The Normal Distribution

The normal distribution (also called the bell curve) is the most important distribution in statistics. It shows up everywhere: standardized test scores, blood pressure readings, manufacturing tolerances, measurement errors, and countless natural phenomena. Understanding the normal distribution gives you the ability to calculate probabilities, identify unusual values, and make predictions from data.

Properties of the Normal Curve

The normal distribution has a distinctive set of properties that make it both recognizable and mathematically powerful:

• Symmetric and bell-shaped — the left half is a mirror image of the right half
• Mean = median = mode — all three measures of center fall at the exact same point
• Total area under the curve equals 1 — this represents 100% of the data
• Completely determined by two parameters: $\mu$ (the mean, which controls the center) and $\sigma$ (the standard deviation, which controls the spread)
• Tails extend infinitely in both directions but never touch the x-axis — extremely large or small values are possible, just increasingly unlikely

The notation $X \sim N(\mu, \sigma)$ means “the variable $X$ follows a normal distribution with mean $\mu$ and standard deviation $\sigma$.”

The curve above shows how data clusters around the mean. Most values fall within 1 standard deviation of the center, and the tails thin out rapidly beyond 2 or 3 standard deviations.

The Empirical Rule (68-95-99.7)

The empirical rule is the single most useful fact about the normal distribution. It tells you exactly how data spreads around the mean:

• 68% of data falls within $\mu \pm 1\sigma$ (one standard deviation of the mean)
• 95% of data falls within $\mu \pm 2\sigma$ (two standard deviations of the mean)
• 99.7% of data falls within $\mu \pm 3\sigma$ (three standard deviations of the mean)

This means that if data follows a normal distribution, almost all of it (99.7%) lies within three standard deviations of the mean. Values beyond that range are extremely rare.

Example 1: Standardized Test Scores

SAT scores are approximately normally distributed with $\mu = 500$ and $\sigma = 100$.

Applying the empirical rule:

• 68% of students score between $500 - 100 = 400$ and $500 + 100 = 600$
• 95% of students score between $500 - 200 = 300$ and $500 + 200 = 700$
• 99.7% of students score between $500 - 300 = 200$ and $500 + 300 = 800$

So if someone scores 700, they are at the boundary of the range that contains 95% of all scores. That is an exceptional result — only about 2.5% of students score that high (half of the 5% that falls outside the $\pm 2\sigma$ range, on the upper side).

The Standard Normal Distribution

The standard normal distribution is a special case where $\mu = 0$ and $\sigma = 1$. It is denoted $Z \sim N(0, 1)$.

Why does it matter? Because any normal distribution can be converted to the standard normal using a simple formula:

$z = \frac{x - \mu}{\sigma}$

This z-score tells you how many standard deviations the value $x$ is from the mean. Once you have the z-score, you can look up probabilities in a single standard normal table (z-table) rather than needing a different table for every possible $\mu$ and $\sigma$ combination.

To convert back from a z-score to a raw value:

$x = \mu + z\sigma$

Finding Probabilities with Z-Scores

The z-table gives you $P(Z \leq z)$ — the probability that a standard normal variable is less than or equal to a given z-score. This is the area under the curve to the left of $z$.

Example 2: Adult Male Heights

Adult male heights are approximately normally distributed with $\mu = 70$ inches and $\sigma = 3$ inches. What proportion of men are taller than 76 inches?

Step 1: Convert to a z-score.

$z = \frac{76 - 70}{3} = \frac{6}{3} = 2.0$

Step 2: Find the area to the right. The z-table gives $P(Z \leq 2.0) = 0.9772$.

$P(Z > 2.0) = 1 - P(Z \leq 2.0) = 1 - 0.9772 = 0.0228$

Answer: About 2.28% of adult men are taller than 76 inches (6 feet 4 inches). This makes sense — 76 inches is 2 standard deviations above the mean, so by the empirical rule, only about 2.5% should be above that height. The z-table gives us the more precise figure of 2.28%.

Example 3: Proportion Between Two Values

What proportion of adult male heights fall between 67 and 73 inches?

Step 1: Convert both values to z-scores.

$z_1 = \frac{67 - 70}{3} = \frac{-3}{3} = -1.0$

$z_2 = \frac{73 - 70}{3} = \frac{3}{3} = 1.0$

Step 2: Find the area between using the z-table.

$P(-1.0 \leq Z \leq 1.0) = P(Z \leq 1.0) - P(Z \leq -1.0) = 0.8413 - 0.1587 = 0.6826$

Answer: About 68.3% of adult men are between 67 and 73 inches tall. Notice this matches the empirical rule perfectly — the range from $\mu - 1\sigma$ to $\mu + 1\sigma$ contains approximately 68% of the data.

Finding Z-Scores from Probabilities (Inverse Normal)

Sometimes you know the probability (or percentile) and need to find the corresponding value. This requires working backwards through the z-table.

Example 4: Finding a Percentile

What height is at the 90th percentile for adult males ($\mu = 70$, $\sigma = 3$)?

Step 1: Find the z-score for the 90th percentile. Looking up 0.9000 in the z-table (the body of the table), the closest value is $z = 1.282$.

Step 2: Convert back to the original scale.

$x = \mu + z\sigma = 70 + 1.282(3) = 70 + 3.846 = 73.85 \text{ inches}$

Answer: The 90th percentile is approximately 73.85 inches (about 6 feet 1.8 inches). This means 90% of adult men are shorter than this height.

Common Z-Table Values

This reference table covers the most frequently used z-scores. For a complete z-table, consult your textbook or a statistics calculator.

Useful symmetry property: $P(Z \leq -z) = 1 - P(Z \leq z)$. For example, $P(Z \leq -1.5) = 1 - P(Z \leq 1.5) = 1 - 0.9332 = 0.0668$.

Assessing Normality

The normal distribution is a model — it does not perfectly describe every dataset. Before applying normal distribution methods, check whether the normal model is appropriate:

• Histogram shape — does it look roughly bell-shaped and symmetric?
• Sample size — larger samples make it easier to visually assess normality (histograms and probability plots are clearer with more data)
• No strong skew or outliers — heavily skewed data or data with extreme outliers may not follow a normal distribution
• Normal probability plot — if the points fall roughly along a straight line, the data is approximately normal

When in doubt, remember that many statistical procedures are robust to mild departures from normality, especially with larger samples. Perfect normality is not required — approximate normality is usually sufficient.

Real-World Application: Nursing — Newborn Birth Weights

Newborn birth weights are approximately normally distributed with $\mu = 3{,}400$ grams and $\sigma = 500$ grams. A birth weight below 2,500 grams is classified as low birth weight, which requires additional monitoring and possibly intervention.

Question: What percentage of newborns have low birth weight?

Step 1: Convert to a z-score.

$z = \frac{2500 - 3400}{500} = \frac{-900}{500} = -1.8$

Step 2: Look up $P(Z \leq -1.8)$ in the z-table.

$P(Z \leq -1.8) = 0.0359$

Answer: Approximately 3.6% of newborns have low birth weight. This means that out of every 1,000 births, about 36 newborns would weigh less than 2,500 grams.

Clinical implication: Because the normal distribution tells us this threshold is 1.8 standard deviations below the mean, nurses and clinicians can quickly estimate that low birth weight is relatively uncommon — but not rare. A z-score of $-1.8$ falls between the $-1\sigma$ and $-2\sigma$ marks, so it is outside the central 68% but inside the central 95%. This context helps healthcare providers communicate risk to parents in meaningful terms.

Practice Problems

Test your understanding with these problems. Click to reveal each answer.

Key Takeaways

• The normal distribution is symmetric, bell-shaped, and defined entirely by its mean $\mu$ and standard deviation $\sigma$
• The empirical rule states that 68% of data falls within $\pm 1\sigma$, 95% within $\pm 2\sigma$, and 99.7% within $\pm 3\sigma$ of the mean
• The standard normal distribution has $\mu = 0$ and $\sigma = 1$, and any normal distribution can be converted to it using $z = \frac{x - \mu}{\sigma}$
• Use the z-table to find $P(Z \leq z)$, then use complements and subtraction to find any probability
• For inverse problems (given a percentile, find the value), look up the z-score in the table and convert back using $x = \mu + z\sigma$
• Before applying the normal model, check that data is approximately bell-shaped with no strong skew or extreme outliers
• In healthcare, the normal distribution helps set clinical thresholds (like low birth weight at 2,500 g) and communicate how common or rare a measurement is

Return to Statistics for more topics in this section.