The 2018 polls are tightening partially due to independents

The generic ballot -- voters are asked which party they'll support as opposed to a specific candidate -- has tightened in the most recent CNN poll, which was released in early May.

Forty-three percent of independents in CNN's most recent poll said they would vote for the Democratic candidate and 40% said they would support the Republican candidate, a small enough difference in the numbers that make them basically statistically equal.

While a generic Democratic candidate has been doing better than the Republican during all points of the 2018 cycle, at the beginning of the year, 50% of independents said they would vote for the Democratic candidate and 35% the Republican -- a 15-point gap in CNN's February poll that has decreased to 3 points in May.

The number of independent respondents saying "other" or "neither" to these options has increased to 11% in May from 4% in March. Independents are the most likely partisan group to say "other/neither" or "no opinion" -- at 11% and 6% respectively. Only 2% of Republicans have no opinion and 4% of Democrats. It's possible that independents are moving from "other/neither" or "no opinion" to make their decision, but it's still unknown at this point.

This dive for Democrats comes on the heels of some attention around the generic ballot tightening, not just in CNN's polling, but in many others. The most recent CNN polling had Democrats at a three-point lead, well within the margin of error. However, in February 2018, Republicans faced a 16-point climb.

Some of the other groups that have changed significantly since the high of the February poll:

White non-college graduates -- Support for a generic Democratic candidate went from 45% to 35% Those under the age of 45 -- A 7-point gain for Republicans Those with an income under $50K -- 9-point loss for Democrats, 7-point gain for Republicans White voters -- (8-point gain for Republicans and loss for Democrats

Some pollsters include an option to say the respondent isn't sure who they're voting for or that they don't plan on voting in the election. Those pollsters will tend to have higher "undecided" numbers than those who let respondents volunteer that answer. That, combined with methodology differences in how polling was conducted, create a wide variance in what polls report. However, when looking at the overall picture, it's hard to not see the ballot tightening, with independents making up their minds.