Test scoring strategy

Why attempting more beats being careful

Same intelligence, same 12 minutes — strategy does most of the work. Here's the math.

The perfectionist
Jordan, 135 IQ
~13 / 50
Answers carefully. Solves 15 questions brilliantly. Leaves 35 blank when time runs out.
The strategist
Sam, 135 IQ
~22 / 50
Attempts all 50. Guesses on 15 hard ones. Never leaves anything blank.
vs

Same intelligence. Same 12 minutes. Score differs by a factor of 1.7. In percentile terms: reject Jordan, hire Sam — on a test that was supposed to measure how they think.

Drag the slider below to see the math yourself.

The strategy curve

Attempted
30
Time each
24.0s
Likely score
23 / 50

The gap between guessing and skipping

Purple (Sam's strategy) beats pink (Jordan's strategy) at every single point along the curve. The shaded area between them is the free points you leave on the table if you skip instead of guess.

Attempt all, guess when stuck Skip questions you don't know
Line chart showing expected score curves across 1-50 questions attempted.

Your strategy

30

How prepared are you?

You've done a few sample questions and know the format.

Notice what the chart shows. The green line (attempt all, guess on the rest) stays above the red line (leave blank) at every single value of n. There is no strategy where leaving questions blank beats guessing. None.

Why these tests are broken as filters

Cognitive tests like the Wonderlic, PI Cognitive, and CCAT don't penalize wrong answers. Guessing a random answer has positive expected value. A 4-option question scores 25% on average from random guessing alone. A 5-option question scores 20%.

This isn't obscure knowledge. It's the first line of every test-prep guide. But companies still use these scores as a first-pass filter and treat them like IQ proxies.

What they're actually filtering for:

Those are real traits. But they're traits about test-taking, not about reasoning, adaptability, or "learning speed in the workplace" — which is what the tests claim to measure.

The hidden bias

People with SAT/GRE/GMAT history already know the strategy. Career switchers, older candidates, people from education systems that don't use this format — don't. Same intelligence, different score. Same job, different outcome.

That's not a cognitive filter. That's a familiarity filter dressed as a cognitive one.

What to do about it

If you have a test coming up, you don't have time to fix the system. Beat it.

1. Never leave a question blank. Ever. Even if out of time, tap random answers in the last 5 seconds. Free expected points.

2. Practice pacing, not content. The question types are finite. Most of your prep should be about not getting stuck — learning to recognize when you're sinking time into a question and need to move on.

I built a free pacing timer that trains exactly this. No signup, no ads on the session — just a timer.

Try the free pacing timer →
This is an illustrative model based on the standard speed-accuracy tradeoff curve from cognitive psychology. Default parameters are plausible but not empirically calibrated to a specific test. Tune the sliders to match your own ability. The directional conclusion — that attempting more questions beats leaving blank on no-penalty tests — is arithmetic, not opinion.

The math: Expected score = n · p(720/n) + (50 − n) · g where n is questions attempted, g is guess rate, and p(t) = Pmax · (1 − e−t/τ) is per-question accuracy as a function of time.