Correlations, Contrasts, and Conceptual Clarity.
Saved in:
| Title: | Correlations, Contrasts, and Conceptual Clarity. |
|---|---|
| Language: | English |
| Authors: | Rosenthal, Robert |
| Peer Reviewed: | N |
| Page Count: | 57 |
| Publication Date: | 2002 |
| Document Type: | Reports - Descriptive Speeches/Meeting Papers |
| Descriptors: | Correlation, Effect Size, Estimation (Mathematics) |
| Abstract: | This paper discusses the Pearson product moment correlation (K. Pearson, 1986). Although this correlational metric is old, published in the 19th century, this paper suggests that it remains the most nearly universally applicable index of effect size. It is difficult to imagine a situation in which a Pearson "r" or its equivalent could not be used appropriately to index the magnitude of an event. And because all Pearson "r"s and their equivalents are based on focused comparisons, or contrasts, rather than on diffuse or omnibus comparisons, there is far greater conceptual clarity in the use of "r" than in the use of some other effect sizes measures. This paper illustrates the claims made for Pearson's "r," and discusses some applications and devices that make it more widely applicable. The paper also describes a new statistic that allows the accurate estimation of an effect size called "r" equivalent. It is also noted that there are actually four "r"s that can be usefully used as effect size estimates. Each of these is discussed. (Contains 11 tables and 52 references.) (SLD) |
| Entry Date: | 2003 |
| Accession Number: | ED473809 |
| Database: | ERIC |
| Abstract: | This paper discusses the Pearson product moment correlation (K. Pearson, 1986). Although this correlational metric is old, published in the 19th century, this paper suggests that it remains the most nearly universally applicable index of effect size. It is difficult to imagine a situation in which a Pearson "r" or its equivalent could not be used appropriately to index the magnitude of an event. And because all Pearson "r"s and their equivalents are based on focused comparisons, or contrasts, rather than on diffuse or omnibus comparisons, there is far greater conceptual clarity in the use of "r" than in the use of some other effect sizes measures. This paper illustrates the claims made for Pearson's "r," and discusses some applications and devices that make it more widely applicable. The paper also describes a new statistic that allows the accurate estimation of an effect size called "r" equivalent. It is also noted that there are actually four "r"s that can be usefully used as effect size estimates. Each of these is discussed. (Contains 11 tables and 52 references.) (SLD) |
|---|
