Media effects is a research paradigm concerned with the consequences of media use or exposure. As it is generally used in the study of communication, media effects refers to the consequences of media use experienced by the individual message recipient or audience member. These effects can be cognitive, affective, attitudinal, or behavioral; they can occur in domains as diverse as health, politics, aggression, sexuality, education, child development, and persuasion. Media effects can be driven by characteristics of media messages (e.g., the effects of violent or sexual content), the characteristics of the medium of communication by which those messages are conveyed (e.g., the effects of screen size), or the unique intersection between message and medium. This entry provides a brief overview of the most common research methods used to examine the consequences of media exposure in addition to central concerns and identifies a number of key considerations in applying those methods to media effects research specifically.
Media effects research is characterized by two key elements. The first element is media. Here, “real life” media and their messages are emphasized. Specifically, media effects research tends to examine the effects of actual messages or message patterns or, if dictated by the research question, to examine the effects of messages that, though constructed strictly for the sake of research, are typical of common media messages. The question, then, tends to be not what effects media might elicit but what effects media actually or probably elicit. The second key concept is that of effects. Here, media effects research is concerned with the effects or consequences of media exposure. As such, determining causality between media and effects is a central for media effects researchers.
To examine the effects of media use and exposure, researchers use a variety of methodologies, including content analyses, controlled experiments, longitudinal surveys, naturalistic observations, meta-analyses, and more recently, neurocognitive analyses using tools such as functional magnetic resonance imaging (fMRI) and event-related brain potentials. The research methodology selected depends on the type of questions media researchers are interested in studying. These methods are briefly discussed herein.
Content analysis is a well-established research method involving the systematic examination and subsequent quantitative description of specified elements of various media content. This type of analysis allows researchers to describe and make inferences about media messages and, when in light of theories of media influence, to predict effects of media content on audiences. Examples of content analyses include the analysis of television commercials during children’s programming, the depiction of female characters on video game cover inserts, and the presence and frequency of acts of aggression in film previews.
Controlled experiments are commonly employed in media effects research to determine the causal relationships between media exposure or use and related outcomes and to identify mediating or explanatory factors. In controlled experiments, researchers manipulate elements of media content and compare the reactions, attitudes, beliefs, or behaviors of audiences exposed to messages with those elements to those exposed to messages without. Examples include comparing the effects of health messages containing fear appeals to comparable messages without fear appeals on audience’s self-efficacy and information-seeking behavior, or comparing viewers’ perceptions of political candidates and voting intentions after exposure to different political advertisements or news stories.
Surveys, both cross-sectional and longitudinal, are also used to analyze relationships between media consumption and audience beliefs, attitudes, and behaviors, as well as to examine moderators of these relationships. Cross-sectional surveys are used to identify relationships among media use and its anticipated consequences. In longitudinal surveys, by collecting data from the same survey respondents across time, researchers are able to assess individual-level changes, the media antecedents of these changes, and the conditions under which these changes are more or less likely to occur. Examples include examining the influence of childhood exposure to violent television content on aggression in adulthood and exploring the effects of exposure to sexual media content on the sexual beliefs, attitudes, and behaviors of adolescents.
Neurocognitive analyses involve the observation of media influence through the measurement of brain activity with fMRI or EEG. By examining real-time neuro-physiological reactions to different types of media content, researchers better understand how neural and psychological processes (e.g., regulation of emotion, storage and retrieval of memories, and motor functioning) occur during media exposure as well as how neural function changes in response to consistent or prolonged exposure to a single type of media content. Examples include observing brain activity during exposure to violent images among individuals with varying experience playing violent video games, or observing neural activity among children exposed to emotional or nonemotional stimuli.
Naturalistic observation is another category of research method used by media effects researchers to examine the effects of exposure to media or their messages. Naturalistic observation involves identifying naturally occurring differences in media availability or use across different groups or populations and observing differences in behavior across those groups. This allows researchers to observe and record behaviors in a natural field setting either through direct observation or by comparing population-level statistics. This avoids potential problems with self-report biases. Examples have included comparing arrest rates for violent crimes between competitor cities after televised sporting contests and observing differences in reading skills among children in towns with limited or unlimited access to television programming.
Meta-analysis is a research method conducted by media effects researchers to synthesize results across a set of studies investigating the same underlying phenomenon, identify general patterns across findings, and determine the strength of media effects. This methodology allows researchers to conduct a rigorous, statistical comparison of published and unpublished studies that examine the same topic but vary in a number of ways (e.g., location, sample size, environment, social and economic conditions, etc.). Examples include analyses of studies on violent video games and their effects on aggression and prosocial behavior and of studies examining the role of media images on women’s body image concerns.
These methods are employed across a wide range of research areas across the social sciences, and are not specific to media effects research. However, in applying these methods to questions of media influence, a number of special issues are generally of relatively greater import.
The central consideration of media effects research is the demonstration of effects. In order to conclusively demonstrate causal relationships, media effects researchers rely heavily on controlled experiments with random assignment to condition. Experimental control, with regard to media content as causal factors, requires that stimuli across conditions be essentially identical except with regard to the causal variable of interest. Experimental stimuli vary in the degree to which this control is achieved and, therefore, the degree to which causality can be conclusively attributed to the construct under investigation. Some experiments employ entirely different television programs, films, or advertisements as experimental stimuli; others selectively edit existing content in order to maintain greater control; and yet others employ stimuli created expressly for the purpose of the experiment.
Media effects research is concerned with documenting effects of actual or typical media messages. To hold implications for real-world contexts, media effects research must employ laboratory analogues and survey measures that allow for generalization beyond the study findings. Important factors with regard to external and ecological validity in media effects research include the nature of the media stimuli (or media use measurement), the amount of exposure, and the nature of the exposure experience.
Because of limitations of the laboratory setting and concern for experimental control, the type of stimuli used in treatment groups and the duration of exposure to this treatment is sometimes less reflective of media exposure in natural settings. In laboratory experiments, real-world media content may be altered to control for potential extraneous variables. For example, a script or narrative text may be used in place of audiovisual media in order to manipulate the variable of interest more easily and limit the influence of other unrelated factors such as lighting, camera angle, and voice. In addition, the duration of exposure to treatment stimuli may be significantly reduced due to sample constraints and/or the need to collect data within a reasonable amount of time (e.g., limited means to conduct repeated measures experiment). For example, an experiment examining the effects of particular film content on viewers may use abbreviated film previews in place of watching entire films. Adapting media content for the purpose of gaining greater control, however, may compromise the extent to which the nature of and exposure to the treatment mimics media exposure in, and therefore generalizes to, natural settings.
The selection of media stimuli must be informed by an awareness of the complexity of media messages. Media selected for one characteristic will also necessarily include many others that may have bearing on study outcomes. For example, a researcher examining the effects of violent content may choose a violent film and a nonviolent film, but a myriad of other characteristics of the violent film may contribute to its influence, including attractiveness of the aggressor, justification of the aggression, the presence or absence of weapons, and so on. For this reason, media effects researchers must be particularly cognizant of the diversity and complexity of media messages. Many researchers address this by including multiple, diverse stimuli.
Another challenge in laboratory studies of media effects lies in attempts to observe the process by which these effects occur by measuring online thoughts, emotions, and/or evaluations during media use. In order to assess real-time cognitions or affective reactions to media content, participants are frequently asked to engage in tasks before, during, or immediately after exposure. These measures might include, for example, clicking, listing thoughts, or indicating liking at some point during treatment exposure. Alternately, participants might be fitted with sensors that measure skin conductance, heart rate, or the contraction of facial muscles indicative of specific affective states. Although these intrusive measures often provide insight into participants’ thoughts and feelings with regard to particular media content, they also interrupt or alter the exposure experience thereby reducing the extent to which exposure matches an actual viewing experience.
Media exposure is largely self-directed and governed by an individual’s needs, wants, interests, and social, physical, and media environment at the time of exposure. Research paradigms that focus on this selection include selective exposure, including mood management, and uses and gratifications. Research in these paradigms has demonstrated that the media content to which audiences are ultimately exposed is neither random nor necessarily representative of the overall media environment. Instead, variation in needs, wants, goals, personality traits, attitudes, predispositions, affect, and other traits and states shapes media choices. For example, people tend to systematically select political messages that are consistent with their existing political attitudes and beliefs; men who are angry choose more negatively skewed media content if they anticipate an opportunity for revenge against the individual who angered them; and young people are more likely to select television programs in which members of their own racial or ethnic group play key roles.
Selective exposure limits the external validity of experimental research into media effects. First, the media content that is being investigated as a cause of whatever outcome may have been entirely avoided by the experimental subjects. Second, if those subjects had selected that content, it likely would have been under a set of circumstances that influenced the experience, its meaning, and, quite possibly, its effects. Research into the effects of sexually explicit media content is illustrative. Some participants in experiments investigating such effects may eschew sexually explicit content entirely in their regular viewing choices. Other participants may ordinarily do so only under limited social, relational, or emotional circumstances that substantially shape the meaning of the experience. Selective exposure concerns are addressed somewhat by research methods that employ observation or self-report measures of media exposure rather than manipulation, such as longitudinal surveys or unobtrusive observation.
The operationalization and assessment of viewers’ media use and consumption are also of fundamental concern for the study of media effects. Self-report measures, in which respondents are asked about the types of and frequency of exposure to particular media, have a long tradition of use in media effects research. However, self-report measures rely on several assumptions that may be problematic and may therefore suffer from systematic error. In order for these measures to be valid, the researcher must assume the respondent can accurately recall their own media use when cued to do so and that they will report that use in an unbiased way.
Self-report measures of media use and consumption often assume that respondents can accurately recall and estimate the frequency and duration of exposure to various types of media content. These assumptions are unrealistic. The use of these measures, then, may lead to retrospective biases, underestimates of effect sizes, lack of control for spurious variables, and reverse causality issues. Media effects researchers have attempted to address these biases by using guided recall and recognition measures rather than free recall; such measures may provide, for example, a list of various television programs, movies, and popular publications and ask participants to indicate their experience with each. Such measures are not without their limitations; they assume an exhaustive list of relevant media content, an assumption that is increasingly unfounded in the age of on-demand media content. Media diaries, in which respondents are asked to record their media consumption as it occurs, are another example of an alternative method to measuring media consumption that may reduce some of the error inherent in self-report measures. Media effects researchers may also reduce uncertainty regarding the actual content of media exposure specified by respondents by conducting content analyses of such media to use in conjunction with self-reports.
In addition, self-report measures are frequently used to analyze why respondents consume particular media. This type of assessment assumes that respondents are either conscious of their reasons for consuming particular media content or can be made aware of their motives for consumption through various techniques. Self-reports may also be influenced by the socioeconomic status of respondents, social desirability, and inconsistent conceptualizations. To address this issue, media effects researchers may examine whether various indicators of motives for media consumption correlate with measures of media exposure and selection to determine the relationship between media use, consumption, and motivations. Developing more precise measures of media consumption is necessary to facilitate progress in media effects research.
In addition, postexposure tasks attempting to assess behaviors and/or behavioral intent often do not measure natural behaviors. For example, participants exposed to violent media content may be asked to engage in tasks such as choosing the amount of hot sauce to allocate to a fellow participant (i.e., measuring aggressive behavior) or interrupting a simulated fight outside the laboratory (i.e., measuring prosocial behavior). Although such measures allow researchers to assess the influence of media content on subsequent behaviors, these artificial measures may be problematic because they are not assessments of actual behaviors that occur in natural settings.
Combining the results of multiple studies that employ different methods, each with differing strengths in terms of internal, ecological, and external validity, allows stronger conclusions to be reached. This approach has been labeled triangulation by many researchers. Triangulation has played an important role in demonstrating media effects in a number of domains including the effect of exposure to violent media on aggression; controlled experiments have demonstrated that violent media play a causal role, surveys have documented that these effects occur with real-world media exposure, and naturalistic observation has documented that these observations are not merely a function of self-report.
Meta-analysis provides an objective, statistical tool for combining the results of studies that share the same question but employ diverse methods.
Ethical considerations around media effects research can be organized around two major concerns. First, experimental protocols that involve exposing individuals to potentially harmful media (e.g., graphically violent material, sexually explicit content) are ethically fraught. Although it seems likely that most effects of experimental exposure to problematic content are relatively brief or can be limited or eliminated by appropriate debriefing, they are not necessarily inconsequential. Various media effects experiments have resulted in arguably harmful outcomes (e.g., some theorists maintain that exposure to pornography leads to a greater tolerance for violence against women and for this reason, media effects research involving exposure to pornography is seen potentially harmful). In addition, a substantial portion of media effects research investigates effects of media on children; during relatively vulnerable or impressionable developmental stages, even relatively brief exposure can have lasting effects. Fear effects in response to media messages among children, for example, have been found to have effects that linger for years for some child viewers.
The second ethical consideration, and the one that is relatively unique to media effects research, deals with exposing people to media content that they may find inherently deeply objectionable. Some types of media content may be perceived as inherently wrong even to view. Viewing any sexually explicit material, for example, is a violation of some religious and moral codes. Depictions of graphic violence are seen as dehumanizing by some viewers. Experimentally exposing individuals to materials they do not wish to view is unethical. For this reason, complete and detailed consent procedures should be employed to prevent such exposure.