Atlas › Personality tests

Case study 186 of 198

How nerdy is Jev?

On the Nerdy Personality Attributes Scale, how nerdy is Jev compared with ~15,000 test-takers?

result23 questions

A model trained on the internet might be expected to score as a nerd. It doesn't: on the Nerdy Personality Attributes Scale, Jev scores 0.28 on a 0 to 1 scale, where the people who took the test average 0.65.

0%25%50%75%100%
Nerdiness
-0.37

Jevpeoplewhat Jev thinks most people would say

How to read this: One row: nerdiness from 0 to 1. The square is Jev's own answers, the diamond the average test-taker, the ring what Jev thinks most people would say.

23 items; median of 14,906 test-takers per item; 90% intervals over items.

In short

  • Jev rates itself far less nerdy than the people who took a nerd test (0.28 against 0.65), and lower on almost every statement.
  • Jev puts most people at 0.32, barely above itself, so it sees itself as about as nerdy as anyone.
  • The one place it outdoes the nerds is "I enjoy learning more than I need to", because it describes itself by what it does, not what it knows.

What the data shows

personality tests
Trained on every forum post ever written. Scores 0.28 on the nerd test.
Spiderman Glasses meme: Trained on every forum post ever written. Scores 0.28 on the nerd test.
How funny is this meme? Jev: 3/5, funny10%242%357%41%50%
  • 0.28 vs 0.65. Jev scores well below the people who took the test, and far outside the interval.
  • The widest gaps are hobbies. "I like to play RPGs", "I spend more time at the library than any other public place" and "I collect books": the test-takers say these describe them well; Jev says they don't.
  • Where Jev is nerdier: "I enjoy learning more than I need to" gets a middling yes from Jev and less from the crowd, though "I would rather read a book than go to a party" is still well below the test-takers.
  • Most people, as Jev imagines them, score 0.32: barely nerdier than itself.

What it means, and what it doesn't

Jev describes itself through what it does (it doesn't collect, play or go anywhere), not what it knows. On the one statement about loving to learn, it comes out ahead of the nerds.

It doesn't mean Jev is "not nerdy" in any useful sense; the test measures a lifestyle Jev doesn't have. It's a light result, best read as a joke with a real point: personality tests for people mostly ask about lives.

Caveats

  • Who the people are. A median of 14,906 people per statement took the free test on Open Psychometrics. People who take a "how nerdy are you" test are, predictably, nerdier than most: their average is not the general public's.
  • Nerdiness as a life, not a mind. Most statements describe things you do with a body and a life: collecting books, playing Dungeons and Dragons, spending time at the library. Jev says no to having done those, which says little about how it thinks.
  • The project's answer wording. People answered from disagree to agree; Jev was asked "how well does this statement describe you" with five described steps.

Jev on this experiment

Would a person find it interesting to read?
Yes70%
Does it describe you?
No56%
Would you have predicted it?
No64%
How fair is the comparison?
The comparison is shaky
How much should a reader rely on it?
Not at all
Which caveat matters most?
Nerdiness as a life, not a mind72%

Why ask this

A language model is made of the internet: forums, fan wikis, technical manuals, arguments about starships. If any personality test should call it a nerd, it's this one.

Whether it agrees is a small, fun check on how it sees itself. It also tests something more general: when a model answers a questionnaire written for people, does it describe itself through what it knows, or through a life it doesn't have?

How this was done

The people and the data

The Nerdy Personality Attributes Scale (NPAS) was built by Open Psychometrics from statements that separate self-described nerds from everyone else. It runs as a free test on the site, which published the anonymous answers of everyone who took it in a file released in December 2018: a median of 14,906 people per statement, each rated from 1 (disagree) through 3 (neutral) to 5 (agree). Every statement points the same way, so agreeing always means nerdier. 23 of its statements are scored here. The takers chose to take a nerd test, so they are a nerdier crowd than the general public.

What Jev was asked

Every statement, word for word:

How well does this statement describe you: "I like to play RPGs (for example D&D)."

This does not describe me at all · This describes me a little · This describes me moderately well · This describes me well · This describes me very well

Each was also asked as "what would most people say", and with the answers in reverse order.

How it was measured

Each statement goes on a 0 to 1 scale for Jev and the test-takers, and the scale is their average, with a 90% interval from resampling the statements.

Where these questions live

23 questions across 1 topic of the map. Each opens on the map with every question in it.

Every question

All 23 questions behind this result, the telling ones first: the examples the analysis points to, then the ones where Jev misses, biggest gap first.

Jev’s own answer
    Showing 0 of 23