---
title: Semi-tail Units in Statistical Testing
url: https://www.emergentmind.com/topics/semi-tail-units
type: topic
---

# Semi-tail Units in Statistical Testing

A semi-tail unit is a quantile-based measure for expressing extremeness, efficiency, and evidence in statistical hypothesis testing. Each semi-tail unit corresponds to a halving of the tail area under a null distribution, and this logarithmic base-2 scale permits direct additivity, uniform critical value progression, and natural efficiency quantification. Semi-tail units have been proposed as a universal, interpretable alternative to p-values, z-scores, and test-specific quantiles, applicable to any ordered statistical distribution [2506.22910].

## 1. Definition and Mathematical Formulation

Let $X$ be a real-valued test statistic, with $P = P(X \ge x_{\text{obs}})$ the right-tail probability under the null hypothesis. The **semi-tail unit**, denoted $s$, is defined by
\[
s = -\log_2 P
\]
which quantifies how many doublings of rarity have occurred in the observed tail probability. For example, $s=3.32$ corresponds to $P=0.10$, $s=4.32$ to $P=0.05$, and $s=5.32$ to $P=0.025$. In other words, each increment of 1 in $s$ represents a halving of $P$. For two-tailed tests, the $\zeta$-value is employed:
\[
\zeta = -\log_2 \left[2P(X \ge |x_{\text{obs}} - m|)\right]
\]
where $m$ is the median under $H_0$.

This formulation enables direct mapping between percentile, tail probability, and semi-tail unit via
\[
s = -\log_2(1-q)
\]
for upper-tail percentile $q$. For instance, a result at the 99th percentile yields $s \approx 6.64$.

## 2. Interpretive Properties and Practical Benchmarks

The primary interpretive property is additivity and halving: every $+1$ increment in $s$ divides the tail probability by two. Table 1, mapping selected probabilities and percentiles to $s$:

| Tail Probability $P$ | Semi-tail $s$ | Percentile |
|---------------------|---------------|------------|
| $0.10$              | $3.32$        | $90\%$     |
| $0.05$              | $4.32$        | $95\%$     |
| $0.025$             | $5.32$        | $97.5\%$   |
| $0.01$              | $6.64$        | $99\%$     |

For two-tailed tests, the two-tailed $p$-value is $2^{-|\zeta|}$. This allows writing all critical values as equally spaced points on the $s$ (or $\zeta$) scale, eliminating the distribution-specific tables common to classical hypothesis testing.

## 3. Conversion from and to Classical Test Statistics

To convert any test statistic:
1. Compute the classical $p$-value under the null.
2. Set $s = -\log_2 p$ (one-tailed) or $\zeta = -\log_2 p$ (two-tailed).

Examples:
- Normal $Z$-test, $z=1.96$: two-tailed $p \approx 0.05 \Rightarrow \zeta \approx 4.32$.
- $t$-test with $\nu=10$, $t=2.228$: two-tailed $p \approx 0.05 \Rightarrow \zeta \approx 4.32$.
- $\chi^2_1$ test, $\chi^2=6.63$: one-tailed $p\approx0.01\Rightarrow s\approx6.64$.
- Poker hand "royal flush": $s \approx 19.3$ (tail prop $\approx1.54\times10^{-6}$).

All critical $\alpha$-levels become $s_c = -\log_2\alpha$, so achieving a more stringent level is an arithmetic progression in $s$.

## 4. Evidence Combination and Additivity

If $p_1,\ldots,p_k$ are independent one-tailed $p$-values, their joint tail probability is $\prod_i p_i$, so the aggregate $s$-value is
\[
s_{\text{total}} = -\log_2\left( \prod_i p_i \right ) = \sum_i s_i
\]
This allows for direct addition of semi-tail units when accumulating evidence across studies or tests. For example, three $p$-values of $0.10$, $0.08$, and $0.05$ yield $s=3.32$, $3.64$, and $4.32$, so $s_\text{total}=11.28$.

## 5. Semi-tail Units and Bahadur Efficiency

Semi-tail units naturally encode efficiency in terms of Bahadur slopes. The Bahadur exact slope for a sequence $T_n$ under an alternative $m$ can be written as
\[
B_n(t) = -2\ln P_{m}(T_n\ge t) = (2\ln2)\, s_n(t)
\]
and the asymptotic Bahadur slope becomes $\beta=(2\ln 2)\sigma$ with $\sigma=\lim_n n^{-1} s_n$. Thus, the $s$-value per sample is an interpretable "bits" or "semi-tail units per observation" rate.

Semi-tail *efficiency difference* between two tests $1,2$ is $\Delta\sigma = \sigma_2 - \sigma_1$. A positive $\Delta\sigma=0.15$ means test $2$ drives the sample $0.15$ units deeper into the tail (i.e., $10\%$ smaller $p$-value) compared to test $1$, an interpretable "distance" in evidence [2506.22910].

## 6. Unification and Advantages over Classical Scales

Semi-tail units provide a universal, interpretable, and additive scale for expressing the extremeness of results across all testing contexts. Key unifying features:
- All decision thresholds are linear and equally spaced ($s$ increases by $1$ when $\alpha$ halves).
- No distribution-specific lookup tables are required.
- Additivity for combining independent evidence (direct sum of $s$).
- Simple efficiency comparison: differences in $s$ per sample correspond to multiplicative changes in tail probability/explanatory power.
- Applicable to any ordered distribution, not just those with tabulated quantiles (e.g., permutation tests, empirical distributions, and discrete combinatorial ranks).

These properties, as demonstrated in applications from normal and $t$-tests to poker hand rankings, distinguish the semi-tail unit from p-values, z-scores, and conventional quantile measures [2506.22910].

## 7. Worked Examples and Applications

### Percentile to $s$-value mapping:
- $q=0.75 \to s=2$
- $q=0.90 \to s=3.32$
- $q=0.95 \to s=4.32$
- $q=0.99 \to s=6.64$

### Discrete ranking (combinatorial, e.g., poker hands):
Let $N$ be total outcomes, $k$ the rank or better, $s = -\log_2(k/N)$. Royal flush: $s\approx19.3$.

### Combining $p$-values:
Three experiments give $s$-values $3.32$, $3.64$, $4.32$; sum is $11.28$, corresponding to $p=2^{-11.28}\approx4.1\times10^{-4}$.

### Efficiency:
The difference in semi-tail slope per sample ($\Delta\sigma$) quantifies how much more rapidly one test accumulates evidence, providing a bits-scale improvement directly interpretable (e.g., $\Delta\sigma=0.15$ is a $10\%$ reduction in $p$ per observation in the asymptote).

Semi-tail units constitute a universal and principled system for standardization and comparison of test statistics, evidence accumulation, and efficiency reporting in modern statistical research [2506.22910].

Source: https://www.emergentmind.com/topics/semi-tail-units