An investigation into the relationships between cloze test scores and informal reading inventory scores of fifth grade pupils
This study investigated the relationship between instructional level scores as determined by a cloze test and instructional level scores as determined by an informal reading inventory. Subjects used in this study were selected from schools in a mid-western community. Five schools were selected from a total of twenty-two elementary schools by means of a random numbers table. Administrative officials acknowledged the fact that the schools selected were representative of the various socio-economic classes serviced by the school system. Fifty male and fifty female subjects were selected from the total fifth grade population of these five schools. This was accomplished by matching assigned pupil numbers with random numbers generated by a computer.A cloze test and an informal reading inventory (IRI) were administered to all subjects over a six-week period during the months of April and May of 1972. For purposes of this study, the IRI was accepted as the criterion with which all other variables were compared. Teachers' estimates of their pupils' instructional levels were also gathered during the testing period. Participating teachers were not able to consult previously determined estimates of their pupils' instructional levels during the data gathering process.Statistical analysis of the data was accomplished by the application of the Pearson Product Moment Correlation. A classification analysis yielded information in terms of the quantities of scores on any two variables which were in exact agreement. Also determined were the quantities of scores which ranged within plus or minus one year of each other, and the quantity of scores which ranged within plus or minus two or more years of each other.The testing of five null hypotheses resulted in significant correlation coefficients at the .01 level between:1. Instructional level scores as determined by an IRI and instructional level scores as determined by a cloze test (.78).2. Instructional level scores as determined by an IRI and teachers' estimates of their pupils' instructional levels (.82).3. Instructional level scores as determined by a cloze test and teachers' estimates of their pupils' instructional levels (.74).4. Instructional level scores as determined bya cloze test and instructional level scores as determined by the word recognition subtest of the IRI (.78).5. Instructional level scores as determined by a cloze test and instructional level scores as determined by the comprehension subtest of the IRI (.69).A classification analysis revealed that while correlations were highly significant, the percentage of pupils' scores which were in exact agreement was 26.2 per cent for all hypotheses tested. Approximately 38.2 per cent of the pupils' scores fell within a range of plus or minus one year of each other. The remaining scores which amounted to 35.6 per cent fell within a range of plus or minus two or more years of each other for all hypotheses tested.These findings seem to indicate that high correlation coefficients are an inadequate criterion on which to accept the cloze procedure as a valid technique for determining the instructional levels of pupils, since only thirty-one per cent of the population sample made identical instructional level scores on both the cloze test and the IRI.An additional finding of this study is that teachers' estimates of their pupils' instructional levels are inaccurate to the point that their continued usecannot be justified. Only twenty-two per cent of teachers' estimates are in exact agreement with the instructional level scores as determined by the IRI.In conclusion, the possibility exists that the cloze procedure may yet provide classroom teachers with a technique for assessing instructional level. However, this practice cannot be recommended based upon the findings of previously conducted correlational studies.