我理解你的意思。
做因子分析的时候，KMO如果太低是不适合做因子分析的。如果KMO值大于0.6尚可接受，象你讲的0.5多似乎有点问题。
看有无问项负荷过低，如有将其删除后再试试。
如果还不行，那就可能是问卷设计有问题或者问卷调查过程出了问题。

如果是高人，非常感谢，我再弄弄我的问卷和数据，看能不能提高KMO值，最好能高0.8以上

高人不敢当。相互学习吧！
其实KMO值不用那么高，如有0.7就已经可以了。不要对此要求过高！

This is a good place in order to see difference between reliability and validity analysis.
http://www.socialresearchmethods.net/kb/relandval.php

Further more there are some kinds of validity where some of them you can test using statistical methods like as discriminant and convergent validity and others not like as content validity. Clearly looking at KMO you can't see validity.