It all depends on our data. Theoretically, you can use Granger causality analysis for your data. Suppose your data, A and B,  are simultaneously recorded long enough, then should be able to run the Granger causality analysis.

If the sample of data is not enough, even you get positive Granger causality, then you would be cautious on the results.

首先要了解格兰杰因果关系和因果关系并不是一码事，可以做格兰杰因果检验，注意避免虚假回归就可以了。

我现在只有4个时间点，但有20个样本，该怎么做？