Search

Evaluations in education often throw away important information because of a penchant for averages. Multiple regression techniques are used to estimate the average effect of policies across schools, and usually school performance is represented by the average score of its students on an achievement test. The author suggests some ways of broadening educational evaluations: (1) to consider "outliers," or exceptional performers among schools; and especially, (2) to consider other statistics of a school's distribution of scores besides the mean, which have an intuitive link to ill-defined but still meaningful educational objectives like equality, mobility, success with exceptional children, and attainment of certain minimum levels of skills. Tables reflecting the author's research are included. (Author/SE)

To the data analyst, outliers can present both a problem and an opportunity. Stray or outlying observations can severly distort estimates of a distribution's central tendency (like the mean) and estimates of one variable's relationship to another (like the regression coefficient). These problems are frequent and serious, and as a result, increasing numbers of statisticians are developing new estimating procedures that are robust in the face of outliers. But outliers may also present an opportunity. An unusual observation may indicate the existence of a process not operating in the rest of the distribution. In education, for example, an outlier may turn out to be an unusually effective school, perhaps one worthy of emulation throughout the educational system. Finding the best by locating outliers may be particularly important in the evaluation of public policies. Can one find exceptionally good police forces and study the causes for their success? What about outstanding rural development projects, exceptional hospitals, and unusally effective manpower training programs? Indeed, one may suggest a worthwhile general rule for policy evaluations: For exceptions to the general rules, include a search for unusual performers. (Author)

Studies were conducted to determine whether measures of intellectual merit that are used as criteria for admission to the University of Karachi in Pakistan have predictive validity. The predictive validity of the intermediate examination score was studied with recent graduates of the pharmacy, medicine, and engineering schools. Additional studies of business school and arts and sciences graduates were also conducted. Regression analyses indicated that the current criterion of merit (intermediate score) explained little of a student's subsequent university performance. Even after including matriculate test scores and several independent variables in the analysis, the regressions had little predictive power. Possible explanations of the findings and a discussion of the pros and cons of merit-based admissions criteria are presented. (SW)

This report deals with new ways to look at achievement test scores as measures of educational outcomes. It attempts to show how existing data might be used more productively by planners and evaluators, by suggesting how test score statistics other than the mean might be used, and how they might indicate success along a wide variety of educational objectives. The report is a theoretical effort in which the empirical behavior of the proposed new statistics is examined briefly. (BB)

It was concluded, in an empirical study of distributions of examination scores within secondary schools in Karachi, Pakistan, that intraschool inequalities in student achievement were relatively large; that variables for socioeconomic background and school policy did not explain the intraschool variations across schools; and that a school which was particularly equal on one test was not particularly equal on another. Data were gathered from a stratified random sample of 207 secondary schools, based on neighborhood socioeconomic status; type of school, including government, nationalized, or private; and the mix of male and female students. Results were analyzed from scores on tenth-grade essay tests in the humanities and sciences. Data on teacher characteristics were used to calculate student-teacher ratio and teacher experience. Schools were classified according to the median income of students' households. Based on results of the study, it was hypothesized that Karachi school personnel were indifferent to the intraschool variations or did not have effective means for closing the gap. Futher research, using longitudinal data, was recommended. (MH)

In an examination of the main issues facing international studies at American universities, demands for more centers, more scholarships, and more research funds are discounted. Close review of the educational and scholarly aims and achievements of international research centers is advocated. It is suggested that useful reviews of this sort should be local--at particular centers and universities--and should focus on new sorts of planning, management, and integration as the key steps for the future. Some of the main issues related to university-based reviews, particularly of area centers, are outlined. Three questions are addressed: (1) What are the objectives of area research? (2) According to insiders, what have been the principal achievements?; and (3) What are some of the main problems and opportunities facing area centers now? Implications for universities are cited regarding undergraduate training, the research agenda, professional schools; information, and management. Suggestions are provided for topics for university-based reviews: e.g., determining the most effective type of coordinating council across centers; investigating whether an associate dean or a provost for international activities would be desirable; and assessing long-term stays by foreign visitors so they could participate in teaching as well as research. A list of references is included. (Author/MSE)