Statistically insignificant. Look at those sample sizes. Every single article interpreting poll results is invalid because they compare changes as if the margin of error is zero. There is no such thing.
The most that can be concluded is: there was no huge swing toward Kamalalala.
And that’s before considering that every single TV pollster also misstates what the MOE is applied to.
Therefore it is not that Harris interview failed and damaged the campaign because of her weakness in the interview, it is that her inadequacies in the interview are clearly insufficient to halt the ongoing deterioration in her campaign's trajectory that has already peaked.
By the time the debate occurs the need for the Harris campaign to stop the hemorrhaging will be apparent not just to those privy to her internal polling but to the world at large. She will be confronted in the debate with the task of turning around the momentum in the election and she will be under much increased pressure and tension that she is not temperamentally suited to handle.
As the stakes in the debate increase for her, her dilemma of repudiating the Biden candidacy on the one hand or opening herself to charges of flip-flopping on the other hand will grow more apparent to everyone and cause her either to lurch into extremes or submerge herself in banalities in an attempt to hold a middle ground.
She will resort to personal attacks against Trump and references to abortion but these will not be sufficient to halt her downhill slide.
She will not be able to perform in a fashion that requires ad lib., or deft footwork under pressure.
On the other hand, if she does not participate in the debate she will inevitably see her chances diminish despite the Herculean efforts of the media to make her more than what she is, an empty pants suit.
What would be a statistically significant sample size?