Null Hypothesis¶
Formula¶
\[
H_0: \theta = \theta_0
\quad \text{(example form)}
\]
Parameters¶
- \(H_0\): baseline/default hypothesis
- \(\theta\): parameter of interest
- \(\theta_0\): hypothesized value
What it means¶
The null hypothesis is the reference claim tested against the observed data.
What it's used for¶
- Formal hypothesis tests.
- Defining p-values and rejection thresholds.
Key properties¶
- Chosen before seeing results in principled analyses.
- Often represents "no effect" or a baseline model.
Common gotchas¶
- Failing to reject \(H_0\) is not proof that \(H_0\) is true.
- Poorly chosen nulls can make tests uninformative.
Example¶
In an A/B test, \(H_0\) may state that the two conversion rates are equal.
How to Compute (Pseudocode)¶
Input: research question and statistical test setup
Output: null-hypothesis specification H0
define the parameter/contrast of interest
state H0 as a baseline/no-effect/no-difference model for that quantity
pair H0 with an alternative hypothesis H1 and a planned test statistic
return the hypothesis specification
Complexity¶
- Time: Not an algorithmic computation; this is a study-design/specification step
- Space: \(O(1)\) for the formal hypothesis statement
- Assumptions: Hypothesis specification precedes data analysis and determines downstream test/p-value computations