Cloze format for fourth year pupils.
Date
1974
Authors
Walmsley, Donald Percy
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Matching the readability level of available print materials with the reading ability of pupils was identified as a major problem of classroom teachers. Cloze procedure was found to be one method of making a suitable match. The advantages of cloze procedure were: (a) it was quick to prepare, administer, and mark and interpret; and (b) it measured the readability of the material and the reading ability of the subject with the same instrument. Cloze test scores could be used to indicate whether or not the material tested was within the subject's reading capability.
The purpose of the present study was to develop a simple, reliable cloze test format which had concurrent validity as a reading test.
Three aspects of cloze format were tested: (a) deletion rate, (b) test length, and (c) number of cloze tests required to adequately test a selection.
Data were used from a total of 132 subjects at the fourth year level in the two phases of the study. Three groups of 34 pupils took part in the first phase. Each wrote 1 set of 2 cloze test forms 300 words long. Each set used a different deletion rate. Reliabilities of the sets were compared. Next, reliabilities for lengths ranging from 30 to 300 words were compared. One group of 30 pupils took part in the second phase of the study. This group wrote three cloze test forms based on one selection and a standardized reading test. Their teacher ranked them according to her perception of their reading abilities. Reliabilities over 1, 2, and 3 forms were compared. As a result of the three comparisons cloze format (CF) consisting of one test form 250 words long with a 1-in-5 deletion rate was determined.
Split-half and alternate form reliabilities of CF were tested for significance and compared with similar reliabilities of a standardized reading test. The standardized reading test scores and the teacher's ranking were considered to be measures of reading ability. Correlations between these and CF were tested for significance.
Split-half reliability was .897, and alternate form reliability varied from .958 to .965. Correlations between CF and the standardized reading test varied from .945 to .952. The Spearman Rho correlation between CF and the teacher's ranking was .925. All results were significant at the .01 level of confidence.
It was concluded that CF was simple and reliable and that it ha d satisfactory concurrent validity as a test of reading ability.
It was noted that, to be useful for pupil placement in suitable reading materials, the individual CF scores had to be translated into readability scores. A table, compiled from the literature and not an aspect of this study, was provided for this purpose.