Welcome to the Normal Distribution!

In this chapter, we are going to explore what many statisticians call the "King of Distributions." The Normal Distribution is incredibly important because it crops up everywhere in real life—from the heights of people and the weights of apples to exam scores and even the errors made by scientific instruments. By the end of these notes, you’ll understand how to describe this "bell-shaped curve," how to calculate probabilities with your calculator, and how to use it to simplify complex problems.

1. What is the Normal Distribution?

The Normal Distribution is a continuous probability distribution. Unlike discrete distributions (where things are counted in whole numbers), continuous distributions deal with measurements that can take any value, like 1.5, 1.55, or 1.5555...

Key Properties of the Bell Curve

  • Symmetrical: If you folded the curve in half at the center, both sides would match perfectly.
  • The Mean, Median, and Mode are equal: They all sit right in the middle at the highest point of the curve.
  • Total Area = 1: The total area under the curve represents the total probability, which is always 1 (or 100%).
  • Asymptotic: The "tails" of the curve get closer and closer to the horizontal axis but never actually touch it.

Notation: We write a normal distribution as \(X \sim N(\mu, \sigma^2)\).
\(\mu\) (mu): The population mean (the center).
\(\sigma^2\) (sigma squared): The population variance (how spread out it is).

Quick Tip: Be careful! In the notation \(N(\mu, \sigma^2)\), the second number is the variance. When you use your calculator, it will usually ask for \(\sigma\) (the standard deviation). Always remember that \(\sigma = \sqrt{\text{variance}}\).

Key Takeaway

The Normal Distribution is defined by two things: its center (mean) and its spread (standard deviation). If you know these, you know the whole distribution!

2. The "Empirical Rule" (2/3, 95%, 99.8%)

Even without a calculator, we know roughly where the data sits in a Normal Distribution. The syllabus requires you to know these specific approximations:

  • Within 1 standard deviation (\(\mu \pm \sigma\)): Approximately \(2/3\) (about 68%) of all observations lie here.
  • Within 2 standard deviations (\(\mu \pm 2\sigma\)): Approximately 95% of all observations lie here.
  • Within 3 standard deviations (\(\mu \pm 3\sigma\)): Approximately 99.8% of all observations lie here.

Real-World Analogy: Imagine the height of adult men. If the mean is 175cm and the standard deviation is 7cm, then 95% of men will be between 161cm (\(175 - 2 \times 7\)) and 189cm (\(175 + 2 \times 7\)). It’s very rare to find someone outside that 3-standard-deviation range!

3. Finding Probabilities and Parameters

In your exam, you are encouraged to use your calculator functions rather than old-fashioned tables. You generally need two main functions:

Normal Cumulative Distribution (Normal CD)

Use this when you know the values (like height or weight) and you want to find the probability (the area under the curve).

Inverse Normal

Use this when you know the probability (e.g., "the top 10%") and you want to find the corresponding value on the x-axis.

Did you know? The "Standard Normal Distribution" is a special version where the mean \(\mu = 0\) and the standard deviation \(\sigma = 1\). We call this the \(Z\)-distribution. We use the formula \(Z = \frac{x - \mu}{\sigma}\) to turn any normal value into a \(Z\)-score. This tells you how many standard deviations away from the mean a value is.

Common Mistake to Avoid

When finding \(P(X < 5)\) or \(P(X \le 5)\) in a Normal Distribution, they are exactly the same! Because it is a continuous distribution, the probability of being exactly 5.000000... is zero. Don't let the "equal to" sign confuse you like it does in the Binomial distribution.

4. The Distribution of the Sample Mean (\(\bar{X}\))

This is a slightly more advanced concept, but it's very logical. If we take a sample of \(n\) items and calculate their average (\(\bar{X}\)), that average will also follow a normal distribution, but it will be "skinnier" (less spread out) than the original population.

If \(X \sim N(\mu, \sigma^2)\), then the sample mean is distributed as:
\(\bar{X} \sim N\left(\mu, \frac{\sigma^2}{n}\right)\)

Why does this happen? Think about it: an individual person might be very tall or very short. But if you take a group of 100 people and find their average height, it’s much more likely to be close to the population mean. Extreme values cancel each other out!

5. Using the Normal to Approximate the Binomial

Sometimes, working with the Binomial distribution (counting successes) is too difficult if the number of trials (\(n\)) is very large. In certain cases, we can "cheat" and use the Normal distribution instead.

When can we do this? (Conditions)

You can use the Normal approximation if:

  • \(n\) is large (usually \(n \ge 20\))
  • \(p\) is close to 0.5 (symmetrical)
  • OR more specifically: \(np > 10\) and \(n(1-p) > 10\).

The Continuity Correction

Because we are moving from a discrete distribution (Binomial: 1, 2, 3...) to a continuous one (Normal: 1.1, 1.2...), we have to adjust our values by 0.5. This is called the continuity correction.

How to do it: Imagine each whole number is actually a "block" that goes 0.5 below and 0.5 above the number.

  • To include 10: \(P(X \ge 10)\) becomes \(P(Y > 9.5)\).
  • To exclude 10: \(P(X > 10)\) becomes \(P(Y > 10.5)\).
  • To include up to 10: \(P(X \le 10)\) becomes \(P(Y < 10.5)\).

Don't worry if this seems tricky at first! Just draw a quick number line. If you want to include the number 10, you need to start your area at 9.5 to make sure the "block" for 10 is fully covered.

Key Takeaway

When approximating the Binomial with the Normal, use \(\mu = np\) and \(\sigma^2 = np(1-p)\), and always apply the +/- 0.5 continuity correction!

Summary Checklist

  • Can you list the properties of a Bell Curve?
  • Do you know the 2/3, 95%, and 99.8% rules?
  • Can you find probabilities using the Normal CD function on your calculator?
  • Do you remember to divide the variance by \(n\) when dealing with sample means (\(\bar{X}\))?
  • Can you apply the continuity correction when approximating a Binomial distribution?