The Use of Media in Media Psychology
Byron Reeves, Leo Yeykelis & James J. Cummings (2016) The Use of Media in Media Psychology, Media Psychology, 19:1, 49-71, DOI: 10.1080/15213269.2015.1030083
Abstract
Media psychologists have theoretical interests in both people and media, yet research investments considerably favor subjects over stimuli. An analysis of 306 studies, taken from the journal Media Psychology over the last 10 years, and from the most cited media experiments in other journals, shows that studies invested in tens of thousands of human subjects but the studies used small samples of media material that were often narrow and unrepresentative. The vast majority of experiments (65%) used single examples of media messages per condition yet they discussed large categories of real world media experiences. Analysis of specific selections showed that media represented in research are less variant, nuanced, and idiosyncratic than media found in the real world. Two categories of solutions are discussed. First, new statistical solutions promote more attention to media repetitions analyzed as random factors. Second, we review the advantages of uncommon research designs that emphasize stimulus investments, including single subject designs that collect intra-individual data and that construct unique models using the entirety, rather than samples, of messages that people experience.
The Importance of Media Selections for Research
Media psychology studies tend to account for people variance by using large number of participants, but the variance among media is ignored.
Three reasons why media selections are important
Stimuli are most often the primary interests. Media psychologist are either aim to influence the media production or look at the influence of current media.
Due to large amount of variance in media (media are almost infinitely describable), focusing on media selection may has the highest return.
Variance in media experience is growing.
State of the Field
Studies from Media Psychology journal and Google Scholar are analyzed. The following attributes are reported:
Number of media stimuli: most of study uses just one. Even studies that uses multiple, they simply averages the effect.
Type of media
Number of citation by stimulus sampling
Discussion
Low media stimuli counts can have consequences in both external validity and construct validity.
Empirical evidence that failure to treat stimuli as random effect can have severe statistical consequences.
Single message stimuli is OK if that is the construct of the study, but often it is not.
We think we picked the messages that are representatives, but often we pick the idea messages that could lead to fake result (representative heuristic problem)
Type I, Type II, and even polar opposite conclusions are possible when failed to considering variance in media
Solution
Mixed Effect Model
Analysis of single subjects over time
Last updated