EART60061 Chap.3 The Normal Distribution and d Values
The Normal Distribution and d Values
The Normal Distribution explains why random variation takes a bell shape and how that shape turns a standard deviation into a probability. The Gaussian equation has two parameters: the mean fixes the position of the peak and the standard deviation its width, which also marks the point where the curve stops steepening.
Every normal distribution holds 68% of its area within one standard deviation and 95% within 1.96, and its tails never reach the axis. The chapter builds a small statistical table by adding the areas of narrow classes, explains why a coarse table disagrees slightly with printed values, and defines d values as the distances that enclose a chosen share.
It then uses a sample's mean and standard deviation as a model of the population to estimate how often a rare measurement, such as a very tall student or a high monitoring result, should occur. The chapter ends by stressing that every probability depends on two stated assumptions, random variation and sample statistics standing in for the population, which is the bridge to confidence limits on means.
What this chapter covers
- 01
The Gaussian equation and its two parameters
- 02
Standard deviation as the inflection point
- 03
Area under the curve and class width
- 04
Distances enclosing 68%, 95% and 99%
- 05
Predicting rare measurements
- 06
Resolution error in a home-made table
- 07
Assumptions behind tail estimates
- 08
Matching a widening curve to a growing standard deviation
Worked example · free
Estimate how often a river temperature exceeds a threshold
- 295% range: 14.2 plus or minus 1.96 times 1.8 = 14.2 plus or minus 3.53, so 10.67 to 17.73 C.
- 1Distance of the threshold: d = (18.0 - 14.2) / 1.8 = 2.11 standard deviations above the mean.
- 2Upper-tail area beyond d = 2.11 is about 0.017, so roughly 1.7% of readings.
Key terms
- Gaussian equation
- The formula for the normal distribution, fully determined by a mean and a standard deviation.
- Unit normal distribution
- A normal distribution scaled so that the total area under the curve equals one.
- Asymptotic tail
- The part of the curve that approaches the axis without touching it, so extreme values are rare but possible.
- d value
- The number of standard deviations either side of the mean that encloses a chosen share of a normal distribution.
- Tail probability
- The share of a distribution lying beyond a chosen distance from the mean, used to judge how rare a value is.
The Normal Distribution and d Values FAQ
What is a d value?
A d value is the number of standard deviations from the mean that encloses a chosen share of a normal distribution. The value 1.96 encloses 95% and is the one used most often in environmental work. The module has you derive such values yourself from areas under the curve.
Why does my own table give 73% instead of 68%?
The table is built from a small number of classes, so each slice of the curve is approximated by a rectangle. With only 50 classes the sums drift from the true values; finer classes bring them closer. The difference is a resolution effect, not a mistake in the equation.
Why do modelled frequencies need multiplying by the class width?
The Gaussian equation gives a frequency per unit of the variable. When classes are five centimetres wide, each class collects five units' worth, so the frequencies must be multiplied by five for their total to equal the sample size, as the heights workbook shows.
Can a normal distribution predict values that have never been measured?
Yes, within its assumptions. Because the curve never touches the axis, it gives a small but real probability to extreme values such as a student over seven feet tall. The estimate is only as good as the assumptions that variation is random and that the sample describes the population.
Where on the curve does the standard deviation sit?
Following the curve up from the axis, it becomes steeper and then begins to flatten. The point where the gradient stops increasing lies exactly one standard deviation from the mean, which gives the number a geometric meaning and is used in later exercises to locate the 68% area.
Exam move
Rebuild the d value table yourself: tabulate the Gaussian equation across narrow classes, multiply by the class width and add the areas within one, two and three standard deviations. Then practise converting in both directions, from a threshold to a probability and from a probability to a range, sketching the curve each time. Finish by stating the assumptions in one sentence for every tail estimate you make.
Keep a short table of the d values you use most, 1.00, 1.96 and 2.58, with the share each encloses, and check your own spreadsheet values against it.