A dead heat in Texas

Recent polling puts Democrat James Talarico and Republican Ken Paxton within points of each other in the U.S. Senate race. Talarico has edged ahead slightly since the May 26 runoff.

“Texas' candidates in this year's U.S. Senate race are neck and neck”

Better numbers than Beto

Talarico's leads in recent polls are larger than anything Beto O'Rourke posted in 2018, when he came close to unseating Ted Cruz. He is also far outraising Paxton.

“Talarico is dominating Paxton in fundraising.”

Who paid for the poll

Campaign or party-sponsored polls deserve skepticism. Nonpartisan shops release everything they find; partisan ones can publish only the results that flatter their candidate.

Bigger samples, steadier results

The more voters a poll surveys, the more likely its numbers reflect what real voters think. Small samples wobble.

Adults, registered, or likely

Polls survey one of three groups. Adult samples are the broadest but least predictive; registered voters are the election-year standard; likely voter screens matter most as Election Day nears.

How people get picked

Probability samples give everyone a known chance of selection. Cheaper non-probability panels can skew, though some now match or beat probability polls for accuracy.

The margin of error caveat

Only probability polls can calculate a true margin of error. Non-probability polls estimate their error with statistical models instead.

How this average is built

The tracker pulls from the New York Times polling database, counts only post-runoff head-to-head polls, and weights each by recency, sample size, and whether a party paid for it.

“Polls sponsored by a political party count half as much as ones without a partisan sponsor.”

Not a forecast

The average describes sentiment right now, not who wins. With few polls released so far, a single outlier can swing the line noticeably.

“The average is not a forecast of the election and who will eventually win.”