Result Generalizability and Detection of Discrepant Data Points: Illustrating the Jackknife Method.
Saved in:
| Title: | Result Generalizability and Detection of Discrepant Data Points: Illustrating the Jackknife Method. |
|---|---|
| Language: | English |
| Authors: | White, Amy E. |
| Peer Reviewed: | N |
| Page Count: | 14 |
| Publication Date: | 2000 |
| Document Type: | Reports - Descriptive Speeches/Meeting Papers |
| Descriptors: | Data Analysis, Educational Research, Estimation (Mathematics), Generalization |
| Abstract: | The jackknife, as refined by J. Tukey (1958), is a valuable tool for the internal replication of a study. The jackknife statistic is particularly useful with small sample sizes. Large samples are labor intensive, and other methods better address this situation. The jackknife procedure involves the use of a single sample drawn from a normally distributed population. The jackknife procedure is a general method for reducing the bias in an estimator while providing a measure of the variance of the resulting estimator by sample reuse. The essence of the jackknife approach is to partition out the impact of a particular subset of the data on an estimate derived from the total sample. The method attempts to determine if any one case or group of cases exerts an inappropriate influence on the overall statistic of interest. To illustrate the value of the jackknife, an example is presented that uses actual educational research data. The study (B. White and L. Daniel, 1999) concerned career motivations of persons planning to teach. (Contains 4 tables and 13 references.) (SLD) |
| Entry Date: | 2001 |
| Accession Number: | ED445079 |
| Database: | ERIC |
| Abstract: | The jackknife, as refined by J. Tukey (1958), is a valuable tool for the internal replication of a study. The jackknife statistic is particularly useful with small sample sizes. Large samples are labor intensive, and other methods better address this situation. The jackknife procedure involves the use of a single sample drawn from a normally distributed population. The jackknife procedure is a general method for reducing the bias in an estimator while providing a measure of the variance of the resulting estimator by sample reuse. The essence of the jackknife approach is to partition out the impact of a particular subset of the data on an estimate derived from the total sample. The method attempts to determine if any one case or group of cases exerts an inappropriate influence on the overall statistic of interest. To illustrate the value of the jackknife, an example is presented that uses actual educational research data. The study (B. White and L. Daniel, 1999) concerned career motivations of persons planning to teach. (Contains 4 tables and 13 references.) (SLD) |
|---|
