Experimental data always contains patterns. Even though this may at times be no more than noise versus signal. Changes in these patterns is a highly valuable, highly sensitive diagnostic tool. It is a sanity check and a problem catcher all in one. See a pattern change, investigate.

Without exception, colleagues/peers who attempt to "improve" results are always exposed by the change in the pattern they inadvertently introduce.

At a broad level there was a significant pattern change in DJT support election night pre/post counting pause.

Hold steady and be righteously angry. The pattern change flags a problem. Investigation typically reveals the details. There is smoke and it will lead to a fire bigger than just localized mallarky.