Jung-Sook Lee Competition
Comprehension and Recall of Internet News: A Quantitative Study
of Web Page Design
by
D. Leigh Berry
Manship School of Mass Communication
Louisiana State University
820 North Blvd., Baton Rouge, LA 70802
Phone: 225.336.9059 (h); 225.388.3887 (w)
E-mail: [log in to unmask]
A paper submitted to the Communication Technology & Policy Division, Jung-Sook
Lee Competition, of the AEJMC National Convention, New Orleans, LA, August
1999 ABSTRACT
Comprehension and Recall of Internet News: A Quantitative Study of
Web Page Design
by D. Leigh Berry
Manship School of Mass Communication
Louisiana State University
This experimental study examined the effects of multimedia on Internet news
readers. Subjects viewed one of two versions of the same Web siteDone with
multimedia and one without. Dependent variables were comprehension, recall, and
response to site. Findings did not support a significant difference in
comprehension, recall, or response arising from presence or absence of
multimedia. Comprehension and recall with regard to items such as current events
knowledge, gender, and advertisements are also discussed. Jung-Sook Lee
Competition
Comprehension and Recall of Internet News: A Quantitative Study of
Web Page Design
Introduction
Broadcast and print media have long been the only sources of news for the
consumer; however, in the past five years, news organizations have moved
on-line, to the Internet, developing Web sites at an amazingly rapid rate. The
result of this rapid evolution is a powerful new medium that has the potential
to provide readers with information at a speed and extent not possible in the
traditional form. Where traditional papers are "bound by the rising costs of
newsprint" (Henderson & Fernback, 1998, p. 115), the space for Internet news is
virtually unlimited.
As of July 10, 1998, the database for Editor and Publisher Online offered links
to 3,679 magazines, 3,014 newspapers, 1,925 radio sites, 136 syndicated news
services and 1,242 television sites (1998), and 36 million users (20% of
Americans) are reported as getting news from the Internet at least once a week
(Event-driven news audiences: Internet news takes off, 1998). The majority falls
in the categories white, male, 18- to 29-year-olds, college graduates and
households with incomes over $50,000 annually (Event-driven news audiences:
Internet news takes off, 1998). Nearly every major news organization from print
to broadcast now has a related Web site. Print publications typically produce
Web sites that contain the same stories as the print version, placed in a very
basic Web design that is easy to navigate. For example, the New York Times and
USA Today both consist of a main page that contains a picture of the print
version's flag, headlines typed in the same font and point size as in the print
version that are linked to stories, which are simply copied, pictures and all,
and reformatted in HTML and published on the Web. Television and radio sites may
also have audio or video files to accompany a presentation of the principal
stories or transcripts of the broadcast news program or show, but for the most
part, they too exhibit a very print-oriented design. For example, the Web sites
for CNN, NPR and MSNBC (NBC news) all contain links to the main news topics
(e.g., news, business, sports, etc.) and to news programs regularly appearing on
these networks; but when the links to these shows and their stories are
followed, what one finds are full-text stories or transcripts with photo
enhancements that look very much like a traditional newspaper or newsmagazine.
Audio and video may be included with these stories, or there may be links to
pages consisting entirely of audio and video files with separate reports on the
same stories. Items not found in the traditional medium that may be included on
these sites range from reporter biographies and programming guides to contact
information via e-mail.
While these designs are traditional in that their physical appearance resembles
a typical newspaper or newsmagazine, the addition of video and audio files to
text-based stories that viewers can access on demand and go back to as many
times as they wish is revolutionary and creates the potential for completely new
methods of news delivery combining print with broadcastDtwo media often
considered at odds with one another. Sites usually offer links to archival
information, "chat" rooms for viewers to respond to either site content or
regular programming and surveys or other interactive devices that attempt to
create a sense of community or involvement with the news organization. Such
enhancements are intended to bring the audience closer to the news source and
provide an interesting, though not entirely reliable, gauge of public opinion.
In view of this Internet revolution, it becomes important to study the effects
on the audience of multimedia enhancements, such as the addition of audio or
video files to traditional, print-based stories.
This study examines comprehension and recall for Web sites that exhibit
traditional print-based design without multimedia compared with Web sites that
incorporate multimedia enhancements. It examines the benefit of using audio and
video with text to convey messages effectively and attempts to give insight to
the optimum method of news delivery via the Internet.
Review of the Literature
Comprehension and Recall of Internet News
The body of research on Web site design and its relation to comprehension and
recall is quite small. In collecting information upon which to base this study,
therefore, it was necessary to consult studies that focused on comprehension and
recall for traditional print and broadcast sources. The literature was organized
into studies that focused on the advantages of using a print or a broadcast
source and the problems with broadcast news in general; studies that critiqued
past research on comprehension and recall of news; studies that focused on
cognitive processing and elements of story structure that increase the
processing of news stories; and studies concentrating on problems in sending and
receiving messages as these problems relate to news stories.
Reading the News
Studies of audience recall and comprehension for news stories have not provided
encouraging findings for the press. As a general rule, comprehension and recall
scores for news presented in any medium are not very high (Bernard & Coldevin,
1985; DeFleur & Cronin, 1991; Gunter, Furnham, & Gietson, 1984; Hyatt, Riley, &
Sederstrom, 1978; Saljo & Cedersund, 1988). The body of research seems to agree,
however, that as a general rule, recall and comprehension are higher for news
received from a print source than from any other medium (DeFleur et al., 1992).
Studies focusing on reading comprehension alone have shown that differences
between print and computer screen reading tests are not significant (Barber,
1988; DeFleur et al., 1992). Facts presented in print and on computer screen
have also been shown to promote a significantly higher level of comprehension
and recall than facts from the same stories presented on radio or television
(DeFleur & Cronin, 1991). For this reason, studies of comprehension and recall
for news appearing in print can be generalized to studies of comprehension and
recall for Internet news in formulating hypotheses.
The advantages of print news are varied. Studies have suggested improved recall
for print news is due to the fact that readers can pace themselves while going
through printed stories, as opposed to broadcast media where the information is
presented quickly and must be processed immediately (Gunter, Furnham, & Gietson,
1984; DeFleur et al., 1992). Studies have also suggested comprehension is
improved because of the level of attention paid to print reports as compared to
broadcast (Browne, 1978; DeFleur et al., 1992). Comprehension is directly
correlated with looking (Anderson & Burns, 1991; Kardes, 1993), and print
presentations require active looking, or reading, for information to be
received. Broadcast news, by contrast, often suffers from "contextual
interference," or interference as a result of the viewer engaging in other
activities while the message is being sent (Anderson & Burns, 1991; DeFleur et
al., 1992; Woodall, Davis, & Sahin, 1983). Finally, studies have suggested print
promotes better recall because it requires the reader to imagine his or her own
visuals, which requires more cognitive processing of information, as opposed to
broadcast where the necessary images are provided (Gunter, Furnham, & Gietson,
1984).
Watching the News
Although most Web news sites demonstrate a print-oriented design, with
improvements in technology, more are incorporating elements of broadcast media
as well. These elements take the form of audio and video files, animated
graphics and moving text, and full-color Web sites as opposed to print's
traditional black-on-white.
Studies of broadcast media have consistently shown broadcasting's inferiority
to print in promoting recall and comprehension. Reasons given for the
inferiority of broadcast news, whether by television or radio, have interesting
implications for Web news because of the technological similarities between the
two. Neil Postman, in his book Amusing Ourselves to Death, criticizes television
news for reducing social and political discourse to the level of entertainment
(1985). Many researchers have pointed to this characteristic of television news
as one of the main hindrances in its ability to make a lasting impression on the
viewer. Studies have suggested that because broadcast news is brief and
entertaining, it is cognitively processed at a lower level than print news,
which could account for the reason it is recalled at a significantly lower level
(Lang, 1989; Wicks, 1992). Other researchers have suggested the problem is that
broadcast news cannot be studied in the same manner as print news (DeFleur et
al., 1992; Gunter, Furnham, & Gietson, 1984). The high scores of subjects given
free-recall tests after treatment with a print source are attributed to the
different cognitive processes used when receiving messages from a print source
as opposed to a broadcast source (Bennett, Swenson, & Wilkinson, 1992; DeFleur
et al., 1992; Gunter, Furnham, & Gietson, 1984). Broadcast news is presented
quickly and with little repetition. This precludes the viewer's ability to focus
on items that may require more cognitive processing and excludes viewers who may
lack initiation with the news story or who may be involved in other activities
while the broadcast is taking place (DeFleur & Cronin, 1991; Lang, 1989; Wicks,
1992).
For this reason, more recent studies have focused on the broadcast source,
hoping to find ways to improve presentation and thereby increase comprehension
and recall. These studies have shown viewers exhibit better retention for
broadcast items that use highly visual information when the visual content is
closely related to the material being spoken than for items that use the
"talking head" scenario (Brosius, Donsbach, & Birk, 1996; Edwardson, Kent,
Engstrom, & Hofmann, 1992). Studies also have shown that when highly redundant
graphic presentations are used with verbal presentations, recall is
significantly improved (Griffin & Stevenson, 1996). Reasons for this improvement
have been attributed to incorporating "multiple channels" of processing
information or presenting the same information or explanatory information using
a different method of delivery (Berry, 1983; Booth, 1970-71). Finally, as is the
case with print news, studies have shown that the level of attention paid to the
broadcast is directly related to the amount of information recalled. Saljo and
Cedersund found that retention of broadcast news was "quite accurate, up to
73.9%, when there was strong reason to assume the viewers had paid attention [to
the broadcast]" (1988, p. 450).
Critiques of broadcast media research have indicated that past studies
incorporated inaccurate measures to examine the way information is coded and
later retrieved from broadcast sources (Berry, 1983; Findahl & H ijer, 1985;
Woodall, Davis, & Sahin, 1983). For example, some researchers have suggested the
use of free-recall tests is invalid when used with broadcast media as it does
not give viewers the necessary prompts to access information that may be stored
in memory (Findahl & H ijer, 1985; Woodall, Davis, & Sahin, 1983). Other
researchers have pointed out that the telephone survey method traditionally used
to measure television news recall does not guarantee the individuals surveyed
were actually paying attention to the broadcast in question (DeFleur et al.,
1992; Saljo & Cedersund, 1988). These critiques have stated the need to focus on
"viewer variables" as well as presentation variables. These studies contend that
print is not necessarily superior to broadcast; rather researchers have failed
to take into account the influence of the viewer, or how he or she affects the
communication process, when studying the two media. Since communication involves
a message that is sent by one source and received by another, defenders of
broadcast media stress the need to focus on the breakdown taking place on the
receiving end, or to study the "listener's contribution," as well as the changes
that can be made in the source (Saljo & Cedersund, 1988; Lang, 1989).
Viewer Variables
Viewer variables, or variables that arise from the viewer and that may
influence comprehension and recall of the message, include interest in the
subject matter (Berry, 1988), amount of general news knowledge (Woodall, Davis,
& Sahin, 1983; Price & Zaller, 1993; Findahl & H ijer, 1985), educational level
(Findahl & H ijer, 1985), gender (Findahl & H ijer, 1985; Gunter, Furnham, &
Gietson, 1984) and even the time of day one watches or reads the news (Gunter,
Jarrett, & Furnham, 1983). General media knowledge and interest are closely
related to each other and usually act together to determine the amount of
attention a specific stimulus will receive.
The increased level of comprehension and recall of news stories produced by
general news knowledge can be explained as a result of expanded access to
knowledge stored in the individual's memory (i.e., the semantic memory network)
and an increased likelihood the new information will find a link to this
existing knowledge. Researchers have shown that general news knowledge is not a
function of educational level; rather, the more current events knowledge
possessed by the individual, the greater the comprehension and recall of future
news (Berry, 1983; Findahl & H ijer, 1985; Griffin & Stevenson, 1996; Price &
Zaller, 1993; Woodall, Davis, & Sahin, 1983). It has, however, been suggested
that higher educational levels foster greater news comprehension because of the
subject's prior conditioning to attend to messages and acquire meaning from them
(Price & Zaller, 1993). It has also been shown that viewers are less likely to
pay attention to broadcast news that requires background information or
"initiation" with the story (Berry, 1988; Findahl & H ijer, 1985; Lang, 1989;
Larsen, 1988; Woodall, Davis, & Sahin, 1983). Viewers, instead, choose to focus
on stories of which they have at least some prior knowledge. This selection
process is often guided by interest.
Several studies have focused on the importance of interest in remembering the
news. Interest can be aroused by the presentation of a story, or by any number
of individual experiences stored in the subject's memory. Interest in stories,
as it applies to the extent to which the subject can relate to the events being
reported, is accepted as perhaps the strongest contributor to the recall and
comprehension of news stories in all media (Berry, 1988; Booth, 1970-71; Findahl
& H ijer, 1985; Saljo & Cedersund, 1988; Scott & Goff, 1988; Woodall, Davis, &
Sahin, 1983). Interest aroused through empathy leads to deeper levels of
cognitive processing and, therefore, improved comprehension and recall. Interest
can correlate to general knowledge, which, as previously discussed, has been
shown to increase viewer recollection of the news, but prior interest is not a
guarantee of increased comprehension or recall. Research has shown that interest
aroused by method of presentation is more likely to increase comprehension and
recall than a subject's existing interest in a topic alone (Berry, 1988); and it
has been suggested that as an individual's interest is aroused by a particular
news item, he or she will "actively select" that particular item or topic to
focus on in future news stories (Berry, 1983; Findahl & H ijer, 1985; Griffin &
Stevenson, 1996; Price & Zaller, 1993; Saljo & Cedersund, 1988). Interest in a
topic such as health, for example, will not automatically imply higher
comprehension and recall for a story on meningitis if the subject has no
particular knowledge with regard to the condition or if the story is not
presented in such a way as to arouse the subject's empathy. Level of interest
also determines level of attention paid to a particular item, and as stated
before, higher levels of attention are directly related to deeper levels of
processing. It has been suggested that people do not even attempt to pay
attention to or remember all items in a news program or paper, and, therefore,
when studying comprehension and recall across news media, only items of interest
to the subject should be evaluated (Saljo & Cedersund, 1988). Interest, it is
suggested, is the filter that determines what stories will be attended to,
deeply processed, and later remembered.
The last two viewer variables to be considered are gender differences and time
of day effects. It has been suggested by past research that males recall news
stories at a significantly higher rate than females (Gunter, Furnham, & Gietson,
1984; Findahl & H ijer, 1985). One suggested reason is the idea that males in
general are more likely to attend to news broadcasts than females (Gunter,
Furnham, & Gietson, 1984). Recent studies on this issue, however, have shown
that "the core audience for nightly network news is mainly older and femaleD55%
of women over age 50" (Event-driven news audiences: Internet news takes off,
1998). Studies of episodic memory and recall have shown that females
consistently outperform males in recall tests measuring episodic memory
(Herlitz, Nilsson, and Backman, 1997); however, studies focusing on semantic
memory processing have not shown a significant difference in recall between
males and females (Herlitz, Nilsson, and Backman, 1997). It is notable that in
the 18 to 25 age group, males lead females in the use of on-line technologyDa
factor that could influence recall scores by gender (Event-driven news
audiences: Internet news takes off, 1998).
Finally, time of day effects have been shown in several studies to influence
comprehension and recall. Studies have consistently shown that information
presented earlier in the day is more likely to be recalled than information that
is presented later (Gunter, Jarrett, & Furnham, 1983). These studies have also
shown that throughout the day, concrete information, or information about the
who and the what of a story, is recalled at a consistently higher rate than
abstract information, or causal information (Gunter, Jarrett, & Furnham, 1983).
Concrete and abstract information will be explained more in the discussion of
presentation variables.
Presentation Variables
Presentation variables, or variables that arise from the method of presentation
and that may influence the viewer's or reader's ability to comprehend and recall
the message, are controlled largely by story structure. Story structure consists
of the type of information presented, the style of writing employed and the use
of emotionally stimulating or neutral images. Item order has also been suggested
as a factor influencing recall; however, studies of item order do not always
agree on the significance of its effect.
The first element of story structure, type of information, is a function more
of recall than comprehension. In the typical news story, two distinct classes of
information are presented. Concrete information, which takes into account the
who, what, when and where of events and which is easily visualized by the
receiver, and abstract information, which involves the causes and consequences,
or the how and the why of the story (Findahl & H ijer, 1985; Gunter, Furnham, &
Gietson, 1984; Woodall, Davis, & Sahin, 1983). The latter information is
considered abstract because it is made up of words that cannot be easily
visualized (e.g., justice, terrorism, poverty, etc.) and, therefore, cannot be
processed as deeply. Researchers have found that for this reason subjects more
easily recall concrete details (e.g., a 42-year-old white male, the White House
lawn, a brown Oldsmobile, etc.) because subjects can form clear pictures of
these details in their minds. Concrete events, along with the other viewer
variables such as interest and prior knowledge, are processed at a deeper level
because they draw on an existing network of information.
The second element of story structure, writing style, can take one of two
forms: the narrative style or the inverted pyramid style, which is the standard
writing style for news. Thomas J. Housel states that "schema theorists from
cognitive psychology and linguistics have found that news readers comprehend,
recall and are more aroused by stories written in the narrative structure than
by stories written in the typical news structure" (1984, p. 505). Housel's study
focuses on linguistic complexity in broadcast news, but the story structure he
refers to is the inverted pyramid style, which is also the main structure used
in most print media. Linguistic complexity is defined as deleting connectives
and transitions, using short sentences, including information that can be easily
visualized but that is irrelevant to the story, arranging paragraphs in
descending order of importance and using ambiguous pronoun-antecedent references
(Housel, 1984). Story structures following the inverted pyramid style are
written so that the most important information, what would be the conclusion of
the story, is presented first, followed by the causes or reasons, with the
general background information (in a narrative, the introduction) presented
last. Proponents of more "scannable" Web pages recommend the inverted pyramid
style as the most effective method of writing for the Internet because of its
ability to capture the viewer's attention quickly (Nielsen, 1997). Studies of
the inverted pyramid style, however, have suggested that it is not highly
conducive to accuracy of recall. The inverted pyramid style divides the reader's
attention between attempting to decode the meaning of the conclusion and
receiving the facts as they are presented instead of focusing the subject on the
content of the report and leading him or her to the conclusion (Lang, 1989;
Magliano, Schleich, & Millis, 1998). Studies that convert stories written in the
inverted pyramid style to a more logical, narrative presentation of events have
shown significant increases in subject comprehension and recall (Berry, 1988;
Larsen, 1988).
The last element of story structure is the use of emotionally stimulating
information or strong negative or violent images to increase recall for specific
events. The majority of studies focusing on this variable have shown that the
use of a highly stimulating imageDdescriptions or pictures of events that elicit
strong feelings, whether positive or negative, on the part of the
viewerDsignificantly increase the subject's ability to recall that particular
news item (David, 1996; Gunter, Furnham, & Gietson, 1984; Lang, Newhagen, &
Reeves, 1996). Researchers have found that the use of these images has a
stronger effect on males using a print medium than on females (Gunter, Furnham,
& Gietson, 1984). Research has also shown that the use of strong negative images
can result in an avoidance reaction on the part of the viewer, as opposed to a
highly stimulating positive image, which has been shown to increase viewer
receptivity to the message (Lang, Dhillon, & Dong, 1995). One study even
suggested that the use of deviant images, or images that do not fit the viewer's
expectations (e.g., "man bites dog" as opposed to "dog bites man"), can increase
the viewer's recall of the story by causing increased interest in the story and,
therefore, increased attention (David, 1996). Both situations increase
comprehension and recall because the viewer spends more time thinking about the
story as a result of the image.
The final presentation variable to be considered is item order. Several studies
have been performed varying the order of items in a newscast, even the order of
sentences in the individual stories, to determine the effect such ordering has
on the individual's ability to recall specific facts (Booth, 1970-71; Lang,
1989; Findahl & H ijer, 1985; Kirriemuir, 1997; Wicks, 1992). For the most part,
these studies have shown that item order has little effect on the individual's
memory for specific facts (Klein, 1978). There have been a few studies, however,
that have shown increased recall for items appearing in the first and last
positions of a broadcast or in a series of advertisements (Booth, 1970-71;
Berry, 1983; Wicks, 1992), and it has been noted that there appears to be more
attention paid to items at the top or beginning of Web pages (Kirriemuir, 1997;
Nielsen, 1996). These first and last positions have therefore become known as
the "favored positions" in the line-up and should be monitored as potentially
confounding variables in measuring viewer recall.
Theory and Hypotheses
In discussing theories of comprehension and recall, one must begin by
differentiating, as several researchers have done, between comprehension, or
understanding of a story, and recall, or memory for the items contained therein.
The two processes are highly correlated; however, they are two separate
cognitive functions (Woodall, Davis, & Sahin, 1983). Recall is the process by
which subjects are able to pull up or remember information to which they have
been exposed. It involves storage, retrieval and access to input information.
Comprehension, or understanding, by contrast, involves an interaction between
new information that is being received and information or knowledge that has
already been stored (Woodall, Davis, & Sahin, 1983; Findahl & H ijer, 1985). It
has been proposed that subjects can remember information that was not
understood, and it has been shown that subjects can accurately understand
information that is not correctly remembered (Findahl & H ijer, 1985; Woodall,
Davis, & Sahin, 1983). As Woodall, Davis and Sahin have pointed out, it is
imperative in studies of comprehension and recall that researchers keep the two
processes separate and make clear distinctions between measures of comprehension
and measures of recall (1983).
Theories of cognitive processing of news are based largely on the
levels-of-processing framework of memory developed by Craik and Lockhart.
Levels-of-processing draws on the model of the brain as a computer, viewing the
human receiver as a processor of information. The semantic network is made up of
information that is encoded as abstract symbols to form a "language" of events
(Alterman & Bookman, 1992). Comprehension and memory are functions of the
individual's ability to encode and link new information to these symbols and
place it in his or her network. Problems in memory are attributed to problems
either in storage capacity or processor speed (Craik & Lockhart, 1972). Either
the information does not fit into the individual's semantic network, or there is
not enough time for the individual to process all of the information being
receivedDan event sometimes referred to as "information overload."
The brain processes the information encountered and determines whether the new
information fits into the individual's network of existing information or
experiences. Individuals subconsciously use this network to decide which news
stories to focus on, and, in turn, this network determines which stories will
make it into the individual's long-term memory (Craik & Lockhart, 1972; Saljo &
Cedersund, 1988; Woodall, Davis, & Sahin, 1983). This subconscious selection or
addition to the existing network takes place through a process called "knowledge
updating" (Larsen, 1988, p. 441). It is this examination of comprehension and
recall using the levels-of-processing framework that causes presentation
variables and viewer variables to become significant.
Theories of Comprehension and Recall for Cyberstories
The focus of this experiment is Web site design and its effect on comprehension
and recall of news stories. As previously mentioned, little research has been
done with regard to user interaction with news stories in an on-line
environment. The studies that have been documented focus for the most part on
elements contributing to usability, for example, bulleted lists, short
paragraphs, minimal scrolling, restricted use of animations and avoidance of
frames (Nielsen, 1996). These theories of usability are useful in the creation
of a valid instrument for measuring comprehension and recall for multimedia
sites.
It has been proposed that individuals do not read information on the
InternetDthey scan (Nielsen, 1997). Emerging trends in Web design, therefore,
call for a home page made up of "scannable" headlines with brief abstracts
beneath them to allow the reader to quickly decide whether the information
interests him or her. Web designers recommend that stories be written in the
inverted pyramid style with only one thought per paragraph (Nielsen, 1997).
Motion on these pages must be kept to a minimum, and frames should be avoided as
they tend to confuse the reader by breaking the cognitive flow of the pages
(Nielsen, 1996). These usability models neglect the presentation variables shown
to contribute to comprehension and recall in traditional media; and multimedia
pages, by virtue of being new, introduce distractions not present in traditional
media, which could influence comprehension and recall (DeFleur & Cronin, 1991).
While a text-only site would also suffer from this novelty effect, multimedia
increases it by presenting new stimuli in a form that has very few semantic
links with the subject. These factors could produce lower levels of processing
for the information contained in the stories on these sites and, therefore,
lower comprehension and recall. The hypotheses tested in this study, therefore,
are as follows:
H1: Comprehension will be better for full-text stories appearing on a
traditional, print-based (plain) site than for the same stories appearing
with audio or video enhancements on a more interactive (multimedia) site.
H2: Recall also will be better for text-only stories on the plain site
than for the same stories accompanied by audio and/or video on the
multimedia site.
H3: User response in terms of likability and level of interest, by
contrast, will be higher for the multimedia site than for the plain site.
Methodology
Eighty-four undergraduate students at Louisiana State University
participated in this study. Educational differences were controlled for
because all subjects were at least sophomores, which implies that the
majority of the students were 19- to 20-year-olds with at least one year
of college education. Each subject was assigned to an individual
IBM-compatible personal computer with a 17" color monitor, and students
assigned to the multimedia sites were provided a set of headphones for
use
with the audio and video components. Internet connection speed was T1 for
all computers to ensure swift connection and to avoid any negative
affects
of downloading time on attention. Subjects participated during regular
class time and were divided into groups of ten to 20 students.
Participants from the same class all viewed the same site, and subjects
were instructed to examine the stories on the site they had been assigned
at a relaxed pace and not to spend excessive time on any particular
story.
Minimal instructions were given to avoid conditioning subjects to the
treatment.
Subjects viewed one of two versions of the same Web site. The Web site
consisted of six stories not heavily covered in the media and that were
at
least six months to a year old. One story was selected from each of the
following news categories: local news, national news (politics),
international news, sports, health and human interest (people). Stories
were selected from the archives of the CNN Interactive Web site, and
effort was taken in the selection process to choose moderately
interesting
stories that were not overly visual or sensational. For example, stories
were selected that focused on the war on drugs, school safety and AIDS
research while stories on triple homicides, car bombings or animal
attacks
were avoided. Stories were selected with the intent that they would raise
similar degrees of interest without biasing the results for or against
any
one story (as much as possible, considering individual differences).
The stories selected for use in this study all originally contained audio
and/or video accompaniment. On the multimedia site these enhancements
appeared in their original format and placement in the story. On the
plain
site these enhancements were transcribed and placed in the same location
in the story as on the multimedia site. Otherwise, the text and images on
both sites were identical. This made it possible to determine the
influence on comprehension and recall of the presence or absence of the
audio and/or video files. Both Web sites contained a weather box giving
current weather information, and both sites contained at least three
advertisements. The ads on the plain site appeared as still banners,
while
the ads on the multimedia site had subtle animation. Both sites were
given
the generic title The Crosstown Reporter, and headline sizes and font
styles were the same for both sites.
The plain site consisted of three linked Web pages with two stories per
page. The first page contained the local and political stories; page 2
contained the international and health stories; and page 3 consisted of
the people and sports stories. All pages had links to the previous and
next pages at the tops and bottoms of each page. Layout on the home page
consisted of the flag extending across the top of the page with the
weather box at the top left corner and a picture of the downtown Baton
Rouge area in the top right. Pages 2 and 3 had the flag repeated in
miniature at the top left and links to the previous and next pages at the
top right. Ads were centered at the bottom of each page, and no ads were
placed in or between stories.
There were no additional graphics on the plain site apart from the flag,
the pictures and the ads. Stories did not contain bullets or text
enhancements such as bold, colored, or italicized text unless italics
were
necessary for grammatical purposes (e.g., titles of books, movies, plays,
etc.). All stories were in black text on white background and were
arranged in columns format.
The multimedia version of the site consisted of seven linked Web pages.
The home page consisted of six headlines, each with a brief one-sentence
abstract beneath it, arranged in two columns of three. Following the
abstracts were links to the full stories. The full stories were contained
on one page and were linked back to the home page. The flag was identical
to the one used on the print-based site, appearing in the top left corner
on each linked page along with the category of the story (i.e., local,
poli
tical, international, health, people or sports). Each page also contained
ads exhibiting subtle animation, and ads were centered at the top of each
page. The same pictures that appeared on the home page were repeated on
the linked pages to illustrate the stories they accompanied. The home
page
also contained a weather box with the same information as the weather box
appearing on the plain site, but with an animated illustration.
Data Analysis
The dependent variables in this study were comprehension, recall and user
response to the site. The independent variable was site design (i.e.,
presence or absence of multimedia). Factors controlled for were item
order, level of current events knowledge and interest. Item order and
interest were measured on a self-reporting basis in the post-test, and
current events knowledge was measured in the pre-test. To monitor the
effects of interest, subjects were asked in the post-test to rank the
stories they found most interesting as well as the general areas of news
that interest them most. Gender and time of day were asked for in the
post-test. To control for possible prior knowledge, subjects were asked
in
the post-test to indicate whether they had any prior knowledge of any of
the stories used in the experiment and if they did, to list those stories
by name or topic.
Subjects were instructed to attend solely to the news stories presented
before viewing the sites, and, in responding to the test questions, they
were asked not to guess at the answers. These instructions were intended
specifically to reduce error in the results.
Subjects were given a timed current events pre-test consisting of 12
short-answer questions. The pre-test was strictly a measure of prior
knowledge and level of media use, and it did not reflect the post-test.
Students were allowed five minutes to complete this portion of the test;
however, the time limit was not announced to the subjects. After the
treatment with the individual sites, subjects were given a two-part
post-test. The first part was a 12-question objective test in which
subjects were asked to indicate the first and last stories they read, to
give individual responses to the design of each Web site, to complete an
interest section in which they indicated on a scale of 1 to 5 (with 1
being "most interested") the category of news that interests them most,
to
indicate which stories they may have already been familiar with, and to
indicate their sex and the time of day the test was taken. Subjects also
indicated using the scale mentioned above the stories on the site they
found most interesting. This portion of the post-test was quite brief and
mainly was used for informational purposes and as a control device.
The second part of the post-test was the measure of comprehension and
recall. The questions were structured to measure these two dependent
variables separately. Questions measuring recall were mainly
identification items phrased as who, what, when and where questions,
while
the questions measuring comprehension were more interpretive and were
phrased as why and how questions. Questions were presented in a 30-item
cued-recall test. The cues were the names of specific people, references
to locations or events or the mention of particular facts or details that
preceded the information sought. These questions differed from free
recall
because subjects were not asked simply to list as much information as
they
could remember; rather, they were given specific cues as to the
information sought.
Subjects were asked five questions on each story: three recall and two
comprehension questions; and answers were scored based on a two-point
system that has been used in several previous studies. Completely correct
items were assigned two points, partially correct items were assigned one
point and answers that were not correct or questions that were not
answered were assigned zero points (DeFleur et al., 1992; Gunter,
Furnham,
& Gietson, 1984). The maximum possible recall score was 36, and the
maximum possible comprehension score was 24. This number of questions was
selected to give a good measure of comprehension and recall for each
story
without causing fatigue in the subjects.
Analysis and Discussion of Findings
Before conducting the experiments, a pilot study was run in which a small
group of graduate students and communications professors participated in
the experimental treatment and took the pre- and post-tests. The pilot
study highlighted potential difficulties in using the computers and
potential problems in setting up the sites for student use. After the
pilot results were examined and the problems solved, the actual tests
were
conducted.
The first round of tests involved approximately 60 undergraduate
journalism students divided into four groups of 14 to 17 students, each
presented with one of the Web sites discussed. Experimental results were
then subjected to preliminary analysis, and it was found that the number
of female subjects outnumbered males two to one in the multimedia setting
and three to one in the plain setting. For this reason, another round of
experiments was conducted to increase the number of male respondents. In
this second round of experiments, all female respondents were discarded
and only the male responses were kept. An additional two female subjects
were discarded from the text-only setting of the first round as they were
observed not participating in the Web site viewing portion of the
treatment. After the second round, male to female ratios were equal and
the final numbers were 44 participants in the plain site to 40
participants in the multimedia site.
The results from all subjects were combined so that the mean scores could
be compared in SPSS for Windows. First, a Pearson correlation analysis
was
conducted to compare the scores for comprehension, recall, response to
site, sex, interest in stories on the site presented, time of day, level
of media use and current events knowledge. The results indicated
significant correlations between comprehension and recall, between time
of
day and recall, between level of current events knowledge and time of
day,
between sex and response to site, between comprehension and response to
site and between media use and level of current events knowledge.
Comprehension and recall were correlated (r = .557; p < .01) as would be
expected. Time of day was shown to be correlated with recall (r = .248; p
< .05) and current events scores (r = .275; p < .05), with subjects who
tested in the morning scoring higher on both tests than those
participating in the afternoon. Total comprehension scores were
negatively
correlated with response to site (r = -.236; p < .05); however, as
response scores were ranked in descending values (i.e., the lower the
score, the higher the response), this actually shows comprehension being
improved by a favorable response to the site. Level of media use and
current events scores were correlated (r = .632; p < .01) as would also
be
expected, and presence of multimedia was shown to be correlated with
favorable response to the site (r = -.190). This last correlation,
however, was not found to be statistically significant (p = .084).
Hypothesis One: Comprehension and Site Design
This hypothesis predicted that comprehension would be greater for stories
appearing on the plain, text-only site than for stories appearing on the
multimedia site. Comprehension scores as a whole were quite low for both
sites, but that finding is consistent with previous research. The average
comprehension score for both sites was about 50%. Mean comprehension
scores are illustrated in Table 1.
Table 1: Descriptives of Comprehension
Site Design
Mean Score
Total Possible
Percentage
S.D.
Plain
12.5
24
52%
5.0
Multimedia
12.2
24
50%
4.2
Mean comprehension scores were subjected to the ANOVA test, and it was
found that there was no significant difference in comprehension scores
for
subjects viewing the plain site compared with subjects viewing the
multimedia site. This hypothesis, therefore, was not supported as it
could
not be proven that differences in scores were not caused by factors other
than chance. Results of the ANOVA test are shown in Table 2.
Table 2: ANOVA for Comprehension
Model
SS
df
MS
F
p
Between
Within
Total
1.061
1776.975
1778.036
1
82
83
1.061
21.670
.049
.825
The plain site was coded as 0, and the multimedia site was coded as 1 for
this test. Comprehension scores were then combined and compared based on
the influence of the independent variable site design, or "group." In the
row for between groups, the variation in comprehension that can be
attributed to site design is not very large. The F ratio was shown to be
insignificant (p = .825), and largest variation in scores (Within) is
attributable to variables that cannot be accounted for. This suggests
that
site design had little effect on the subjects' ability to comprehend the
information contained therein.
Hypothesis Two: Recall and Site Design
The second hypothesis predicted that recall scores for stories appearing
on the plain site would be higher than for the same stories appearing on
the multimedia site. As with comprehension, recall scores were also not
very high for the subjects in each group, and again, this is consistent
with past research. Mean recall scores are illustrated in Table 3.
Table 3: Descriptives of Recall
Site Design
Mean Score
Total Possible
Percentage
S.D.
Plain
16.4
36
45.5%
6.0
Multimedia
16.6
36
46.1%
4.9
Mean recall scores were also subjected to the ANOVA test, and it was
found that although recall scores for the multimedia site were higher
than
recall scores for the plain site, the difference was not significant.
This
hypothesis, therefore, was also not supported. Results are shown in Table
4.
Table 4: ANOVA for Recall
Model
SS
df
MS
F
p
Between
Within
Total
.782
2524.170
2524.952
1
82
83
.782
30.783
.025
.874
The variation in recall that can be attributed to site design was again
not very large, and the F ratio was shown to be insignificant (p = .874).
This suggests that site design had little effect on the subjects' ability
to recall the information presented.
It is notable that the standard deviation among scores on the multimedia
site was much lower than scores on the plain site for both comprehension
and recall. Also notable is that the mean recall score for subjects
assigned to the multimedia site is slightly higher than for the plain
site, implying that multimedia may slightly improve recall; however, as
shown in Table 4, the difference in mean recall scores for the two sites
was not significant.
Hypothesis Three: Subject Response and Site Design
The third and final hypothesis predicted that subjects viewing the
multimedia site would report a higher, more positive response in terms of
likability and level of interest than subjects viewing the plain site.
Interest was tested through a Likert scale in which students indicated
level of interest in each of the six stories from 1, meaning high
interest
in the story, down to 5, meaning very little interest in the story.
Likability was measured in a similar manner with students ranking the
site
as either 1: "I would visit this sight regularly" (excellent), 2: "I
would
visit this site again" (good), 3: "I might visit this site again" (fair),
or 4: "I probably would not visit this site again" (poor).
Frequency analysis of likability showed that subjects in the plain group
indicated lower response, with 52% giving a 3 or 4 and only 48% giving a
1
or 2 rating, compared to the multimedia site where 65% gave a 1 or 2 and
only 35% gave a 3 or 4 rating. Results of the Pearson correlation
analysis, however, showed the correlation between site design and
likability was not significant (p > .05). The ANOVA test was also run for
likability scores; results are shown in Table 5.
Table 5: ANOVA for Likability
Model
SS
df
MS
F
p
Between
Within
Total
1.691
45.261
46.952
1
82
83
1.691
.552
3.064
.084
Likability scores were compared based on the influence of the independent
variable site design, with plain coded as 0 and multimedia coded as 1.
The
F ratio was high, indicating a relationship between likability and
multimedia design; however, this ratio was not shown to be statistically
significant (p = .084).
Interest rankings for the multimedia site were also more pronounced than
for the plain site. The percentage of subjects who fell into the neutral
category was much higher for the plain site than for the multimedia site.
The percent indications of interest for all stories broken down by site
are illustrated in Table 6.
Table 6: Percent Interest in Stories by Site Design
Site
High Interest
Neutral
Low Interest
Plain
13.6%
72.9%
13.5%
Multimedia
22.5%
52.5%
25%
The mean interest score for all stories combined on the plain site was
14.9, and for the multimedia site it was 15.4; however, the lower the
score, the higher the interest indicated. Interest in stories on the
multimedia site, therefore, was actually shown to be slightly lower. The
correlation between total interest and site design was not found to be
statistically significant (p = .506) as was the case with likability;
therefore, this third hypothesis was also not supported.
Discussion
The lack of support for the first two hypotheses in this study may be
explained by the research on "multiple channels" discussed earlier in the
literature review. This research focused on improving recall scores for
broadcast news using many different methods of presentation. Examination
of the Web sites used in this study shows that the multimedia items were
embedded in a text setting, and although in three of the stories the
multimedia items did not repeat or paraphrase the text, the enhancements
were still closely related to the content of the story and gave
additional
information on the topic being discussed. The disadvantages of broadcast,
therefore, may have been balanced by the advantages of having a text
accompaniment. Also, in this experimental procedure, subjects were
directed to attend solely to the information presented, which probably
ruled out the "contextual interference" past studies have attributed to
lowered recall scores for broadcast media. Although in this study
multimedia did not increase comprehension and recall scores above those
for the plain site, further study may show that multimedia enhances the
user's ability to remember textual information by reinforcing that
information following the "multiple channels" approach.
The results of this study present interesting considerations for Web site
developers as well as media outlets. While the hypotheses were not
supported, it is interesting to note what was suggested. Comprehension
and
recall appear to be unaffected by the presence or absence of multimedia
as
does user response to the individual stories and to the Web site as a
whole. For the Web developer or media outlet considering the addition of
multimedia or currently using multimedia, or even for those who may be
concerned about the absence of multimedia in their sites or their
inability to produce such enhancements, these findings have several
implications.
The first implication is that the use of subtle animations, links and
multimedia may not reduce the reader's understanding or ability to recall
the information presented. This disagrees with existing research on Web
design that has suggested avoiding such visual illustrations. It suggests
that organizations need not be concerned that these enhancements will
damage the effectiveness of their message, but it also suggests to those
smaller organizations that may not be financially or technologically able
to use multimedia that their omission will not detract from these
organizations' ability to convey an effective message.
The second implication of these results is that the use of animated
visuals or high-tech enhancements like audio or video may not increase
the
viewer's overall response to the Web site. While these findings do not
suggest that the use of these enhancements decreases response, it may
again give comfort to the smaller organization concerned about its
ability
to compete effectively in the Internet community without using advanced
technology. It is important to keep in mind, however, that the frequency
analysis indicated a higher response to sites that did include
multimedia,
and while this difference was not shown to be statistically significant,
it is possible that a larger study using more subjects might result in a
different conclusion.
The third implication, and perhaps the most compelling for Web
researchers, is that interest in the content of the Web site may not be
significantly affected by multimedia. As discussed in the methodology,
stories used in this study were purposely selected based on moderate
interest levels. This was done in order to avoid any one story's being
significantly interesting over another and to highlight any heightened
curiosity and, therefore, increased interest due to the presence of
multimedia enhancements. This may have affected likability ratings;
however, it should have made the impact of the additional technology on
interest more pronounced. What is suggested by the findings is that a
moderately interesting story is still considered only moderately interes
ting whether it is technologically enhanced or not. The Web developer or
media outlet that may consider using multimedia to increase the interest
level for a particular story or for the content of a Web site may find
that practice unsuccessful. By contrast, the smaller media organization
that may have a particularly compelling message and that may be concerned
it will not be noticed due to its plain appearance, may find their fears
to be unfounded.
It is again proper to point out that frequencies indicated more neutral
responses to the plain site than for the multimedia site; however, the
mean scores, though not statistically significant, showed interest
indicated at a higher frequency for stories on the plain site than on the
multimedia site. Larger studies, however, using more subjects and
subjects
more representative of the general population may lead to different
conclusions.
Problems with the Study
As with most experimental studies, this study suffers from a lack of
external validity, meaning the results cannot be generalized to real-life
situations. For example, in real-life situations, individuals are rarely
required or even motivated to read every story on a Web site in a set
amount of time without being allowed to follow links to other sites or
engage in other activities. This study also suffers from the college
sophomore problem, or the problem that "a rather large proportion of
research findings . . . come from studies of white rats and college
students" (Myers & Hansen, p. 374). This is due to the difficulty of
securing more representative subjects that are not committed to other
responsibilities, which, of course, take priority over participation in
experimental research. This problem can also be attributed to the
financial constraints facing the typical graduate researcher.
The most that can be hoped for in any experimental research is an
indication of causal relationships among variables. In the case of this
study, there was also the problem of subjects not being extrinsically
motivated to participate. The students used in this study were required
to
appear in class; however, once the experiment was announced, they were
told by their instructors that their performance was not related to their
grade in the class. Analysis of scores often showed subjects who skipped
large portions, failed to answer any questions, or provided silly
responses (e.g., "Monica Lewinsky" as the principal of Belen High School,
etc.). While such findings cause considerable frustration on the part of
the researcher, they suggest that the use of motivation techniques, such
as the offer of money, extra credit or some other type of reward may have
an impact on the results of future studies on this topic.
A concern going into the study was number of news items used and the
effect this would have on comprehension and recall scores. While the use
of six news stories is consistent with past research (Bennett, Swenson, &
Wilkinson, 1992; Brosius, Donsbach, & Birk, 1996; Lang, Newhagen, &
Reeves, 1996; Sundar, 1998), and the rationale behind having one story
representing each area of news was to account for individual differences
in interest, it is possible that the use of fewer stories and more test
questions would have yielded different results. A few past studies have
been conducted in which subjects were presented with only one to four
news
items and then evaluated using longer post-tests (Gunter, Furnham, &
Gietson, 1984; DeFleur et al., 1992; DeFleur & Cronin, 1991; Wicks,
1992).
Future studies on this topic may consider this option when constructing
the test sites and evaluation measures.
Finally, it was discovered after the treatment that two of the four
questions testing response to site were double-barreled, and, therefore,
the results of these questions could not be used in the analysis. Future
study would require the omission or restructuring of such items to allow
increased precision in measuring this dependent variable.
Additional Findings and Suggestions for Future Research
This study briefly examined several other factors in addition to the
proposed hypotheses. Questions on the pre- and post-tests gathered data
on
such influences as time of day, style of writing, interest, sex of the
subject, stories read first and last, and number of advertisements
remembered from each site. Given below are short discussions of the
responses to these questions and suggestions for future research.
Time of Day
Time of day and recall were found to be significantly correlated (r =
.248; p < .05). This is consistent with existing research that has shown
recall levels to be higher in the morning and lower in the afternoon
(Gunter, Jarrett, & Furnham, 1983). The eight classes used in the study
were equally distributed, with two plain groups and two multimedia groups
tested before noon and two plain groups and two multimedia groups tested
after noon to control for this variable. Time of day was also
significantly correlated with current events scores, with subjects who
participated in the morning scoring higher on the current events pre-test
than their peers in the afternoon (r = .275; p < .05). This suggests that
information obtained a day or even a week before was recalled at a
significantly higher level in the morning than in the afternoon. Level of
media use was significantly correlated with current events knowledge (r =
.632; p < .01), meaning subjects who indicated a high level of media use
(one or more times a day by either watching, reading or listening to the
news) performed better than their low-use counterparts on the current
events pre-test.
Gender
Analysis further indicated that sex and response to site were
significantly correlated (r = .223; p < .05). Males, according to this
study, were more likely to give the site a low rating than were females.
The reasons for this finding are not clear, but it may suggest that
females are more reluctant to give a low rating than males. It could also
suggest that females are more responsive to multimedia than males. Mean
comprehension and recall scores for both sites combined were higher for
females than for males-a finding that is consistent with past research on
interest and increased comprehension and recall. The mean comprehension
score for females was 13.1 (55%) compared with 11.7 (49%) for males, and
the mean recall score for females was 17.1 (48%) compared with 16.0 (44%)
for males. Future researchers may want to examine the response by gender
to multimedia or to Web news in general, the differences in male and
female rating patterns, or the content of stories selected for use in a
particular site and its relation to male and female likability rankings.
Abstract v. Concrete Information
Redundancy in audio and video accompaniment and whether the information
contained in the story was abstract or concrete was also noted for each
story. Redundancy was determined by whether or not the information
presented in the audio or video clip repeated or paraphrased the same
information given in text. Stories were classified as abstract if the
majority of the information in the story focused on the causes and
consequences of an event and if the information was considered difficult
to visualize. Comparisons of these factors and mean recall scores for
each
of the stories used on the multimedia site are illustrated in Table 7.
Table 7: Mean Scores by StoryDMultimedia
Story
Redundant
Abstract
Mean Recall
S.D.
Mean Comp.
S.D.
Local
yes
no
2.38
1.79
2.38
1.50
National
yes
yes
1.98
1.46
1.28
1.11
International
yes
yes
1.45
1.26
1.58
1.47
Health
no
yes
4.53
1.18
2.25
1.24
Sports
no
yes
3.88
1.52
2.70
1.32
People
no
no
2.43
1.20
2.10
1.01
Simply looking at the descriptive statistics indicates that stories
containing highly redundant audio or video accompaniment had lower mean
recall and comprehension scores than stories containing unique audio or
video. This is not consistent with the research and perhaps warrants
further study. Also, stories containing abstract information appeared to
have little effect on comprehension or recall scores. This is also not
consistent with the research and perhaps warrants further investigation.
Interest
The health and sports stories had the highest recall and comprehension
scores and were also indicated as the most interesting stories on the Web
site by students in both groups. This finding is consistent with research
suggesting higher interest will lead to higher comprehension and recall.
Also notable is that when indicating general news interest, more subjects
listed sports and national news as the news categories of highest general
interest; however, while the mean sports scores were high, comprehension
and recall scores for the national story were lowest and second lowest of
all scores respectively. This tends to support the proposal by Berry
(1988) that interest aroused by the way a story is presented is a
stronger
predictor of recall than prior interest in a general news category.
Item Order
Item order was noted by having each subject indicate the stories read
first and last on the Web site. It was interesting that the national
story
was the first story read on both Web sites by the majority of subjects
(98% for plain; 93% for multimedia). The stories read last by site were
the sports (84%), people (11%) and health (5%) stories on the plain site;
and the people (75%), sports (23%) and international (2%) stories on the
multimedia site. This order was to be expected on the plain site as the
national story was the first full-text story presented and the sports
story was the last, and subjects would have had to do considerable
scrolling and navigation to read the stories in a different order. This
was surprising, however, on the multimedia site where subjects were
presented headlines and abstracts of all six stories and were allowed to
choose which story to read first based on this information. On the main
page for the multimedia site, the information about the national story
was
placed in the top left position and the people story was at bottom right,
which may suggest that Web readers still follow the traditional reading
pattern of left to right, top to bottom in spite of being given a choice.
This behavior may also have been the result of the instruction to read
all
stories, and had subjects been given the option, they may have chosen
instead to read only one or two of the stories and in a different order.
Still, as mentioned previously, recall scores for the national story were
lowest of all, indicating the "favored" first and last positions may have
little influence on recall of Web news, and though mean recall scores
were
higher for the sports story, this may have been more a function of
interest than position on the Web page.
Advertisements
Recall of advertisements by site was added to the post-test as a result
of post-pilot interviews. It was added as a self-reporting item in which
subjects listed by product name or slogan all the ads they could remember
from the site they viewed. Simple frequency analysis found higher ad
recall among subjects viewing the plain site than subjects viewing the
multimedia site. Nineteen of the 44 subjects assigned to the plain site
remembered at least one ad, with two remembering all three, compared to
six subjects remembering only one of the six ads from the multimedia
site.
Reasons for these results are unclear, but a possible cause for the
lowered recall of ads on the multimedia site could be the presence of
competing visual stimuli represented by the audio and video files. It
could mean that stories containing audio or video enhancements are better
at holding the user's attention. It could also mean advertisers should
rethink their placement of banners on pages containing multimedia
enhancements. This is an interesting finding, and one that should be
researched further.
Subject Computer Knowledge
An unexpected observation was the number of participants with low
computer skills. In each group, subjects were given step-by-step
instructions on opening the Web browser, following links to the stories
and even operating the multimedia files; however, in each group several
students still required individual assistance, and it was observed that a
few participants in the plain group only read two stories because of
apparent unfamiliarity with operating hyperlinks. This was interesting in
light of the popular perception of young people as having the highest
rates of computer literacy. While this may still be true, it was
surprising the number of computer novices at the college sophomore level.
Scores for participants who were not observed finishing the Web sites
were
not included in the analysis.
Multimedia and Response to Site
Response to site and use of multimedia showed a positive correlation that
was not found to be statistically significant when subjected to more
stringent analysis. Future research involving more subjects and more
representative samples may suggest a significant correlation between
audio
and video enhancements and likability for a particular site. The same is
true for the relation between use of multimedia and interest in the
content of a Web site. One male subject reported in a post-experimental
interview that he would probably not read a text-only story unless he was
especially interested in the content. He said he would be more inclined
to
read a story containing multimedia; however, he indicated that time was a
major factor in this decision rather than increased interest in
multimedia
over text. This is an issue worth exploring and one that might be useful
to investigate for future Web developers.
In other post-experimental interviews, several subjects from the
multimedia group stated that hearing a quote was more believable to them
than simply reading one. The reason given was a belief that it is easier
to alter text than audio or video, in which subjects felt tampering would
be more noticeable. Studies have been conducted on similar phenomena,
most
notably by Sundar on source attribution and increased credibility ratings
for Web stories (1998). The views communicated by these students,
however,
indicated a slightly different angle of research and highlighted possible
considerations in the decision to use additional audio or video in
addition to text in reporting the news. Future research into accuracy and
reliability of Web news may consider this credibility issue.
Finally, replication of this study is recommended to verify these
results, or perhaps find information that may lead to different
conclusions. Replication would ideally incorporate larger numbers of
subjects who are more representative of the general population, and
future
study on this topic should consider the use of fewer news stories and
more
post-test items. The combination of text and video could function in the
same manner as the combination of text and pictures, and text
incorporating illustrative and somewhat redundant audio and video might
lead to higher recall, as was suggested by research in broadcast media
(Brosius, Donsbach, & Birk 1996). In any event, the combination of
multimedia and text is a new and exciting method of news presentation abo
ut which the body of knowledge is very small. Future research into this
new arena can only serve to increase knowledge and chart new paths
through
this relatively unexplored territory of message delivery. References
Alterman, R., & Bookman, L. A. (1992). Reasoning about a semantic memory
encoding of the connectivity of events. Cognitive Science, 16, 205-232.
Anderson, D. R., & Burns, J. (1991). Paying attention to television. In J.
Bryant & D. Zillmann (Eds.), Responding to the Screen: Reception and
Reaction Processes (pp. 3- 25). Hillsdale, NJ: Erlbaum.
Barber, P. 1998. Applied cognitive psychology: An information-processing
framework. New York: Methuen.
Bennett, E. M., Swenson, J. D., & Wilkinson, J. S. (1992). Is the medium
the message? An experimental test with morbid news. Journalism
Quarterly,
69, 921-928.
Berry, C. (1983). Learning from television news: A critique of the
research. Journal of Broadcasting, 27, 359-370.
Berry, C. (1988). Memory studies and broadcast messages. In M. M.
Gruneberg, P. E. Morris, & R. N. Sykes (Eds.), Practical Aspects of
Memory: Current Research and Issues. (Vol. 1, pp. 434-439). New York:
John
Wiley and Sons.
Bernard, R. M., & Coldevin, G. O. (1985). Effects of recap strategies on
television news recall and retention. Journal of Broadcasting and
Electronic Media, 29, 407-419.
Booth, A. (1970-71). The recall of news items. Public Opinion Quarterly, 34
, 604-610.
Brosius, H. B., Donsbach, W., & Birk, M. (1996). How do text-picture
relations affect the informational effectiveness of television newscasts.
Journal of Broadcasting and Electronic Media, 40, 180-195.
Browne, K. (1972). Comparison of factual recall from film and print
stimuli. Journalism Quarterly, 55, 350-353.
Craik, F. I. M., & Lockhart, R. S. (1972). Levels of processing: A
framework for memory research. Journal of Verbal Learning and Verbal
Behavior, 11, 671-684.
David, P. (1996). Role of imagery in recall of deviant news. Journalism
and Mass Communication Quarterly, 73, 804-820.
DeFleur, M. L., Davenport, L., Cronin, M., & DeFleur, M. (1992). Audience
recall of news stories presented by newspaper, computer, television and
radio. Journalism Quarterly, 69, 1010-1022.
DeFleur, M. L., & Cronin, M. M. (1991). Completeness and accuracy of
recall in the diffusion of the news from a newspaper vs. a television
source. Sociological Inquiry, 61, 148- 166.
Editor and Publisher Online. (July 1998). MediaINFO links on-line media
directory [On-line]. MediaInfo Links Database Statistics. Available:
http://www.mediainfo.com/ephome/npaper/nphtm/statistics.htm
Edwardson, M., Kent, K., Engstrom, E., & Hofmann, R. (1992). Audio recall
immediately following video change in television news. Journal of
Broadcasting and Electronic Media, 36, 395-410.
Findahl, O., & H ijer, B. (1985). Some characteristics of news memory and
comprehension. Journal of Broadcasting and Electronic Media, 29, 379-396.
Griffin, J. L., & Stevenson, R. L. (1996). The influence of statistical
graphics on newspaper reader recall. [On-line]. Visual Communication
Quarterly, 51. Available: http://sbweb3.med.iacnet.com/infotrac/session
Gunter, B., Furnham, A., & Gietson, G. (1984). Memory for the news as a
function of the channel of communication. Human Learning Journal of
Practical Research and Applications, 3, 265-271.
Gunter, B., Jarrett, J., & Furnham, A. (1983). Time of day effects on
immediate memory for television news. Human Learning Journal of Practical
Research and Applications, 2, 261-267.
Henderson, B., & Fernback, J. (1998). The campus press: A practical
approach to on-line newspapers. In D. L. Borden & K. Harvey (Eds.). The
Electronic Grapevine: Rumor, Reputation and Reporting in the New On-Line
Environment. (pp. 113- 121). Hillsdale, NJ: Erlbaum.
Herlitz, A., Nilsson, L. G., & Backman, L. (1997). Gender differences in
episodic memory. Memory and Cognition, 25, 801-811.
Housel, T. J. (1984). Understanding and recall of TV news. Journalism
Quarterly, 61, 505-508.
Hyatt, D., Riley, K., & Sederstrom, N. (1978). Recall of television
weather reports. Journalism Quarterly, 55, 306-310.
Kardes, F. R. (1993). Consumer inference: Determinants, consequences and
implications for advertising. In A. A. Mitchell (Ed.), Advertising
Exposure, Memory and Choice. (pp. 163-191). Hillsdale, NJ: Erlbaum.
Kirriemuir, J. (1997). The professional Web-zine and parallel publishing
[On-line]. Ariadne: The Web Version. Available:
http://www.dlib.org/dlib/february97/ariadne/02kirriemuir.html
Klein, A. (1978). How telecast's organization affects viewer retention.
Journalism Quarterly, 55, 356-359.
Lang, A. (1989). Effects of chronological presentation of information on
processing and memory for broadcast news. Journal of Broadcasting and
Electronic Media, 33, 441- 452.
Lang, A., Newhagen, J., & Reeves, B. (1996). Negative video as structure:
Emotion, attention, capacity and memory. Journal of Broadcast and
Electronic Media, 40, 460- 477.
Lang, A., Dhillon, K. & Dong, Q. (1995). The effects of emotional arousal
and valence on television viewers' cognitive capacity and memory. Journal
of Broadcast and Electronic Media, 39, 313-327.
Larsen, S. F. (1988). Remembering reported events: Memory for news in
ecological perspective. In M. M. Gruneberg, P. E. Morris, & R. N. Sykes
(Eds.), Practical Aspects of Memory: Current Research and Issues. (Vol.
1,
pp. 440-445). New York: John Wiley and Sons.
Magliano, J. P., Schleich, M. C., & Millis, K. K. (1998). Discourse
process and its relevance to the Web. In C. Forsythe, E. Grose, & J.
Ratner (Eds.), Human Factors and Web Development (pp. 45-61). Hillsdale,
NJ: Erlbaum. 45-61.
Myers, A., & Hansen, C. 1993. Experimental Psychology. (3rd ed.). Pacific
Grove, CA: Brooks/Cole.
Nielson, Jakob. (1996). Top ten mistakes in Web design [On-line]. Alertbox
. Available: http://www.useit.com/alertbox/9605.html
Nielsen, Jakob. (1997). How users read on the Web. [On-line]. Alertbox.
Available: http://www.useit.com/alertbox/9710a.html
Postman, N. 1985. Amusing ourselves to death: Public discourse in the age
of show business. New York : Viking.
Price, V., & Zaller, J. (1993). Who gets the news? Alternative measures
of news reception and their implications for research. Public Opinion
Quarterly, 57, 133-164.
Saljo, R., & Cedersund, E. (1988). The retention of news: On empirical
results and implicit assumptions of human communication. In M. M.
Gruneberg, P. E. Morris, & R. N. Sykes (Eds.), Practical Aspects of
Memory: Current Research and Issues. (Vol. 1, pp. 446-451). New York:
John
Wiley and Sons.
Scott, R. K., & Goff, D. H. (1988). How excitation from prior programming
affects television news recall. Journalism Quarterly, 65, 615-620.
Sundar, S. S. (1998). Effect of source attribution on perception of online
news stories. Journalism and Mass Communication Quarterly, 75, 55-68.
Wicks, R. H. (1992). Improvement over time in recall of media information:
An exploratory study. Journal of Broadcasting and Electronic Media, 36,
287-302.
Woodall, W. G., Davis, D. K., & Sahin, H. (1983). From the boob tube to
the black box: Television news comprehension from an information
processing perspective. Journal of Broadcasting, 27, 1-23.
|
