cognitive access to negatively arousing news

COMMUNICATION RESEARCH • February 2000Grabe et al. • Negatively Arousing News

MARIA ELIZABETH GRABEANNIE LANGSHUHUA ZHOUPAUL DAVID BOLLS

Cognitive Access toNegatively Arousing NewsAn Experimental Investigationof the Knowledge Gap

Over the past 30 years, survey researchers have documented the existence of aknowledge gap and expressed concern that people with little education are fal-ling behind because they do not acquire the information necessary to partici-pate in socioeconomic spheres. This study is the first to offer (a) experimentalevidence for the existence of the knowledge gap and (b) explanations for it interms of varying levels of information processing capacities, or cognitiveaccess. Participants from higher and lower educational backgrounds paidequal levels of attention to television news stories, but they did not display thesame recognition memory for facts. Moreover, participants in the higher edu-cation group were physiologically more aroused by news than those in thelower education group. These findings do not pinpoint whether cognitiveaccess is learned or innate, but they do suggest that the biological systems ofpeople from higher educational backgrounds are particularly alert in prepar-ing for information processing.

The idea that people with little education have less knowledge about impor-tant social issues than people who achieve higher levels of education chal-lenges democratic ideals of equality. A number of scholars (Beltran, 1975;Bennett, 1988; Childers & Post, 1975; Deli Carpini & Keeter, 1994; Donohue,Olien, & Tichenor, 1987; Graber, 1993; Olien, Donohue, & Tichenor, 1983)have argued that knowledge, particularly about public affairs, is a majormeans to empowerment. Put simply, social systems favor people who areinformed, whereas the uninformed become disenfranchised, often without

3

COMMUNICATION RESEARCH, Vol. 27 No. 1, February 2000 3-26© 2000 Sage Publications, Inc.

being aware of it (Deli Carpini & Keeter, 1994). Although people from lowersocioeconomic strata do accumulate knowledge, it is typically gossip, rumor,and folklore rather than knowledge that is socially and economically empow-ering (Childers & Post, 1975; Moore, 1987; Price & Zaller, 1993).

Tichenor, Donohue, and Olien (1970) first alluded to the mass media’s con-tribution in perpetuating knowledge inequities when they formally posed theknowledge gap hypothesis:

As the infusion of mass media into a social system increases, segmentsof the population with higher socioeconomic status tend to acquire thisinformation at a faster rate than lower segments, so that the gap inknowledge between these segments tends to increase rather thandecrease. (pp. 159-160)

From the time of this formulation to 1997, approximately 100 studies,each examining some dimension of the knowledge gap hypothesis, have beenpublished (for comprehensive reviews of the literature, see Gaziano, 1983,1997; Viswanath & Finnegan, 1996). Without exception, existing research onthe knowledge gap hypothesis has employed survey methodology to examinecorrelational and causal relationships between knowledge and such vari-ables as education, media exposure, attention, interest and involvement inpublic affairs media, salience of the message to audience members, and chan-nel use (radio, television, newspaper) of audience members. The surveyapproach to studying the knowledge gap certainly has utility. Yet, there isreason to argue for methodological reevaluation when a theorem such as theknowledge gap hypothesis, which focuses on individual-level informationacquisition, has only been investigated through the aggregate-level surveymethod. To date, not a single experimental study has been conducted toexamine the information processing mechanisms that might underlie vari-ance in information gain. The study reported here attempts to fill this void bybeing the first to do so.

Physical Versus Cognitive Access

Whereas physical access to information has been identified as a central con-tingent condition in knowledge gap research (see Gaziano, 1983, 1997; Vis-wanath & Finnegan, 1996), the matter of cognitive access or informationprocessing competence has virtually been ignored as a possible explanationfor knowledge inequities. Although not specifically tied to the knowledge gap,the distinction between physical and cognitive access is starting to enjoyattention in information science studies of the Internet. Physical availability

4

COMMUNICATION RESEARCH • February 2000

of equipment through which citizens can search and locate information in amedia environment is distinguished from the ability of media consumers toaccess and make use of online information (Kling, 1999).1

The distinction between physical and cognitive access to mediated infor-mation, particularly television, also deserves attention as it might shed lighton why this medium fails to rectify information inequities. The popularassumption that television messages are processed without much effortmight account for why knowledge gap researchers have ignored issuesrelated to cognitive access and instead investigated television news for itspotential to level knowledge inequities (see Galloway, 1977; Gantz, 1978;Neuman, 1976; Shinghi & Mody, 1976; Tichenor et al., 1970; Torsvik, 1972).Investigations of television’s capacity to create a more evenly informed citi-zenry have generally been inconclusive. In fact, a number of knowledge gapstudies have contradicted the notion that television viewing eliminates theknowledge gap (Gandy & El Waylly, 1985; Griffin, 1990; Gunter, 1987;Horstmann, 1991; McLeod & Perse, 1994; Simmons & Garda, 1982). As theinformation processing literature has demonstrated, viewing television newsis not a simple cognitive task. The cognitive system is burdened with process-ing multiple audio-visual channels (voice, ambient sound, visual material) ofmostly survival-related information in a living environment crowded withdistraction (see Drew & Grimes, 1987; Graber, 1990; Newhagen, 1998).These findings fuel the motivation for an experimental investigation of theknowledge gap with focus on television news.

Approximately 99% of homes in the United States have at least one work-ing television set and 70% of homes have more than one set (U.S. Bureau ofthe Census, 1997). These figures indicate virtual saturation, and in essence,full physical access to television. In terms of television news viewing, 64% ofAmericans say they rely on national and local television newscasts for infor-mation (Radio Television News Director’s Association, 1999). Knowledge gapresearch confirms that people on the lower socioeconomic strata tend todepend more on television than print media for information (Bogart, 1981;McLeod & Perse, 1994). Physical access or exposure to television news is cen-tral to this investigation because different levels of familiarity with televi-sion news formats might affect how people with different educational back-grounds process television news. Hence,

Research Question 1: Does exposure/physical access to television news dif-fer by educational level?

Although the notion of cognitive access has not yet been employed in tele-vision news research, a number of studies in the information processing

5

Grabe et al. • Negatively Arousing News

tradition point to the logic of testing this concept in the context of the knowl-edge gap hypothesis.

The limited capacity model of information processing (Geiger & Newha-gen, 1993; A. Lang, Dhillon, & Dong, 1995) proposes that television viewersdo not retain all the information they are exposed to during a broadcast butinstead select portions of messages for encoding and storage into long-termmemory. At the same time, previously stored information needed to makesense of the current message is retrieved. These subprocesses (selection,encoding, storage, and retrieval) occur continuously and simultaneously dur-ing viewing, taxing the viewers’ information processing resources. Thus, notall audio-visual information is selected, encoded, stored, and retrieved withthe same efficiency.

The allocation of processing resources, whether controlled or automatic,plays an important role in storing information. Controlled processes, as theterm implies, are under the control of the viewer. Several factors may acti-vate controlled resource allocation, including the viewers’ goals, intentions,interests, and motivations. Automatic processes, on the other hand, are trig-gered by characteristics of the message. Structural features (attention-grabbing video editing and camera techniques), as well as compelling mes-sage content, elicit involuntary allocation of processing resources.

The motivation to attend to a message, or controlled processing, variesgreatly from viewer to viewer, whereas the basic mechanics of automaticinformation processing are shared by most human biological systems. Thus,differences among viewers in terms of controlled information processing arelargely explainable by varying levels of interest in the message; differencesamong viewers in terms of automatic processing are best explained by vari-ance in aptitude to process information efficiently. When automatic process-ing is activated, researchers have an opportunity to study the underlyingprocesses of information gain without interference from motivational factors.

Among the variables used to study information processing, physiologicaland emotional arousal, attention, and recognition memory are the mostrevealing (A. Lang et al., 1995; A. Lang & Basil, 1998; Newhagen & Reeves,1992). These three concepts constitute important catalysts of informationprocessing and, for the purposes of this study, are treated as determinants ofcognitive access and ultimately information gain.

Arousal

Research has established associations between arousing television messagesand a number of outcomes of information processing. First, arousing televi-sion content elicits more attention than calm television messages (A. Lang

6


et al., 1995; A. Lang, Newhagen, & Reeves, 1996; Newhagen & Reeves, 1992,Zillmann, 1982). Second, viewers allocate more processing resources (mea-sured using secondary-task reaction times) to arousing messages, comparedto calm messages (A. Lang et al., 1995; Lang et al., 1996). Third, viewers havebetter recognition memory (A. Lang, Bolls, Potter, & Kawahara, 1999;A. Lang et al., 1995; A. Lang, Zhou, Schwartz, Bolls, & Potter, in press), cuedrecall (A. Lang, Bolls, et al., 1999), and free recall or comprehension (Gure-vitch & Levy, 1986; Lang et al., 1995) for arousing messages, compared tocalm messages. Thus, when messages are arousing, viewers (a) automati-cally pay more attention to media content, (b) automatically allocate process-ing resources to encoding and storage, and, as a result, (c) experience anincrease in sympathetic nervous-system activation. Most importantly, thesethree consequences of arousing messages improve the ability of viewers togain information from viewing television news.

Negatively arousing messages appear to have particularly powerful out-comes for information gain. Reeves, Newhagen, Maibach, Basil, and Kurz(1991) and A. Lang and colleagues (1996) report that memory is better fornegative than positive messages. P. J. Lang (1985), Newell (1990), and Shoe-maker (1996) explain the influence of negative messages on informationprocessing from the perspective of evolutionary psychology: In the face of asurvival threat, emotional conditions are automatically activated to enhancephysical performance (attack or flight). Newhagen and Reeves (1992), New-hagen (1998), and Shoemaker (1996) argue that negatively compelling televi-sion news images resemble a nonmediated survival threat to the degree thatthe biological systems of viewers automatically prepare for premium per-formance, hence the increase in physical and emotional arousal, attention,and memory associated with negatively compelling messages (see also Basil,Schooler, & Reeves, 1991; Bradley, Greenwald, & Hamm, 1993; Bradley,Greenwald, Petry, & A. Lang, 1992; A. Lang & Friestad, 1993; Newhagen &Reeves, 1992; Reeves, A. Lang, Thorson, & Rothschild, 1989; Reeves et al.,1991).

Negatively arousing stimuli could be expected to optimize automaticinformation-processing operations so that the underlying mechanisms mightbe studied in people from different educational backgrounds. No previousresearch suggests that people with different educational backgrounds shouldexperience different levels of physiological or self-reported arousal whenthey watch negatively compelling news. In fact, a few knowledge gap studieshave indirectly suggested that the arousal response to negatively compellingmessages should not differ by educational background (see Frazier, 1986;Galloway, 1977; Genova & Greenberg, 1979). Donohue, Tichenor, and Olien(1975) conclude that knowledge gaps about local news issues tend to narrow

7


because these issues are typically about conflict and therefore should be sali-ent to all people, regardless of educational background. This salienceassumption leads to a second research question:

Research Question 2: Do arousal responses to negatively arousing newsstories differ by education level?

Attention

Weenig and Midden (1997) suggest that less-educated people pay less atten-tion to media messages than people who are more educated. From an infor-mation processing perspective, different levels of attention are caused by dif-ferences in the allocation of processing resources on account of eithermotivation or information processing competency. The first possibility, thatthe difference lies with the motivation of viewers, is more a question aboutinterest and behavior than pure cognitive access. In knowledge gap research,motivation emerges as one of the most prominent individual-level variablesto explain information differences (see Gaziano, 1997; Viswanath & Finne-gan, 1996). Motivation to attend to messages has been tested in relation toknowledge gain using variables such as viewer interest, involvement, con-cern, participation, and perceived message importance (see Chew & Palmer,1994; Ettema & Kline, 1977; Ettema, Brown, & Luepker, 1983; Fredin, Mon-nett, & Kosicki, 1994; Genova & Greenberg, 1979; Kwak, 1999; McLeod &Perse, 1994). Ettema and Kline (1977) and Ettema and colleagues (1983) con-clude that knowledge gaps exist in part because of differing levels of interestin and use for information. They dismiss information processing explana-tions for the knowledge gap by arguing that public affairs issues do notrequire highly developed processing abilities. Instead, they argue that publicaffairs issues are merely more important and useful to people in higher socio-economic groups, which leads to the formation of knowledge gaps.

To assess the contribution of motivation to news information acquisition,both the information and entertainment interests of viewers were consideredin this study. This leads to a third research question:

Research Question 3: Do evaluations of the informativeness and enjoy-ment of negatively arousing news stories vary by education level?

The second reason why viewers might allocate varying amounts of atten-tion resources to news viewing involves their cognitive processing compe-tence. To maximize the chances of measuring the information processing

8


mechanisms at work, rather than the motivation of viewers to attend to mes-sages, the effect of motivation on allocation of processing resources could becontrolled in the experimental setting. First, participants could be instructedto pay close attention to the messages because their memory for the contentwill be tested. Second, as already discussed, negatively arousing news storiestrigger automatic allocation of processing resources, thereby overwhelmingindividual differences in motivation to attend to the news. This leads to thefourth research question:

Research Question 4: Does attention to negatively arousing televisionnews stories vary by education level?

Encoding of Information

Viewers with higher education levels may remember more from the mediabecause the process of reasoning is more elaborated for educated people thanfor the less educated (Park & Kosicki, 1995). It is generally thought thatgreater elaboration leads to a higher level of processing and an increase inthe amount of information stored (Craik & Lockhart, 1972).

Similarly, Rogers (1976), Robinson (1967), and Chew and Palmer (1994)argue that less-educated people may have less-developed cognitive abilitiesto select, store, and retrieve information, suggesting that they may gain lessknowledge from the same amount of information exposure than higher edu-cated people. Stauffer, Frost, and Rybolt (1978) attribute memory differencesto the possibility that people in higher and lower education groups do notobtain information from television news with the same efficiency. A possibleexplanation is that the educational processes that develop specific readingand writing skills may enhance the general ability to process visual andaural information (Park & Kosicki, 1995).

These authors suggest that memory is better for people with higher levelsof education because their cognitive processes involved in producing memoryfunction more efficiently. To investigate whether education impacts the effi-ciency of information storage, this study tests how well participants withvarying levels of education fare in encoding and retrieving information. Thisleads to the fifth and sixth research questions:

Research Question 5: Does the encoding of information from negativelyarousing news vary by education level?

Research Question 6: Does memory sensitity vary by education level?

9


Method

Participants with different educational backgrounds were recruited for thisstudy. One group had no more than a high school education, whereas theother group had at least some postgraduate education. The participants wereinvited to view eight broadcast news stories. During viewing, participants’heart rates and skin conductance were measured. Immediately after viewingeach story, respondents rated it for enjoyment and information value andindicated their own emotional response. A verbal recognition task, testingrecognition memory for information in the story narratives, was adminis-tered after participants had viewed all eight stories.

Design

This experiment used a mixed education (2)´ story (8)´ order (4) repeatedmeasures design. Education and order were between-participants factors,whereas story was a within-participants factor. Education had two levels,higher and lower. The repetitious story factor comprises the eight news sto-ries that participants viewed. The four levels of order represent the fourquasi-randomized orders of the stimulus tape that were produced. Partici-pants were randomly assigned to view one of the stimulus tape orders.

Stimulus Materials

The stimuli for this experiment consisted of nine stories. One story was usedas a practice message for participants to familiarize themselves with theexperimental situation. The practice story was viewed first in every experi-mental order. The other eight stories were used for data collection. The storytopics included a drive-by shooting; a hostage situation; a tornado; a flood;conflict between abortion protesters, the police, clinic workers, and patients;a residential fire; violent clashes between KKK and Black Panther protest-ers; and conflict during a flag- burning rally. Each of these stories containednegatively compelling images, which are most likely to automatically com-mand attention from viewers.

Raw video material for the stimuli was obtained from WISH-TV, the localCBS affiliate. Edited stories ranged from 90 to 120 seconds in duration. Fourmale newscasters, all with professional experience in broadcast news, wereused for voice-over narration. Each newscaster narrated two stories. The sto-ries were produced on an AVID nonlinear editing system.

10


Dependent Variables

Media use/physical access. Before the experiment, participants recordedon the questionnaire how often they watched local news, national newscasts,and television news magazine programs. Ratings were made using a 3-pointscale: once a week, 2 to 3 times a week, or more than 4 times a week.

Education. Two groups of participants were used in this experiment. Par-ticipants were either in the higher education group (n = 20) or in the lowereducation group (n = 20). Participants with some graduate education wererecruited from interdisciplinary academic programs at a large midwesternuniversity. Potential participants with academic training in journalism ortelecommunications were excluded from the participant pool to minimizefamiliarity with the stimulus material. The lower education group consistedof university employees with no more than a high school education. Partici-pants in the lower education group included janitors, cooks, and physicalplant workers.

Like the majority of investigators of the knowledge gap, we were inter-ested in participants with different levels of education and ended up withparticipants who also differed in terms of social status. However, we do notview these differences in the social statuses of participants as confoundingthe education variable because education level is a key dimension in measur-ing the very complex notion of social class. In fact, knowledge gap researchershave documented a strong correlation between education and other socialclass variables such as income and occupation (see Gaziano, 1997). There isalso a large body of literature in sociology that documents the high correla-tion between education and social class (e.g., see Gorman, 1998; Kinloch,1987; Shanahan, Miech, & Elder, 1998; Yeakey & Bennett, 1990).

Arousal

Both self-reported and physiological arousal were measured in the experi-ment. Self-reported arousal was measured using the Self-Assessment Man-nequin (SAM) (P. J. Lang, Greenwald, Bradley, & Hamm, 1993). The SAMscale is composed of three 9-point pictorial scales that measure three dimen-sions of emotion (arousal, valence, and dominance). For this study, we usedonly the arousal scale. SAM has been shown to be a valid measure of emo-tional response to many types of media messages, including television(Detenber & Reeves, 1996; A. Lang et al., 1995; Lombard, Reich, Grabe, Cam-panella, & Ditton, 2000). Lower scores on the arousal dimension of SAMreflect greater self-reported arousal (1 = arousing, 9 = not arousing).

11


Skin conductance has been reliably employed as an indicator of sympa-thetic nervous system activation, signaling arousal (Cacioppo & Tassinary,1990; Hopkins & Fletcher, 1994). Activation of the sympathetic nervous sys-tem leads to an increase in sweat gland activity, causing greater skin conduc-tivity. Skin conductance responses were collected 20 times per secondthroughout the presentation of each story and for a 5-second baseline periodprior to each story’s onset. Each participant’s skin conductance during thenews stories was coded for the frequency of spontaneous skin conductanceresponses. A skin conductance response was defined as an increase in skinconductance level of at least 0.5 micro siemens (see Cacioppo & Tassinary,1990). The greater the frequency of skin conductance responses, the greaterthe activation of the sympathetic nervous system.

Motivation measures. Single-item semantic differential rating scales wereused to obtain participants’ subjective evaluation of each news story.Between stories, participants rated how enjoyable and informative eachstory was on 10-point semantic differential scales (1 = low, 10 = high).

Attention. Participants’ heart rates were used as a measure of how muchattention was being paid to the news stories. Research has shown that car-diac deceleration is a physiological indicator of increased attention paid to anexternal stimulus (Cacioppo & Tassinary, 1990; A. Lang, 1990), includingtelevision messages (A. Lang, 1994; A. Lang, Bolls, et al., 1999; A. Lang,Zhou, et al., in press; A. Lang et al., 1996). Although the heart is under thedual control of the sympathetic and the parasympathetic nervous system,heart rate decelerates during attention due to activation of the parasympa-thetic nervous system (Cacioppo & Tassinary, 1990).

Heart rate data were collected throughout the presentation of each storyand for a 5-second baseline period immediately prior to the onset of the story.Heart rate was averaged over 5-second intervals in the stories and trans-formed into change scores by subtracting the 5-second averages from heartrate during the last second of the baseline period. The length of news storiesvaried, but 16 to 26 heart-rate change scores were computed for each story.Analysis was performed on change scores computed during the first 40 sec-onds of each story. Because attention can fluctuate over the course of a mes-sage, heart rate data were analyzed over time. Thus heart rate data wereanalyzed using an education (2) ´ story (8) ´ order (4) ´ time (8) mixedrepeated-measures ANOVA. The first three factors (education, story, order)are the factors described above. For the heart rate data, another factor, time(8), was included. This factor refers to the eight 5-second periods that

12


comprise the first 40 seconds of each message, allowing analysis of thechange in heart rate over the course of the messages.

Encoding of information. Recognition of news story content was testedusing a forced-choice recognition test. Participants were presented with 112two-second audio clips. Half of the clips were from news stories the partici-pants had viewed, and half were foils from stories they did not view. Sevenaudio clips were taken from each of the eight news stories. Clips wereselected to test recall memory for information related to the five Ws and Hjournalistic formula (who, what, where, when, why, and how) in addition tohow the local news event fit national trends. Four randomized orders of theaudio recognition test were produced. Participants were instructed to indi-cate whether the clip was or was not from one of the news stories they viewed.

Recognition data were analyzed both in terms of accuracy (percentage ofcorrect answers) and signal detection analysis. Signal detection analysis isbased on the theory that finding a memory in the brain is similar to detectinga weak signal in the environment (see Shapiro, 1994). Two major componentsaffect a person’s decision about whether a signal or memory event is detected.One dimension of that decision is how sensitive or good a person’s memory is,called sensitivity or (d prime). The second dimension of the decision aboutwhether a signal/memory is detected relates to how willing a person is toguess. This dimension is called the criterion bias: how liberal or conservativea person’s decision-making strategy is.

To perform a signal detection analysis, four values are computed for eachparticipant: (a) the percentage of hits—the percentage of items participantssay they have seen before they have in fact seen it; (b) the percentage ofmisses—the percentage of items they say they had not seen before that theyhad seen; (c) the percentage of correct rejections—the percentage of itemsthey say they had not seen before that they had not seen; and (d) the percent-age of false alarms—the percentage of items they say they had seen beforethat they had not seen. These values are combined to compute a participant’ssensitivity (d prime) and criterion bias. The greater a participant’s sensitiv-ity, the more accurate their memory, both in terms of hits and correct rejec-tions. Criterion bias is determined by the number of false alarms and missesand is interpreted as how confident a participant needs to feel about havingseen an item before being willing to say the item was seen before. Signaldetection analysis allows researchers to attribute the percentage of correctmemories to either a sensitive memory or a willingness to guess.

13


Procedure

Informed consent was obtained from all participants. The informed-consentform stated that the purpose of the study was to learn more about how peoplelearn from mediated messages. It also informed participants of the place-ment of electrodes for the collection of physiological data. Each participantwas paid $10 for partaking in the experiment.

Participants viewed the news stories sitting approximately 5 feet from a20-inch color television. Prior to viewing the news stories, five Beckman AG/AGCL standard electrodes were attached to each participant’s forearm andpalm of their nondominant hand for the collection of physiological data.Directions for participating in the experiment were recorded onto the audiotrack of the videotapes containing the stimuli. These narrated instructionspreceded the stimuli material, explained the self-report scales used in theexperiment, and directed participants between stories to complete a ques-tionnaire containing the evaluative and SAM scales.2 Participants firstviewed a practice story about riverboat gambling. Once the participants indi-cated that they understood the self-report scales, they were shown the eightstimulus stories. After completing the scales for the last news story, the elec-trodes were removed and participants were given a recognition test. Theywere then debriefed and thanked for their time.

Results

Physical Access

The first research question asked whether physical access to television newsdiffered by education level. From the items measuring how often participantswatched local and national broadcasts, as well as news magazine programs,no significant differences emerged between the lower and higher educationgroups. Yet, the means for the lower education group (local news, M = 2.21,SD = 1.03; national news, M = 1.84, SD = 0.96; news magazine programs, M =1.17, SD = 0.69) were higher than for the higher education group (local news,M = 1.7, SD = 0.98; national news, M = 1.8, SD = 1.06; news magazine pro-grams, M = 0.9, SD = 0.55) across all three news genres. It is thereforeunlikely that participants in the lower education group had less physicalaccess to television or that they were less familiar with television news thanthe higher education group.

14


Cognitive Access

Arousal. The second research question asked whether arousal responsesto television news stories differed by education level. Arousal was measuredboth by viewers’ self-reported arousal ratings and the frequency of nonspe-cific skin conductance responses during the messages. As Table 1 shows,there was a significant main effect for education on the skin conductancedata, F(1, 30) = 6.00, p < .02, e2 = .09. The lower education group had fewerskin conductance responses during the collection period (M = 1.29, SD =2.18), indicating lower arousal than the higher education group (M = 2.28,SD = 2.35).

The main effect for education on the self-report data was not significant,F(1, 37) = 1.20, p < .28, although the mean ratings were in the same direc-tion as the skin conductance data, with the lower education group reportingless arousal (M = 5.1, SD = 1.95) than the higher education group (M = 5.7,SD = 1.13).

Attention. The third research question assessed the motivation of partici-pants to attend to messages. It asked whether evaluations of the informative-ness and enjoyment of negatively arousing news stories varied by educationlevel. The main effect (see Table 1) of education on the enjoyment rating wassignificant, F(1, 37) = 5.07, p < .03, e2 = .07, with the higher educated groupreporting greater enjoyment of the stories (M = 4.09, SD = 1.64) than thelower educated group (M = 2.91, SD = 1.61). There were no significant effectsfor education on the informativeness scale.

The fourth research question asked whether attention to negativelyarousing news stories also varied by education. Heart rate was used as ameasure of attention. The greater the decrease in heart rate during viewing,

15


Table 1Summary of Significant Results

Education Level

Dependent High School SomeVariable or Less Graduate School F df p e2

Skin conductance 1.29 2.28 6.00 1, 30 .020 .09Accuracy of

recognition 62.00% 68.00% 10.11 1, 32 .003 .19d prime 0.695 1.05 3.40 1, 38 .004 .18Enjoyment 2.91 4.09 5.07 1, 37 .030 .07

the greater the attention of the viewer. The main effect for education on theheart rate data was not significant, F(1, 27) = 1.19, p < .28. Although therewas a significant main effect for time, F(7, 189) = 16.46, p < .000, e2 = .35,there was no interaction between time and education. Thus, both groupsshowed the characteristic significant decrease in heart rate indicative ofattention to the stories, but there was no difference between the high- andlow-education groups in the amount of attention paid to the stories.

Encoding of information. The fifth research question asked if there are dif-ferences in how people with higher and lower education encode informationfrom the news. The percentage of accuracy for the two groups on the recogni-tion test was used to address this research question. The main effect for edu-cation was significant, F(1, 32) = 10.11, p < .003, e2 = .19. The higher educa-tion group’s mean accuracy was 68% (SD = 6.00) compared to the lowereducation group’s mean accuracy of 62% (SD = 7.00).

The accuracy of recognition results is useful as an indication of memory.Yet, it assumes that recognition memory for information is either at hand ornot, when many times people are not sure if they remember something(Shapiro, 1994). The final research question therefore probed more detailabout how people from higher and lower educational backgrounds encodeinformation. Signal detection analysis enables a sophisticated measure of

16


Figure 1. Education and Attention

information encoding by revealing memory strength (d prime) and judg-ments related to recalling information (criterion bias). Signal detectionanalysis for the recognition results showed a significant main effect for edu-cation on d prime or memory sensitivity, F(1, 38) = 9.40, p < .004, e2 = .18, butno significant main effect for education on criterion bias. Higher educationgroups had higher sensitivity scores (M = 1.05, SD = 0.32) than the lower edu-cation group (M = 0.69, SD = 0.41).

These signal detection results suggest that participants in the higher edu-cation group had better memory for information presented during the newsstories, and that this difference in memory was due to greater memory sensi-tivity, not to a difference in their willingness to guess. The two groups did notdiffer in how certain they had to feel about an item before judging whetherthey had seen it before. But, after a single viewing, the higher educationgroup had successfully encoded more of the items at that level of certainty.Thus, the greater accuracy of the higher education group was due to betterrecognition memory for what they had seen.

Discussion

Similar to many survey research studies over the past 30 years, this experi-ment supports the existence of a knowledge gap: People with higher levels ofeducation scored better on the verbal recognition tests than those with lowerlevels of education. Physical access to news did not seem to contribute mark-edly to the information processing disparities found in this study. Peoplefrom lower educational backgrounds had as much daily physical access/expo-sure to television news as people with higher educational backgrounds. Cog-nitive access, or information processing proficiency, emerged as a plausiblecontributing factor to knowledge gaps caused by the media. Specifically, test-ing the operation of three information processing mechanisms (arousal,attention, and encoding of information) in people from higher and lower edu-cational backgrounds produced some insights into the relationship betweeneducation level and information gain. Figure 2 summarizes the variablesincluded in this study as well as a number of variables common to surveystudies of the knowledge gap. The diagram presents cognitive access, whichcomprises information-processing aptitude, as a central explanation for howknowledge gaps emerge. Physical access and characteristics of the messageand audience are illustrated as interconnected, contributing sources of vari-ance in cognitive access and resulting knowledge inequities.

Findings related to the notion of cognitive access deserve specific consid-eration. Skin conductance data revealed a difference in physiological arousalduring news viewing between the two education groups. In particular,

17


participants with more education showed greater sympathetic nervous sys-tem activation while watching news stories than those with less education.Because sympathetic activation is generally associated with increases in theallocation of resources to both encoding and storage, this increased arousalmay be partially responsible for the higher education group’s superior recog-nition of information presented in the news stories.

It remains unclear why people with more education experienced greaterarousal during news viewing, but at least two possible explanations could beconsidered. First, the topics in the news stories might simply be more inter-esting and arousing to people with higher education than they were to thosewith lower education. This explanation is possible, but not likely. Most of thestimulus stories featured crime, disasters, and conflict, topics that can beexpected to arouse participants from both education groups. Second, and

18


Figure 2. Explanatory Diagram for the Knowledge Gap Phenomenon* Variables investigated in the study reported here.

theoretically perhaps more plausible, is the explanation that those withhigher education have the ability to mobilize their sympathetic nervous sys-tems in support of cognitive tasks. Such a faculty could either be learned orinnate, conscious or subconscious, but it might predispose a person to pursuehigher education. The idea that people mobilize physiological resources withvarying levels of efficiency to support cognitive functions deserves moreresearch attention.

From the standpoint of evolutionary psychology, it is particularly worthnoting that the biological systems of people with more education preparedthem better for the surveillance of negatively compelling, and thereforepotentially threatening, audiovisual information than the biological systemsof people with less education. It appears that people with higher educationlevels display physiological characteristics that as Shoemaker (1996) notes,may have benefited watchmen in primitive societies who surveyed the envi-ronment and disseminated information to other tribe members in advance ofan oncoming survival threat. Interestingly, higher educated participantsalso reported greater enjoyment of news stories. Perhaps this is a matter ofthe functional division of labor; people with better cognitive skills enjoy andexcel at fulfilling the ancient surveillance role in human societies, whereasthose with less-developed cognitive skills tend to contribute to society onother levels such as physical labor or hunting.

Results of the heart rate analysis provide evidence that education did notaffect attention to the news stories. Participants in both education groupspaid equal amounts of attention to the messages. The heart rate curve,shown in Figure 1, indicates a relatively large decrease in cardiac activityduring news viewing, accounting for 35% of the variance in the heart ratedata. This deceleration confirms that participants in both groups were pay-ing close attention to the stories during viewing.

Taken together, these results suggest that attention to messages mightnot be a fundamental cause of information inequities. This does not meanthat varying levels of controlled attention paid to information could not con-tribute to the knowledge gap phenomenon. Rather, when people are equallymotivated to pay attention (e.g., by viewing messages that compel attention)those with more and less education pay the same, relatively high, level ofattention to the messages.

Also important to note is that these findings do not bolster explanations ofthe knowledge gap as a result of differing levels of interest and perceived use-fulness of media information. Ettema and Kline (1977), as well as Ettemaand colleagues (1983), argue that televised public affairs issues are not socomplicated as to necessitate highly developed information processing skillsbut that these issues are typically more important and useful to people in

19


higher socioeconomic groups. They therefore conclude that the knowledgegap exists in part because of differing levels of interest and use for informa-tion. Likewise, Donohue and colleagues (1975) conclude that knowledge gapsabout local news issues tend to narrow because these issues are typicallyabout conflict and therefore salient to people from all levels of the socioeco-nomic hierarchy. This study did not use public affairs or national stories asexperimental stimuli. Instead, participants were shown negatively compel-ling local news stories that delivered survival-relevant information (crime,disaster, and combative protests) considered salient to all members of societyand therefore likely to command the attention and thus the automatic alloca-tion of processing resources of viewers (Newhagen & Reeves, 1992). Eventhese stimuli, with equal salience to people in the lower and higher educationgroups, yielded a significant knowledge gap.

The results of recognition data demonstrate three things. First, partici-pants with more education remembered more facts from the news stories.Second, participants in both education groups used the same criteria todecide whether they recognized a test item. Third, participants with moreeducation had more sensitive memories than less educated participants.Together, these results suggest that people with more education are qualita-tively and quantitatively better at encoding information from audio-visualmedia than participants with less education. Signal detection analysis wasinstrumental in explaining why participants with higher education levelsfared better in the information recognition tests. They did not outscore thelower education group because they were more willing to guess if they recog-nized information; rather, they remembered more because they encodedinformation more effectively.

The results of this investigation do not give credence to the argument thattelevision, because it is supposedly more accessible and easier to process thannewspapers, has the potential to be the great leveler of the knowledge gap(Brantgarde, 1983; Chew & Palmer, 1994; Galloway, 1977; Gantz, 1978;Kleinnijenhuis, 1991; Neuman, 1976; Tichenor et al., 1970). In this experi-ment, participants saw compelling television messages with universal rele-vance. Results reveal a significant knowledge gap in recall of that informa-tion. An important counterargument to consider is that the mediumgenerally believed to pose relatively small challenges to the cognitive pro-cessing system might in fact require more developed cognitive skills to suc-cessfully encode information than previously believed.

On a methodological note, this study took the first experimental step inexamining the knowledge gap hypothesis. The survey method, which hasproduced a valuable body of research on the knowledge gap phenomenon, hasnoteworthy limitations. Namely, respondents in survey studies often

20


deliberately conceal honest responses to questions due to social desirabilitybiases or are simply not able to accurately explain or assess their own mediaconsumption patterns (Messaris, 1977). The physiological measures used inthis experiment provided relatively untainted data about people’s arousal,attention, and encoding of information associated with media exposure. Onthe other hand, like all experiments, the one reported here is vulnerable tothe standard criticisms of the experimental method.

Experiments are more effective for testing theoretical ideas than for mak-ing population estimates. This study identified differences in the informationprocessing mechanisms of people with different levels of education, but gen-eralizing these findings to a large population would be inappropriate. Theorigin of information processing differences between people from varyingeducational backgrounds remains unexplained. Because it is unclearwhether a viewer’s cognitive-access aptitude—especially his or her arousal toand encoding of information in a news message—is innate or learned, it isimpossible to make recommendations about how people in lower educationgroups can become more empowered by both survival-relevant and public-affairs news. This matter takes on a certain urgency for communicationresearch, given that the World Wide Web is predicted to replace television asthe medium most burdened with the mass dissemination of information.

To date, policy makers and media researchers have been preoccupied withcitizens’ physical access to information technology (Kling, 1999; National Tele-communications and Information Administration, 1998). Perhaps it is equallyimportant to understand how obstacles to cognitive access might widen theknowledge gap that the Internet and World Wide Web seem likely to perpetu-ate. At the same time, the capacity to efficiently process mediated informa-tion in large volume might increasingly become the means whereby literatesocieties stratify power. As Tichenor and colleagues (1970) remarked whenthey formulated the knowledge gap hypothesis, “ . . . more highly educatedpersons are at the vanguard of social and technological change, their acceler-ated acquisition of mediated knowledge may be socially functional” (p. 170).

Notes

1. In the case of the Internet, researchers are now considering cognitive access as afactor in usability studies that typically assess the effectiveness, efficiency, and satis-faction of online experiences. Kling (1999) reports that ordinary people consideraccessing information on the Internet as “too hard,” whereas Borgman (1989) foundthat choice of academic major (e.g., engineering, psychology, English) plays a signifi-cant role in online information retrieval abilities.

21


2. The order of the evaluative measures and Self-Assessment Mannequin scaleswere alternated across stories.

References

Basil, M., Schooler, C., & Reeves, B. (1991). Positive and negative politicaladvertising: A study of television commercials. In F. Biocca (Ed.), Televi-sion and political advertising, Volume 1, Psychological processes(pp. 197-220). Hillsdale, NJ: Lawrence Erlbaum.

Beltran, L. R. (1975). Research ideologies in conflict. Journal of Communica-tion, 15(2), 187-193.

Bennett, S. E. (1988). Know-nothings revisited: The meaning of politicalignorance today. Social Science Quarterly, 5(69), 476-490.

Bogart, L. (1981). Press and the public: Who reads what, when, where, andwhy in American newspapers. Hillsdale, NJ: Lawrence Erlbaum.

Borgman, C. L. (1989). All uses of information retrieval systems are not cre-ated equal: An exploration into individual differences. Information Pro-cessing & Management, 25(3), 237-251.

Bradley, M. M., Greenwald, M. K., & Hamm, A. O. (1993). Affective pictureprocessing. In N. Birbaumer & A. Ohman (Eds.), The Structure of emotion:Psychophysiological, cognitive, and clinical aspects (pp. 48-65). Toronto,Canada: Hogrefe and Huber Publishers.

Bradley, M. M., Greenwald, M. K., Petry, M. C., & Lang, P. J. (1992). Remem-bering pictures: Pleasure and arousal in memory. Journal of Experimen-tal Psychology, 18(2), 379-390.

Brantgarde, L. (1983). The information gap and municipal politics in Swe-den. Communication Research, 10, 357-373.

Cacioppo, J. T., & Tassinary, L. G. (1990). Principles of psychophysiology:Physical, social, and inferential elements. New York: Cambridge Univer-sity Press.

Chew, F. P., & Palmer, S. (1994). Interest, the knowledge gap, and televisionprogramming. Journal of Broadcasting and Electronic Media, 38,271-287.

Childers, T., & Post, J. A. (1975). The information-poor in America. Metu-chen, NJ: Scarecrow.

Craik, F. I., & Lockhart, R. S. (1972). Levels of processing: A framework formemory research. Journal of Verbal Learning and Verbal Behavior, 11,671-684.

Deli Carpini, M. X., & Keeter, S. (1994). The public’s knowledge of politics. InK. J. David (Ed.), Public opinion, the press, and public policy (pp. 19-40).Westport, CT: Greenwood.

Detenber, B., & Reeves, B. (1996). A bio-informational theory of emotion:Motion and image size effects on viewers. Journal of Communication,46(3), 66-84.

Donohue, G. A., Olien, C. N., & Tichenor, P. J. (1987). Media access andknowledge gaps. Critical Studies in Mass Communication, 4, 87-92.

22


Donohue, G. A., Tichenor, P. J., & Olien, C. N. (1975). Mass media and theknowledge gap: A hypothesis reconsidered. Communication Research, 2,3-23.

Drew, D., & Grimes, T. (1987). Audio-visual redundancy and TV news recall.Communication Research, 14(4), 452-461.

Ettema, J. S., Brown, J. W., & Luepker, R. V. (1983). Knowledge gap effects ina health information campaign. Public Opinion Quarterly, 47, 516-527.

Ettema, J. S., & Kline, F. G. (1977). Deficits, differences and ceilings: Contin-gent conditions for understanding the knowledge gap. CommunicationResearch, 4, 179-202.

Frazier, P. J. (1986). Community conflict and the structure and social distri-bution of opinion. Unpublished doctoral dissertation, University of Min-nesota, Minneapolis.

Fredin, E., Monnett, T. H., & Kosicki, G. M. (1994). Knowledge gaps, sociallocators, and media schemata: Gaps, reverse gaps, and gaps of disaffec-tion. Journalism Quarterly, 71, 176-190.

Galloway, J. J. (1977). The analysis and significance of communicationeffects gaps. Communication Research, 4, 363-386.

Gandy, O., Jr., & El Waylly, M. (1985). The knowledge gap and foreign affairs:The Palestinian-Israeli conflict. Journalism Quarterly, 62, 777-783.

Gantz, W. (1978). How uses and gratifications affect recall of television news.Journalism Quarterly, 55, 664-672, 681.

Gaziano, C. (1983). The knowledge gap: An analytical review of media effects.Communication Research, 11, 412-423.

Gaziano, C. (1997). Forecast 2000: Widening knowledge gaps. Journalism &Mass Communication Quarterly, 74(2), 237-264.

Geiger, S., & Newhagen, J. (1993). Revealing the black box: Informationprocessing and media effects. Journal of Communication, 43(4), 42-50.

Genova, B.K.L., & Greenberg, B. S. (1979). Interests in news and the knowl-edge gap. Public Opinion Quarterly, 43, 79-91.

Gorman, T. J. (1998). Social class and parental attitudes toward education:Resistance and conformity to schooling in the family. Journal of Contem-porary Ethnography, 27(1), 10-44.

Graber, D. A. (1990). Seeing is remembering: How visuals contribute tolearning from television news. Journal of Communication, 40(2), 134-155.

Graber, D. A. (1993). Mass media and American politics. Washington, DC:CQ Press.

Griffin, T. (1990). Energy in the eighties: Education, communication and theknowledge gap. Journalism Quarterly, 67, 554-566.

Gunter, B. (1987). Poor reception: Misunderstanding and forgetting broad-cast news. Hillsdale, NJ: Lawrence Erlbaum.

Gurevitch, M., & Levy, M. (1986). Information and meaning: Audience expla-nations of social issues. In J. Robinson & M. Levy (Eds.), The main source:Learning from television news (pp. 159-175). Newbury Park, CA: Sage.

Hopkins, R., & Fletcher, J. E. (1994). Electrodermal measurement: Particu-larly effective for forecasting message influence on sales appeal. InA. Lang (Ed.), Measuring psychological responses to media (pp. 113-132).Hillsdale, NJ: Lawrence Erlbaum.

23


Horstmann, R. (1991). Knowledge gaps revisited: Secondary analyses fromGermany. European Journals of Communication, 6, 77-93.

Kinloch, G. C. (1987). Social class and attitudes toward education. Journal ofSocial Psychology, 127(4), 399-401.

Kleinnijenhuis, J. (1991). Newspaper complexity and the knowledge gap.European Journal of Communication, 6, 499-522.

Kling, R. (1999). What is social informatics and why does it matter? D-LibMagazine, 5(1) [Online journal]. Available: http://www.dlib.org/dlib/january99/kling/01kling.html

Kwak, N. (1999). Revisiting the knowledge gap hypothesis. CommunicationResearch, 26(4), 385-413.

Lang, A. (1990). Involuntary attention and physiological arousal evoked bystructural features and mild emotion in TV commercials. CommunicationResearch, 17(3), 275-299.

Lang, A. (1994). What can the heart tell us about thinking? In A. Lang (Ed.),Measuring psychological responses to media messages (pp. 99-112). Hills-dale, NJ: Lawrence Erlbaum.

Lang, A., & Basil, M. (1998). Attention, resource allocation, and communica-tion research: What do secondary task reaction times measure anyway? InM. Roloff (Ed.), Mass communication yearbook (Vol. 21, pp. 443-474). Bev-erly Hills, CA: Sage.

Lang, A., Bolls, P., Potter, R., & Kawahara, K. (1999). The effects of produc-tion pacing and arousing content on the information processing of televi-sion messages. Journal of Broadcasting and Electronic Media, 43, 451-475.

Lang, A., Dhillon, K., & Dong, Q. (1995). The effects of emotional arousal andvalence on television viewers’ cognitive capacity and memory. Journal ofBroadcasting and Electronic Media, 39(3), 313-327.

Lang, A., & Friestad, M. (1993). Emotion hemispheric specialization and vis-ual and verbal memory for television messages. CommunicationResearch, 20(5), 647-670.

Lang, A., Newhagen, J., & Reeves, B. (1996). Negative video as structure:Emotion, attention, capacity, and memory. Journal of Broadcasting andElectronic Media, 40, 460-477.

Lang, A., Zhou, S., Schwartz, N., Bolls, P. D., & Potter, R. F. (in press). Whenan edit is an edit can an edit be too much? The effects of edits on arousal,attention, and memory for television messages. Journal of Broadcastingand Electronic Media.

Lang, P. J. (1985). The cognitive psychophysiology of emotion: Fear and anxi-ety. In A. Tuma & J. Maser (Eds.), Anxiety and the anxiety disorders(pp. 151-170). Hillsdale, NJ: Lawrence Erlbaum.

Lang, P. J., Greenwald, M., Bradley, M. M., & Hamm, A. O. (1993). Lookingat pictures: Evaluative, facial, visceral, and behavioral responses. Psycho-physiology, 30, 261-273.

Lombard, M., Reich, R. D., Grabe, M. E., Campanella, C. M., & Ditton, T. B.(2000). The impact of television: Screen size and viewer responses topoint-of-view movement. Human Communication Research, 26, 75-98.

24


McLeod, D. M., & Perse, E. M. (1994). Direct and indirect effects of socioeco-nomic status on public affairs knowledge. Journalism Quarterly, 71,433-442.

Messaris, P. (1977). Biases of self-reported functions and gratifications ofmass media use. Et cetera, 34(3), 316-329.

Moore, W. E. (1987). Political campaigns and the knowledge gap hypothesis.Public Opinion Quarterly, 51, 186-200.

National Telecommunications and Information Administration. (1998). Fal-ling through the Net II: New data on the digital divide. [Online report].Available: http://www.ntia.doc.gov/ntiahome/net2/falling.html

Neuman, R. (1976). Patterns in recall among television news viewers. PublicOpinion Quarterly, 40, 115-123.

Newell, A. (1990). Unified theories of cognition. Cambridge, MA: HarvardUniversity Press.

Newhagen, J. E. (1998). TV news images that induce anger, fear, and disgust:Effects on approach-avoidance and memory. Journal of Broadcasting andElectronic Media, 2(2), 265-276.

Newhagen, J. E., & Reeves, B. (1992). The evening’s bad news: Effects of com-pelling negative television news images on memory. Journal of Communi-cation, 42(2), 25-41.

Olien, C. N., Donohue, G. A., & Tichenor, P. J. (1983). Structure, communica-tion, and social power: Evolution of the knowledge gap hypothesis. InE. Wartella & D. C. Whitney (Eds.), Mass communication review yearbook(Vol. 4, pp. 455-462). Beverly Hills, CA: Sage.

Park, E., & Kosicki, G. M. (1995). Presidential support during the Iran-Contra affair: People’s reasoning process and media influence. Communi-cation Research, 22(2), 207-236.

Price, V., & Zaller, J. (1993). Who gets the news? Alternative measures ofnews perceptions and their implications for research. Public OpinionQuarterly, 57, 133-164.

Radio Television News Director’s Association. (1999). Issues and trends.[Online report]. Available: http://www.rtnda.org/issues/survey.htm

Reeves, B., Lang, A., Thorson, E., & Rothschild, M. (1989). Emotional televi-sion scenes and hemispheric specialization. Human CommunicationResearch, 15(4), 493-508.

Reeves, B., Newhagen, J., Maibach, E., Basil, M., & Kurz, K. (1991). Negativeand positive television messages: Effects of message type and messagecontext on attention and memory. American Behavioral Scientist, 34,679-694.

Robinson, J. P. (1967). World affairs and media exposure. Journalism Quar-terly, 44, 23-31.

Rogers, E. M. (1976). Communication and development: The passing of thedominant paradigm. Communication Research, 3, 213-240.

Shanahan, M. J., Miech, R. A., & Elder, G. H. (1998). Changing pathways toattainment in men’s lives, historical patterns of school, work, and socialclass. Social-Forces, 77(1), 231-256.

25


Shapiro, M. A. (1994). Signal detection measures of recognition memory. InA. Lang (Ed.), Measuring psychological responses to media messages(pp. 133-148). Hillsdale, NJ: Lawrence Erlbaum.

Shinghi, P., & Mody, B. (1976). The communication effects gap: A fieldexperiment on television and agricultural ignorance in India. Communi-cation Research, 3, 171-190.

Shoemaker, P. J. (1996). Hardwired for news: Using biological and culturalevolution to explain the surveillance function. Journal of Communication,46(3), 32-47.

Simmons, R. E., & Garda, E. C. (1982). Dogmatism and the “knowledge gap”among Brazilian mass media users. Gazette, 30, 121-133.

Stauffer, J., Frost, R., & Rybolt, W. (1978). Literacy, illiteracy, and learningfrom television news. Communication Research, 5(2), 221-232.

Tichenor, P. J., Donohue, G. A., & Olien, C. N. (1970). Mass media flow anddifferential growth in knowledge. Public Opinion Quarterly, 34, 159-170.

Torsvik, P. (1972). Television and information. Scandinavian Political Stud-ies, 7, 215-234.

U.S. Bureau of the Census (1997). Statistical abstract of the United States.Washington, DC: Author.

Viswanath, K., & Finnegan J. R., Jr. (1996). The knowledge gap hypothesis:Twenty-five years later. Communication Yearbook, 19, 187-227.

Weenig, M. W., & Midden, C.J.H. (1997). Mass-media information cam-paigns and knowledge gap effects. Journal of Applied Social Psychology,27(11), 945-958.

Yeakey, C. C., & Bennett, C. (1990). Race, schooling, and class in Americansociety. Journal of Negro Education, 59(1), 3-18.

Zillmann, D. (1982). Television and arousal. In D. Pearl, L. Bouthilet, &J. Lazar (Eds.), Television and behavior: Ten years of scientific progressand implications for the eighties (pp. 53-67). Washington, DC: Depart-ment of Health and Human Services.

26


cognitive access to negatively arousing news

Documents