SEM为大样本分析法.样本愈大愈好
至少希望100个样本以上
至于要多少变数其实多少是没有关系
不过建议在20个变量以内
但其实40~50个变量也没问题
只是变量惫多,模型拟合度就可能变差

只有将近三十个样本可不可以做呢？

Best way to determine if you have a large enough sample is to conduct a power analysis.
Look at the following link it is very helpful, but you must know or hypothesize some things like as effect size

http://www.danielsoper.com/statcalc3/calc.aspx?id=89

可從power的角度理解，PLS bayesian bootstraping對小樣本可較為有效處理

学习，谢谢