The null hypothesis is a reference statement used in statistical testing, often representing no difference, no association or no treatment effect of the type being examined. The core aim of Science mastery is not to teach students that the null hypothesis is the claim scientists secretly want to prove wrong. It is to help them understand why statistical inference needs a clearly defined baseline model against which the observed data can be compared.
For students and parents searching for null hypothesis, null hypothesis in Science, null hypothesis examples, H0, statistical significance, p-value or null vs alternative hypothesis, the most useful principle is this: the null hypothesis is a testable reference model. Statistical evidence is then used to ask how compatible the observed data is with that model.
The null hypothesis gives statistical reasoning something precise to challenge.
The 60-Second Null Hypothesis
A null hypothesis may state:
- there is no difference between two group means;
- there is no association between two variables;
- a treatment has no effect;
- a parameter equals a specified reference value.
It is often written as H0.
Wait, What? Failing to Reject the Null Does Not Mean the Null Is Proven True?
Correct.
This distinction matters enormously.
If the data is not sufficiently inconsistent with the null model, the usual conclusion is:
We fail to reject the null hypothesis.
That does not prove:
- the effect is exactly zero;
- the groups are identical;
- no relationship exists.
The study may simply lack enough evidence to reject the null model.
Null Hypothesis vs Research Hypothesis
Research question:
Does fertiliser concentration affect plant growth?
Null hypothesis:
There is no difference in mean growth between the specified fertiliser conditions.
Research or alternative hypothesis:
At least one condition differs in mean growth.
The null provides the reference against which statistical evidence is evaluated.
Why Statistics Uses a Null Model
Experimental data always varies.
Even when no real effect exists, two sample means will rarely be exactly equal.
The null model asks:
Could a difference this large arise from ordinary random variation if the underlying effect were absent?
This is the logic behind many significance tests.
A Worked Example: Plant Growth
Group A receives no fertiliser.
Group B receives fertiliser.
Observed mean growth differs by 2.1 cm.
The null hypothesis might state:
The true mean growth difference is zero.
The statistical test then asks how unusual a 2.1 cm or more extreme difference would be under the null model, given the design and variability.
A Worked Example: Correlation
A study examines sleep duration and attention score.
Null hypothesis:
There is no population-level linear association of the type being tested between sleep duration and attention score.
If the observed correlation is strong relative to its expected random variation, the null may be rejected.
But statistical association still does not prove causation.
The Null Hypothesis and p-Values
A p-value is calculated under the null model.
Conceptually, it asks how unusual the observed result—or a more extreme one—would be if the null model were correct and the test assumptions held.
A small p-value indicates the data is relatively incompatible with the null model.
The Null Hypothesis Is Not Always “Nothing Happens”
A null hypothesis may specify a particular reference value rather than literal zero.
Examples:
- mean difference = 0;
- correlation = 0;
- risk ratio = 1;
- new measurement equals a standard value.
The exact null depends on the parameter and test.
Reject vs Fail to Reject
Two common outcomes are:
Reject H0 — the data is sufficiently inconsistent with the null model under the chosen test.
Fail to reject H0 — the evidence is insufficient to reject the null model.
Avoid writing:
“Accept the null hypothesis.”
That wording can imply stronger certainty than the analysis provides.
Null Hypothesis and Type I Error
A Type I error occurs when the null hypothesis is rejected even though it is true.
This is sometimes called a false positive.
The significance threshold controls the long-run probability of this error under the model assumptions.
Null Hypothesis and Type II Error
A Type II error occurs when a real effect exists but the test fails to reject the null hypothesis.
This is a false negative.
Low statistical power makes Type II errors more likely.
See Statistical Power.
Null Hypothesis and Effect Size
A null-hypothesis test asks whether the effect differs sufficiently from the null value.
It does not tell you whether the effect is large enough to matter.
Always examine:
- effect size;
- confidence interval;
- scientific relevance.
See Effect Size.
Null Hypothesis and Confidence Intervals
Confidence intervals show which effect sizes remain reasonably compatible with the data under the model.
For many common tests, if the confidence interval excludes the null value, the corresponding significance test rejects the null at the matching level.
See Confidence Intervals.
Primary Science Foundations
Primary learners do not need formal null-hypothesis testing.
They can build the underlying reasoning by asking:
- Could this difference happen just by chance?
- What would we expect if the treatment made no difference?
- How much evidence would make us reconsider that expectation?
Secondary Science Null Hypothesis
Secondary students should increasingly understand:
- H0 as the reference model;
- reject vs fail to reject;
- p-values;
- Type I and Type II errors;
- effect size and confidence intervals.
How to Practise Null-Hypothesis Reasoning
For each study, write:
- research question;
- null hypothesis;
- alternative hypothesis;
- null value;
- what evidence would count against H0.
Common Null-Hypothesis Mistakes
- treating failure to reject as proof of no effect;
- writing a vague null with no measurable parameter;
- confusing the null with the research question;
- ignoring effect size;
- assuming p-values measure the probability that H0 is true.
Frequently Asked Questions
What is the null hypothesis?
The null hypothesis is a reference statistical model, commonly representing no effect, no difference or no association of the type being tested.
What does H0 mean?
H0 is the standard symbol for the null hypothesis.
Does failing to reject H0 prove no effect?
No. It means the study did not provide sufficient evidence to reject the null model.
How is the null hypothesis related to p-values?
The p-value is calculated assuming the null model and asks how unusual the observed result would be under that model.
Useful eduKateSG Routes
The Core Aim
The null hypothesis gives statistical inference a reference point.
Define the no-effect model clearly. Compare the evidence with it. Reject only when the data justifies it. Never confuse “not rejected” with “proved true”.
That is the core aim: make statistical decisions precise without pretending uncertainty has disappeared.
Properly taught kids shine a bright light into the future.