Normal Approximations of Exact Tests in Configural Frequency Analysis
β Scribed by Dr. L. R. Bergman; A. von Eye
- Publisher
- John Wiley and Sons
- Year
- 1987
- Tongue
- English
- Weight
- 429 KB
- Volume
- 29
- Category
- Article
- ISSN
- 0323-3847
No coin nor oath required. For personal study only.
β¦ Synopsis
The most common testa for types and ontity-pea in configural frequency analysis are normal approximations of exact bats. I n the paper such statistics under the complete independence model and under the fixed margins model are discussed. It turns out that these test statistics are not acceptable when the number of simultaneously performed testa is large or when the expected frequencies are small. I n them cases, the use of exact testa is advocated and mme existing computer programs for such bats are indicated. A normal approximation based on the strong version of the De Moivre-hplace limit theorem is alw discussed. Empirical examples are given from longitudinal data describing psychological development of boys.
π SIMILAR VOLUMES
## Summa y The test statistics u d until now in the CFA have been developed under the sssumption of the overell hypothesis of totml independence. Therefore, the multiple test procedurea bssed on theae stetistics sre reslly only different teats of the oversll hypotheah. If one likm to test a epecie
## A continuity correction ie proposed for the teet etatistice of H A B ~X A N (1978) and LBHMAOHES (1981) for identifying overfrequented (or underfrequented) cells in the three-dimensional Configural Frequencybalyeie (CFA). Its quality is shown by compariaon with the teat baed on the exact distri