Welcome to the Normal World!

In this chapter, we are going to explore the "Queen of Distributions": the Normal Distribution. This bell-shaped curve is everywhere—from the heights of people to the weights of cereal boxes. Later, for our HL students, we will look at how we can use small samples to make very clever guesses about entire populations using estimators and confidence intervals.

Whether you are at SL or HL, the most important tool for this chapter is your Graphic Display Calculator (GDC). We won't be doing the heavy lifting by hand; we'll be teaching the GDC to do it for us!

1. The Normal Distribution (SL & HL)

The normal distribution is a continuous probability distribution. When we graph it, it forms a perfectly symmetrical bell shape.

Key Properties of the Curve

  • Symmetry: The curve is perfectly symmetrical about the center.
  • The Center: The mean (\( \mu \)), median, and mode are all equal and located at the exact center.
  • Total Area: The total area under the curve is exactly \( 1 \) (representing a total probability of 100%).
  • Asymptotic: The "tails" of the curve get closer and closer to the horizontal axis but never actually touch it.

Notation

We write that a random variable \( X \) follows a normal distribution as:
\( X \sim N(\mu, \sigma^2) \)

  • \( \mu \) (mu) is the mean (where the peak is).
  • \( \sigma^2 \) (sigma squared) is the variance. Note that \( \sigma \) is the standard deviation (how wide the bell is).

Using Your GDC for Probabilities

In your exam, you will use the NormCD (Normal Cumulative Distribution) function to find the probability between two values.

  • Lower Bound: The starting value. If there is no lower bound, use a very small number like \( -9 \times 10^{99} \).
  • Upper Bound: The ending value. If there is no upper bound, use a very large number like \( 9 \times 10^{99} \).
  • Input: You will plug in your \( \mu \) and \( \sigma \).

The Inverse Normal Calculation

Sometimes you know the probability (the area) but you don't know the \( x \)-value. For example: "What is the minimum height required to be in the tallest 10% of the population?"

Use the Inverse Normal function on your GDC. You usually provide the Area (always measured from the left-hand side tail in most calculators), \( \mu \), and \( \sigma \).

Quick Tip: If a question asks for the "top 10%", remember that the "left area" is \( 0.90 \) (100% - 10%). Always draw a quick sketch of the bell curve to visualize which side you are calculating!

2. Unbiased Estimators (HL Only)

In the real world, we rarely know the true mean (\( \mu \)) or variance (\( \sigma^2 \)) of a whole population. Instead, we take a sample and use it to estimate these values.

The Sample Mean

The best estimate for the population mean \( \mu \) is the sample mean \( \bar{x} \). It is an unbiased estimator, meaning that on average, it hits the target perfectly.

The Sample Variance

This is a bit tricky! If you just calculate the variance of a sample using the standard formula (dividing by \( n \)), you will usually underestimate the true population variance. To fix this, we use the unbiased estimate of the population variance (\( s_{n-1}^2 \)):

\( s_{n-1}^2 = \frac{n}{n-1} \times (\text{sample variance}) \)

GDC Tip: When you look at "1-Var Stats" on your calculator, look for \( s_x \). That is the unbiased standard deviation! The symbol \( \sigma_x \) is usually the sample standard deviation (which we rarely use as an estimator).

3. The Central Limit Theorem & Sample Means (HL Only)

If we take many samples of size \( n \) from a population and calculate the mean of each sample, those sample means form their own distribution!

The Sampling Distribution

If the original population is \( X \sim N(\mu, \sigma^2) \), then the distribution of the sample mean \( \bar{X} \) is:
\( \bar{X} \sim N(\mu, \frac{\sigma^2}{n}) \)

Notice that the mean stays the same, but the variance gets smaller (we divide by \( n \)). This makes sense: the average of 100 people's heights is much more likely to be close to the "true" average than one single person's height.

The Central Limit Theorem (CLT)

This is a bit of mathematical magic. The CLT states that if your sample size \( n \) is large enough (usually \( n > 30 \)), the distribution of the sample mean will be approximately normal, even if the original population was not normal at all!

Did you know? This is why the Normal Distribution is so important. Even if we are studying something weird and non-normal, the averages we calculate will behave normally.

4. Linear Combinations of Normal Variables (HL Only)

Sometimes we need to add or subtract normal variables. For example, if a lift carries 4 people, what is the distribution of their total weight?

If \( X \) and \( Y \) are independent normal variables:

  • Means: \( E(aX + bY) = aE(X) + bE(Y) \)
  • Variances: \( Var(aX \pm bY) = a^2Var(X) + b^2Var(Y) \)

Important Warning: Even if you are subtracting variables (\( X - Y \)), you add the variances! Variation always increases when you combine independent variables; it never cancels out.

5. Confidence Intervals (HL Only)

A Confidence Interval (CI) is a range of values that we are reasonably sure contains the true population mean \( \mu \).

What does "95% Confidence" mean?

It does not mean there is a 95% chance the mean is in this specific interval. It means that if we repeated our sampling process 100 times, we would expect 95 of the resulting intervals to contain the true population mean.

Calculating the Interval

While there are formulas involving \( z \)-scores, the AI curriculum expects you to use your GDC:

  1. Go to the Statistics menu.
  2. Select Intervals (or INTR).
  3. Select Z-Interval (if you know the population standard deviation) or T-Interval (if you only have the sample data). Note: In many AI contexts, if \( n \) is large or the population is normal, we focus on the Z-interval approach.
  4. Input your Confidence Level (e.g., \( 0.95 \)), the mean, the standard deviation, and the sample size \( n \).

Key Takeaway:

  • Higher Confidence (e.g., 99%): Results in a wider interval (to be more sure, you need a bigger net).
  • Larger Sample Size (\( n \)): Results in a narrower, more precise interval.

Common Mistakes to Avoid

  • Standard Deviation vs. Variance: Double-check if the question gives you \( \sigma \) or \( \sigma^2 \). Your GDC usually asks for \( \sigma \) (standard deviation).
  • Left vs. Right Tail: In Inverse Normal questions, always check if the area is "less than" or "greater than."
  • Dividing by \( n \) in the CLT: When dealing with the mean of a sample, remember the new standard deviation is \( \frac{\sigma}{\sqrt{n}} \).
  • Adding Variances: When finding the variance of \( X - Y \), remember to add the variances, not subtract them!

Don't worry if the HL concepts like CLT and Estimators feel a bit abstract at first. Just remember: the SL content is about finding areas under the bell curve, and the HL content is about using that bell curve to make predictions about averages!