The Role of News Media Use and Demographic Characteristics in the Prediction of Information Overload
International Journal of Communication 8 (2014), 699–714 1932–8036/20140005
Copyright © 2014 (Qihao Ji, Louisa Ha, & Ulla Sypher). Licensed under the Creative Commons Attribution Non-commercial No Derivatives (by-nc-nd). Available at http://ijoc.org.
QIHAO JI
Qihao Ji: qji@fsu.edu Louisa Ha: louisah@bgsu.edu Ulla Sypher: Ulla.sypher@cci.fsu.edu Date submitted: 2013-08-06
Florida State University, USA
LOUISA HA Bowling Green State University, USA
ULLA SYPHER Florida State University, USA
Drawing on the information overload theory, this study investigates how news media use relates to the probability of information overload. Our broad goal is to explore how typical media use outside the working environment impacts information overload. Through a large mail survey conducted in northwest Ohio (N = 661), the study combined resident samples and college student samples and examined several variables regarding demographic characteristics, news media use, and information searching efficiency. Multinomial logistic regression was used to analyze the data. Results confirmed that age, gender, income, traditional news media use, and information searching efficiency can partially predict one’s probability of experiencing information overload. Theoretical explanations for these outcomes are presented, and implications for information overload research are discussed.
Keywords: information overload, news use, survey, digital literacy, information processing
About 25 years ago, sociologist Orrin Klapp (1986) wrote in Overload and Boredom, “This is a high-input society” (p. 2). The claim still remains true. Thanks to the development of electronic information communication technology, people today are exposed to thousands of pieces of information. The Internet in particular changed the way people create and exchange information. Consumers benefit from the abundance of information, but they may also suffer from too much information when they lack the capacity to process the wealth of information. 700 Qihao Ji, Louisa Ha, & Ulla Sypher International Journal of Communication 8(2014)
While information overload (IO) has been a prominent phenomenon attracting tremendous amount of attention ever since the beginnings of the printing industry, comprehensive academic interest in this phenomenon was not ignited until IO was proved to exist by Jacoby, Speller, and Berning’s (1974) pioneering experiment, in which subjects were presented with various brands and asked to rate them and choose the best one. Results showed that many subjects could not successfully pick out the brand they rated highest when too many brands were presented, indicating that people tend to make poorer decisions when they are presented with increasing amounts of information. A study by Eppler and Mengis (2004) surveyed nearly all academic studies on IO and found that the issue has been investigated by scholars from disciplines including sociology, psychology, marketing, business management, and accounting.
Against this background, it is rather surprising that among those myriad studies on IO, few approached it from the communication perspective. Hargittai, Neuman, and Curry (2012) observed that the relationship between typical media consumption and IO is rarely examined outside the workplace. Hence, researchers are not able to answer questions about whether and how different types of media use might impact IO at home. To address this issue, Hargittai and her colleagues conducted qualitative interviews on people’s perception of news use on TV and the Internet. They found that IO is not as pervasive as they expected in the American home. The researchers noted that their methodology limited the results’ generalizability, and they called for survey research to further examine this issue. Indeed, other qualitative studies have suggested news consumption actually can lead to IO in some cases (Aldoory & Van Dyke, 2006). Therefore, the primary purpose of the current project was to continue the investigation of news use and IO in both traditional and new media outlets.
Additionally, although a handful of studies on IO have revealed that many personal characteristics and demographic variables are associated with IO, a systematic examination of all demographic characteristics and IO is absent. We intend to fill this gap with a study focusing on individual characteristics.
Another important path of inquiry involves the ways in which people cope with IO. The topic of coping strategy represents a substantial proportion of the existing IO literature (Eppler & Mengis, 2004). It has been stated that people are not passive when confronting IO; rather, they became more selective and efficient in their information searching behavior (Janssen & Poot, 2006; Reuters, 1998; Schultze & Orlikowski, 2004). In this regard, information searching efficiency seems helpful for alleviating IO. From a slightly different perspective, literature on the notion of digital literacy also noted information searching efficiency as an important construct (Bawden, 2008), and some argued that there is a clear theoretical connection between digital literacy and IO (e.g., Koltay, 2011). Therefore, we intend to test the impact of information searching efficiency with our quantitative empirical evidence.
Theoretical Considerations
The Theory of Information Overload
Despite various definitions that have been suggested for the concept of information overload, the general notion refers to a person receiving more information than s/he can process in a certain period of International Journal of Communication 8 (2014) News Media Use and Information Overload 701
time; “it occurs when [information] supply exceeds the [individual’s processing] capacity” (Eppler & Mengis, 2004, p. 326). The underlying assumption is that a person can only digest a certain amount of information in any given time. In addition, Davis (2011) emphasized that IO happens within a domain (e.g., computer screen or workplace), and it triggers failure of intention, which can lead to lower working efficiency.
Broadly speaking, scholars distinguish between two main types of IO: objective and subjective (Eppler & Mengis, 2004; Malhotra, 1984). Objective IO refers to the characteristics or quality attributes of information (including the amount of information, processing time, level of information complexity, and intensity) that are affiliated with overload. This particular perspective drove many experimental IO studies that primarily were administered by consumer and marketing researchers (Merz & Chen, 2006). In contrast, subjective IO refers to each individual’s personal feeling when confronted with information, such as confusion, cognitive strain, and other similar dysfunctional consequences. Experiments are not appropriate with regard to subjective IO, because time constraints in laboratory settings do not always happen in real-life situations (Bock, Mahmood, Sharma, & Kang, 2010). Research often adopted surveys and qualitative interviews to capture the complexity of perception (Bakker, 2007; Bock et al., 2010; Hargittai et al., 2012).
In the context of news media use, the subjective definition of IO appears to be more appropriate. This is because news use (as well as other typical media use) is not identical to information use in the working environment, which requires information to be processed in a very limited time span. Rather, news media use is consumed on an as-needed basis. Therefore, we operationalized IO as a perception of feeling overwhelmed by information. Accordingly, we chose survey over experiment to compensate for the shortcomings of the qualitative method in IO research.
Although the notion of IO is well established in the working environment (e.g., Schultze & Vandenbosch, 1998), it is important to understand the effect of IO generally. From a sociopsychological perspective, Klapp (1986) proposed the key concept of boredom as a result of IO. He argued that too much information can be noise-like, because it is hard to extract meaning or interest from the flood of information. Individuals, then, would feel cluttered, dissatisfied, lonely, isolated, tedious, or distracted, all of which cause boredom. In other words, IO affects people’s psychological well-being. In that sense, understanding whether and how typical media use (news media use, in this case) influences IO is valuable in terms of guiding people’s daily media consumption.
IO undoubtedly is a complicated perception that can be caused and influenced by various factors. According to Jackson and Farzaneh’s (2012) conceptual IO model, IO is generally associated with two groups of factors. Intrinsic factors refer to the fundamental elements of IO, including the amount of information, information-processing capacity, and available time. Extraneous factors have an indirect impact on IO but a direct impact on intrinsic factors. These variables typically involve characteristics of information, quality of information, task and process parameters as well as personal factors. Clearly, although both intrinsic and extraneous factors are causes of IO, intrinsic factors influence IO more directly. Moreover, Jackson and Farzaneh (2012) argued that “personal factors directly affect the information processing capacity, [and] these [personal] factors cannot be entirely distinguished from 702 Qihao Ji, Louisa Ha, & Ulla Sypher International Journal of Communication 8(2014)
information processing capacity” (p. 527). Thus, quantity of information, personal factors, and time are the three key elements of IO. Due to these theoretical considerations, we decided to focus on the role of the first two elements in the news use context: quantity of information (news media use) and personal factors (demographic variables and information searching efficiency).
Demographic Factors and Information Overload
Several studies have provided empirical evidence regarding IO and individual differences. One study regarding radio content and cognitive effort took age into account (Lang, Schwartz, Lee, & Angelini, 2007). It found that participants’ cognitive overload did not occur for radio arousal content. The study also found that teens and college students showed different levels of cognitive effort while exposed to arousing radio messages. In other words, the study established age as a significant demographic variable in influencing people’s information processing. Similarly, in a study commissioned by the Associated Press in 2007 (Nordenson, 2008), researchers examined young adults’ news use patterns around the world and concluded that young generations are suffering from news fatigue—a condition closely related to IO. Another preliminary survey study found that IO among college students is positively correlated with gender (female), age, and education (Williamson & Eaker, 2012). Therefore, our first set of hypotheses is:
H1a: Controlling for gender, young adults are more likely to encounter IO than older adults.
H1b: Controlling for age, women are more likely to encounter IO than men.
H1c: Controlling for age, gender, and education, people with relatively higher household incomes are more likely to perceive IO than those who have lower incomes.
H1d: Controlling for age, gender, and income, people with relatively higher levels of education are less likely to perceive IO than those who have relatively lower levels of education.
News Media Use and Information Overload
News consumption today is far beyond a singular behavior that merely involves newspapers, television, and radio. Channels such as computers, mobile phones, and social media join the “media repertoire” (Yuan, 2011) from which people select news information. Perhaps the term “technostress” (Rosen & Weil, 1997) captures the relationship between increased information channels and IO, because it refers to the possibility that the changing patterns of information and communications technology might lead to a stressful psychological state when people confront countless pieces of information.
However, as we have suggested earlier, empirical studies of typical media use and IO are scarce among the IO literature. Hence, we first turned to some early media effects literature that examined the relationship between different media usage and cognitive overload (usually a substitute term for IO). In a conceptual study of how people process television messages, Lang (2000) proposed a limited capacity model, which suggested that message processors have limited capability (in terms of recognition memory, resource allocation, orienting behavior, reaction time, etc.) when they are exposed to media content. In International Journal of Communication 8 (2014) News Media Use and Information Overload 703
other words, people’s ability to process the messages they receive is limited, because processing mediated message requires users to constantly devote cognitive effort, a finite capacity in everyone. Over decades, many empirical studies have supported Lang’s model. For instance, Lang, Bolls, Potter, and Kawahara (1999) found a curvilinear relationship between camera cuts (whether related or unrelated cuts) and cognitive overload in visual media. Interestingly, in a group working environment, Schultze and Vandenbosch (1998) found that, as information load increases, one’s processing capacity also increases to protect oneself from being overwhelmed, but only up to a certain point. Similar results were observed at the individual level as well (Jacoby, 1984; Schroder, Driver, & Streufert, 1976). Based on these results, we hypothesize that, as news consumption increases, the probability of information overload occurrence (and the intensity of IO) will increase accordingly. However, as the overload reaches a relatively high level, people will refuse to—or will simply be unable to—process more information, which, in turn, decreases the likelihood of encountering IO. Therefore, the relationship between time spent on news media (an indicator of the quantity of information) and the probability of experiencing IO can be shown in a bell-shaped curve (see Figure 1), where mid-level news media use predicts a higher probability of IO occurrence than lower or higher levels of news media use.
Figure 1. The proposed relationship between news use and probability of IO occurrence.
In light of the notion of media repertoire, which emphasizes the coexistence of multimedia usage, and the fact that people rely on multichannel news sources, we divided news media use into two components: use of traditional news media and use of Internet news media. Consequently, we hypothesized that mid-level traditional news media use on traditional media predicts a higher probability of encountering IO than lower or higher levels of news media use. 704 Qihao Ji, Louisa Ha, & Ulla Sypher International Journal of Communication 8(2014)
However, since the effects of information transmitted through computer-mediated communication channels is different from those transmitted through traditional mass media channels (Lang, Borse, Wise, & David, 2002), we suspect that news use on the Internet might not predict the probability of encountering IO in the same pattern as it would for traditional news media.
H2: Controlling for demographic variables and Internet news use, those with mid-level traditional news media use will be more likely to perceive IO than those with lower or higher traditional news media use.
RQ2: How does a person’s Internet news media use predict one’s probability of encountering IO, controlling for demographic variables and for traditional news use?
Information Searching Efficiency and Information Overload
As noted earlier, previous IO literature has established information searching efficiency as a coping strategy among those perceiving IO (Janssen & Poot, 2006; Reuters, 1998; Schultze & Orlikowski, 2004). Studies on digital literacy—a notion that refers to the personal capability of identifying, processing, and creating information through all media platform (Bawden, 2008)—also addressed the importance of information searching capability. Koltay (2011) argued that there is a clear theoretical connection between digital literacy and IO, so the relationship between information searching efficiency and IO can be assumed. Moreover, the IO model also suggested that a person’s information-processing capability plays an active role in decreasing the probability of IO occurrence (Jackson & Farzaneh, 2012). Hence, we hypothesized that:
H3: Controlling for demographic variables and news media use, people with relatively higher information searching efficiency are less likely to perceive IO than those who have relatively lower levels of information searching efficiency.
Method
This study is part of a larger mail survey and parallel online survey conducted in northwest Ohio in 2011. Samples were comprised of two parts: students from a public university in northwest Ohio and residents of northwest Ohio. By using the tailored design method (Dillman, 2007), randomly selected residents of northwest Ohio (N = 1,500) were sent a questionnaire package including a cover letter, a visually attractive questionnaire booklet, a stamped reply envelope, and a fresh one-dollar bill as incentive for participation. The nonrespondents of the first mailing were sent a postcard reminder 3 weeks after the initial contact. We recruited college students from among 32 general education and large introductory lecture classes with various majors and class standings. Students received extra credit for participating in the study online. A total of 661 responses were received, of which 216 were residents of northwest Ohio and 445 were college students.
Respondents were asked about their media use and attitudes. The questionnaire contained two parts: The first part asked about use of various media, such as how much time one spends reading International Journal of Communication 8 (2014) News Media Use and Information Overload 705
newspapers every week, how many years of experience one has using the Internet, what kind of news one prefers, one’s attitude toward different media, and one’s perception of the probability of information overload. The second part included demographic variables.
Measures
Demographic variables: gender, income, education, and age. Respondents were asked to report their age and gender. Respondents selected annual income from among the following categories: (1) under US$30,000, (2) $30,001–$60,000, (3) $60,001–$90,000, (4) $90,001–$150,000, (5) over $150,000. Education was measured by (1) completion of grade 8 or less, (2) completion of grades 9 to 11, (3) high school graduate or equivalent, (4) 1 to 3 years of college or technical school, (5) college graduation (4 years), (6) attended or completed graduate school.
Probability of IO occurrence. Based on the subjective IO definition, respondents were asked whether they had perceived information overload via this question: “Have you ever felt overwhelmed in terms of the information provided by various media?” Possible answers were (1) yes, frequently; (2) sometimes, or (3) not at all. The scale was ordinal in nature, which enabled us to compare different levels of IO to the null IO group. The decision of assessing the probability of IO occurrence with a single-item measure was made based on several considerations. First and foremost, a similar one-item measure of IO has been used in related research using an ordinal scale (Dubosson & Fragniere, 2009). Second, as Misra and Stokols (2012) have pointed out, most studies measure IO with context-specific items. An instrument with proper levels of universality was not available when we administered the study. Third, our primary goal was to find how certain demographic characteristics and news media use related to people’s perception of the probability of encountering IO. In other words, we were interested in the subjective perception of IO rather than the actual IO, which involves many other considerations (Haksever & Fisher, 1996). Finally, because this study is more interested in media-based sources of IO as opposed to place-based sources (Misra & Stokols, 2012), instruments containing place-based items were eliminated.
Traditional and Internet news media use. Traditional news media use was measured by the number of hours per week that people used traditional news media (television, radio, news magazines, and newspapers). Internet news media use was measured by the number of hours per week that people used digital news media (online newspapers, online portals, social media, and mobile phones). It is important to note that time spent on news media is in fact a better measure of the quantity of news information in the survey setting because: (1) Unlike information use in the workplace, news consumption is more active, which means that time spent on news use reflects people’s actual interest in news (Yuan, 2011). (2) Compared to asking how much news information people use every day, time spent on news media use is more easily recalled by respondents.
Information searching efficiency. Because we did not find a scale for information searching efficiency, we asked respondents to what extent they believed they had devoted too much time to searching for valuable information. Possible answers for this item were: (1) frequently, (2) sometimes, (3) never. We developed this item based on theoretical discussion about the concept from both the IO and 706 Qihao Ji, Louisa Ha, & Ulla Sypher International Journal of Communication 8(2014)
digital literacy literature (e.g., Bawden, 2008; Janssen & Poot, 2006). Reverse coding was applied for this variable.
Data Analysis and Findings
Respondents’ Profiles and Perceived IO
Data analysis was based on the merged samples of residents and students to create a sample with more age variance, because the data from the resident sample turned out to contain mostly older adults and seniors (Mage = 55.6). Among the 661 college and resident respondents, 59.5% were women. The average age was 30.79 after the two samples were merged. A total of 354 participants (51.4%) had attended 1 to 3 years of college or technical school, 172 (27.2%) had completed high school or a lower level of education, and 135 (21.3%) had a 4-year college degree or higher. For household income, nearly half of the respondents (48.9%) reported income of less than $30,000 per year, 111 (16.8%) respondents’ household income was between $30,001 and $60,000 annually, 15.2% respondents reported income between $60,001 and $90,000, and 106 (17.5%) respondents had an income higher than $90,000.
In regard to respondents’ perceived IO, 141 (24.5%) respondents said they had never experienced IO; 434 (75.5%) either said yes (31.3%) or sometimes (44.2%).
We used multinomial logistic regression (MLR) to test the hypotheses. Since our dependent variable is categorical in nature, this data analysis technique is appropriate. MLR divides the regression into several binary regressions and compares them separately to a baseline group (Long, 1997). It allows the predicted variable to have more than two levels of response. Also, it is possible to consider multiple predictor variables simultaneously, just as in other regression analyses. The advantage of this technique, as Visser (2004) suggests, is that “it does not require the assumptions associated with many other tests (such as normality and homogeneity of variance) to be met” (p. 62). However, the technique assumes the existence of well-populated tables, an adequate sample size, the absence of significant outliers, and independence of observations. To conduct the regression, each predictor measure was recoded into three levels based on percentile values. Specifically, the data were categorized into education (high school, 3 years college, 4 years college or higher); total news media use (1–8 hours, 9–16 hours, 17 hours or more); Internet news media use (1–4 hours, 5–10 hours, 11 hours or more); traditional media use (0–2 hours, 3–6 hours, 7 hours or more); and Internet experience (0–8 years, 9–11 years, 12 years or more). Respectively, respondents in each of the three groups are considered “lower level,” “mid-level,” and “higher level” on the measures. Considering the young age of college students, we divided age into two categories: 25 and younger, and 26 and older.
For all hypotheses tests, the reference category for the dependent variable was “no, never felt overwhelmed about information.” Each hypothesis was tested twice by a two-level predicted variable IO (yes, no) and a three-level predicted variable IO (frequently, sometimes, never). Two-level MLR tests the existence of relationships, and three-level MLR determines the extent to which the variables predict consistent future intentions. For each multinomial regression, odds ratios (ORs), p values, and confidence International Journal of Communication 8 (2014) News Media Use and Information Overload 707
intervals (CIs) are reported. As a measure of effect size, ORs reveal to what extent a trait in one group is higher or lower than the baseline group.
Hypotheses Tests and Results
H1a considered age and IO. The overall models are statistically significant (two-level comparison: log likelihood = 21.642, df = 2, p < .001; three-level comparison: log likelihood = 41.867, df = 4, p < .05). Controlling for gender, young adults are 1.4 times (p < .05, 95% CI, OR = 0.9–2.1) more likely than older adults to experience IO. Young adults are 1.5 times (p < .05, 95% CI, OR = 1–2.3) more likely than older adults to experience IO sometimes. Therefore, H1a was partially supported.
H1b considered gender and IO. The overall models are the same as H1a. Controlling for age, men are 0.6 times (p < .01, 95% CI, OR = 0.4–0.9) less likely to experience IO than women. Men are 0.5 times (p < .01, 95% CI, OR = 0.4–0.8) less likely than women to experience IO sometimes. Therefore, H1b was partially supported.
H1c predicted that people with lower household income would have lower perceived IO. The overall models are statistically significant (two-level comparison: log likelihood = 272.308, df =14, p < .01; three-level comparison: log likelihood 141.043, df =7, p < .01). Controlling for age, gender, and education, people with a household income of $30,001 to $60,000 are 0.4 times (p < .01, 95% CI, OR = 0.2–0.8) less likely to perceive IO than people who have household incomes of more than $90,000. People who have household incomes of $60,001 to $90,000 are 0.5 times (p < .05, 95% CI, OR = 0.3–1.1) less likely to experience IO than people who have household incomes over $90,000. People who have household incomes of $60,001 to $90,000 are 0.4 times (p < .01, 95% CI, OR = 0.2–1) less likely to experience IO sometimes than people who have household incomes of more than $90,000. Similarly, people who have household incomes of $30,001 to $60,000 are 0.4 times (p < .01, 95% CI, OR = 0.2–0.8) less likely to sometimes experience IO than people who have household incomes of over $90,000. Therefore, H1c was partially supported.
H1d considered education and IO. The overall models are the same as H1c. However, two-level MLR tests suggest no existing relationship (p > .05). Therefore, H1d was not supported. 708 Qihao Ji, Louisa Ha, & Ulla Sypher International Journal of Communication 8(2014)
Table 1. MLR Analysis: Using Demographic Variables and Traditional News Use to Predict Encountering IO.
| |||||||
Variables
|
Probability of encountering IO
| ||||||
p
|
Exp(B)
|
SE B
|
β
| ||||
Two-level MLR, “yes”
| |||||||
Male
|
< .05
|
.656
|
.239
|
0.421
| |||
Education, 3 years of college
|
< .05
|
.542
|
.359
|
0.613
| |||
Income, $30,001–$60,000
|
< .05
|
.406
|
.402
|
0.902
| |||
Traditional news use, 0–2 hours
|
< .05
|
1.756
|
.310
|
0.563
| |||
Traditional news use, 3–6 hours
|
< .05
|
1.695
|
.311
|
0.528
| |||
Three-level MLR, “frequently”
| |||||||
Traditional news use, 3–6 hours
|
< .05
|
1.810
|
.363
|
0.593
| |||
Three-level MLR, “sometimes”
| |||||||
Male
|
< .001
|
.530
|
.262
|
0.635
| |||
Education, lower than high school
|
< .05
|
.464
|
.419
|
0.767
| |||
Education, 3 years of college
|
< .05
|
.469
|
.379
|
0.757
| |||
Income, $30,001–$60,000
|
< .001
|
.359
|
.427
|
1.024
| |||
Income, $60,000–$90,000
|
< .05
|
.393
|
.453
|
0.934
| |||
Traditional news use, 0–2 hours
|
< .05
|
1.838
|
.337
|
0.609
| |||
Note. All nonsignificant results are omitted for simplicity. Tables displaying full results are available from the first author upon request.
| |||||||
No comments:
Post a Comment