WJ ElectionsThe Waterloo Journal ↗
2026 / UNITED STATESCONNECTING TO POLLING FEEDINDEPENDENT. SOURCE FIRST.
ELECTION INTELLIGENCE / NATIONAL ENVIRONMENT

The midterms.
In focus.

The race for Congress, through the polling. A transparent view of the national mood ahead of November 3.

ELECTION DAY / NOV 03, 202609DAYS

National generic congressional ballot
Polling ≠ a seat forecast

Loading attributed polling data…
WJ 30-DAY SUMMARY—One latest poll per pollster · simple mean
DEMOCRATIC SUPPORT—Raw reported vote share
REPUBLICAN SUPPORT—Raw reported vote share
INCLUDED POLLSTERS— / 0 pollsMixed voter populations
01 / NATIONAL SIGNAL

The generic ballot

Trailing 30-day mean of each pollster’s most recent poll at each date.

DemocraticRepublican
30%40%50%60%
Percent of respondents · Undecided/other responses retained outside these two series. No uncertainty interval is estimated.
03 / THE EVIDENCE

Behind the numbers.

LAST 30 DAYS / 0 POLLS
POLLSTER / SOURCEFIELDWORKSAMPLEDEMREPMARGIN

Showing up to 30 recent polls. LV = likely voters; RV = registered voters; A = adults. Each row links to the original release recorded in the source dataset.

04 / OPEN METHODOLOGY

Show the work.
Keep the uncertainty.

These graphics are a descriptive polling summary, not an election prediction. National support cannot be converted directly into a number of seats.

Silver Bulletin polling database ↗
  1. Collect & validate. Read the publicly linked generic-ballot CSV. Require valid dates, source links and plausible vote shares. Refresh the server cache hourly when visited.
  2. Deduplicate & filter. Keep one question per poll ID, preferring LV, then RV, then adults. Population filters apply before this selection. Do not add pollster grades or house-effect adjustments.
  3. Calculate transparently. For each date, use each pollster’s latest poll ending in the preceding 30 calendar days, with equal weight per pollster. The trend repeats that calculation daily. Ties use the lowest question ID.
  4. Retain limitations. Mixed populations, different questions, rounding, house effects and correlated errors remain. No margin of error for this summary is claimed. If collection fails, display the dated October 25 snapshot with a warning.