Hello Robert Tulip,

I would first suggest you listen to what JohnW and the others are saying, they know what they're talking about.

I would also caution against testing a specific hypothesis, out of a multitude of hypotheses you consider, for statistical significance. What you really want to do is conduct a joint test of all of the hypotheses you consider. If you test enough hypotheses individually (how is the rain one day before conjunction, two days before, three days before, and so on), it is virtually guaranteed that some of them (in a particular data sample) will be statistically significant, even if there is no true relationship. You need a joint test of all of your hypotheses.