Appendix: Evaluating Evidence
Published online by Cambridge University Press: 28 July 2009
In this appendix, we outline some aspects of how the results of an experiment are evaluated. Although this clearly cannot be a comprehensive treatment of data analysis techniques, even as it applies to the results presented in this book, it at least may acquaint the reader with some of the relevant concepts. For readers with some background in statistical methods, the appendix also documents some procedures for calculating the likelihood ratios we use to compare models.
In the vocabulary of experimental design, a manipulated independent variable is referred to as a factor, and each possible value of that variable is a factor level. For example, if one presents readers with two different versions of a story, one would say that the factor of story version has two levels. Often experiments have more than one factor. In a factorial experiment, each possible level of one factor is combined with each possible level of the other factors, and each combination of factor levels determines a particular experimental condition. For example, suppose the factor of story version was factorially combined with the factor of reading goal with two levels: reading to identify the narrator's point or reading to identify the plot events. In such a design, there would be four conditions: version 1 read for narratorial point; version 1 read for plot events; version 2 read for narratorial point; and version 2 read for plot events. Two types of results can be examined in a factorial experiment.
- PsychonarratologyFoundations for the Empirical Study of Literary Response, pp. 261 - 274Publisher: Cambridge University PressPrint publication year: 2002