Follow this and additional works at: https://docs.lib.purdue.edu/clcweb
Part of the Comparative Literature Commons, and the Critical and Cultural Studies CommonsDedicated to the dissemination of scholarly and professional information, Purdue University Press selects, develops, and distributes quality resources in several key subject areas for which its parent university is famous, including business, technology, health, veterinary medicine, and other selected disciplines in the humanities and sciences.CLCWeb: Comparative Literature and Culture, the peer-reviewed, full-text, and open-access learned journal in the humanities and social sciences, publishes new scholarship following tenets of the discipline of comparative literature and the field of cultural studies designated as "comparative cultural studies." Publications in the journal are indexed in the Annual Bibliography of English Language and Literature (Chadwyck-Healey) The study of literature attaches much worth to putting theory to the test. In metaphor theory also a great deal of experimentation is found, particularly among cognitive psychologists with an interest in (literary) language and among literary scholars with an interest in psychological methods. Although the experiments may obtain valuable information, much of its value is lost or interpreted wrongly by improper use of the statistical tools of analysis. In this article, I explain that many experiments that supposedly have a 2x2 design, actually used more factors or factor levels, and thus contain a higher-order interaction. For the research question, such interactions seem unimportant. They often are resulting from trivial manipulation: The order of presenting the stimuli, or the order of using the preferred hand in a reaction-time (RT) choice task. Yet, I will point out that higher-order interactions may cast a completely different light on what seems a straightforward result. The rationale of my article is quite simple. Suppose a group of expert readers is contrasted with a group of novices to see whether education improves the understanding of postmodern poetry. It turns out that experts find postmodern poems better to comprehend than novices. Of course, experts are better educated, and practice facilitates comprehension. However, this interpretation is based purely on analysis of a main effect (experts vs. novices). Suppose, however, that the results show no difference for sex in novices but do show that female experts comprehend postmodern poetry much better than expert men do. It now turns out that comprehensibility only increased for expert female readers. In other words, education only facilitates comprehension for women. Men stay as stupid as they were. Thus, the interaction between expertise and sex strongly limits the generalization that education improves comprehensibility. Naturally, most researchers do not neglect the possible effects of such important factors as sex and age. However, less obvious factors are ignored all too often, although they may limit findings in quite the sa...