# Sources — House Incumbency Reelection Rate
## "Nothing Moves the Reelection Rate"
### Book: *Why Do You Think the Government Doesn't Work?* — Chapter 2, Observation (incumbency)
### Last updated: August 2026 (reframed — see Version History)

---

## What This Chart Shows

The percentage of U.S. House incumbents who sought reelection and won, by election year, 1970–2024, plotted on a **full 0–100% scale**.

**The finding is the flatness.** Across twenty-eight elections and fifty-four years the line moves within a 13-point band, spending twenty-five of those elections at or above 90%. The three deepest anti-incumbent elections of the era — 1974, 1992, 2010 — are marked, and at full scale none of them reads as a visible dip. That is the intended result, not a rendering problem: the chart is built so that a reader can see the three biggest "throw them out" moments in living memory annotated on a line that does not visibly move.

This is the reelection rate among those who *chose to run*. See "Who is counted" below — it is a real limitation and the chart states it on its face.

---

## Primary Sources

### Brookings Institution — Vital Statistics on Congress, Table 2-7
- **URL:** https://www.brookings.edu/wp-content/uploads/2022/11/2-7-Full.pdf
- **Coverage:** 1946–1994 (chart uses 1970–1994)
- **Notes:** The authoritative historical source for House incumbency data. Tracks total seeking reelection, defeated in primaries, defeated in general election, total reelected, and reelection rate.

### OpenSecrets — Reelection Rates Over the Years
- **URL:** https://www.opensecrets.org/elections-overview/reelection-rates
- **Coverage:** 1996–2022
- **Notes:** Compiled from FEC filings and official election results. Consistent methodology with Brookings for comparability.

### Ballotpedia — Election Results 2024: Incumbent Win Rates
- **URL:** https://ballotpedia.org/Election_results,_2024:_Incumbent_win_rates_by_state
- **Coverage:** 2024
- **Notes:** Reports ~97% of congressional incumbents reelected in 2024. Used for the final data point.

---

## Methodology Notes

- **Who is counted:** Only incumbents who actively sought reelection. Retirements and members who chose not to run are excluded from both numerator and denominator. This is the standard political-science definition of incumbency advantage.
- **Average:** Across the plotted period (1970–2024, 28 elections), the mean reelection rate is **94.0%**. This is the second callout stat. (The dashed on-chart reference line that previously carried it was removed in the 2026-08 reframe — see Design decisions.) (Over the longer 1954–present span the figure is in the same 93–94% range.)
- **The floor:** Across the plotted period the rate never falls below **85.4%** (2010). In 25 of 28 elections it stayed at or above 90%.
- **The range:** highest 98.3% (1988 and 1998), lowest 85.4% (2010) — a total span of 12.9 points across 28 elections.
- **Who is counted (stated on the chart):** the denominator is incumbents who *sought* reelection. Members who retired, died, or ran for another office are excluded from both numerator and denominator. Total chamber turnover is therefore higher than this line shows, and some of the apparent stability is at-risk members exiting before Election Day rather than losing on it. This runs in the conservative direction — it understates entrenchment — which is why the chart can carry it in plain sight rather than burying it here.

---

## Design decisions (2026-08 reframe)

Recorded here because several were judgment calls rather than data choices.

**Y-axis changed from 80–100% to 0–100%.** This is the core change. A share out of 100 belongs on a full scale; the truncated axis magnified variation that is small in absolute terms and made ordinary election-to-election noise look like drama. On the full scale the line sits high and nearly flat, which is the honest picture and also the argument.

**The 94% average reference line was dropped.** On a 0–100 axis it lands within a few pixels of the data line for most of the series — it reads as a printing error rather than a reference. The average survives as the second callout stat, where it is legible.

**The area fill was kept, and strengthened** from a fade to a near-flat tint. At full scale it becomes a solid block occupying most of the plot, which is the right visual for the claim: held seats are a mass, and the thin band above the line is everything that has ever changed hands.

**Wave labels moved below the points.** On a full scale the line hugs the top of the plot, leaving no room above it. Each marked election now drops a short dashed leader into the filled area with its label beneath, so the annotation is legible without the underlying point having to be a visible trough.

**Title changed** from "Share of U.S. House Incumbents Reelected, 1970–2024" to "Nothing Moves the Reelection Rate." The old title described the metric; house style is for the title to state the finding, and the finding is now the flatness rather than the level. The title is deliberately rhetorical — the rate does move, between 85% and 98% — and the subtitle and stat row supply the actual numbers immediately beneath it. If that reads as overclaiming next to the surrounding chapter, "Even the Biggest Waves Barely Move the Reelection Rate" says the same thing with the hedge built in.

**Red is on the 85% stat rather than the 97% one.** Under the old framing the notable number was how high the rate is now; under the flatness framing it is how little even the worst year gave up. House convention reserves red for the point the reader should notice.

---

## On the "dips" (what the chart labels)

The chart marks only the three elections in the plotted period where incumbent reelection fell **below 90%** — the genuine troughs:

| Year | Rate | Label | Driver |
|------|------|-------|--------|
| 1974 | 87.7% | Watergate | Post-Watergate backlash |
| 1992 | 88.3% | Redistricting | Post-1990 redistricting, House Bank scandal, record retirements |
| 2010 | 85.4% | Tea Party | 2010 midterm wave |

**Deliberately NOT labeled as "waves":** Several elections often called "waves" in partisan-control terms did **not** produce low incumbent-reelection rates and are therefore not marked on this chart:
- **1980** (90.7%) and **1994** (90.2%) — the 1994 "Republican Revolution" flipped House control but still reelected ~90% of incumbents who ran; the turnover came heavily from open seats, not defeated incumbents.
- **2006** (94.0%) and **2018** (91.3%) — party-control "waves" with reelection rates near or above average.

This distinction is the point: even the most famous "throw-them-out" elections rarely unseat more than ~1 in 7 incumbents. Marking only the true reelection troughs keeps the chart honest and consistent with the prose.

---

## What the chart does NOT show

The chart does not show the rate at which vulnerable incumbents **retire** rather than face a difficult election. Effective incumbency protection is therefore somewhat higher than the reelection rate alone suggests, because some at-risk members exit before Election Day. This is a limitation in the conservative direction (it understates entrenchment), and the chart does not attempt to quantify it. **As of the 2026-08 reframe this appears on the chart itself**, in the "Who is counted" box, rather than only here — a reader who spots the denominator problem unaided will discount the whole chart, so it is better stated up front.

It also does not show open-seat turnover, which is where most chamber change actually comes from. 1994 is the clean example: the Republican Revolution flipped control while reelecting 90.2% of incumbents who ran.

---

## Note for drafting (not for the published chart)

This metric is **ambiguous on its own** — a high reelection rate could indicate voter satisfaction. In the book it is presented as *half of a paradox*: near-total reelection sits next to near-record-low approval (the approval observation). Neither number alone is damning; the pair is. The chart should describe the data; the interpretation lives in the prose.

---

## Version History

| Date | Change | Author |
|------|--------|--------|
| May 2026 | Initial publication | TSB |
| June 2026 | Retitled to descriptive form; reduced wave labels to the three genuine sub-90% troughs (1974, 1992, 2010); corrected average line to plotted-period mean (94%); replaced redundant "15% replaced" callout with a stability stat; repointed from Essay #107 to Book Ch2 | TSB |
| August 2026 | **Reframed around flatness.** The argument this chart supports changed from "incumbents win a lot" to "the rate has barely moved in fifty years, and even the biggest anti-incumbent waves didn't move it much." Y-axis changed 80–100% → 0–100%; 94% reference line dropped; area fill strengthened; wave labels moved below the line; "Who is counted" note added to the chart; callouts reordered to lead with "25 of 28"; title changed to state the finding. No data changed — same values, same sources, same three marked elections. Full rationale in "Design decisions" above. | TSB |

*If you believe a data point is incorrect, please open an issue at https://github.com/The-Statecraft-Blueprint/tsb-assets or contact us. We will review, correct if warranted, and document the change in this version history.*