ACTUARIAL RISK ASSESSMENT
THE DALLAS COHORT
This analysis was commissioned to assess mortality patterns among individuals identified as material witnesses, persons of interest, or subjects of investigation in connection with the assassination of President John F. Kennedy on November 22, 1963.
The cohort ("Dallas Cohort") was assembled from Warren Commission witness lists, subsequent congressional inquiry records, and investigative files provided by the commissioning party. Individuals were included based on documented connection to the investigation, not speculation regarding involvement.
Cohort Parameters:
- Initial cohort size: 67 individuals
- Observation period: November 22, 1963 — December 31, 1976
- Mean age at observation start: 41.3 years
- Gender distribution: 58 male, 9 female
Purpose: To determine whether observed mortality within this population deviates significantly from actuarial expectations, and if so, to characterize the nature and magnitude of that deviation.
Expected mortality was calculated using standard life tables (U.S. Population, 1960-1970, age-adjusted). Cause-of-death distributions were benchmarked against CDC vital statistics for the corresponding periods.
Statistical significance was assessed at the p < 0.05 threshold. Variance from expected values is expressed as percentage deviation and standard deviation from mean.
Deaths were categorized according to official cause of death as recorded on death certificates or medical examiner reports. In cases where official determinations were disputed or revised, the original ruling was used for consistency.
| Period |
Cohort Size |
Expected Deaths |
Actual Deaths |
Variance |
| 1963-1966 |
67 |
2.3 |
14 |
+508% |
| 1967-1970 |
53 |
1.8 |
8 |
+344% |
| 1971-1974 |
45 |
1.5 |
5 |
+233% |
| 1975-1976 |
40 |
0.7 |
6 |
+757% |
Note: Cohort size decreases each period due to prior deaths and natural attrition.
33 Deaths
OBSERVED VS. 6.3 EXPECTED (1963-1976)
Cumulative mortality over the observation period: 33 deaths observed against 6.3 expected. This represents a variance of +424% from actuarial projections.
The probability of this variance occurring by chance in a random population sample is less than 1 in 1017.
Expected Distribution (U.S. General Population, 1960-1976):
| Cause |
Expected % |
Dallas Cohort % |
Ratio |
| Natural Causes |
72% |
30% |
0.42x |
| Accident |
15% |
27% |
1.8x |
| Suicide |
8% |
21% |
2.6x |
| Homicide |
5% |
22% |
4.4x |
Anomalies Flagged:
- Homicide rate: 4.4x general population
- Suicide rate: 2.6x general population
- Accident rate: 1.8x general population
- Suicides occurring within 72 hours of scheduled testimony: 3 cases
- Deaths ruled "natural" with disputed or absent autopsy: 4 cases
• • •
The following entries represent cases with notable timing, circumstance, or cause-of-death anomalies. This is not a complete list of cohort mortality.
KILGALLEN, Dorothy Mae
DOD:
November 8, 1965
Age:
52
Cause:
Barbiturate/alcohol combination (accidental)
Connection:
Only journalist to conduct private interview with Jack Ruby
Pending:
Book manuscript on assassination (never located)
Found in bed she never used, wearing clothes she had not worn the previous evening. Reading book she had finished days prior. Interview notes never recovered.
FERRIE, David William
DOD:
February 22, 1967
Age:
48
Cause:
Berry aneurysm (natural causes)
Connection:
Garrison investigation primary subject; CAP association with Oswald
Pending:
Grand jury subpoena served February 21, 1967
Two typed suicide notes found at scene. Death ruled natural. Subpoena served less than 24 hours before death.
GIANCANA, Salvatore "Sam"
DOD:
June 19, 1975
Age:
67
Cause:
Gunshot wounds (7), .22 caliber, to head and neck
Connection:
CIA-Mafia assassination plots (Cuba); Robert Maheu contact
Pending:
Church Committee testimony scheduled June 20, 1975
Shot from behind while cooking sausage in basement kitchen. Shooter entered through unlocked door. No signs of struggle. No arrest made.
ROSELLI, John
DOD:
August 7, 1976 (approx.)
Age:
71
Cause:
Asphyxiation, dismemberment; body in oil drum, Dumfoundling Bay, FL
Connection:
CIA-Mafia plots; Church Committee witness
Pending:
Scheduled follow-up testimony to Church Committee
Had testified to Church Committee in 1975. Told associates he believed CIA was "using the Mafia as a cover" for Kennedy assassination. Missing 10 days before body discovered.
DE MOHRENSCHILDT, George
DOD:
March 29, 1977
Age:
65
Cause:
Shotgun wound to head (ruled suicide)
Connection:
Oswald's primary Dallas contact; introduced Oswald to Paine household
Pending:
HSCA interview scheduled March 30, 1977
Received HSCA investigator Gaeton Fonzi's card hours before death. Had written to George H.W. Bush at CIA seeking "ichearing" from surveillance. Found by daughter.
RUBY, Jack Leon (né Rubenstein)
DOD:
January 3, 1967
Age:
55
Cause:
Pulmonary embolism secondary to lung cancer
Connection:
Killed Lee Harvey Oswald on live television, November 24, 1963
Pending:
Appeal of murder conviction; new trial granted
Cancer diagnosis came 30 days before scheduled new trial. Told family and visitors "they" were injecting him with cancer cells. Denied access to independent physicians. Died in custody.
SULLIVAN, William C.
DOD:
November 9, 1977
Age:
65
Cause:
Gunshot wound (hunting accident)
Connection:
FBI Domestic Intelligence chief; oversaw Oswald file pre-assassination
Pending:
HSCA testimony scheduled
Shot near his home in Sugar Hill, NH. Shooter claimed to have mistaken him for a deer. Sullivan was walking to meet fellow hunters. Case closed as accident.
BOWERS, Lee Edward Jr.
DOD:
August 9, 1966
Age:
41
Cause:
Single-vehicle automobile accident
Connection:
Railroad tower operator; witnessed "flash of light" and "smoke" from grassy knoll
Pending:
None (had already testified to Warren Commission)
Vehicle veered into bridge abutment on empty road. Witness reported Bowers appeared to be in "strange shock" before impact. Attending physician noted he was in "a strange state of shock" despite minimal blood loss.
• • •
Observed mortality within the Dallas Cohort exceeds actuarial expectations by a factor of 5.2x over the study period. This variance is not distributed evenly across time; mortality spikes correlate with:
- Immediate post-assassination period (1963-1966): Investigation and Warren Commission
- Garrison investigation (1967-1969)
- Church Committee hearings (1975-1976)
The cause-of-death distribution is profoundly anomalous. Violent death (homicide, suicide, accident) accounts for 70% of cohort mortality, compared to 28% in the general population of similar age and demographic profile.
1 in 1017
PROBABILITY OF OBSERVED VARIANCE BY CHANCE
The timing correlation between scheduled testimony and death—particularly the three cases of death within 72 hours of subpoena or scheduled appearance—represents a statistical anomaly that cannot be explained by standard actuarial models.
It is beyond the scope of this analysis to determine causation. We can only report that the observed mortality pattern in this cohort is, in actuarial terms, impossible.
This analysis is subject to the following limitations:
- Cohort definition may introduce selection bias; individuals were included based on investigative connection, which may correlate with higher-risk occupations or associations
- Official cause-of-death determinations were accepted as given; independent verification was not possible
- Some cohort members' post-1963 activities may have introduced mortality risk unrelated to the investigation
- This analysis does not imply causation or criminal conduct
Notwithstanding these limitations, the magnitude of variance from expected mortality is so extreme that methodological adjustments cannot account for it. Even with aggressive correction factors applied to account for potential cohort bias, observed mortality remains at minimum 3.1x expected values.
Further investigation is recommended.
• • •
I've been doing this work for twenty-three years. Exposed populations—asbestos workers, combat veterans, coal miners. I know what elevated mortality looks like.
I've never seen numbers like this.
The model doesn't break. The model can't break. That's the point of a model—it accounts for variance, for outliers, for the thousand small ways that life deviates from the mean. That's what the confidence intervals are for.
But these numbers break the model.
I don't know what happened in Dallas. I only know that whatever happened, it didn't stop on November 22nd. And it didn't stop with Lee Oswald. The killing went on. The dying went on. And nobody seems to have noticed the pattern.
Or maybe they noticed. Maybe that was the point.
— R.E.H.
This analysis was submitted to the House Select Committee on Assassinations in March 1977. The Select Committee did not include it in their final report.
Richard Hartwell died of a heart attack on May 2, 1977. He was fifty-one years old. His own actuarial tables gave him another twenty-six years.
The model doesn't break. — CDJ