polling for the us congress has small sample sizes, poor methodology and large margins of error. results are highly inconsistent and poorly predictive. averaging it out into aggregate is not a good methodological approach.

the way to read these polls is like this:

1. if the polls all say the same thing, the outcome is clear.

2. if the polls are variable, it's a toss up.

the polling in texas, alaska, iowa, maine and ohio is fluctuating but the republicans are always ahead. there is little to no reason to expect the democrats to be competitive in these states.

rather, it is michigan, new hampshire, north carolina and georgia that are producing variable results, in wide swings, with huge margins, which makes the result less certain.