Mysteries Department
Sign inRequest access
Run · Marconi scientists' deaths, 1982–1990

Base-rate test of the Marconi cluster, 1982–1990

Deterministic; the seed is unused.

0.250

Result

Result. The lists name 8 GEC employees who died by suicide, open verdict or asphyxia in 1982–1990; ordinary rates predict about 32 such deaths among GEC's UK graduates alone and about 161 among all its UK staff. The cluster is a sample of ordinary deaths, not an excess over them.
Test · estimate [95% interval]Against the null and the answersVerdict
Listed GEC suicides and open verdicts / expected among GEC UK science and engineering graduates, 1982–1990ratio · against the two answers
+0.25 [+0.05, +0.45]
NO EXCESSEXCESS−0.250+0.25+0.5+0.75+1+1.25
Excludes “excess”consistent with “no excess”
Same, at half the population rate (allowing for lower suicide rates among professionals)ratio · against the two answers
+0.51 [+0.12, +0.90]
NO EXCESSEXCESS−0.250+0.25+0.5+0.75+1+1.25
Excludes “excess”consistent with “no excess”
Listed GEC suicides and open verdicts / expected among all GEC UK employeesratio · against the two answers
+0.05 [+0.01, +0.09]
NO EXCESSEXCESS−0.250+0.25+0.5+0.75+1+1.25
Excludes “excess”consistent with “no excess”
All listed defence-worker suicides and open verdicts (15) / expected among UK industry directly employed on MoD workratio · against the two answers
+0.04 [+0.01, +0.07]
NO EXCESSEXCESS−0.250+0.25+0.5+0.75+1+1.25
Excludes “excess”but differs from GEC graduates by 2.1σ
Estimate, 68% interval (thick) and 95% (thin). Rule declared with the run: rule out 'excess' if the 95% interval lies below 1; rule out 'no excess' if it lies above 1

Across the evidence every measurement of the same quantity

Listed suicides and open verdicts among GEC's UK staff, 1982–1990, as a ratio to the number expected from England and Wa4 measurements
MEASUREMENTRATIOESTIMATE [95% CI]WEIGHTNO EXCESSEXCESS−0.250+0.25+0.5+0.75+1+1.25This run, GEC graduatesthis department+0.25 [+0.05, +0.45]1%This run, GEC graduates, half this department+0.51 [+0.12, +0.90]0%This run, All GEC UK staffthis department+0.05 [+0.01, +0.09]28%This run, UK defence industrythis department+0.04 [+0.01, +0.07]71%Pooled+0.05 [+0.03, +0.07]100%Pooled by inverse variance. Heterogeneity: I² 70%, Q 10.0 on 3 df.
RATIO · ESTIMATE [95% CI]NO EXCESSEXCESS−0.250+0.25+0.5+0.75+1+1.25This run, GEC graduates+0.25This run, GEC graduates, half +0.51This run, All GEC UK staff+0.05This run, UK defence industry+0.04Pooled+0.05
Pooled+0.0 [+0.0, +0.1]
Disagreement (I²)70%
Against the answers“excess” excluded
Every measurement excludes “excess”, so it is ruled out whichever is right. They disagree with each other.

What it used and found

Figures 1

Figure 1

Ordinary suicide rates predict far more deaths than the lists contain

Suicides and open verdicts, 1982–1990: expected at England and Wales rates (ages 20–64) for three workforces, against those on the published lists

0100200300400deaths, 1982–1990GEC UK graduates (17,500)All GEC UK staff (126,600)UK defence industry (285,000)ExpectedOn the lists
0100200300400deaths, 1982–1990GEC UK grad…All GEC UK …UK defence …
ExpectedOn the lists
Expected counts use ONS rates including deaths of undetermined intent (most open verdicts). The lists are not complete counts, so the gap shows how many ordinary deaths a list like this leaves out, not a low suicide rate.
Drawn by the site

Parameters

ages = 20-64, years = 1982-1990, counted = ['suicide', 'open verdict', 'accidental asphyxia'], populations = {'GEC UK employees': 126623, 'GEC UK science and engineering graduates': 17500, 'UK industry directly employed on MoD work': 285000}, rate_multipliers = [1, 0.5]

Cite it

Mysteries Department. (2026). Base-rate test of the Marconi cluster, 1982–1990 [Run RUN0DGGDGRHYP78]. https://mysteriesdept.com/r/RUN0DGGDGRHYP78
Private beta Downloads are for beta members

We're in a private beta while we get things right. Everything here is free to read; downloading datasets, sources and code is open to members for now.

It takes a couple of minutes. We'll reply by email.