Null and alternative hypotheses (H₀ H₁)
H₀ is the null hypothesis, the statement a hypothesis test starts from and looks for evidence against, usually that there is no difference. H₁, also written Hₐ, is the alternative hypothesis, which the test concludes if H₀ is rejected.
How to read it
OpenStax’s Introductory Statistics describes H₀ as “a statement of no difference between the variables” and Hₐ as a claim about the population that contradicts H₀, “what we conclude when we reject H₀”. Penn State’s STAT 500 course notes write the alternative as Hₐ or H₁ and call it the research hypothesis; NIST’s Engineering Statistics Handbook writes H₀ and Hₐ. Both hypotheses are statements about the population, so they are written with the population’s letters: OpenStax’s examples include H₀: μ = 2.0 against Hₐ: μ ≠ 2.0 for a mean (μ), and H₀: ρ = 0 for a correlation (ρ).
Reject or fail to reject
A test ends in one of two decisions, “reject H₀” or “do not reject H₀” (OpenStax). NIST’s handbook explains why the second is not called accepting it: to accept a hypothesis does not mean that it is true, only that there is no evidence to believe otherwise. Fisher put it more strongly in 1935: the null hypothesis “is never proved or established, but is possibly disproved”. How strong the evidence against H₀ is, is given by the P-value (p).
Which sign goes in H₀
OpenStax’s rule is that H₀ always has a symbol with an equal in it (=, ≤ or ≥) and Hₐ never does (≠, < or >). It adds that many researchers write = in the null hypothesis even when the alternative has > or <.
Look-alikes
The small 0 is a zero: Unicode’s SUBSCRIPT ZERO, U+2080, with SUBSCRIPT ONE, U+2081, for H₁. The α that appears beside H₀ in a test is the chance of rejecting H₀ when it is true, the Type I error.
Read next
- Next in Statistics symbolsP-value (p)A p-value (p) is the probability of getting a result at least as extreme as the one observed if the null hypothesis were true. The smaller it is, the stronger the evidence against the null hypothesis; p < 0.05 has long been the usual line for calling a result “statistically significant”.
- α is the chance of rejecting H₀ when it is true.AlphaAlpha (Α, α) is the first letter of the Greek alphabet. In statistics α is the chance of a Type I error; in physics it is the fine-structure constant and the alpha particle; in machine learning it is often the learning rate or the strength of regularisation.
More statistics symbols
Browse symbols drawn with letters and numbers.
Sources
- Introductory Statistics 2e, 9.1 Null and Alternative Hypotheses, OpenStax (Rice University), Section 9.1; Table 9.1; Examples 9.2 and 9.4
- Introductory Statistics 2e, 12.4 Testing the Significance of the Correlation Coefficient, OpenStax (Rice University), Section 12.4
- STAT 500 Applied Statistics, 6a.1 Introduction to Hypothesis Testing, Penn State online course notes (archived copy), Penn State Eberly College of Science, Department of Statistics, Null hypothesis; Alternative hypothesis
- 1.3.5. Quantitative Techniques, NIST/SEMATECH e-Handbook of Statistical Methods, National Institute of Standards and Technology (NIST), Hypothesis Tests; Practical Versus Statistical Significance
- Earliest Uses of Symbols in Probability and Statistics, Jeff Miller, Earliest Uses of Various Mathematical Symbols, MacTutor History of Mathematics, University of St Andrews, Symbols in Statistics; Symbols associated with the normal distribution; Symbols associated with testing hypotheses
- Earliest Known Uses of Some of the Words of Mathematics (H), Jeff Miller, MacTutor History of Mathematics, University of St Andrews, HYPOTHESIS and HYPOTHESIS TESTING in Statistics
- Superscripts and Subscripts: Range 2070–209F (code chart), The Unicode Standard, Version 18.0, Unicode Consortium, 2080, 2081
Researched, edited and approved by Rory Hansen before publishing. Last reviewed October 8, 2026. How we research
Added .