Abstract
Explosive growth in the number of options prompts media researchers to consider how selection behavior changes under higher choice conditions. Two experiments demonstrate that choice environments offering options in smaller sets lead users to be more likely to select news content, in particular “hard news” content. A third study incorporates theories of information processing to explain the observed effects of choice environment. The study provides evidence that smaller sets of options lead users to compare the merits of each option, whereas larger sets of options prompt users to quickly scan the environment for an acceptable option.
Introduction
Over the past two decades, most American citizens have gained access to a vast resource of knowledge about current events via the Internet. Although one might expect media users with access to more information to have greater knowledge of their world, evidence from experimental and survey analyses (Prior, 2007) shows otherwise. According to these studies, newly available options siphon off media users with minimal interest in news who were incidentally exposed to news under the lower choice conditions of the 20th-century media environment. Although these studies offer some insight into how media selection behavior differs between lower choice and higher choice conditions, they do not fully explore the differences in types of selections made by users under differing conditions nor do they offer insight into the cognitive processes implicated in media selections under higher choice conditions.
The present research presents the results of three experiments designed to address these shortcomings. The first study determines whether lower choice environments foster greater news consumption even when the ratio of news options to other options is held equal across conditions. The second study shows that preference for so-called “hard news” stories (e.g., news about international politics) over “soft news” (e.g., entertainment news stories) is affected by the number of options made available. The third study shows evidence of a difference in cognitive processing under lower choice and higher choice conditions. The research concludes by examining selection behavior under lower choice and higher choice conditions to better explain the dynamics of news selection behavior and choice environment.
Theories of Media Selection in Lower Choice and Higher Choice Environments
The field of media choice research enjoyed a period of significant growth over the past three decades, yielding several important insights about selective exposure to media (for a review, see Hartmann, 2009). Starting with the basic premise that individual preferences guide media selection, several studies from this period (e.g., Marewski, Galesic, & Gigerenzer, 2009; Zillmann & Bryant, 1985) acknowledge that media selection is often made under “sub-optimal” conditions, that is, conditions in which individuals cannot assess the subjective values of all available options and select the best one. Other studies make the case that most media selections are, in some sense, habitual and involve automatic (as opposed to controlled) mental processes (e.g., LaRose, 2010). Still others assess the interplay between structural characteristics of the media choice environment and users’ attributes and how this interaction affects selection behavior (e.g., Dennis & Taylor, 2006).
More broadly, research on decision making (Biehal & Chakravarti, 1983; Iyengar & Lepper, 2000; Iyengar, Wells, & Schwartz, 2006) supports the notion that the structure of the choice environment can contribute to a kind of unconscious biasing of selection behavior. Decisions are the product of interplays between the choice environment and perceptual and inferential cognitive systems that selectively attend to and interpret certain aspects of the options in particular ways. In instances in which a decision is made quickly (a category to which most instances of media selection belong), perceptual and inferential “habits of mind” heavily influence our decisions. These cognitive shortcuts, or “heuristics,” evolved over millions of years in response to relatively stable characteristics of the environment and are thus not necessarily likely to produce optimal outcomes in a rapidly changing choice environment (Evans, 2003; Kahneman, 2011). Collectively, this research highlights the importance of understanding selective media exposure not exclusively in terms of access to options or conscious decisions to gratify desires, but also in terms of more subtle effects of characteristics of the choice environment and in terms of unconscious (or less conscious) mental processes.
Within the past decade, the rapid rise in the number of information and entertainment options available to media users has prompted communication scholars to speculate about the effects of abundance on selection behavior. Some express concern over the possibility that increased choice may lead to an expansion of the political knowledge gap between information seekers and entertainment seekers (Prior, 2007). Others worry that it may lead to greater political and ideological polarization via selective exposure (Mutz, 2006; Pariser, 2011). Although many of these studies establish links between new media and new selection habits, they fail to isolate particular characteristics of new media (e.g., the number of options present to users) that are responsible for the changes; thus, claims that abundance, in particular, causes specific changes in media selection behavior are largely speculative. Of the few empirical studies to isolate the effects of the number of options on media selection behavior (e.g., Heeter, 1985; Wise, Bolls, & Schaefer, 2008; Yuan & Webster, 2006), most do not fully explore how higher choice environments affect the types of content chosen by media users.
An exception is Markus Prior’s (2007) studies which attempted to isolate the effects of higher choice on television viewers’ tendencies to select news programs. Using survey experiments, Prior was able to show that when individuals were provided with a menu of options similar to the ones offered television viewers in the United States at 6:00 p.m. on a weeknight during the broadcast television era (consisting of four different news programs) as well as the option not to watch television, they were significantly more likely to watch news than individuals provided with options similar to those offered television viewers during the postbroadcast television era (consisting of the same original four news options plus five other entertainment-related television options). Prior’s design of the lower choice experimental condition was intended to re-create the common broadcast-era television programming practice of “road-blocking”: simultaneous scheduling of one type of programming on all or most channels during a block of time.
Prior (2007) invokes Klein’s (1971) Least Objectionable Program theory as a means of explaining the observed behavior. Klein states that many broadcast-era television viewers did not especially enjoy the programs they selected so much as they preferred the activity of viewing television to other activities. Viewers made do with the least objectionable option, which, at certain times of day, happened to be a news program. Prior’s (2007) experiment and Klein’s theory explain changes in the dynamics of selection behavior that resulted from one particular change in the media choice environment: the expansion of available television channels. However, Prior’s (2007) experimental design does not re-create another important facet of change to the media choice environment: the shift from scheduled, sequentially available television content to simultaneously available online content. The shift from sequential availability, in effect, takes many options that were competing against smaller sets of other options for media users’ attention and forces them to compete with every other option simultaneously (see Table 1).
Three Media Choice Environments.
Another theoretical perspective on selection behavior lends credence to the premise that incidental exposure to news is less likely to occur when options are abundant, even when the practice of roadblocking is not used. Behavioral economic theories of choice suggest that the greater the number of available options, the more likely an individual will be to find an option suiting his or her desires at that moment rather than having to make do with an unappealing option (Baumol & Ide, 1956; Boatwright & Nunes, 2001; Simon, 1987). Researchers examining television choice find evidence that television viewers select programs in a manner consistent with this theory: Viewers provided with more television channels are more likely to view programs of a type that coincide with their stated preferences (Youn, 1994).
This review shows that speculation about the effects of the new media environment on media selection behavior as well as experimental analysis of choice both support the notion that the shift from sequentially available smaller sets of options to simultaneously available larger sets of options results in a decline in exposure to news. Despite the convergence of these theoretical perspectives on the matter, the effect of this particular change in the choice environment has never been isolated from the practice of roadblocking. News selection under roadblocking choice conditions—when television viewers at 6:00 p.m. had no other option than to watch news—only demonstrates that the desire to watch television at 6:00 p.m. is stronger than the viewers’ desire to avoid news content. Whether individuals in a higher choice environment are more likely to select news options from a comparable set of options to those offered in a lower choice environment has yet to be tested. Based on these theories of selection behavior and prior studies of media choice, the following hypothesis is put forth:
Study 1
Method
Design
This study uses a randomized experimental design. The number of available options (higher vs. lower) is the independent variable. The type of content selected by users (news vs. nonnews) is the dependent variable.
Participants
The sample consisted of 144 volunteer participants from an introductory communication studies course at a large Midwestern university. The sample was 55% male, 69% White, and had a mean age of 18.83 years (SD = 0.97).
Materials and measures
Menu options
The menus of media options from which participants chose (see Table 2) were constructed with the intention of maintaining verisimilitude between the experimental setting and the real-world settings in which these media users made selections. To assemble menus that were popular among the sample population, multiple sources were used: 30 guided interviews conducted with undergraduate students regarding their daily media use habits, the Kaiser Family Foundation’s (2011) report on teen and adult media use from 2009 to 2011, and the web traffic tracking site Alexa (http://www.alexa.com/), which tracks the most popular websites among 18- to 24-year-olds living in the United States. The menus of options included episodes of popular television shows, video games, social media websites, as well as software (e.g., Microsoft Word) and websites (e.g., a course-related website) that students use to complete schoolwork. The five news options were websites for the New York Times, CNN, Yahoo! News, Fox News, and The Huffington Post. For the purposes of the present analyses, options were either coded as news options of nonnews options.
Study 1 Media Options.
Number of options
The numbers of available options in higher choice and lower choice experimental conditions vary among studies of choice. Iyengar and Lepper’s (2000) study of food choice uses six options for the lower choice condition and 24 and 30 for the higher choice conditions. Providing 30 options is both feasible and reasonably approximates the size of a personal media repertoire from which individuals in higher choice media environments typically choose (Yuan & Webster, 2006). Given extant research on media choice, it was deemed appropriate to limit the number of choices in the lower choice conditions to six so as to maintain some consistency with the existing research on lower choice versus higher choice selection behavior and to create a condition which is similar to the broadcast-era choice environment. Individuals in the lower choice condition were randomly assigned to see one of five subsets of the 30 options shown in Table 2. Each subset consisted of one news option, one school-related option, and four randomly selected entertainment/social media options.
Procedure
When participants arrived for the study, they were each seated at a computer and asked to fill out a survey. The first question of the computer-based survey informed participants that there would be 25 min of spare time during the study and asked participants to select from a list of websites, television shows, movies, software, and video games with which to spend this spare time. Participants were randomly assigned to select an activity from a list of six options in the low-choice condition and a list of 30 options in the high-choice condition. Condition assignment was double-blinded. The order of all options in all menus was randomized. After making selections, participants answered some basic demographic questions. At the conclusion of the study, participants were debriefed and thanked for their participation.
Results
To test
Discussion
The results of this experiment suggest that the shift from sequentially available smaller sets of options to simultaneously available larger sets of options is directly responsible for a change in media selection behavior. The observed difference between the selection behavior of those selecting from more options and those selecting from fewer options was consistent with the expected pattern: Individuals who chose from a greater number of options selected fewer news options. When individuals had fewer options from which to choose, they appeared to make do with news options.
In this experiment, news is never presented as the only kind of website or television program, so news selection in lower choice conditions cannot be explained using Klein’s (1971) explanation that individuals’ desire to use a particular medium trumps their lack of affinity for the content. This finding is not medium-specific; regardless of whether the options are television content or web content, media users are more apt to select news when their options are restricted. In addition, the findings are not the result of greater availability of news relative to other options. In all cases, participants had one news option for every five other options.
One possible explanation for this finding is that many news options presented in the experiment each contain greater choice with the website than many other options. If users decide to visit The Huffington Post or Buzzfeed websites, they may assume that they can subsequently select from a variety of content about celebrities, sports, or about international trade policy. This is in contrast to content-specific options, such as an episode of the television show Family Guy, that restrict users to a very specific type of content. To further isolate the effect of choice environment on news selection, a follow-up study was conducted in which online news consumers were asked to select from “hard news” headlines and “nonhard news” headlines.
Study 2: Selective Exposure to Hard News
Wilbur Schramm (1949) was among the first communication researchers to suggest that there were two types of news offering either delayed gratification (“hard news”) or immediate gratification (“soft news”). Subsequently, both journalists (see Tuchman, 1973) and scholars (e.g., Baum, 2002) have made distinctions between these two types of news which can be identified by their topics or the extent to which they refer to a larger context or social issue (for a more extensive review, see Reinemann, Stanyer, Scherr, & Legnante, 2012). Although television news programs and magazines tended to specialize in one type of news or the other, popular websites such as Yahoo! News, Buzzfeed, and The Huffington Post present users with a mix of entertainment and hard news. Given this characteristic of news websites, it is appropriate to designate headlines or links, rather than websites, as the unit of analysis for Study 2.
This study is primarily concerned with selective exposure to the kind of news stories found by Prior (2003, 2007) to be associated with gains in political knowledge, that is, news stories clearly relating to a civic, economic, or political topic. Therefore, “hard news” is operationalized based on topic. As in Study 1, the larger menu of options in higher choice environment presents users with a greater array of alternatives which are, collectively, more likely to include options that are better fitted to users’ individual preferences. Users need not make do with a hard news story in the higher choice environment, but can instead find a less objectionable alternative. Building on the results of Study 1, this study seeks to find evidence of a difference in news selection habits due to the shift from sequentially available smaller sets of options to simultaneously available larger sets of options. The following hypothesis is proposed:
Method
Participants
Although Study 1 presented users with several options that could not be accessed online (e.g., the console video game Batman: Arkham City), Study 2 seeks to re-create the experience of visiting a news website. For purposes of external validity, it was deemed appropriate to recruit participants online using Amazon’s Mechanical Turk website. 2 Participants received monetary compensation (50 cents) for completing the online survey. Two-hundred sixty participants completed the survey. The sample was 55% male, 82% White, and had an average age of 36.16 years (SD = 14.01).
Materials
Menus of articles from which participants chose (see Figure 1) were comprised of headlines drawn from the Yahoo! News website (http://news.yahoo.com) during the time of the study. Headlines were accompanied by small “thumbnail” images related to the stories. Some of the headlines were drawn from the website’s national, international, and business sections (e.g., Associated Foreign Press, 2012); these were categorized for the purpose of this study as hard news stories. The other headlines were drawn from the website’s entertainment or sports sections (e.g., Access Hollywood, 2012); these were categorized as nonhard news stories. The higher choice condition consisted of 30 headlines, whereas the lower choice condition consisted of six. Headlines in the lower choice condition were a randomly selected subset of the headlines in the higher choice condition. Thus, the survey question in the lower choice condition was no more likely to have a greater or fewer number of hard news headlines than the question in the higher choice condition.
Study 2 menu.
Procedure
An online survey was made available to users via the Mechanical Turk website. Participants were randomly assigned to receive either the menu of 30 options or the menu of six options. The first question of the survey asked participants to choose a story to read. Participants then answered some basic demographic questions, and were thanked for their participation.
Analyses and Results
To test
Discussion
Results of this study provide additional evidence that individuals selecting from a greater number of options are less apt to select hard news options than users selecting from fewer options even when the type of options (news headlines) is held constant across conditions. The design of Study 2 effectively rules out the possibility that users’ selections reflect a desire for more choice; each option represents a single piece of media content. This finding is especially relevant to researchers seeking to understand the news consumption habits of Internet users. Knowing what characteristics of websites encourage users to click on hard news headlines is an essential part of understanding the causes of the gap between media users with more political knowledge and those with less.
Although Study 1 and Study 2 offer compelling evidence that higher choice media environments result in less news consumption and less hard news consumption, respectively, neither study offers evidence as to what specific psychological mechanism is responsible for the observed between-group differences. To shed light on this aspect of news selection, a third study was conducted.
Study 3: How Users Choose
Theories of information processing start with the assumption that individuals have limited cognitive resources to utilize when selecting and processing information (Lang, 2000). By observing how media users’ allocate this limited resource under various conditions, we gain a better understanding of the processes involved in media selection. Within the context of news headline selection, it has been observed that additional news headlines lead to a physiological response—increased heart rate—that is indicative of greater allocation of cognitive resources (Wise et al., 2008). However, this particular study presented users only with unpleasant or upsetting news. In this affect-specific context, the authors assert that each option represented potential threats to users, invoking a particular motivational response that may not occur if the news headlines were more varied in character. To understand media users’ efficient allocation of limited cognitive resources in a more general information choice environment, I turn to dual-processing models of cognition.
Dual-process models consist of more conscious, controlled, slower processes and less conscious, automatic, faster processes (Evans, 2003; James, 1890). These models serve as the foundation for theories of information processing as well as attempts to identify the strategies individuals use to search their environments for relevant or gratifying information. Sternberg (1966) posited a careful, exhaustive approach to search that consists of serial comparison of options as distinct from a quick scanning of the environment in which the search stops when a match to an ideal type is found. Sternberg’s theory, like other well-established theories of information processing (e.g., Petty & Cacioppo, 1986), suggests that the implementation of one or the other type of cognition or search strategy depends in part on contextual factors.
Subsequent research brought these theories of cognition and search strategies to bear on the response of individuals to environments with abundant information, establishing links among types of cognition, search strategies, and the characteristics of the information environment in which they are likely to be enacted. Schneider and Shiffrin (1977) found that when searching within a small set of options, individuals use the former strategy and type of processing (“compare” strategy/controlled processing); when the environment contains an amount of information greater than the brain’s limited capacity (i.e., “information overload”), individuals use the latter strategy and type of processing (“scan” strategy/automatic processing).
Schneider and Shiffrin (1977) provided evidence of the link between search environment (specifically, the number of options in the search environment) and information processing by assessing the effect of additional items to a search environment on deliberation time (i.e., the amount of time needed to make a selection). They found that when an individual is engaged in controlled processing of information, the amount of time that one spends deliberating among the options is sensitive to the addition of other options. Specifically, when additional options are added to the environment of someone engaged in controlled processing, the chooser uses a more efficient form of selection behavior; they spend significantly less deliberation time per option. 4 This is not the case for an individual engaged in automatic processing. When an additional option is added to the environment of someone engaged in automatic processing, the deliberation time per option does not change; the selection process is already being performed in as efficient a manner as possible.
These theories can be used to help explain the effects observed in Studies 1 and 2. Under lower choice conditions, media users may be engaged in controlled processing, comparing each option with the other, considering the qualities of each and ultimately deciding that news content, specifically hard news stories, is worth their time. Only when the user is not overburdened with options can he or she consider these qualities of an option and make a judgment on their value relative to those of the other available options. Conversely, media users under higher choice conditions may not be engaging in careful comparisons and instead may be scanning the environment for an appealing option. Evidence of such a difference would manifest itself in self-reported information selection strategies as well as the extent to which the addition of options to the environment prompts users to become more efficient in their selection behavior. Based on these understandings of search strategies and information processing under higher choice and lower choice conditions, the following hypotheses are made:
Method
Design
To test the extent to which additional options influence the time spent making a selection, four experimental conditions were created. The two lower choice conditions presented participants with either four or six options randomly selected from a larger set of 30 headlines. The two higher choice conditions presented participants with either 28 options randomly selected from the complete set or the complete set of all 30 options. This permitted a comparison between the deliberation differential in the lower choice conditions (average deliberation time per option in the six-choice condition minus average deliberation time per option in the four-choice condition) and the deliberation differential in the higher choice conditions (average deliberation time per option in the 30-choice condition minus average deliberation time per option in the 28-choice condition).
Participants
As in Study 2, participants were recruited via Amazon Mechanical Turk. Each participant was compensated with US$1 for their time. The sample consisted of 792 participants and was 57% male, 84% White, and had an average age of 33.58 years (SD = 11.24).
Measures
Search strategy
To determine whether they used a comparison search strategy or a scanning search strategy, participants were asked to rate the extent to which the following statements described how they made their choices: “I scanned through the options looking for one to choose”; “I compared the options and chose one” (1 = doesn’t describe it at all; 5 = describes it perfectly).
Deliberation time per option
“Deliberation time per option” was assessed by tracking the amount of time (to the hundredth of a second) participants spent on the webpage before clicking on a headline. This number (the total deliberation time) was then divided by the number of options available to the participants, yielding the deliberation time per option (e.g., a total deliberation time of 10 s for a participant in the four-option condition yield a deliberation time per option of 2.5).
Procedure
Participants were randomly assigned to see four, six, 28, or 30 headlines. After clicking on a news headline of their choice, participants were asked how they made their decision as well as a series of demographic questions. They were then thanked for their participation.
Analysis and Results
To compare the self-reported search strategies of those selecting in higher choice or lower choice conditions, participants in the lower choice conditions (four options and six options) were aggregated as were participants in the higher choice conditions (28 options and 30 options). The self-reported search strategies of these two groups were then compared using an independent-sample t test.
The difference between deliberation differentials in the lower choice and higher choice conditions exhibited the pattern predicted by
Deliberation time per option in 4-, 6-, 28-, and 30-option conditions.
Discussion
Study 3 provides evidence that different search strategies and different cognitive processes are implemented in lower choice and higher choice media environments. Self-report accounts of users’ search strategies indicate that users in lower choice conditions are more likely to use the comparison search strategy, whereas those in the higher choice conditions are more likely to use the scanning search strategy. Although this particular measure of search strategy possesses the limitations of all self-report measures (namely, that users may not be fully aware of how they make a selection), the deliberation differentials corroborate the differences in strategies between those in lower choice and those in higher choice conditions. The time-insensitivity of selections observed in the higher choice conditions is the signature of the scanning search strategy, whereas the time-sensitivity of selections observed under lower choice conditions is the marker of the comparison search strategy.
These differences may at least be partially responsible for the differences in selection behavior observed in Studies 1 and 2. Previous research on decision making (Khan, Dhar, & Wertenbroch, 2005; Shiv & Fedorikhin, 1999) finds that automatic processing is associated with the selection of hedonic options (e.g., social networking sites, entertainment news) while controlled processing is associated with the selection of utilitarian options (e.g., international news). According to this line of research, direct comparison of options with one another prompts individuals to take long-term consequences into account when deciding. Media users may not see news stories about conflict in the Middle East as immediately gratifying, but they may see the long-term utility of being a well-informed citizen. Findings from Study 3 indicate that changes to a choice environment can invoke the kinds of information search strategies and cognitive processing associated with the consideration of the long-term consequences of one’s choices, facilitating exposure to hard news.
General Discussion
These three studies provide a clearer picture than previously existed of how the processes and outcomes of media selection change as a result of changes to the choice environment. Specifically, the present research sought to explore the effect of the shift from lower choice sequential availability to a media choice environment in which all options must simultaneously compete with one another. The findings indicate that the likelihood of the selection of news options, particularly hard news options, is greater in the more constricted choice environment.
Although the difference in self-reported search strategies and information processing observed in Study 3 offer one plausible explanation of the observed differences, further considerations of the differences between the lower choice and higher choice experimental conditions reveal another possible contributing factor.
Exploring Differences in Selection Behavior in Lower Choice and Higher Choice Environments
The rich data set yielded from these three studies offers opportunities for post hoc analyses of choice patterns as they differ by condition. A cursory examination of the distribution of selections across the 30 options available as they vary among choice conditions in Study 3 (see Figure 3) reveals an intriguing pattern. Under higher choice conditions, fewer options account for a greater overall share of the users’ selections. A comparison of the variance in selections in higher choice conditions (SD = 2.58; variance = 6.67) and the variance of selections in lower choice conditions (SD = 1.59; variance = 2.55) provides further evidence of this difference in selection behavior, as does a breakdown of the percentage of total selections accounted for by the most popular choices in higher choice or lower choice conditions (see Table 3).
Distribution of selections in four-, six-, 28-, and 30-option conditions.
Percentage of Selections Accounted for by Top 10, Top 5, and Top 2 Choices in Higher Choice and Lower Choice Conditions in Study 3.
The design of the studies and characteristics of the media choice environments which the studies seek to re-create evidently contribute to the likelihood of clustering of users on several highly popular options. Lower choice environments, in effect, hold back especially appealing options, giving them access to a smaller share of potential users than would be the case under higher choice conditions. These circumstances also allow less appealing options to flourish, benefiting from the absence of a more appealing competition. As in Klein’s conceptualization of the broadcast television choice environment, “least objectionable options” flourish in lower choice media environments.
This pattern of selection behavior has direct implications on the overall diversity of selections made by media users. When users are asked to select from smaller sets of options, they collectively select a more diverse set of options, resulting in a “fatter-tailed” distribution of selections (see Figure 3). Although the virtually unlimited bandwidth of the Internet offers users more choice, the fact that all options must compete with one another at all times appears to lead to less diversity of selection behavior and less exposure to options that would, under lower choice conditions, be considered the least objectionable. Evidence from these studies suggests that the clustering seen at the top of the media selection distribution curve that many lament (e.g., Hindman, 2009) is not solely a consequence of the larger budgets of dominant players (as previously theorized by Hindman) nor a result of the “viral” promotion and distribution of popular options by other users, but is an inevitable outcome of an environment in which all options compete with one another simultaneously.
Limitations
The findings presented here must be accompanied by several caveats. First, the design used in all three studies treats users of lower choice option sets as entirely discrete, which does not necessarily reflect the reality of use of sequentially scheduled media offerings. There is no doubt that some portion of media users base their selections on whatever happens to be available when they have free time, but there are also users who would wait for a particularly desirable piece of content scheduled at a later time. Participants in the lower choice conditions did not have the option of waiting for a more desirable option at a later time or on a subsequent webpage; they were forced to choose among a limited number of options. In this sense, fidelity to the broadcast television environment in the design of the experiments was sacrificed to isolate the effects of sequential-small-set environments and simultaneous-larger-set choice environments on selection behavior. This does not, however, prevent the findings from being used to design future websites and software applications that facilitate exposure to a broader range of media content options including hard news content.
In addition, the studies did not account for user preference for news or hard news stories. Prior (2007) has shown this to be a robust predictor of news consumption in higher choice media environments. Although accounting for preference may help explain some of the variance in users’ selections, it is unlikely to explain the effects observed in the studies. Participants were randomly assigned to conditions in all three studies; therefore, a user with a strong preference for news was no more likely to in the higher choice experimental condition as they were in the lower choice experimental condition.
Similarly, the studies did not account for particular characteristics of the options other than whether or not the options were news or hard news. Hard news options may have been less likely to be selected under higher choice conditions because they lacked some arousing characteristic, because they were less relevant to the user, or because they evoked less pleasure. Although behavioral economic theories of cognition and choice lead us to conclude that it is the lack of hedonic value of hard news options that accounts for their failure in higher choice media environments, the precise characteristic that is to blame for their lack of popularity in higher choice environments was not explored.
The measure of information selection strategy used in Study 3 relied on participants’ abilities to reflect on their thinking and report on it accurately. Such assumptions may not be warranted given the fact that such behavior is often automatic and always quick. Although the self-report measures of search strategy possessed face validity, they were not significantly negatively correlated with one another (r = −.06; p = .11). This could indicate that such strategies are not mutually exclusive, that users can at least perceive themselves to be doing both or neither at the same time, or it could indicate that the questions were not properly understood by the users. Future studies should either refine this self-report measure or replace it with another objective measure of search strategy, such as eye tracking. Finally, it remains unclear as to whether or not (or to what extent) differences in search strategy and cognition explain the differences in selection behavior observed in Studies 1 and 2. Future research should seek to include other measures of search strategies and cognition to further isolate the contribution of these factors to media selection behaviors.
Conclusion
The results of this study provide empirical evidence that a simple change to website design can influence the likelihood that users are exposed to a greater diversity of news stories. Subtle changes in design can be implemented without users perceiving that they were being guided toward selecting a greater diversity of stories, resembling a kind of “nudge” (Thaler & Sunstein, 2003) rather than an overt attempt to get users to click on certain links. If websites or users were to implement changes such as making fewer options available on a single webpage, the effects on users’ exposure to certain kinds of news stories as well as their levels of political knowledge and civic engagement over time (which are known to be associated with consumption of news about civic and political topics) could be profound.
Popular and academic discourses on the topic of selection behavior in higher choice media environments tend take either the optimistic view that increased competition is breaking up the previously unchallenged oligopoly of a few dominant content producers (Anderson, 2006), or the skeptical view that such environments cultivate the pursuit of individual users’ specific preferences to the exclusion of all else (e.g., Pariser, 2011). Despite the abundance of options and the hopes and fears they inspire, a small number of popular options continue to dominate the media marketplace. This is evidence of a “Superstar” effect, described by Rosen (1981) and Brynjolfsson, Hu, and Smith (2010) as the tendency of one or two extremely popular options to dominate marketplaces, even in higher choice environments. Results of the present research show that the abundance offered by higher choice media environments does not lead to diversity in selection behavior but instead to greater dominance of so-called “Superstars.” It also suggests that a kind of compromise in the design of media choice environments in which the overall quantity and diversity of options that are the hallmarks of the digital media environment are preserved while not making every option available to all users at once could produce greater diversity in selection behavior.
Going forward, the goal of media researchers should be to further specify claims about the relationship between choice environments and media selection behavior and look for evidence of these claims “in situ” rather than making grand pronouncements about the entire media choice universe (which the user never encounters at one time). The present research is also intended to help media researchers and practitioners to think more creatively about the concept of “choice” as it relates to actual selection behaviors of users. We have yet to fully explore and understand the diversity of the ways in which options can be presented to users and the impacts of presenting options in these various ways on a wide variety of outcomes.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.