EFFECTS OF ALTERNATIVE DELETION PATTERNS,BLANK CONDITIONS,AND SCORING SYSTEMS ON INTERMEDIATE LEVEL CLOZE TEST PERFORMANCE

The majority of cloze research has focused on a traditional test format whereby (1) deletions occur on an every‐nth word basis, (2) standard length blank spaces are provided, and (3) only exact word replacements of the original text receive credit. In this study, an analysis was made of the effects of selected deletion patterns (every‐nth versus total random), blank conditions (standard versus cued), and scoring modes (verbatim versus synonymic) on the cloze test performance of 64 sixth grade readers. Neither the deletion pattern nor blank condition main effects achieved significance; however, as expected, increased performance was observed for synonymic scoring. Most importantly, all first‐order interaction effects attained significance. In the total random/cued/ verbatim cloze format, context distances of less than four words generated correct responses as effectively as context distances of four, the equivalent of an every‐5th word deletion strategy. Further, this format exceeded all others in internal consistency and in its ranking of subjects' performance relative to their synonymic cloze test scores.