A Comparison of Bayesian and Traditional Indices for Measuring Information Gain Sensitivity in a Cloze Test

摘要: Standardized reading comprehension tests have been criticized for their inability to assess what a reader has gleaned from the passage during process (Tuinman, 1973). Tuinman reported scores as high 65 percent correct on multiple-choice which subjects answered without having had access passages questions were based. Tuinman’s paper, among others, caused researchers focus distinction between and information gain (IG). In general, concept of IG involves comparing reader’s state knowledge before after test passage. One method assessing IG, suggested by Coleman Miller (1967), is obtain cloze an opportunity read in its original, unmutilated form; difference posttest pretest considered be score.