Again, the scientific and not quite
So, we still continue to talk about and not very scientific.
If the result is very different from the expected (and actually, even if it is not much different!) You must make sure that it is still real results. For example, an independent study. Best of all, that the study did not have the same scientist, and the other, using a radically different method of work.
In the example described, it would be reasonable to see how this rainwater react indicator paper. Because the sensor can not be calibrated, could be damaged cord connecting the sensor to the measuring unit etc.
Here we come to a very important first step. And the next few notes, I dedicate calibration errors, measurements, etc.