Content-Type: text/html Memory Decay and the Agenda-Setting Effect: An Examination of Three News Media The mental processing of information obtained through the mass media is complex and elaborate. Individuals attend to, process and retain a wide variety of information in many different ways. At the same time, the mass media provide individuals with an endless stream of issue coverage. Issues covered today replace issues covered yesterday, and individuals somehow must make sense of this coverage. The end result of this mental processing of issue information is the mass media's agenda-setting effect. Coverage of issues in the mass media gives individuals salience cues by which they judge the perceived importance of these issues. Of concern here is how the rise and fall of coverage of issues on the mass media agenda is related to the public agenda, given the fact that individuals have a limited ability to retain information. Obviously, individuals do not remember everything they read in the newspaper or everything they view on television, especially as new information becomes available to replace old information. In other words, an individual's memory of information gained from the news media decays over time. Watt, Mazza and Snyder (1993) took memory decay into consideration in their investigation of the optimal agenda-setting time-lag. Basing their study on the memory decay curve first investigated by Ebbinghaus (1885), Watt et al. reasoned that the effects of exposure to media messages will decay exponentially over time. Today's news will have a stronger effect on individuals than yesterday's news, and much more influence than news covered days earlier. They found that the memory decay time-lags differed for different issues: from 30 days for Iran to 600 days for inflation. While the agenda-setting effect of various types of issues may decay differently over time, the influence of different media also may differ. For example, the influence of a visual medium, such as television, may decay more rapidly than a verbal medium, such as newspapers. In addition, a national medium, such as a national television newscast, may have a different decay time period than a local medium, such as a local newscast. Thus, the present study will attempt to answer the following question: Given the fact that an individual's memory of information gained from the mass media decays over time, what is the time period in which agenda-setting effects disappear for three different news media? The three media examined are local newscasts, national network newscasts and a local newspaper. The answer to this question has several important implications. First, this study addresses an important topic for research in the general area of media effects. Since agenda-setting is "social learning," how individuals learn about society is an important area of research. Second, this study has important ramifications for agenda-setting research. The results may offer insights into how individuals process and retain information from different media. Third, finding the memory decay time-lag in agenda-setting is imperative for future researchers. It would give researchers guidelines for plotting future studies, especially for studies in which a specific type of news medium is included. In addition, the time-lag for agenda-setting investigations is taking on greater significance in light of recent attempts to employ complex statistical analyses, such as time-series analyses (Zhu, Watt, Snyder, Yan and Jiang 1993) and path analyses (Wanta and Hu, 1994). As statistical analyses increase in complexity, the precision needed for measurements of media content is amplified. Theoretical framework One of the most important considerations that agenda-setting researchers must address is what time frame they will employ in their analyses. In other words, researchers must decide how far back in time they will go to analyze media content prior to their field work. As Winter and Eyal (1981) note, "Since most of these studies measure and compare the media and public agendas over time, the temporal variable would appear to be crucial." Time-lag selection is especially important in agenda-setting research, since studies in this area investigate a causal hypothesis. Chaffee (1972) argues that a time-lag that is too short will not capture the causal relationship, but a time-lag that is too long is also a serious problem because "there is always the danger that a causal effect will 'dissipate' over time if the researcher waits too long to measure it." Salwen (1988) believes the time-lag question is important because researchers need to confine their measures of media coverage to as short a time period as possible because "any time discrepancies in the measurement of the public agenda may affect the public's evaluations of issue salience." Despite the importance of a precise time frame, many discrepancies remain regarding the optimal time period to include in agenda-setting studies. Studies have examined issues in time frames as short as one week (Mullins, 1977; Becker and McCombs, 1977), and as long as nine months (Sohn, 1978). Funkhouser (1973) compared media coverage across a decade with public concern in the same time period. Furthermore, even the few studies that have dealt specifically with the time-lag question have produced inconsistent results. Table 1 lists the results of five studies that specifically examined the optimal time-lag in agenda-setting. The optimal time-lags varied from a four-month period stretching from two to six months before the survey period (Stone and McCombs, 1981) to zero to two weeks before the survey period (Eaton, 1989). The varied results on the optimal agenda-setting time-lag, however, could be due to methodologies employed. Stone and McCombs (1981) examined two news magazines: Time and Newsweek. Winter and Eyal (1981) studied one issue (civil rights) from 1954-1976 and front-page coverage in the New York Times. Salwen (1988) also looked at one issue (the environment) and coverage in the three largest daily newspapers serving Lansing, Michigan. Zucker (1978) investigated four issues across time and coverage on the three national networks. Eaton (1989) examined complete agendas based on bi-weekly data collected from three network broadcasts, four newspapers and three newsmagazines combined. In addition, most of the research dealing with the time-lag question examined one issue, or a series of single issues, across time. But given the large number of "Type I" studies (McCombs, 1981) -- which examine an entire agenda of issues covered by the news media and an entire agenda of issues perceived as important by members of the public -- an investigation comparing complete media and public agendas is sorely needed. This type of examination allows for an investigation of the inter-relatedness of i ssues, an important consideration given the recent research that suggests that issues compete with each other in a "zero-sum" game (Zhu, 1992). Previous research also has noted the possibility of differences between the news media. Several researchers (Tipton, Haney and Basehart, 1975; McClure and Patterson, 1976) have found that newspapers correlate better than television with voter agendas. Shaw and McCombs (1977) argue that television news might have a stronger short-term impact, but newspaper content may have a more consistent effect across longer periods of time. Zucker (1978), on the other hand, argues that at the national level, the public may be more influenced by the three networks' newscasts than by newspapers because of television's accessibility. A number of researchers also have suggested that national news media, to a large degree, set the agenda of issues covered by local media. Breed (1955), for example, suggests that news flows downward from the elite dailies. In other words, small dailies learn coverage patterns from larger newspapers. Crouse (1972) similarly argues that the elite media, such as the New York Times, influence the national agenda. Front-page coverage in the New York Times, he concludes, means prominent coverage in every other paper in America. Thus, if smaller, more localized media react to the national media, the national media may have a more immediate effect on the public. In other words, if local media follow national media coverage, issues in the local press will take longer to reach the general public than ones in the national or regional media -- and consequently, will take longer to influence the public agenda. However, since the national media may first influence local media and then the public, the national media may show a slower decay in effect over time. In addition, national media also devote more coverage to international and national issues than do local media. Indeed, the traditional agenda-setting question, which was employed here, asks respondents "What is the number one problem facing our country today?" Given that this question addresses a national agenda, national media should have a stronger agenda-setting effect than local media, which must devote significant coverage to issues of only local concern. As a result, individuals may demonstrate a slower memory decay for national issues covered by the national media. Added to the mix of factors surrounding the time-lag question is the role of memory decay. Individuals do not remember equally as well media content across several days. Thus, the time period when agenda-setting effects disappear is as important to researchers as a precise optimal time-lag. In other words, at what point do individuals fail to recall the issue information transmitted by the news media and accumulated in media consumers' memories? The study of memory and its decay dates back to before the turn of the century. In a series of experiments conducted by Ebbinghaus (1885), lists of words were learned and relearned on successive days. The amounts of time necessary for successful learning each day were noted and later compared to each other in order to discuss the relationship between information retention and time. From the resulting graph, Ebbinghaus -- and subsequent researchers who further upheld his findings -- concluded that the "main characteristic (of human memory decay) is a rapid fall immediately after learning and a gradual flattening out as the interval is prolonged. Forgetting becomes more and more gradual as time advances (Woodworth and Schlosberg, 1954, p. 726)." Further, when Ebbinghaus plotted his data against the logarithm of time, he observed that "retention declines approximately in proportion to the log of time (Woodworth and Schlosberg, 1954, p. 726)." Watt, Mazza and Snyder (1993) use this exponential expression of memory decay over time when analyzing three issues presented in television newscasts and their agenda-setting effects. In order to account for past coverage of an issue in addition to current stories, Watt et al. modified the Ebbinghaus curve. The resulting formula analyzes memory decay of all issue coverage and its effect on audience issue salience, while also accounting for issue prominence (i.e., where it was placed within the newscasts) and issue obtrusiveness (i.e., personal experience with an issue by audience members). In addition to altering the Ebbinghaus curve, Watt et al. use the concept of a "time window" to account for past, accumulated coverage and current news stories about an issue. These time windows, or data sampling periods of variable length, are used to determine "how long ... people continue to be affected by past stories in the media (Watt et al., 1993, p. 409)." As with Ebbinghaus' memory loss hypothesis, all issue coverage and salience within the time window should proportionately decay. Given the ma thematical fact that an exponential decay never reaches zero, Watt et al. assume an eventual memory loss of 95 percent as the maximum loss of story impact in each time window. Their results show that different issues show their strongest agenda-setting correlations when analyzed within different time windows. Their memory decay time windows ranged from 12-60 days for Iran to 600 days for inflation. Our study builds upon the groundwork laid by Watt et al. (1993). Instead of examining individual issues in time windows of up to two years, however, the present study investigates three different news media for shorter time frames. Logically, a news report should not have an influence on individuals 600 days later -- as found by Watt et al. Our study examines daily coverage for six months for two news media and for 50 days for a third. Method A telephone survey was conducted in a U.S. city in February 1994. The area surveyed has a population of approximately 200,000 and is home to a large state university. Interviewers were students at this university. The response rate was 60 percent. The public agenda was determined by responses to the traditional agenda-setting question: "What is the number one problem facing our country today?" The 12 issues that were mentioned most often by respondents were included in the study. The percentage of respondents who mentioned each issue as being the most important problem determined where the issues ranked on the public agenda. The issues and the percentage of respondents naming them as the number one problem facing our country are listed in Table 2. Coverage of the 12 issues were then examined in three news media: the national broadcasts of ABC World News Tonight, the local broadcasts of the evening news for the station with the highest ratings in the area surveyed, and a local morning daily newspaper serving the area. The ABC broadcasts were coded through the Vanderbilt Television News Abstracts for the six months preceding the first day of our survey. In a few cases, always on either a Saturday or Sunday, ABC did not have a national broadcast. To maintain consistent time intervals, a broadcast of either CBS or NBC was coded on these days. The local television broadcasts were coded through station logs for the 50 days preceding the first day of the survey. Only 50 days of logs were available for this study. Finally, the front pages of the main news section were coded for the local daily newspaper for the six months preceding the first day of our survey. Intercoder reliability using Scott's pi averaged .89. Each news story was weighted according to the distance in time from the date of broadcast to the start of the survey period. The weights were determined by the memory decay curve proposed by Ebbinghaus (1885). (For a more detailed description of the memory curve developed by Ebbinghaus, see Woodworth and Schlosberg, 1954). The weights are based on two main assumptions: 1. Individuals eventually retain only about five percent of all knowledge that they learn. In other words, 95 percent of an individual's memory decays. 2. Memory decay is not linear. That is, memory decays rapidly at first, then slows. The effects of the accumulated weighted coverage were then examined for each day of the study. For example, for Day 1, the day before the beginning of the survey period, the coverage of issues televised on that day was weighted by .05, again, under the assumption that memory decays 95 percent. For Day 2, the coverage of the day before the survey was weighted by .10 and coverage of two days before the survey was weighted by .05. These weighted scores were then summed. For Day 3, weights ranged from .12 to .08 to .05. Each subsequent day was weighted similarly, so that coverage the farthest from the survey period was weighted by .05 and the other dates were weighted based on the Ebbinghaus curve. Thus, the coverage closest to our survey period received a heavier weight than coverage farther back in time. Indeed, if Ebbinghaus' memory decay curve is correct, information gained recently should be recalled more efficiently than older information and thus should have a stronger agenda-setting effect on individuals. Spearman rank-order correlations were then computed examining the relationships between the accumulated weighted coverage of the 12 issues for each day in our analysis and the public agenda. In other words, in the case of ABC newscasts, Spearman rhos examined the agenda-setting effect on the public for time-lags ranging from one day before our survey period to 180 days before our survey period. If agenda-setting effects are most pronounced after one day of coverage, for example, the Spearman rho for one day of coverage and the public agenda will be the largest in the study. The Spearman rhos also will suggest where the public's memory decay occurs. If the Spearman rhos are no longer statistically significant after 30 days, for example, the results will suggest that the public's memory of issue information decays after 30 days. Methodological strengths and weaknesses Several methodological aspects are important here. First, the methodology here employs an entire agenda of issues. This was the presupposition underscoring the original agenda-setting hypothesis -- that an "agenda" of issues in the news media would influence an "agenda" of issues that the public perceives as important. Thus, this analysis returns to the area proposed in the original agenda-setting hypothesis. Second, the analysis allows for the examination of several news media. Logically, information from different media will be processed differently among individuals. The passive processing of visual information from television should produce results that are different from the active processing of verbal information from newspapers. Local information also should be processed differently than national information. Third, the analysis allows for the examination of several memory decay time lags. In other words, our analysis allows us to pinpoint to the day when the agenda-setting effect eventually decays in our study -- from one day to 180 days. The major shortcoming of this study is that the analysis, by design, uses aggregate data -- that is, data from an entire "public" rather than from individuals. Thus, the memory decay examined here is not an individual's decay, but rather a decay from an entire population of individuals. Indeed, mental processes take place within individuals, not within a mass of individuals. However, agenda-setting is a societal effect (see Lowery and DeFleur, 1988). Thus, memory decay, while taking place within indivi duals, should be apparent in an analysis of a population of individuals in which this process is occurring. Results Spearman rank-order correlation coefficients for the three news media in our study are listed in Tables 3 to 6 and are plotted in Figures 1 to 4. As Figure 1 and Table 3 detail, the ABC news broadcasts did not produce any statistically significant Spearman rhos for the 12 issues in our study. In other words, ABC news apparently did not have an agenda-setting effect on the respondents in our study. This was contrary to what we initially believed. We originally thought that national media would best match the public agenda because the agenda-setting question asks what the respondent believes is the most important problem facing "our country" today. The local media's concentration on local issues, we thought, would lessen their agenda-setting effect. A secondary analysis of our data revealed that one issue -- international problems -- received an extensive amount of coverage, but ranked low on the public agenda (a tie for tenth). Indeed, the nature of national network news implies that the networks provide extensive coverage of news around the world, which apparently did not catch the attention of respondents in our study. Despite the quandary caused by the issue of international problems, several trends are apparent. As Figure 1 and Table 3 show, the Spearman rank-order correlations for the national news broadcasts of ABC reached one of their highest levels at Day 4 and Day 5 (r = .32), before decreasing. Besides the first few days of the study, the lowest rho occurred after 14 days. The rhos showed a large increase again at Day 38 ( r = .36) and reached their peak at Day 62 (r = .39). They remained at the r = .39 level until Day 92, when they dropped to r = .36 and remained at this point until the end of the study (Day 180). To further examine the network television news-public relationship and to guard against the results of our study being suppressed because of coverage patterns for this one issue, we re-analyzed the data after dropping international problems from the analysis. The results of the 11 remaining issues mirrored the results of the original 12, except for the fact that statistically significant findings resulted at several points. As Figure 2 and Table 4 show, the Spearman rhos reached statistical significance after only four days (r = .63), and dropped below the p < .05 level of statistical significance after Day 6. The lowest rho was again at Day 14 (r = .41). The rhos reached statistical significance again at Day 38 (r = .65) and reached their peak at Day 62 (r = .68), before decreasing slightly at Day 92. Figure 3 and Table 5 detail the results of the Spearman rhos for local television broadcasts. The rhos show several differences from the rhos for national news. Here, the Spearman rhos reached statistical significance at Day 3 (r = .63) and peaked at Day 6 and Day 7 (r = .67). The rhos then decreased slightly, dropping below the level of statistical significance at Day 11. The rhos again reached statistical significance from Day 15 to Day 17 and again at Day 20. All of the other rhos were not statistically significant at the p < .05 level. In addition, the rhos were unchanged from Day 42 through the end of our study period at Day 50 (r = .51). The Spearman rhos again were different for the local newspaper. As Figure 4 and Table 6 show, the rhos reached statistical significance at Day 8 (r = .62). The rhos reached significance at the p < .01 level at Day 10 (r = .71), and peaked at both Day 15 and Day 18 (r = .81). The rhos dropped back to the p < .05 level at Day 28 (r = .67), before leveling off at Day 34 through Day 59 (.61). The rhos were no longer statistically significant at Day 84 (r = .54). The rhos were unchanged from Day 88 through the end of our study (r = .51). Discussion The purpose of this study was to examine how the agenda-setting effect dissipates over time as individuals' memory of issue information decays. The findings are summarized in Table 7. The results show how differently memory decay affects the agenda-setting process for the three news media. A few similarities between the three media did emerge. First, agenda-setting effects were evident early for all three media. Effects appeared after four days for national network news (though only after the issue of international problems was dropped from the analysis), after six days for local news and after eight days for the local newspaper. Second, the Spearman rank-order correlations showing agenda-setting effects dropped below statistically significant levels relatively quickly for all three news media. The rhos were no longer statistically significant after six days for ABC news and after 11 days for local newscasts. While the rhos for the local newspaper were statistically significant for a much longer time -- until Day 85 -- the rhos did drop below the p < .01 level at Day 28. Thus, memory of issue information for local newspapers did decay slightly after four weeks. All of these results seem logical on the surface. Since individuals' long-term memory is limited, a memory decay of a few weeks after an initial agenda-setting effect of four to eight days appears plausible. However, the agenda-setting effects, as demonstrated by the Spearman rank-order correlations, showed several differences for the three news media. Especially evident were differences in long-term memory decay between the three news media examined here. The agenda-setting effect for the local newspaper, for example, first appeared on Day 8 and peaked at Day 15 and Day 18. The agenda-setting effect decayed slightly at Day 28, and did not completely disappear until Day 85. Thus, the optimal time-lag for agenda-setting effects to occur was longer for the local newspaper than for either local or national newscasts. Moreover, the local newspaper produced the strongest Spearman rank-order correlation in the study (r = .81), indicating that in our study, newspapers had a more powerful agenda-setting effect than television, as both Tipton, Haney and Basehart (1975) and McClure and Patterson (1976) found. The agenda-setting effect also decayed much more gradually for the local newspaper, since the agenda-setting effect for both local and national broadcasts decayed only a short time after an early initial effect. This finding supports the argument of Shaw and McCombs (1977) that television news has a stronger short-term impact, but that newspapers have an effect across longer periods of time. The agenda-setting effect decayed at much different rates for both local and national newscasts. The agenda-setting effect for local news broadcasts disappeared in Day 11, though there was a minor effect for Days 15-17 and 20. The agenda-setting effect disappeared completely by Day 21. For national news, the agenda-setting effect was not evident at all, until the issue of international problems was removed from the study. Then, the agenda-setting effect was evident for Days 4 and 5, but decayed by Day 6. However, the agenda-setting effect re-emerged at Day 38, then remained through the end of our study period, 180 days in all. Two important points should be made about the results for the national newscasts. First, while an agenda-setting effect occurred early (Day 4 and 5), a consistent agenda-setting effect didn't appear until Day 38, or more than five weeks before the survey period. Thus, the accumulated coverage of five weeks led to the most consistent agenda-setting effect for the ABC newscast. In addition, the optimal time-lag for the strongest agenda-setting effects -- when memory decay was accounted for -- was between 62 and 92 days, or about two to three months before the survey period. Second, individuals' memory decay for national news broadcasts was very gradual. In fact, our findings suggest that the public's long-term memory decay of issue information went beyond the six months of news coverage examined in the present study. One other possible explanation about the memory decay time period for the national network news should be noted: Perhaps agenda-setting effects really peaked at between four and five days, and memory of the issues decayed at Day 6. The consistent agenda-setting effect found after Day 38, then, may be due to the fact that coverage patterns finally stabilized here. Indeed, the media agenda changed little from Day 38 to Day 62 and did not change at all from Day 62 through Day 92 and from Day 92 through Da y 180. It seems eminently logical that individuals' abilities to recall news coverage of issues should not last for two to three months. One week seems like a more logical time period for agenda-setting effects to occur, and two weeks seems like a logical time period for the memory of issue information to decay. The results of local television news coverage, meanwhile, appear more easily decipherable. Here, the optimal time-lag for the strongest agenda-setting effects to occur was six to seven days -- or about one week. The memory decay of issue information followed one week later -- with the Spearman rank-order correlations reaching one of their lowest points on Day 14. (when r = .46). While agenda-setting effects again reached statistical significance on Day 15, none of the Spearman rhos after Day 21 were statistically significant. Thus, taken as a whole, both local and national television news media and the local newspaper produced agenda-setting effects in a relatively short time before our survey period. The memory of the issue information gained by respondents in our study also decayed in a relatively short time -- much shorter than the Watt et al. (1993) study found. The long-term memory decay of agenda-setting effects, however, showed a wide discrepancy between the results of the local and national newscasts. This discrepancy, as with the lack of agenda-setting effects found for ABC news, again could be due to the types of coverage patterns of national news. Besides the extensive coverage of international news, national network news also may be more consistent in its coverage of all national issues, while local news covers a wider range of issues. Since network new s covers a more consistent agenda of issues, this agenda eventually reaches a point where it stagnates -- where issues no longer go up or down on the agenda. In other words, the differences between issue ranks reaches a stage where only a major event that receives a staggering amount of coverage can more an issue up past another issue. This point of stagnation may have occurred after two to three months of our study. Local news, on the other hand, may demonstrate more flexibility in its coverage of issues. Thus, the agenda of issues covered on local newscasts may not stagnate as the agenda for national news might. Overall, it should be noted that the findings here should be tempered. Historical factors obviously could cause time-lag and memory decay differences at any given time. As Stone and McCombs (1981) conclude, "It cannot be assumed that a neat pattern will be established in any single specific week during which the content of the news media will yield the same, or a highly similar, pattern of salient issues corresponding to the public agenda." The optimal time-lags and memory decay time periods found here, then, could be unique to the present study. Regardless, the results here point to a potential fruitful area for future research. Researchers should examine why agenda-setting effects seem to drop off after a certain amount of time. Apparently, mental processing of the news may be limited by time. As Zucker (1978) concluded, old news may be forgotten. Mental processing of the news obviously is very complex. Overall, it is hoped that the findings here will offer future agenda-setting researchers a guide to follow when plotting time-lags between survey periods and media coverage. The results demonstrate that memory decay can vary across different media, and these differences should be noted when determining time frames for agenda-setting studies. Greater precision in measuring time-lags and memory decay time periods is imperative for future agenda-setting research. Table 1. Previous studies examining the optimal time-lag for agenda-setting effects to occur. Researchers News media Optimal time-lag Eaton (1989) ABC, NBC, CBS, zero to two weeks New York Times, Los Angeles Times, Chicago Tribune, Wall Street Journal, Time, Newsweek, U.S. News & World Reports Stone and Time and four-month period McCombs (1981) Newsweek stretching from two to six months before the survey period Salwen (1988) Three largest five to seven weeks dailies serving Lansing, Mich. Winter and New York Times four-to-six week Eyal (1981) period immediately prior to fieldwork Zucker (1978) Three national two to six weeks networks before the survey period Table 2. Responses to "most important problem" question. Issue Percent naming it number one problem 1. Crime 40.5 2. Health care 13.5 3. Economy 12.2 4. Dissatisfaction with government 6.3 5. Social problems 5.4 6. Budget deficit 5.0 7. Poverty/homelessness 4.5 8. Morality 3.6 9. Human rights 2.7 10. (t) International problems 2.3 10. (t) Drug abuse 2.3 12. AIDS 1.8 Table 7. Optimal time-lags and memory decay time periods for three news media. Medium Optimal Memory decay time-lags time periods National network 4-5 days & 14 days & news broadcast 62-91 days beyond 180 days Local news 6-7 days 14 days & broadcast 21 days Local 15 & 18 days 28 days & newspaper 84 days References Becker, Lee B. and McCombs, Maxwell E. (1977), "U.S. Primary Politics and Public Opinion: The Role of the Press in Determining Voter Reactions," paper presented to the International Communication Association annual conference, Berlin, W. Germany. Breed, Warren (1955), "Newspaper Opinion Leaders and the Process of Standardization," Journalism Quarterly,, 32:277-284, 328. Chaffee, Steven H. (1972), "Longitudinal Designs for Communication Research: Cross-Lagged Correlations," paper presented to the Association for Education in Journalism annual conference, Carbondale, Ill. Crouse, Timothy (1972), The Boys on the Bus, New York, Random House. Eaton, Howard (1989), "Agenda-Setting With Bi-weekly Data on Content of Three National Media," Journalism Quarterly, 66:942-948, 959. Ebbinghaus (1885), Hermann. Uber das Gedachtnis. Leipzig: Dunker and Humblot. Funkhouser, G. Ray (1973), "The Issues of the Sixties: An Exploratory Study in the Dynamics of Public Opinion," Public Opinion Quarterly, 37:62-75. Lowery, Shearon A., and DeFleur, Melvin L. (1988), Milestones in Mass Communication Research, second edition. New York: Longman. McClure, Robert D. and Patterson, Thomas E. (1976), "Print vs. Network News," Journal of Communication, 26:23-28. McCombs, Maxwell E. (1981), "The Agenda-Setting Approach." In Dan D. Nimmo and Keith R. Sanders, K.R., (eds.) Handbook of Political Communication, Beverly Hills, CA, Sage. Mullins, L. Edward (1977), "Agenda-Setting and the Young Voter," In Donald L. Shaw and Maxwell E. McCombs (eds.), The Emergence of American Political Issues: The Agenda-Setting Function of the Press, St. Paul: West Publishing, pp. 133-148. Salwen, Michael B. (1988), "Effects of Accumulation of Coverage on Issue Salience in Agenda-Setting," Journalism Quarterly, 65:100-106. Shaw, Donald L. and McCombs, Maxwell E. (1977), The Emergence of American Political Issues: The Agenda-Setting Function of the Press, St. Paul, West Publishing. Sohn, Ardyth Broadrick (1978), "A Longitudinal Analysis of Local Non-Political Agenda-Setting Effects," Journalism Quarterly, 55:325-333. Stone, Gerald C. and McCombs, Maxwell E. (1981), "Tracing the Time Lag in Agenda-Setting," Journalism Quarterly, 58:51-55. Tipton, Leonard; Haney, Roger D. and Baseheart, John R. (1975), "Media Agenda-Setting in City and State Election Campaigns," Journalism Quarterly, 52:15-22. Wanta, Wayne, and Hu, Yu-Wei (1994), "The Effects of Credibility, Reliance, and Exposure on Media Agenda-Setting: A Path Analysis Model,' Journalism Quarterly., 71:90-98. Watt, James H.: Mazza, M.; and Snyder, L.B. (1993), "Agenda-Setting Effects of Television News Coverage and the Memory Decay Curve," Communication Research, 20:408-435. Weaver, David H.; Graber, Doris A.; McCombs, Maxwell E. and Eyal, Chaim H. (1981), Media Agenda-Setting in a Presidential Election: Issues, Images and Interest, New York, Praeger Publishers. Winter, James P. and Eyal, Chaim H. (1981), "Agenda-Setting for the Civil Rights Issue," Public Opinion Quarterly, 45:376-383. Woodworth, Robert S. and Schlosberg, Harold (1954), Experimental Psychology , revised edition. New York: Henry Holt & Company. Zhu, Jian-Hua (1992), "Issue Competition and Attention Distraction: A Zero-Sum Theory of Agenda-Setting," Journalism Quarterly, 69:825-836. Zhu, Jian-Hua; Watt, James H.; Snyder, Leslie B.; Yan, Jingtao; and Jiang, Yansong (1993), "Public Issue Priority Formation: Media Agenda-Setting and Social Interaction," Journal of Communication, 43:8-29 Zucker, Harold G. (1978), "The Variable Nature of News Media Influence," in B.D. Ruben (ed.) Communication Yearbook No. 2, New Brunswick, Transaction Books, pp. 225-245. Memory Decay and the Agenda-Setting Effect: An Examination of Three News Media By Wayne Wanta School of Journalism and Communication 1275 University of Oregon Eugene, OR 97403-1275 (503) 346-3752 [log in to unmask] and Melissa J. Roy School of Journalism and Communication 1275 University of Oregon Eugene, OR 97403-1275 **Manuscript submitted to the Communication Theory & Methodology Division for consideration of presentation at the Association for Education in Journalism and Mass Communication annual convention, Washington, D.C. **Wayne Wanta is an assistant professor and Melissa Roy is a graduate student in the School of Journalism and Communication at the University of Oregon. Memory Decay and the Agenda-Setting Effect: An Examination of Three News Media Abstract This study examined how individuals' memory of information about issues covered in three news media decays over time. Local newscasts, national network newscasts and local newspaper coverage showed agenda-setting effects in a relatively short time (four to eight days), but the effect decayed shortly thereafter (14 to 28 days). The results also suggest that the agenda-setting effect is much stronger and memory decay is much more gradual for the local newspaper.