The primary purpose of this research was to investigate, using Monte Carlo procedures, the stability and bias of canonical results as a function of sample size, average within-set correlations, average between-set correlations, and number of variables and to determine the feasibility of using matrix sampling in the context of canonical correlational analysis. The major findings from this research indicated that: / Under the conditions examined, the standard errors associated with the canonical correlations and the redundancy index were small and the standard errors of canonical weights and variate-variable correlations were large. The effect of sample size was moderate for canonical correlations and the redundancy index and large for canonical weights and variate-variable correlations. The effect of within-set correlations was small for the first and second canonical correlations and the redundancy index, large for canonical variate-variable correlations, and extremely large for canonical weights. The effect of between-set correlations was small for the second canonical correlation, moderate for the first canonical correlation and the redundancy index, large for canonical variate-variable correlations, and extremely large for canonical weights. The effect of number of variables was moderate for the first and second canonical correlations and the redundancy index, extremely large for canonical weights, and inconsistent for canonical variate-variable correlations. / It was recommended that the sample size for conducting a canonical correlational analysis should be determined in view of the primary purpose of investigation. If stability of the canonical correlations or the redundancy index is desired, from 45 to 60 subjects are necessary to obtain reliable results. If the stability of canonical weights is desired, from 500 to 6000 subjects are necessary to obtain reliable results. It was also pointed out that the subjects per variable index was misleading for the purpose of determining the sample size. / Source: Dissertation Abstracts International, Volume: 43-04, Section: A, page: 1121. / Thesis (Ph.D.)--The Florida State University, 1982.
| Identifer | oai:union.ndltd.org:fsu.edu/oai:fsu.digital.flvc.org:fsu_74809 |
| Contributors | ATASH, MOHAMMED NADIR., Florida State University |
| Source Sets | Florida State University |
| Detected Language | English |
| Type | Text |
| Format | 153 p. |
| Rights | On campus use only. |
| Relation | Dissertation Abstracts International |
Page generated in 0.0021 seconds