What Does Z Alpha (zα) Mean?
The notation zα names a point on the z curve by the area in its right tail: zα is the value with exactly α of the distribution above it. It looks like alphabet soup the first time; it is actually the most honest naming system in statistics, because the name states the definition.
The definition: P(Z > zα) = α. So z₀.₀₅ is the z value with 5% of the area to its right: 1.645. Every critical value you know is a zα in disguise — the subscript is the tail area, never a percentile and never a probability to the left.
The standard set, decoded
- z₀.₁₀ = 1.282 — 10% in the right tail; the one-tailed 10% critical value, and the 80% confidence two-tailed building block.
- z₀.₀₅ = 1.645 — the one-tailed 5% critical value; also half of every 90% confidence interval.
- z₀.₀₂₅ = 1.960 — written zα/2 when α = 0.05: the 95% confidence critical value, because a two-tailed split puts α/2 = 0.025 in each tail.
- z₀.₀₁ = 2.326 and z₀.₀₀₅ = 2.576 — the 1% one-tailed line and the 99% confidence critical (α/2 = 0.005).
To find any zα from the cumulative table, convert: the table reads left areas, and zα has left area 1 − α. For z₀.₀₄, search the table body for 0.96: the answer is 1.751. That one-step conversion — right-tail naming, left-tail table — is the whole skill, and it is exactly the reverse lookup mechanised on the find z score from probability page. The collected values, with their test and interval roles, live on the critical z values page.
Why the notation exists
Hypothesis testing runs on tail areas: α is chosen first (the false-positive risk the analyst accepts), and the critical value is whatever z makes that tail real. Naming the value after its tail keeps the logic visible in formulas — a rejection rule written "reject when z > zα" says what it does. The same convention generalises across distributions (tα,df and χ²α,df name their tables' critical values identically), so learning it here pays rent on the t table and chi square table too. Its close cousin z* — the star notation used by many textbooks for the same critical values in confidence intervals — gets its own post: what is z star?
One more member completes the working set: z₀.₂₀ = 0.842, the value with a fifth of the distribution above it — it appears in 60% confidence intervals (α/2 = 0.20) and in power calculations. With z₀.₂₀, z₀.₁₀, z₀.₀₅, z₀.₀₂₅, z₀.₀₁ and z₀.₀₀₅ memorised or derivable in seconds, essentially every formula in an introductory course that names a zα becomes arithmetic rather than a table hunt.
Also asked as: what does the notation za indicate · what does z alpha indicate · what is z alpha in statistics · z alpha meaning.
Related questions
Frequently asked questions
Is zα the same as a percentile?
It is the mirror of one: zα sits at the (1 − α) × 100th percentile. z₀.₀₅ is the 95th percentile. Same point, named from opposite tails — tests name from the right because α lives there; percentiles name from the left because ranks accumulate there.
What is zα/2 for a 95% confidence interval?
α = 0.05, so α/2 = 0.025 and z₀.₀₂₅ = 1.96. The interval leaves 2.5% in each tail, 95% in the middle — the subscript arithmetic is the interval's design stated in one symbol.