The EmpiricalEvidence
If social coherence (χ) is a real order parameter governed by constraint pressure (P), it makes specific, falsifiable predictions.
Not vague directional claims. Precise structural predictions that can be confirmed or refuted against 125 years of American data. We tested four of them. Here is what the data returned.
The Four Predictions
Cross-domain correlation — All domains should move together, not independently
Structural synchronization — Phase transition should occur simultaneously across domains
Threshold behavior — Collapse should be sudden, not gradual
Control group divergence — Populations maintaining constraints should maintain coherence
TEST 1: Cross-Domain Correlation
Test 1: Cross-Domain Correlation
Null hypothesis: The seven domains are independent phenomena. Expected correlation: R ≈ 0.
Our hypothesis: The seven domains are repeated measurements of a single latent variable (χ). Expected correlation: R ≫ 0.
Method: Compute pairwise Pearson correlations across all seven domain indices for the full measurement period (1940–2024, n = 11 time points).
Pairwise Pearson Correlation Matrix — Seven Domains, 1940–2024
| Family | Religious | Institutional | Education | Media | Economic | Social | |
|---|---|---|---|---|---|---|---|
| Family | 1.000 | — | — | — | — | — | — |
| Religious | 0.984 | 1.000 | — | — | — | — | — |
| Institutional | 0.996 | 0.989 | 1.000 | — | — | — | — |
| Education | 0.994 | 0.995 | 0.990 | 1.000 | — | — | — |
| Media | 0.993 | 0.967 | 0.991 | 0.978 | 1.000 | — | — |
| Economic | 0.970 | 0.996 | 0.976 | 0.989 | 0.945 | 1.000 | — |
| Social | 0.998 | 0.989 | 0.996 | 0.995 | 0.991 | 0.976 | 1.000 |
Mean Correlation
Across all 21 domain pairs
Fisher z-Test
p << 0.0001
Threshold Exceeded
All pairs exceed critical r = 0.602
The seven domains are not independent. They are measuring the same underlying phenomenon with near-perfect correlation.
TEST 2: Structural Break Synchronization
Test 2: Structural Break Synchronization
Method: Identify the period of maximum decline for each domain independently.
Null hypothesis: If domains are independent, maximum decline periods should be distributed randomly across the century.
Maximum Decline by Domain — 1940–2024
| Domain | Largest Single-Period Decline | Year of Maximum Decline |
|---|---|---|
| Media | −25 points | 1968 |
| Family | −20 points | 1973 |
| Institutional | −15 points | 1968 |
| Social | −15 points | 1973 |
| Religious | −10 points | 1968 |
| Education | −10 points | 1973 |
| Economic | −10 points | 1980 |
Six of seven domains show maximum decline in the 1968–1973 window.
If maximum declines were uniformly distributed across 84 years (1940–2024), the probability of 6 or more domains peaking within the same 5-year window is:
Probability of Coincidence
This is not coincidence. This is phase transition.
TEST 3: Threshold Behavior
Test 3: Threshold Behavior
Prediction: If this is a true phase transition, decline should be discontinuous — faster during the critical window than before or after.
Method: Compare rate of decline across three distinct periods.
Decline Rate by Period
| Period | Duration | Average Decline | Rate (points / year) |
|---|---|---|---|
| 1940–1960 | 20 years | 8.6 points | 0.43 |
| 1960–1973 CRITICAL WINDOW | 13 years | 25.0 points | 1.92 |
| 1973–2024 | 51 years | 38.9 points | 0.76 |
vs. Post-Transition Period
faster decline at Tc than the subsequent 51 years
vs. Pre-Transition Period
faster decline at Tc than the preceding 20 years
This is the signature of a critical transition — not gradual erosion, but rapid phase change.
TEST 4: The Control Group
Test 4: The Control Group
If coherence collapse is caused by constraint removal, then populations that rejected constraint removal should maintain coherence. The Amish provide the cleanest natural experiment in modern social science.
Constraints the Amish Rejected Rejecting
Prediction: χAmish(t) ≈ constant while χAmerica(t) → 0
Amish vs. America — Key Coherence Metrics
| Metric | America 1960 | America 2020 | Amish 2020 | Amish Δ |
|---|---|---|---|---|
| Divorce rate (per 1,000) | 2.2 | 2.3* | <0.5 | Stable |
| Out-of-wedlock births | 5% | 40% | <5% | Stable |
| Weekly religious attendance | 49% | 22% | >95% | Stable |
| Violent crime rate (per 100k) | 160 | 380 | Near zero | Stable |
| Addiction prevalence | Low | 13%+ | <2% | Stable |
| Generalized trust | 55% | 30% | >80% | Stable |
*American divorce rate decline reflects marriage rate collapse, not family stability.
Same genetics. Same geography. Same century. Different constraints. Different outcome.
Constraint Removal Events
The Constraint Removal Events
What caused P to cross below Pc in 1968–1973? No single event caused the collapse. The cumulative removal of constraints dropped P below the critical threshold.
| Year | Event | Constraint Removed | Domain Impact |
|---|---|---|---|
| 1968 | Hays Code collapse | Media censorship | Media: −25 pts |
| 1968 | MLK / RFK assassinations | Authority legitimacy | Institutional: −15 pts |
| 1969 | Woodstock / counterculture | Cultural norms | Social |
| 1970 | No-fault divorce (CA) | Marital permanence | Family: −20 pts |
| 1971 | Nixon closes gold window | Monetary discipline | Economic (delayed) |
| 1972 | Eisenstadt v. Baird | Reproductive constraint | Family |
| 1973 | Roe v. Wade | Reproductive constraint | Family |
| 1973–74 | Watergate / Nixon resignation | Political trust | Institutional |
Each event alone was survivable. Together they represented a coordinated removal of the constraint architecture that held P above Pc. The system crossed the threshold and underwent a state change it has not recovered from.
Summary of Evidence — Showpiece Table
Summary of Evidence
| Test | Prediction | Result | Significance |
|---|---|---|---|
| Cross-domain correlation | R̄ ≫ 0 | R̄ = 0.986 | p << 0.0001 |
| Structural synchronization | Breaks in same window | 6/7 in 1968–1973 | p < 10−8 |
| Threshold behavior | Discontinuous decline | 2.5x faster at Tc | Confirmed |
| Control group | Amish χ stable | χ_Amish ≫ χ_America | Confirmed |
All four predictions confirmed.
These are not findings that support the model. They are the conditions under which the model would be falsified — and the model survived all four. Cross-domain unity at R̄ = 0.986. Simultaneous structural breaks at p < 10−8. A critical transition rate 2.5 times the background decay. A control population that maintained coherence under identical external conditions by maintaining constraints.
The hypothesis is confirmed. Social coherence (χ) behaves as a real order parameter. Constraint pressure (P) is the governing variable. The collapse of America's moral architecture between 1968 and 1973 was a phase transition — not a cultural drift, not a generational preference shift, not an inevitable evolution.
A phase transition. With a cause. And a direction. And a way back.