Daniel Sheehan

Effects of Technology Immersion on Middle School Students’ Learning Opportunities and Achievement

ABSTRACT An experimental study of the Technology Immersion model involved comparisons between 21 middle schools that received laptops for each teacher and student, instructional and learning resources, professional development, and technical and pedagogical support, and 21 control schools. Using hierarchical linear modeling to analyze longitudinal survey and achievement data, the authors found that Technology Immersion had a positive effect on students’ technology proficiency and the frequency of their technology-based class activities and small-group interactions. Disciplinary actions declined, but treatment students attended school somewhat less regularly than control students. There was no statistically significant immersion effect on students’ reading or mathematics achievement, but the direction of predicted effects was consistently positive and was replicated across student cohorts.

Coursework Mastery and School Success: Gender, Ethnicity, and Poverty Groups Within an Urban School District:

Unique data containing coursework-mastery test scores for all middle school students in a large urban school district are employed to test the narrowly defined meritocratic hypothesis that course-grade differentials for gender, ethnicity, and poverty groups are accounted for by the differential coursework mastery of these groups. A broadly defined meritocratic hypothesis is also tested by a model which includes measures of student absenteeism as well as “frog pond” and “bad school” contextual effects. Both the narrowly and broadly defined meritocratic hypotheses are rejected. With coursework mastery and the other variables held constant it is found that girls receive higher course grades than boys, Asians receive higher course grades than Anglos, and nonpoor youths receive higher course grades than poor youths. Black/Anglo and Hispanic/Anglo differentials are mixed. Implications are drawn for future studies of the educational stratification process.

A Comparison of the Selected- Response Strategy and the Constructed-Response Strategy for Assessment of a Third-Grade Writing Task

AbstractThe performance of third grade students on selected-response test items and constructed-response test items, intended to measure the identification and construction of complete sentences, was compared by factor analysis. Two factors associated with the two testing strategies were found, indicating that the strategies measure different tasks. In addition, two criterion variables, teachers’ judgment of mastery of the task and a written composition, were used to validate the factors. Teachers’ judgments associated with both factors, although somewhat more with the factor marked by the constructed-response test items. Mastery of the written composition was primarily associated with the factor marked by the constructed-response test items. From the results of this study, it is suggested that selected-response measures, having predictive validity, may lack construct validity.

Book Reviews : Jon Clark Marshall and Loyde Wesley Hales. Essentials of Testing. Reading, Massachusetts: Addison-Wesley Publishing Company, 1972. Pp. ix + 162.

differences, but it gets lost among the repeated references to accepting the null hypothesis. (The reviewer has a preference for never accepting the null hypothesis-only failing to reject it, perhaps discussing /3’s for various differences of practical consequence.) Confidence intervals for means are not even discussed. The foregoing should not be interpreted as approval by the reviewer of practices, such as reporting &dquo;almost significant&dquo; results, which the author quite rightly condemns. Instead, the mistake to which the author’s approach may lead is one of naively accepting and rigidly adhering to a statistical model which may have begun to show inadequacies. We have had too many researchers dogmatically conclude (within the limits of probability) that A is better (or worse) than B or that A is as good as B on the basis of a single statistical test, and this book does not help at all in alleviating this situation. The above criticisms should certainly not deter use of this book