Tobias S, Carlson J E
Multivariate Behav Res. 1969 Jul 1;4(3):375-7. doi: 10.1207/s15327906mbr0403_8.
Bartlett's test of sphericitg: was applied to a correlation matrix computed on random normal deviates by Armstrong and Soelberg (1968), and returned a chi square value indicating that the matrix could have been generated from a population where the correlation coefficients are zero. These results re- emphasize the desirability of computing this test prior to proceeding t o factor extraction, and in accord with the findings of other writers, indicate this test to be sensitive in detecting results which can be ascribed to chance.
应用于阿姆斯特朗和索尔伯格（1968年）根据随机正态偏差计算的相关矩阵，得到的卡方值表明该矩阵可能是从相关系数为零的总体中生成的。这些结果再次强调了在进行因子提取之前计算此检验的必要性，并且与其他作者的研究结果一致，表明该检验在检测可归因于偶然的结果时很敏感。
