Mortality Salience Effects on Selective Exposure and Web Browsing Behavior

Abstract

Issue-related attitudes appear to drive individuals' selective Website browsing behavior, as a match between issue-related attitudes and Website content results in a greater number of page views. However, this relationship between content domain involvement and selective exposure is more complex than current theories hold, as the relationship emerges only when individuals are not primed to reflect on their mortality. Reminders of mortality, which are common in everyday life, lead to a defense motivation, which influences selective exposure behavior. In the control condition, issue-related attitudes predicted Website browsing, but under a mortality salience induction, these attitudes did not predict browsing behavior. Further, response latencies for the measure of individuals' attitudes toward the Website were influenced by their issue-related attitudes in the mortality salience condition but not in the control condition. The patterns of interactions produced in this experiment (N=215) illustrate the relationship between dispositional and situational factors that drive selective exposure in an online context. The results also underscore the importance of using behavioral and process-dependent measures in addition to self-report measures.

Introduction

Despite an abundance of information on the World Wide Web, people tend to limit their browsing behavior to content that matches their pre-existing interests. When they are in a defensive mode of thinking, this restrictive pattern is even stronger. The present study examined how inducing a defensive cognitive mode can affect Web browsing behavior.

Terror Management Theory (TMT) maintains that when a person is made aware of his or her own mortality, he or she has the potential to experience anxiety, which he or she then tries to avoid.^1–3 People cope with this by defending their cultural worldview and by reinforcing their sense of self in a variety of ways.⁴ Coping efforts depend on opportunities available to them,² and this tendency should be reflected in online browsing behavior. The Heuristic-Systematic Model (HSM)^5,6 outlines three factors that motivate users to process content, one of which is a defense motivation. This motivation is driven by the goal of maintaining or bolstering one's material interest via beliefs and attitudes that are closely linked to one's sense of self.⁶ To access supportive messages, individuals seek content that reinforces attitudes they already hold and avoid attitude-inconsistent messages.⁷ Therefore, mortality-aware individuals likely engage in different browsing practices, depending on their cultural worldview and their stake in the content available.

The cognitive processes through which individuals defend their sense of self and their cultural worldview have intrigued scholars,⁸ who have examined differences in processing tendencies employed under mortality salience conditions. Both the HSM and Cognitive-Experiential Self-Theory (CEST)⁹ hold that people generally process information in two distinct ways: (a) systematic or effortful processing entails careful attentive scrutiny of information; while (b) experiential heuristic processing involves low-effort use of mental shortcuts to evaluate information in an associative fashion. In the context of HSM, inducing mortality salience (using the standard induction) before participants' viewing of advertisements with strong and weak arguments (taken from previous studies) was shown to reduce systematic, or effortful analytic, processing.¹⁰ Individuals reminded of their mortality (vs. a control) did not distinguish between strong and weak arguments presented in advertisements. Likewise, testing TMT's connection to CEST showed that mortality salience led to worldview defense only under associative processing.¹¹ Thus, mortality salience should result in less systematic processing, and the types of attitudes implicated in systematic processing should have less influence.

Selective exposure

Many individual factors may play a role in determining how individuals select and attend to media content, including mood, ideology, past experience, and social identity.¹² Studies across media platforms have shown that media consumers make media choices to regulate their affective state,^12,13 and that these effects do not influence media choices in a uniform fashion. For example, satisfaction with a given dimension of one's life can impact how related content is processed; lower satisfaction with professional status led to more online browsing of college and career content, while higher satisfaction with romantic life led to more browsing of romantic content.¹⁴ Pre-existing attitudes toward issues may also be a key determinant of exposure. The idea that individuals select and attend to media that reinforce their existing beliefs¹⁵ has long been considered a key component of increased polarization and differential media use.¹⁶ One explanation holds that humans minimize cognitive dissonance¹⁷ by avoiding information which conflicts with their pre-existing beliefs, although some studies^18,19 have shown no evidence for this being a motivation. A related, but distinct, explanation is that individuals choose content to reinforce their social identity.^20–22

Though selective exposure theories originated to explain patterns of television and newspaper consumption, the Web offers a much greater option for content selection, as it affords a non-linear selection of specific content.²³ In an online browsing context, users are drawn to information that reinforces their existing attitudes. Broad political ideology influences content selection on the basis of like-leaning media outlets,²⁴ and opinion reinforcement, rather than opinion-challenging content, is a greater predictor of exposure to content on online news sites.²⁵

Since selective exposure can be a function of selecting attitudinally congruent information, people who express strong concern for an issue should browse more pages of a Website that presents pro-issue information. This pattern is also consistent with dual-process models, which suggest that people who have relatively high levels of involvement in a given topic, such as the environment, will process information with greater depth, which is characteristic of systematic processing. As a result, we predict that:

H1: A participants' issue-related attitudes will influence their browsing behavior of an issue-congruent Website, such that participants who tend to have pro-issue attitudes will view more of the site's Web pages.

Since mortality salience reduces systematic processing, participants will be less motivated to engage in the attitudinally congruent browsing behavior that is characteristic of high-involvement processing. Thus, participants' issue-related attitudes will have less influence on browsing behavior when they have been reminded of their own mortality. As a result, we predict the following:

H2: Mortality salience will moderate the effect of participant's issue-related attitudes on selective exposure, such that their issue-related attitudes will be a stronger predictor of the number of pages of an issue-congruent Website which they view in the control condition than in the mortality salience condition.

H3: Mortality salience will moderate the effect of participants' issue-related attitudes on their attitude toward an issue-congruent Website, such that their issue-related attitudes will be a stronger predictor of their attitudes toward the Website in the control condition than in the mortality salience condition.

However, since mortality salience leads to a reduction in systematic processing, individuals might still rely on their issue-related attitudes in forming impressions under heuristic processing. This type of processing is relatively effortless, which means that individuals will spend less time engaged in cognitive elaboration, reaching a judgment as a part of a snap process. If so, then under a mortality salience induction, individuals who have favorable issue-related attitudes should report a judgment more quickly than individuals whose attitudes are less favorable toward the issue. Therefore, we predict the following:

H4: Mortality salience will moderate the effect of participants' issue-related attitudes on their response latencies for the measure of their attitude toward an issue-congruent Website, such that response latencies will be faster in the mortality salience treatment condition than in the control condition.

Methods

Participants

Participating were 215 undergraduate students (M_age=20.14 years), 60 percent of whom were women and 80 percent of whom were white, who received course credit for their participation.

Procedure and materials

The hypotheses were tested in a between-subjects post-test only experiment with mortality salience (treatment vs. control) as a manipulated factor, and the New Ecological Paradigm²⁶ (NEP; M=4.61, SD=0.70; Cronbach's α=0.77) as a measured independent variable. The mortality salience induction followed the widely used procedure²⁷ in which participants in the treatment condition wrote responses to the following two questions: “Please briefly describe the emotions that the thought of your own death arouses in you” and “Jot down, as specifically as you can, what you think will happen to you as you physically die and once you are physically dead.” In the control, participants completed similar questions about experiencing dental pain: “Please briefly describe the emotions that the thought of experiencing dental pain arouses in you” and “Jot down, as specifically as you can, what you think happens to you physically as you experience dental pain.” The induction's effects are stronger after a brief period of time has elapsed in which participants are prevented from coping with the potential for anxiety; so, as a part of the standard procedure,²⁷ a distractor task is included. The NEP is a widely used measure²⁶ of an individual's beliefs about the environment. A high score reflects the belief that humans should exercise caution to care for the environment, while a low score reflects the belief that nature can be modified readily to accommodate humans' needs. Sample items include The balance of nature is delicate and easily upset, and The earth has plenty of natural resources if we just learn how to develop them. To control for differences in participants' browsing tendency, the Need For Cognition short form²⁸ (NFC; M=4.53, SD=0.80; Cronbach's α=0.88) was included. The data, part of a larger unreported dataset, were collected using MediaLab software²⁹ and millisecond-accurate keyboards.

The NEP was administered at the beginning of the experiment to make participants' environmental attitudes salient, followed by the NFC, along with another unrelated dispositional measure. After an ostensible first study was completed, the study was closed on the computer, and a second study was opened, which began with the mortality salience induction, followed by the distraction task. Participants then browsed a Website for Keep America Beautiful (www.kab.org), which promotes litter prevention, recycling, and landscape beautification. To measure browsing behavior, software recorded the number of site pages (M=2.40, SD=2.03) that each participant viewed.

After visiting the site, participants completed an 18-item semantic differential scale³⁰ that assessed their impression of the site. The scale included adjectival pairs such as likeable-dislikeable, attractive-unattractive, boring-interesting, and valuable-worthless, (M=5.08, SD=1.01; Cronbach's α=0.94).

Mean response latencies (in milliseconds) for the NFC items formed a baseline measure of each participant's response latency. Response latency data were also collected for the attitudes toward the Website items. The response latency baseline measure was subtracted from the attitudinal measure's mean latency, and 8,000 milliseconds was added to this figure to produce a positive value. Thus, the measure (M=5,167.01 milliseconds, SD=1,272.60) reflected the change in latency that was relative to each individual's baseline.

To test for differences in negative affect between conditions, the Positive and Negative Affect Scale³¹ was administered. After participants had completed several demographic measures, they were debriefed and thanked.

Results

A model was constructed that included the NEP, the mortality salience induction, and their interaction terms, with the NFC included as a simple covariate, for each dependent variable: pages viewed, attitudes toward the Website, and response latency for attitudes toward the Website. The models were tested in a stepwise fashion, with the main effects and simple slopes tested before adding the interaction term. Before proceeding, it was determined that negative affect, sometimes considered a potential confound with mortality salience, did not vary significantly, t(199)=1.47, p=0.142, across the mortality salience conditions (M_control=2.25, SE=0.10; M_treatment=2.46, SE=0.10).

The analysis of covariance (ANCOVA) for the number of pages viewed was marginally significant in Step 1, F(3, 208)=2.41, p=0.068, Adj. R²=0.02, but significant in Step 2, with the interaction term included, F(4, 207)=3.24, p=0.013, Adj. R²=0.04. In Step 1, the NEP's slope was a significant predictor (Table 1), providing support for H1. In Step 2, when the two-way interaction was added, the NEP's slope was marginally significant (Table 1). As expected, the interaction between mortality salience and the NEP was significant in predicting the number of pages viewed (Table 1), providing support for H2.

Table 1.

Regression Weights for Page Views (N=212)

Variable	B	SE	β
Step 1
NFC	0.26	0.17	0.10
MS	−0.01	0.14	−0.01
NEP	0.40	0.20	0.14^*
Step 2
NFC	0.29	0.17	0.12
MS	−0.01	0.14	−0.01
NEP	0.36	0.20	0.13
MS×NEP	0.47	0.20	0.16^*

F(4, 207)=3.24, p=0.013.

Step 1, Adj. R²=0.02.

Step 2, ΔAdj. R²=0.02.

p<0.05.

NFC, Need For Cognition short form; NEP, New Ecological Paradigm; MS, mortality salience.

In a post-hoc simple-slope analysis, the NEP predicted the number of pages viewed in the control condition (β=0.27, p=0.004), but not in the treatment condition (β=−0.05, p=0.610). The pattern (Fig. 1) showed that, in the control, participants who had greater levels of environmental concern viewed more of the site's pages, but this effect was absent under mortality salience.

FIG. 1.

Browsing behavior as a function of the interaction of mortality salience and environmental attitudes. Best-fitting regression lines. Solid line=control; dashed line=treatment.

The ANCOVA for attitude toward the Website was not significant, F(8, 203)=0.86, p=0.552. No significant effects on attitude toward the Website emerged (all ps>0.20), leaving H3 unsupported.

The ANCOVA for response latencies for Website attitudes was non-significant in Step 1, F(3, 210)=1.09, p=0.356, and marginally significant in Step 2, F(4, 209)=2.32, p=0.058, Adj. R²=0.03. In Step 1, no significant effects emerged, but in the full model, the interaction between mortality salience and the NEP was significant (Table 2), providing support for H4.

Table 2.

Regression Weights for Response Latencies of Attitude Toward Websites (N=214)

Variable	B	SE	β
Step 1
NFC	103.36	109.19	0.06
MS	−72.41	87.52	−0.06
NEP	−162.82	126.63	−0.09
Step 2
NFC	123.84	108.28	0.08
MS	−70.34	86.51	−0.06
NEP	−190.99	125.69	−0.11
MS×NEP	305.25	125.04	0.17^*

F(4, 209)=2.32, p=0.058.

Step 1, Adj. R²=0.00.

Step 2, ΔR²=0.02.

p<0.05.

In a post-hoc simple-slope test, the NEP predicted response latencies in the mortality salience treatment condition (β=−0.268, p=0.010), but not in the control condition (β=0.054, p=0.565). The interaction pattern (Fig. 2) shows that under conditions of mortality salience, individuals who had greater levels of environmental concern took less time to report their attitude toward the Website.

FIG. 2.

Response latencies as a function of the interaction of mortality salience and environmental attitudes. Best-fitting regression lines. Solid line=control; dashed line=treatment.

Discussion

An individual's issue-related attitudes were made salient at the beginning of the experiment, and these attitudes appeared to drive subsequent Website browsing behavior. A match between an individual's issue-related attitudes and Website content resulted in a greater number of page views, or conversely, incongruence between issue-related attitudes and Website content predicted a lower number of page views. Thus, pre-existing issue-related attitudes are an important driver of selective exposure.

However, this relationship between content domain involvement and selective exposure occurred only when individuals were not primed to reflect on their mortality. Consistent with previous research, a mortality reminder appeared to attenuate systematic processing. In the control, participants' issue-related attitudes predicted their Web browsing behavior, but not under a mortality salience induction (Fig. 1). Defensive processing can involve high levels of systematic processing—as one attends carefully, and selectively, to information to refute it or develop counterarguments—or high levels of heuristic processing—as one employs heuristics that bias one's thinking in favor of merely reinforcing existing attitudes—or a combination of both modes.⁶ Here, issue-related attitudes guided browsing behavior only when individuals could process systematically, but the type of defensive processing that mortality salience induces reduced systematic processing, and, thus, issue-related attitudes played no role in selective exposure behavior.

Nevertheless, under mortality salience-induced heuristic processing, issue-related attitudes influenced participant responses. Issue-related attitudes predicted response latencies for Website attitudes in the mortality salience condition but not in the control (Fig. 2). Thus, individuals did not engage in deliberative cognitive elaboration when they reported their evaluation of a Website that was congruent with their attitudes. Likewise, individuals for whom issue-related attitudes did not match the Website content took longer to report their evaluations. In the control—wherein individuals could process systematically—issue-related attitudes did not influence response latencies. Significantly, these factors had no influence on individuals' reported attitude toward the Website in either condition. So, although mortality salience appeared not to influence individuals' reported impression of the site, response latencies suggested that individuals who had been reminded of their mortality, and whose attitudes were congruent with the site, reported their impression more quickly.

The model tests of behavioral dependent variables (browsing behavior and response latencies) were significant, while the test of the self-report dependent variable (attitude toward the Website) was non-significant. It is important to note this difference, because many theoretical claims are staked solely on self-report data. While self-report measures have much to commend them, scholars should remain aware of their limitations and, when possible, seek to include behavioral or process measures to provide additional evidence for their conclusions.

The results also suggest that the relationship between pre-existing attitudes and the browsing of attitude-congruent information is more complex than current theories hold. If reminding participants of their mortality can negate the influence of pre-existing attitudes on selection, the results have implications regarding the processes through which specific persuasive appeals compete for attention. Given the potential for existential anxiety, when certain worldview dimensions may be more accessible,³² this experiment found that pro-issue attitudes no longer drove individuals to seek issue-related information. The types of knowledge structures an individual might draw on during systematic processing might be different than worldview-related knowledge structures that are more accessible when under a worldview defense. A death prime, therefore, might lead to different browsing patterns by changing the accessibility of different types of attitudes.

Although the NEP has been described as a measure of a pro-ecological worldview, or paradigm,²⁶ further research is needed to bolster that claim. Human–nature relations has been identified as a core dimension of a cultural worldview,³³ and TMT research suggests that human–nature relations is, indeed, a part of a cultural worldview, if the environment is tied to one's sense of self.³⁴ However, unlike studies that have found increased accessibility for worldview-related scales under a mortality-salience induction, the NEP's accessibility was attenuated, which suggests that the environment-related attitudes tapped by the NEP do not necessarily form a part of a person's core worldview. Indeed, the reporting of pro-environment attitudes for the NEP might suffer from a social desirability bias. Young adults, it should be noted, are more likely to report ecological concern but less likely to engage in pro-environmental behavior.³⁵

Nevertheless, the present study is an important examination of the interaction of situational and dispositional factors that can influence selective exposure in an online context. People encounter reminders of their mortality nearly every day, whether through news programming, life insurance advertising, or magazine articles on health care. Any one of these reminders can induce a temporary change in an individual's cognitive processes, with important implications for selective exposure on the Web.

Footnotes

Author Disclosure Statement

No competing financial interests exist.

References

Greenberg

, Solomon

, Pyszczynski

. Terror management theory of self-esteem and cultural worldviews: empirical assessments and conceptual refinements. Advances in Experimental Social Psychology, 1997; 29:61–139.

Pyszczynski

, Greenberg

, Solomon

et al. Why do people need self-esteem? A theoretical and empirical review. Psychological Bulletin, 2004; 130:435–468.

Solomon

, Greenberg

, Pyszczynski

. 2004. The cultural animal: twenty years of terror management theory and research. Greenberg

, Koole

, Pyszczynski

. Handbook of experimental existential psychology. New York: Guilford, 13–34.

Pyszczynski

, Cox

. Can we really do without self-esteem? Comment on Crocker and Park. Psychological Bulletin, 2004; 130:425–429.

Chaiken

. Heuristic versus systematic information processing and the use of source versus message cues in persuasion. Journal of Personality and Social Psychology, 1980; 39:752–756.

Chen

, Chaiken

. 1999. The heuristic-systematic model in its broader context. Chaiken

, Trope

. Dual-process theories in social psychology. New York: Guilford, 73–96.

Knobloch-Westerwick

, Meng

. Reinforcement of the political self through selective exposure to political messages. Journal of Communication, 2011; 61:349–368.

Arndt

, Cook

, Routledge

. 2004. The blueprint of terror management: understanding the cognitive architecture of psychological defense against the awareness of death. Greenberg

, Koole

, Pyszczynski

. Handbook of experimental existential psychology. New York: Guilford, 35–53.

Epstein

, Pacini

. 1999. Some basic issues regarding dual-process theories from the perspective of cognitive-experiential self-theory. Chaiken

, Trope

. Dual-process theories in social psychology. New York: Guilford, 462–482.

10.

Magee

, Kalyanaraman

. 2011. Death reminders and advertisements: effects of mortality salience on systematic processing. Paper presented to the Mass Communication Division at the Annual Convention of the International Communication Association, Boston.

11.

Simon

, Greenberg

, Harmon-Jones

et al. Terror management and cognitive-experiential self-theory: evidence that terror management occurs in the experiential system. Journal of Personality and Social Psychology, 1997; 72:1132–1146.

12.

Zillmann

, Bryant

. 1985. Selective exposure to communication. Mahwah, NJ: Erlbaum.

13.

Bryant

, Zillmann

. Using television to alleviate boredom and stress: selective exposure as a function of induced excitational states. Journal of Broadcasting, 1984; 28:1–20.

14.

Knobloch-Westerwick

, Hastall

, Rossmann

. Coping or escaping? Effects of life dissatisfaction on selective exposure. Communication Research, 2009; 36:207–228.

15.

Klapper

. 1960. The effects of mass communication. New York: Free Press.

16.

Sunstein

. 2001. Republic.com. Princeton, NJ: Princeton University.

17.

Festinger

. 1957. A theory of cognitive dissonance. Stanford, CA: Stanford University Press.

18.

Meffert

, Chung

, Joiner

et al. The effects of negativity and motivated information processing during a political campaign. Journal of Communication, 2006; 56:27–51.

19.

Sears

. 1968. The paradox of de facto selective exposure without preference for supportive information. Abelson

, Aronson

, McGuire

et al. Theories of cognitive consistency: a sourcebook. Chicago: Rand McNally, 777–787.

20.

Harwood

. Age identification, social identity gratifications, and television viewing. Journal of Broadcasting and Electronic Media, 1999; 43:123–136.

21.

Slater

. Persuasion processes across receiver goals and message genres. Communication Theory, 1997; 7:125–148.

22.

Slater

. Reinforcing spirals: the mutual influence of media selectivity and media effects and their impact on individual behavior and social identity. Communication Theory, 2007; 17:281–303.

23.

Althaus

, Tewksbury

. Agenda setting and the news: patterns of issue importance among readers of the paper and online versions of The New York Times. Communication Research, 2002; 29:180–207.

24.

Iyengar

, Hahn

. Red media, blue media: evidence of ideological selectivity in media use. Journal of Communication, 2009; 59:19–39.

25.

Garrett

. Politically motivated reinforcement seeking: reframing the selective exposure debate. Journal of Communication, 2009; 59:676–699.

26.

Dunlap

, Van Liere

, Mertig

et al. New trends in measuring environmental attitudes: measuring endorsement of the new ecological paradigm: a revised NEP scale. Journal of Social Issues, 2000; 56:425–442.

27.

Greenberg

, Arndt

, Simon

et al. Proximal and distal defenses in response to reminders of one's mortality: evidence of a temporal structure. Personality and Social Psychology Bulletin, 2000; 26:91–99.

28.

Cacioppo

, Petty

, Kao

. The efficient assessment of the need for cognition. Journal of Personality Assessment, 1984; 48:306–307.

29.

Jarvis

. 2008. MediaLab(version 2008.1.33)New York: Empirisoft Corporation.

30.

Bruner II

, Hensel

, James

. 2005. Marketing scales handbook: a compilation of multi-item measures for consumer behavior & advertising, vol. IV. Mason, OH: Thomson.

31.

Watson

, Clark

, Tellegen

. Development and validation of brief measures of positive and negative affect: the PANAS scales. Journal of Personality and Social Psychology, 1988; 54:1063–1070.

32.

Magee

, Kalyanaraman

. Effects of worldview and mortality salience in persuasion processes. Media Psychology, 2009; 12:171–194.

33.

Koole

, van den Berg

. 2004. Paradise lost and reclaimed: a motivational analysis of human-nature relations. Greenberg

, Koole

, Pyszczynski

. Handbook of experimental existential psychology. New York: Guilford, 86–103.

34.

Vess

, Arndt

. The nature of death and the death of nature: the impact of mortality salience on environmental concern. Journal of Research in Personality, 2008; 42:1376–1380.

35.

Twenge

, Freeman

, Campbell

. Generational differences in young adults' life goals, concern for others, and civic orientation, 1966–2009. Journal of Personality and Social Psychology, 2012; 102:1045–1062.