Skip to content
ConceptReviewed

Hypothesis

Name variants

English
Hypothesis
Kanji
仮説

Quality / Updated / COI

Quality
Reviewed
Updated
COI
none

TL;DR

A hypothesis is a testable statement that can be evaluated with data, often framed as null and alternative hypotheses.

Definition

A hypothesis defines a specific claim about a population or relationship that can be tested through evidence. In statistical testing, the null hypothesis represents no effect, while the alternative represents a meaningful difference or relationship. Clear hypotheses guide experiment design, sample size decisions, and interpretation of results.

Decision impact

  • It determines the experiment design and what data are required.
  • It shapes which metrics and thresholds indicate success or failure.
  • It influences how confidently results can be acted upon.

Key takeaways

  • State hypotheses in measurable terms with defined variables.
  • Specify null and alternative hypotheses before testing.
  • Choose sample sizes that can detect meaningful effects.
  • Interpret results in context, not just by p-values.
  • Document assumptions so others can replicate the test.

Misconceptions

  • A hypothesis is not a casual guess; it is a testable statement.
  • Failing to reject the null does not prove the null is true.
  • Changing hypotheses after seeing data undermines validity.

Worked example

A product team tests whether a new onboarding flow increases activation. The null hypothesis states there is no difference, and the alternative states activation increases by at least 5%. They run an A/B test with a sample size large enough to detect the effect. Results show a statistically significant increase, and the team rolls out the change while documenting assumptions and limitations.

Citations & Trust

  • Introductory Statistics 2e 9.1 Null and Alternative Hypotheses (OpenStax)