If Three People Each Name Three Contacts Among Four, Someone Is Wrong
Gepubliceerd op 13-11-2026 · 3 min leestijd · Dagelijkse calculators
Sofia Nunes — Redacteur gezondheid en welzijn bij OneKitly
Voeding · Vochtinname
Getoetst aan 2 bronnen
Ask four people how many of the other three they were in contact with and the four answers have to be consistent with an actual network. The list 3, 3, 3, 1 is not: if three people each met everyone else, the fourth met three people too, not one. The list 3, 3, 2, 2 is fine and describes a network with five links. There is an even simpler check before that one — the answers must add to an even number, because every contact is counted twice, once at each end. A list summing to 3, such as 1, 1, 1, is impossible on that ground alone.
The list 3, 3, 3, 1 cannot describe any real network. 3, 3, 2, 2 can, with five links. Two centuries-old rules catch survey errors before any analysis starts.
The handshake rule and what it catches
Every link has two ends, so the answers always add to twice the number of links and can never be odd. It follows that the number of people reporting an odd number of contacts is itself always even — a statement that sounds like a party trick and is in fact the first thing a network analyst checks. In contact tracing, an odd total means at least one person's answer is missing or wrong, and it says so before anything has been modelled. Cheap consistency checks that run on the raw answers are worth more than sophisticated ones that run on the conclusions.
The second test, and why it needs a computer
An even total is necessary and not sufficient: 3, 3, 3, 1 sums to ten and is still impossible. The Havel–Hakimi procedure settles it by taking the largest number, connecting it to the next largest ones and repeating on what remains; if a degree ever goes negative, no network exists. Doing that by hand for four people takes a minute and for four hundred takes a machine, which is why the check is built into survey software rather than left to the analyst. It also produces the link count as a by-product, which is the first number any report needs.
| Answers | Possible? | Why |
|---|---|---|
| 1, 1, 1 | No | Sum is odd |
| 3, 3, 3, 1 | No | Nobody can be left with one |
| 3, 3, 2, 2 | Yes | Five links |
| 4, 3, 3, 2, 2 | Yes | Seven links |
Veelgestelde vragen
- Why can 3, 3, 3, 1 not work?
- Because in a group of four, a person with three contacts has met everybody else. If three separate people have each met everybody else, then the fourth person has been met by all three of them and must report three, not one. The contradiction is visible without any theory once it is stated that way, and it is the smallest example where an even total is not enough. Larger groups hide the same contradictions behind arithmetic nobody can do by eye.
- Does a valid sequence mean the data is correct?
- No, only that they are not impossible. A sequence can pass both tests and still describe a network nobody reported, since several different networks usually share the same list of degrees — 3, 3, 2, 2 fits more than one arrangement of five links. The check rules out contradictions rather than confirming truth, which is the ordinary position of a validation rule: it tells you when to stop and ask, not when to relax.
Artikelen die je misschien interesseren
Alle gidsen →Vergelijkbare tools
Bronnen
Een fout in dit artikel gezien?