Klein-Braley, C. (1997). C-tests in the context of reduced redundancy: an appraisal. Language Testing, 14(1), 47–84.
Population: Students at the University of Duisburg, Germany.
Size: 81 completed the entire battery of tests. The control criteria are DELTA scores.
Details: The purpose of this large-scale comparison of the various reduced redundancy approaches was to compare the university’s own C-test to the rest of the tests. The battery of tests consisted of 2 cloze tests, 1 four-part C-tests, 2 four-choice cloze tests, 2 cloze-elide tests, and the dictation portion of the DELTA test.
CLOZE1 used the fixed 7th word deletion rate; CLOZE2 deletions were made using a random number table. (Exact scoring was used. Each had 20 items.)
The 100 possible points on the C-test were derived from four short paragraph-style texts with 25 mutilations each (C1, C2, C3, & C4).
Of the multiple-choice cloze tests (MC1 & MC2), one used the 26 blanks specified by Jonz (1976); the other used the 22 blanks format used by Manning in his (1986) study.
The cloze-elide tests (CE1 & CE2) used ANIMAL MIGRATION and ARCHEOLOGY, which were also taken from Manning’s (1986) study.
The DELTA dictation test (DICT) allocates 50 points, then subtracts 1 point for each erroneous deletion or non-deletion.
Conclusions: Difficulty rankings (P = .5 is ideal): (high > low)
C-TEST (.52), ALLCE (.46), ALLMC (.60), ALLCLOZE (.27)
Reliability (KR-21): (high > low) DICT (.94), DELTA (.93),
C-TEST (.85), ALLCE (.75), ALLCLOZE (.66), ALLMC (.55)
Rank of Usability (performance and ease of use): (high > low)
C-TEST, DICT, ALLMC, and ALLCLOZE tied with ALLCE.
Ease of construction: (high > low)
DICT, ALLCLOZE, C-TEST, ALLCE, ALLMC.
Scoring: (high > low) ALLMC, C-TEST, ALLCLOZE, DICT.
Overall: (high > low) C-TEST followed by DICTATION.
Comments: To enhance the areas where the C-test ranked behind, I created ClozeOnline™ which now offers instant construction and scoring. Create your own at clozeonline.us.