|
Abstract:
|
The minimum number of observations necessary in factor analytic studies to produce latent variable stability is inconsistent and highly debatable. Research studies recommend minimum observations ranging from 100 to 1000 observations or observation-to-variable ratios of 2:1 to 20:1. Research has also demonstrated that the number of observations is clearly not sufficient in itself to produce latent variable stability. A Monte Carlo simulation method is used to manipulate the number of observations, number of factors, ratios of errorless-to-random data, and the number of variables-to-factors. MANOVA statistics are used to determine latent variable stability given the various conditions. Results indicate an intricate relationship between observations, factors, degree of error, and variables. Under certain conditions, 16 observations are sufficient to produce latent variable stability.
|
