminary data screening procedure was conducted to check on the normal distribution of the data.54 Three variables had a minor skewness problem: international news, orientation and play. Five variables had a kurtosis problem: political news, international news, national news, age and education. Even though efforts were made to transform the data, the transformed scores were not as satisfactory as expected in eliminating skewness and kurtosis. In order to keep the findings easier to interpret, it was decided to use the raw data despite its non-normality.

Data Analysis

This study tested two alternative structural models with five latent variables using the correlation matrix. It is suggested that when one is mainly concerned with the pattern of the relationships among variables, the correlation matrix is a proper choice for Structural Equation Modeling (SEM) testing.55 Besides, the correlation matrix rescales all variables to have unit variance so the interpretation of the results is more simplified.56 The study followed a two-step procedure in estimating the parameters in the model. First, with all the latent variables correlated, the measurement model was tested to check for the fit to the data. Second, when a satisfactory fit was obtained for the measurement model, the structural components or paths were estimated. Both the measurement and structural models were estimated with LISREL 8.25 by the iteration technique of unweighted least squares, because unweighted least squares iteration is the choice for a sample that is small and has possible departure from multivariate normality due to the ordinal nature of the data.57

Findings

A total of 1,063 email questionnaires were sent over a two-week period at the beginning of 2002. One hundred fifty two emails were returned as failed deliveries and ten respondents refused to participate. Out of 901 email surveys delivered to the respondents, 147 responses were received, a response rate of 16.3 percent. Five incomplete surveys (truncated or uncompleted) were eliminated, leaving 142 valid questionnaires for analysis.

The sample�s constituent group distribution was faculty (27 percent), civil service (25 percent), students (23 percent), staff (19 percent) and others including retired members (7 percent). The sample had 59 percent females and 41 percent males. The sample respondents consisted of 1 percent with high school and 32 percent with some college, 12 percent with bachelor�s degree, 5 percent with graduate school, 25 percent with master�s degree, and 25 percent with a doctoral degree. In terms of ethnicity, the sample had 85 percent Caucasians, 13 percent non-Caucasians including African Americans, American Asians, American Indians, Latinos and Internationals while the rest (2 percent) did not indicate their racial identity. The minority representation of the sample was not far from the university minority representation of 17 percent. Income distribution was as follows: 23 percent low-income (below $24,999), 45 percent middle-income group ($25,000 to &75,999), 13 percent high-income group (above $80,000 but below $99,999) and 10 percent extreme high-income group (above $100,000). The remaining 10 percent provided no information on their income. The age distribution of respondents was very close to a normal distribution. More than 30 percent were between the ages of 18-34, 50 percent were between 35-54, and 13 percent were above 55. The average age of respondents was forty-one.

Measurement Model

Means, standard deviations and correlations among the manifest variables are presented in Table 2.

Table 2. Correlation and Descriptive Statistics for Observed Variables
Variables0102030405060708091011
01pol1.000
02itl.6431.000
03ntl.718.6901.000
04fai.137.067.1641.000
05acc.151.140.267.7241.000
06bel.202.189.268.643.6691.000
07und.424.421.562.139.172.2181.000
08ori.122.074.202.170.124.154.6831.000
09pla.115.118.219.211.203.145.597.6861.000
10age-.091-.004-.156-.065-.065-.096-.335-.332-.4651.000
11edu.015.078-.093-.016-.017.081-.341-.258-.351.5201.000
Means2.972.593.113.523.523.622.883.042.8841.1313.21
SD1.381.431.47.64.61.64.83.731.0013.211.65
skewness-.03.499-.062-.016.109.131-.463-.281-.194-.060-.101
kurtosis-1.385-1.148-1.422-.206-.238-.328-.323-.495-.880-.719-1.657

The correlation matrix was used as the input for the measurement and structural model testing. Table 3A presents a summary of model fit indices for the measurement model while standardized parameter estimates for the model are presented in Table 3B. As shown, model parameters were all significant (p<.01) and explained substantial amount of item variance (R2 ranged from .52 to .96). In terms of the absolute fit, standardized RMR was .06, not far removed from .05, a RMR value indicating a good fit to the data. GFI at .98 and AGFI at .96 exceeded .90, a value considered to indicate a good fit to the data. In terms of the comparative fit indices, NFI at .96, IFI at 1.00 and RFI at .94 all suggest a good fit to the data. However, the chi-square value of 108.43 indicates a poor fit to the data, but because the data are based on a small sample and the distribution of several variables in the sample is not normal, chi-square may not be a good index to rely on (Hu & Bentler, 1995).58 Hence, based on majority of indices, the measurement model can be reasonably accepted as a good fit to the data.59

Table 3A. Summary of Goodness-of-Fit Indices for Measurement Model
Measurement Model
Absolute Fit IndicesChi-square108.43 (df=36)
Standardized RMR.06
GFI.98
AGFI.96
Comparative Fit IndicesNFI.96
IFI1.00
RFI.94
Table 3B. Standardized Parameter Estimates for the Factors of Perceived Credibility of Internet News, Internet Dependency Relations and Internet Current Affairs News Use
Factors
IndicatorsPerceived Credibility of Internet NewsInternet Dependency RelationsInternet Current Affairs News UseR2
Fair.79*.63
Accurate.86*.74
Believable.82*.67
Understanding.95*.90
Orientation.72*.52
Play.75*.56
Political.77*.60
International.73*.53
National.98*.96

Note. * indicates p <.01

Structural Equation Model

In the second step of model testing, the structural components of the model were estimated, i.e. we attempted to answer the research question we had posed. The full model with the correlation between credibility and IDR was estimated first, and then the model without the correlation (also called nested model in SEM language) was tested. As noted, the strength of the SEM model fit lies in comparing model fits of �competing and theoretically plausible� models rather than assessing the model fit for one model in isolation.60

Almost all the indices indicated the full model provided a marginally better fit to the data than the nested model (Table 4). The nested model was found to be a significantly worse fit to the data when the correlation between credibility and dependency relations was constrained to zero because the chi-square difference at 28.91(df=1) was significant at .01 level. In other words, the correlation path between Internet credibility and dependency relations was so important that dropping it out of the model made the fit significantly worse. The standardized RMR, AGFI, NFI, IFI and RFI also indicated the full model fit the data marginally better than the nested model. All the evidence suggests that the full model with correlated perceived credibility and Internet dependency relations explain Internet current affair news usage in a better way.

Table 4. Assessing and Comparing Full Model & Nested Model Fit Indices
Full ModelNested ModelComparison
Chi-square151.60 (df=37)180.51 (df=38)28.91 (df=1)
Standardized RMR.09.10
GFI.95.95
AGFI.92.91
NFI.92.91
IFI.96.94

Specific Path Estimates

The full model provided a much better fit to the data, therefore specific path parameters were assessed on the basis of the full model. Figure 3 provides the information to confirm or reject the eight hypotheses. H1a (age has a negative effect on Internet current affairs news use) is rejected. H1b (education has a positive effect on Internet current affairs news use) is also rejected. H2 (perceived credibility of the Internet news has a positive effect on Internet current affairs news) is supported (path coefficient gamma at .15 significant at .05). H3 (Internet dependency relations has a positive effect on Internet current affairs news use) is strongly supported (gamma = .50 significant at .01). Both H4a (age has negative effect on the perceived credibility of Internet news) and H4b (education has a negative effect on the perceived credibility of Internet news) are not supported. H5a (age has a negative effect on Internet dependency relation) is supported but H5b (education has a positive effect on Internet dependency relation) is not supported. Contrary to the prediction, education influences Internet dependency relation negatively.

FIGURE 3. Path findings on the Path model of the perceived credibility, Internet dependency relations and demographics on Internet current affairs news use with the correlation between perceived credibility and Internet dependency relations. Note. ** significant at .01; * significant at .05

While the direct impact of age and education on Internet current affairs news use is non-significant (H1a and H1b are both rejected), the indirect impact of age and education through perceived credibility of Internet news and Internet dependency relations is significant (-.24 for age, -.19 for education) confirming the speculation that the relational variable mediates demographics impact on Internet current affairs news use (see Table 5).

Table 5. Influence of Exogenous Variables and Endogenous Variables on all Subsequent Endogenous Variables: Gammas, Betas for Direct, Indirect and Total Effects
Source of Influence
Endogenous
Variables
EffectAgeEducationPerceived
Credibility of Internet News
Internet Dependency RelationsMultiple R2
Perceived Credibility of Internet NewsDirect—-—-.01
Internet Dependency RelationsDirect-.46**-.39**.36
Internet Current Affairs News UseDirect—-—-.15*.50**.22
Indirect-.24**-.19**
Total—-—-

Overall, the basic structural equation model explained 22 percent of Internet current affairs news use variance suggests that it is a reasonable approach to Internet current affairs news use from demographics, perceived credibility and Internet dependency relations perspectives.

Discussion

Consistent with the findings from other researchers,61 perceived credibility of Internet news and Internet dependency relations exert positive influence on Internet current affairs news use. Demographic variables performed very poorly in terms of direct influence on Internet current affairs news use but the indirect impact of demographics on Internet current affairs news use via perceived credibility of Internet news and Internet dependency relation is negative. Moreover, a strong correlation between two mediating variables (perceived credibility of Internet news and Internet dependency relations) was one of this study�s more important findings.

The main objective of the study was to explain the influence on Internet current affairs news use of perceived credibility of Internet news, Internet dependency relations and demographic variables from a holistic approach integrating direct and indirect effects to present an overview of Internet news behavior. While research has extensively addressed the influence of the three perspectives on media use individually, seldom has the research incorporated various influences in one framework. Using structural equations modeling technique, our study furnishes some support for the adequacy of the integrated framework. Internet current affairs news users tend to be those who not only have a stronger dependency relation with the Internet but also rate the Internet news as more credible. Demographic variables, however, have no direct play in Internet current affairs news consumption. The speculation on the relationship between two mediating variables of perceived credibility of Internet news and Internet dependency relation leads to favor the argument that the more credible Internet news is perceived to be, the closer the dependency relations are between the individuals and the Internet and vice versa. Unfortunately, the correlation between the two has long been ignored as a significant research area. This study reveals that besides the respective contributions of perceived credibility and Internet dependency relations to media use, the correlation between them also functions significantly in the complicated network of media use.

The study confirms many of the hypotheses proposed in the model. Perceived credibility and Internet dependency relations both facilitate Internet current affairs news use. The findings are consistent with previous studies identifying perceived credibility62 and media dependency relations63 as good predictors of media use. However, the study fails to attest the importance of age and education in influencing Internet news use directly. Internet current affairs news use, a barometer of how people utilize the new technology to monitor U.S. domestic and world affairs, concentrates in groups featured by being more trusting of the Internet as a news tool and being more dependent on the Internet for various goals. If we attempt to boost exposure to Internet current affairs news use simply by focusing on the demographic classification of users, we will miss the target completely.

The secondary concern of the study is the indirect impact of demographics on Internet current affairs news. It is found that the indirect impact of age and education is significantly negative, suggesting (1) younger people get more exposure to Internet current affairs news by securing a stronger dependency relationship with the Internet, and (2) more educated people get less exposure to Internet current affairs news by having a weaker dependency relationship with the Internet. The completely mediating nature of age and education via IDR in contribution to Internet current affairs news use reveals that the operational influence of demographics in regards to Internet current affairs news use has become more subtle and complicated. Internet dependency relations is the key mediating variable to infuse demographic influence. While the study conforms to most industry surveys and academic findings that younger people heavily depend on the Internet to fulfill various goals and therefore access more Internet news, it found, opposite to traditional wisdom, that the more educated people depend less on the Internet to fulfill life goals and therefore access less Internet news. It is possible that for this highly educated university sample, the educated people at the highest end (50 percent have master and above degrees) hold a more suspicious attitude to the Internet and thus depend less on the Internet for various personal goals accomplishment, which accounts for less frequent access to Internet current affairs news.

Applying the integrative framework of attitudinal, relational and demographic variables in the online news behavior investigation, the study found that the approach to the Internet news behavior has to be multilevel and take both direct and indirect effect into account. Our understanding of the Internet current affairs news use can be furthered and widened by this comprehensive and integrative approach. However, the study had several limitations. The small sample (response rate of 16.3 percent) derived from a university population could make the factor loadings and parameter estimates unstable. Subjective self-reports on the email survey may not reflect the real situation of the respondents. The non-normal distribution of some variables may make the findings biased and hard to generalize. Therefore the findings might be more conclusive if we had a larger sample with a higher response rate. However, for an email survey, this response rate is well above average, and considering the exploratory nature of the investigation, the study may be an acceptable one. In addition, the primary purpose is to test the validity of the model and to obtain a holistic view of the integrated perspectives rather than to identify specific empirical values in a general population.

Future studies should re-examine the conceptual validity of the Internet current affairs news use model proposed here and apply the model to other types of news such as local news or presidential election news. More importantly, future research ought to expand the integrative framework and construct more complex models of current affairs news exposure including other variables, such as political interest, personality, political knowledge, and attitude toward civic duty. Besides, other consequential variables of online current affairs news use, such as political participation, voting, satisfaction from news use, political knowledge gained from the use, may be examined in an attempt at comprehensive model building.. Although structural equation modeling explores causal relationships rather than mere empirical association among variables, readers should be cautious in interpreting the implications of the causal effects. These causal effects should not be treated the same as the end products of the experimental design manipulation but rather as a timely association among variables, a necessary but not efficient condition for establishing cause and effect relation.

Finally, this study used frequency of visits to current affairs news web sites as the endogenous variable. Future studies may consider using the amount of time spent on Internet current affairs news web sites to promote the level of measurement from nominal to ratio.

This study�s contribution lies primary in establishing, and testing, a conceptual framework within which to investigate the growing phenomenon of online news use. Research to date has examined the influence of perceived credibility, Internet dependency relations, age and education on Internet current affair news independently. By integrating these separate approaches into one model, more of the behavior of Internet current affairs news use is revealed. Online news consumption is a news behavior restrained by attitudinal, relational and demographical factors all at the same time. Simply focusing on one aspect will only provide a partial explanation, maybe a biased explanation of the news phenomenon. To reach the online news audience by zooming in on the traditional demographic variables will definitely miss the target, and it is certainly a challenging task for webmasters, news organizations, online advertisers, journalistic professionals to comprehend more in-depth features of their audience. This study suggests that audience�s relational experience with the Internet and audience�s attitude toward using the new medium as a news and information tool are the keys to understand Internet current affairs news use. Moreover, demographics function indirectly via IDR and perceived credibility and their contribution to the Internet current affairs news is mediated by IDR and perceived credibility. This multi-dimensional approach to Internet current affairs news is a first step in extending our understanding of variables that impact online news consumption through an integrative approach, which not only furthers our knowledge of online news consumption and but also provides a close-up view of what dictates that consumption.

Jin Yang is an assistant professor of journalism at the University of Memphis. Padmini Patwardhan is an assistnat professor of communication at Winthrop University. An earlier verson of this paper was presented at the 2003 AEJMC convention.

Leave a comment

Your email address will not be published. Required fields are marked *