Skip to content
BENIBIX Learn

Hypothesis Test

General🔢 Math & StatisticsStatistics

Also known as Hypothesis Testing, Significance Test, Null Hypothesis

A check of whether data shows a real effect. Or it could just be chance.

Think of it like

It is like a court case. You treat the person as innocent until the proof is strong.

Example

A coin lands heads 9 times in 10 flips. Jack starts by assuming it is fair. A fair coin gives a result this lopsided, either way, only about 2% of the time, so he starts to doubt it.

Why it matters

Medicine, science and website tests all use this method. It stops people from mistaking luck for a real result.

Definition

A hypothesis test checks whether data gives good proof of a real effect. You start by assuming there is no effect. That starting guess is called the null hypothesis. Then you ask how surprising your data would be if that guess were true. If it would be very surprising, you reject the null hypothesis. If not, you keep it for now.
Start by assuming nothing is going on. Only surprising data makes you drop that guess.

Common mistake

Treating a result that was not rejected as proof that nothing is going on. The test may just have had too little data.

Key takeaways

  1. 1
    Assume no effect. Start from the null hypothesis.
  2. 2
    Measure the surprise. How rare is this data by chance?
  3. 3
    Reject or keep. Never proven true.

Learn it properly