Did October 7, 2023 change how Congress votes on Israel and Palestine? This research-style analysis scores every dated roll-call vote and bill sign-on on a pro-Israel to pro-Palestine axis, compares each member's record before and after October 7, and tests whether the change is statistically distinguishable from noise — with bootstrap confidence intervals, permutation tests, multiple-comparison correction, and a placebo cutoff a year earlier. Members without enough activity in both periods are labeled insufficient data, never given a fabricated score. The findings describe voting records, not private beliefs.
Published July 13, 2026 · Updated July 24, 2026
Of 1665 members analyzed, 505 had enough recorded actions on both sides of October 7, 2023 to measure a shift.
Scores run from −1 (every action pro-Israel) to +1 (every action pro-Palestine). Positive Δ = record moved toward the pro-Palestine position after October 7, 2023. Adjusted Δ is the agenda-adjusted (centered) shift: each action is scored relative to everyone acting on the same measure, so a member only earns it by moving relative to colleagues facing the identical choice (Methods §3). The verdict is computed on the raw Δ. Sort any column; the table is the full membership, not a curated cut.
The members profiled in the trajectory cards below, ordered by the strength of the statistical evidence, then shift size. The hollow dot is the mean score of all pre-October 7 actions; the filled dot the mean after.
Solid dots are roll-call votes, faded dots are bill sign-ons; hover any dot for the bill and date. Larger outlined dots are the highest-disagreement key votes from the grid below. The amber dashed line marks October 7, 2023; the green dashed line is the member’s own data-detected change point.
Distribution of shift Δ across all eligible members. Gray = no statistically detectable shift; green = members whose shift clears the suggestive or significant bar. Mass to the right of zero means the chamber-wide drift ran toward the pro-Palestine position.
Each eligible member’s single best split date — the point that maximizes the before/after difference in their own record, found with no knowledge of October 7. Green = members whose shift clears the suggestive or significant bar; gray = members whose best split is not statistically distinguishable from noise, shown for context only. The amber dashed line marks October 2023: a green spike there means the data itself picks out the war as the turning point. Click any bar to see exactly which members turned that month.
Running count of members who have taken at least one scored pro-Palestine action (vote or sign-on). A bend after the amber October 7 line means new members crossing that threshold, not just existing supporters acting more often.
Mean stance of all scored actions per party over a 3-month rolling window (each point pools that month plus the two before it). To keep one bill from moving the band by itself, actions are averaged within each bill first and each bill’s weight in a window is capped at 10 actions — so a single bill picking up 100+ same-direction sign-ons reads as one strong bill, not a party-wide swing. Windows with fewer than 3 actions are blanked, not faked or interpolated, and the tooltip shows how many distinct bills back each point (single-bill windows are flagged). The dashed green line is the profiled top-shifter cohort: where it pulls away from its own party band, those members moved
October 7. A row that runs red left of the line and green right of it is the shift in miniature.
A fake cutoff one year earlier. If the method were inventing shifts, names would clear the bar here too — the expectation is (near) zero.
This section always reflects all scored bills — the bundled-bills toggle does not apply here.
How money is counted: the same canonical figure as each member’s profile headline — direct contributions and independent expenditures in support from hardline pro-Israel PACs (raw FEC ledger where available, curated records otherwise), plus bundled donations from the named pro-Israel donor network. Money spent
We score every dated congressional action on Israel/Palestine — roll-call votes and bill sign-ons — on a single −1 (pro-Israel) to +1 (pro-Palestine) axis, split each member’s record at October 7, 2023, and test whether the change is statistically distinguishable from noise: bootstrap confidence intervals, permutation tests, multiple-comparison correction, and a placebo cutoff a year earlier. The result is a defensible, fully downloadable answer to a question usually argued by anecdote. Records shifted; we make no claim about hearts.
The analysis data failed to load. This is a loading error, not an empty dataset — please refresh, or try again shortly.
Bundled bills excluded. Every number and chart below that supports this view has been recomputed from scratch; sections that always use all scored bills are marked.
Classified floor statements are too thin to score (a handful of pre-war rows site-wide). They appear on trajectory cards as unscored corroborating markers and contribute nothing to any number.
The bills on the floor changed after October 7, so a raw shift could reflect the agenda rather than the member. Two controls are computed alongside (the centered shift is the “Adjusted Δ” column in the ranked table; both controls are in the CSV; the raw shift always leads and the verdict is computed on it). The
Independent of the fixed cutoff, each eligible member’s event sequence is scanned for the split date that maximizes the before/after difference (binary segmentation, minimum 8 events) — the
The centerpiece: every member, both directions, sortable, with the uncertainty attached to every number. Positive Δ = moved toward the pro-Palestine position.
Shifters only mean something in context. This is every eligible member at once: most of Congress sits on the no-change diagonal — that stability is what makes the off-diagonal names informative.
The evidence layer. Two period averages can hide a lot; these cards show the actual dated record — every vote and sign-on, the rolling trend through them, the member’s own detected change point against the October 7 line, and how durable the turn proved. Only statistically backed shifters are profiled; the full population is in the ranked table and scatter above.
Where the whole chamber’s shifts fall — and whether the statistically-backed shifters are outliers or the tip of a broader drift.
The change-point detector is given no hint that October 7 matters — it just finds each member’s best split date. If those dates pile up at October 2023, the data itself is pointing at the war.
Individual shifts sit inside party-wide currents. The bands show each party’s mean stance over a 3-month rolling window; the dashed line is the top-shifter cohort against its own party.
The receipts. These are the roll calls where Congress genuinely split — near-unanimous votes reveal nothing about who moved — laid out in time order around the October 7 line, one row per statistically backed shifter. Members are ranked by how consistently they took the pro-Palestine side on these contested votes after October 7; the two columns on the right give each member’s before/after tally.
The follow-the-money check. Each member’s career pro-Israel funding — the same canonical figure shown on their profile page — lined up against the shift verdicts above: the statistically backed shifters, the suggestive tier, eligible members whose record didn’t move, and the chamber as a whole. Medians, not averages, so a couple of million-dollar outliers can’t move the story.
One file, every member, every number on this page — scores, CIs, p- and q-values, both agenda controls, change points, timing and persistence. Recompute anything.
Every number here describes public voting and sponsorship behavior. A member may have changed positions without floor opportunities to show it, or changed behavior for district-political reasons without changing views. We measure the record and stop there.
Post-October 7 Congress voted on different bills than before. The centered and difference-in-differences controls address this, but no statistical control fully substitutes for the counterfactual of identical bills in both periods.
A symbolic resolution and a binding arms-transfer vote both score ±1. The key-vote grid partially compensates by spotlighting the contested votes, but the scalar shift treats all actions equally.