The Capabilities of four types of statistical models were examined to identify effective and ineffective schools through regression analysis. The models are referred to as incomplete since they included the following sets of predictors: unmatched longitudinal achievement and demographic (ULAD), cross-sectional scores and demographic (CSAD), demographic alone (ULDA), and longitudinal scores alone (MLA 3). The superiority of each model to other was determined by comparing results from each model with results from a complete model which included matched longitudinal achievement and demographic predictors. Regression equations with different numbers of predictors were investigated to determine the effect of the number of predictors on the ability of the incomplete models to produce results similar to those of the complete models. The analysis was based upon residuals from equations for each model computed for 277 schools. Residuals were examined by using three tools for classification of schools: confidence intervals based upon standard errors, performance index, and binomial distribution of scores from two different years. The summary of results of analysis are: (1) The two incomplete models, ULAD and CSAD, produced fewer misclassifications of schools than the other incomplete models. (2) No consistent effect of number of predictors was found. The number of different predictors in equations has a lesser effect in prediction than the type of predictors used in equations. (3) Each of the complete and incomplete equations showed a greater number of consistently overachieving schools than chance alone would allow. When such schools were identified, about 10 percent of the schools seemed able to raise their students, on average, by an amount equal to an increase from 50th to the 72nd percentile. (4) Confidence intervals of two standard errors were sufficient to avoid incorrect identification of schools; performance indices produced many classifications. / Source: Dissertation Abstracts International, Volume: 46-12, Section: A, page: 3561. / Thesis (Ph.D.)--The Florida State University, 1985.
| Identifer | oai:union.ndltd.org:fsu.edu/oai:fsu.digital.flvc.org:fsu_75700 |
| Contributors | SHRESTHA, GAMBHIR MAN., Florida State University |
| Source Sets | Florida State University |
| Detected Language | English |
| Type | Text |
| Format | 204 p. |
| Rights | On campus use only. |
| Relation | Dissertation Abstracts International |
Page generated in 0.0036 seconds