UCL v UTL

This dataset $[0.1, 0.2, 0.05]$ is the sample set used in Hewett, 2006 so we’re going to use it here too. It perfectly illustrates why the Upper Tolerance Limit (UTL) (the defensive choice for occupational hygiene), and the Upper Confidence Limit (UCL) (serves a different purpose than ‘most exposed worker in a group’) both fail when sample sizes are small.

To paraphrase multiple regulators I’ve spoken to:

“We are not as interested in the average exposure to a work group as the potentially highest exposure end of the workgroup exposure distribution”.

This is where the langauge gets loose and terms are used interchangably; specfically UCL and UTL.

The Core Difference

• UCL (of the Mean): Estimates where the average exposure lies, or the limit below which you are confident it sit under.

NORM.DIST formula: $mean+ 1.645 \times sd$
NORM.DIST formula: $\bar{x} + 1.645 \times \sigma$

LOGNORM.DIST: $GM \times GSD^{1.645}$

• UTL (95/95): Estimates the limit where the worst-case (95th percentile) of the worse-case of exposures of the population – not the sample set- lie with high confidence.

---

The $k$-factor for small datasets

The $k$-factor is a statistical multiplier used to calculate one-sided tolerance limits (like the 95/95 UTL) when the true population mean and standard deviation are unknown. Unlike a standard Z-score—which assumes you have perfect knowledge of the entire population—the $k$-factor explicitly accounts for the sampling error inherent in small datasets. It acts as a “safety margin” or uncertainty penalty: because we are estimating the distribution from only a few data points, $k$ must be significantly larger than a Z-score to ensure we maintain our desired confidence level (e.g., 95%) that a specific proportion of the population is actually covered.

As the sample size ($n$) increases and our estimate of the population becomes more reliable, the $k$-factor decreases, eventually converging toward the standard normal quantile ($1.645$ for the 95th percentile) as $n$ approaches infinity.

---

Comparison of $k$ vs. $z$ (95% Confidence / 95% Coverage)

Sample Size (n)	k-factor	Z-score (z0.95​)	The “Uncertainty Penalty”
3	7.656	1.645	+365%
10	2.911	1.645	+77%
30	2.220	1.645	+35%
$\infty$	1.645	1.645	0%

---

The Worked Example

With only three data points, the uncertainty is massive. Even though your “best guess” (point estimate) for the 95th percentile is low, the statistical safety margin required for 95% confidence is huge.

Metric	Calculation Logic	Result	Context
Point Estimate (95th %ile)	$\exp^{(\bar{y} + 1.645 \cdot s_y)}$	0.31	“The most likely 95th percentile.”
UCL95 (of the Mean)	Land’s H-statistic calculation	~2.8	“We are 95% sure the average is below 2.8.”
UTL 95/95	$\exp^{(\bar{y} + 7.656 \cdot s_y)}$	20.2	“We are 95% sure 95% of shifts are below 20.2.”

---

How to Calculate It

The “jump” from 0.31 to 20.2 happens because of the $k$-factor, which penalizes the small sample size ($n=3$).

Log-Transform the data

$\ln(0.05) = -3.0$,

$\ln(0.1) = -2.3$,

$\ln(0.2) = -1.6$

Descriptive statistics

$\bar{y} = -2.3$, $GM = 1.0$

$s_y = 0.693$, $GSD = 2.00$

Select the K-Factor

Account for n=3. For a 95/95 limit with 3 samples, use $k = 7.656$. (For $n=20$, this would drop to ~2.4).

Calculate the Limit

$UTL = \exp^{(-2.3 + (7.656 \cdot 0.693))} = \exp^{(3.0)} \approx \mathbf{20.2}$.

---

The “Sample Size Penalty” Simulator

You can use this tool to see how increasing your sample size ($n$) collapses the massive gap between your point estimate (0.31) and your defensible 95/95 limit (20.2).