244 reading comprehension items taken from 28 GRE forms comprised the total item sample. The total number of reading passages represented was 110. Each test form consisted of four reading passages, two long (400 words or more) and two short (200 words or less).

Boston University

Experimenters identified a set of variables from previous literature that are predictive of comprehension difficulty. By obtaining evidence that confirms that these variables are predictive of GRE comprehension item difficulty, it can support the claim that GRE is a measure of passage comprehension

Predicting reading item difficulty and assessing multiple-choice format tests

Freedle, R., & Kostin, I. (1991). The prediction of GRE reading comprehension item difficulty for expository prose passages for each of three item types: Main ideas, inferences and explicit statements. ETS Research Report Series, 1991(2), i-53.

The prediction of GRE reading comprehension item difficulty for expository prose passages for each of three item types: Main ideas, inferences and explicit statements

The following variables are expected to significantly influence reading item difficulty as determined within a multiple-choice testing format: 

- Negations (Carpenter and Just, 1975)
- Referentials (Abrahamsen and Shelton, 1989)
- Rhetorical organizers (Hare et al. 1989)
- Fronted structures (Freedle, Fine, and Fellbaum, 1981)
	- cleft structures (eg. “lt is true that she found the dog,”)
	- marked topics consisting of several subtypes
	- combinations of the above structures (eg. "But, briefly, this didn't stop him”)
- Vocabulary (Graves, 1986)
- Sentence length (Klare, 1974-1975)
- Paragraph length (Hites, 1950)
- Number of paragraphs (Freedle, Fine, & Fellbaum, 1981)
- Abstractness of text (Paivio, 1986)
- Location of relevant text information (Kieras’, 1985 and Hare et al., 1989)

Hypothesis 1

Many of the 10 categories of variables provide independent predictive information in accounting for reading item difficulty. This applies to naturalistic texts as well as the more controlled texts

Hypothesis 2 

Study Dataset

Research provides considerable evidence that multiple-choice tests of reading comprehension yield results consistent with those obtained from controlled experimental studies dealing with language comprehension. These findings also apply specifically to Graduate Record Examinations (GRE) data sets, supporting the validity of the multiple-choice format.

Learn Before

Related