This page is dedicated to a special type of continuous random variable, those having a particular probability density function called the normal distribution. Recall that given a variable X we use the notation \mu=\mathbb{E}(X) (expectation of X), \sigma(X) (standard deviation), and Var(X) (variance) is sometimes represented as \sigma^2(X).

Defining the Normal Distribution

Definition (Normal Distribution): A normally-distributed random variable is given by two parameters (two numerical descriptive measures), the mean \mu and the standard deviation \sigma. In this case, we say X\sim N(\mu,\sigma^2) and the PDF is given by:

    \[f_X(x)=\frac{1}{\sigma\sqrt{2\pi}}e^{-\frac{1}{2}\cdot\left(\frac{x-\mu}{\sigma}\right)^2}\]

This function has the following graph, called a Bell curve:

Rendered by QuickLaTeX.com

The probability density function is an adaptation of the function e^{x^2}. This function does not have an elementary primitive. That is, we cannot express its integral just as some simple function like we have done for polynomials or trigonometric functions.

Therefore, since it is too complicated at this level to integrate, there are two options.

  • We are allowed a technological tool to calculate probabilities.
  • We are asked for very specific probabilities which we know from the graph above.

When we say X\sim N(\mu,\sigma^2) the first number represents the expectation and the second the variance. E.g.: X\sim N(0,4) we should understand that the expectation is 0 and the standard deviation is 2 (because the variance is 4). That is: \mu=\mathbb{E}(X)=0 and \sigma(X)=2. Some sources might use the notation X\sim N(\mu,\sigma), in which case the standard deviation would be 4, so make sure you know beforehand which notation is used in your case.

Why the Normal Distribution is so Special

The Normal Distribution is uniquely important in statistics, physics, data science, and nature primarily because of a special theorem called the Central Limit Theorem. This theorem states that if you take a large number of independent random variables and add them together, their normalised sum tends towards a normal distribution, regardless of the original distribution of the original variables.

In other words, taking enough of independent variables, they begin to behave like a normal distribution. Take for example an experiment you likely have seen in previous years: Roll a dice n times and add the results. Below are the distributions for n=1,n=2, and n=3.

1 Die Roll

Rendered by QuickLaTeX.com

2 Dice Rolls (Sum)

Rendered by QuickLaTeX.com

3 Dice Rolls (Sum)

Rendered by QuickLaTeX.com

As you can see, the more dice we roll, the more the shape of the distribution resembles that of the normal distribution.

Another reason why normal variables are so important is that all the information about X is given by \mu and \sigma. Once you know these, you can draw its distribution easily.

Another interesting property of the normal distribution comes from its density function. That function is symmetric with respect to its mean. Therefore, if \mu represents the mean of a normally-distributed random variable X, then:

    \[\mathbb{P}(X<\mu)=\frac{1}{2}\]


and the same for \mathbb{P}(X>\mu). That is, the mean marks what we understand as the “centre”.

Calculations with Normal Distribution

Let us see a few examples with and without calculator.

Example: A marine biology study monitors the migration weights of a specific salmon population. The data shows that their weights are normally distributed with a mean (\mu) of 12\text{ kg} and a standard deviation (\sigma) of 3\text{ kg}. Draw the distribution graph for this random variable, marking the values at 1, 2, and 3 standard deviations away from the mean alongside their precise interval percentages. Then, use the graph to calculate the probability P(6 \le X \le 15).

We could calculate this probability because it was exactly within so many standard deviations from our mean. For any other value that is not those special values, we would need a calculator. What we calculate above is that the proportion of salmon weighing between 6kg and 15kg is 81.85%.

Example: The daily charge capacity of a solar battery array at a clean energy research farm is normally distributed with a mean (\mu) of 450\text{ kWh} and a standard deviation (\sigma) of 40\text{ kWh}. Use a calculator to evaluate the following probabilities directly:

  • 1. The probability that the array stores less than 415\text{ kWh} on a randomly selected day.
  • 2. The probability that the array stores between 430\text{ kWh} and 518\text{ kWh}.

The second the exercise gives us the mean and standard deviation, every probability can be easily found using a calculator. You may notice that at times we use \mathbb{P}(X<x) and at times \mathbb{P}(X\leq x). Remember that for a continuous random variable, we have \mathbb{P}(X=x)=0 (the likelihood of being an exact value is always 0) so that these two probabilities are always equal.

Understanding the Graph of the Distribution

A useful exercise to understand probabilities without a calculator is to draw the graph of the normal distribution given the mean and standard deviation.

Example: The shelf life (in days) of a particular organic dairy product is normally distributed with a mean (\mu) of 15 days and a standard deviation (\sigma) of 2 days.

  • Complete the bell curve for this distribution using the given standard deviation benchmarks.
  • What proportion of products last more than 17 days?
  • What proportion of products last less than 11 days?
  • What proportion of products last between 13 and 19 days?

Normal Quantiles

By now we have found probabilities such as: \mathbb{P}(X<k) and \mathbb{P}(X>k). What happens when we know the probability but we want to understand the value of k? That is the point of normal quantiles.

We know, from the symmetry of the density function, that \mathbb{P}(X<\mu)=0.5. Therefore, if we are told that: \mathbb{P}(X<20)=0.5 we can conclude that 20 must be the expectation of X. Similarly, we can understand other values (with or without a calculator) based on our knowledge of what the probability is.

Example: Let X be a normally distributed random variable with a mean (\mu) of 15 and an unknown standard deviation (\sigma). Given that \mathbb{P}(X < 17) = 0.8413, determine the value of the standard deviation \sigma.

Your calculator should be able to find normal quantiles when they do not match a particular benchmark.

Example: The length of newborn sea turtles in a wildlife sanctuary is normally distributed with a mean (\mu) of 62\text{ mm} and a standard deviation (\sigma) of 4\text{ mm}. Biologists want to identify the threshold length for the smallest 10\% of the population to give them extra nutritional monitoring. Find the 10\text{th} percentile length x such that \mathbb{P}(X \le x) = 0.10.

We could have guessed, based on the answer, that x should be below 58 (at least one standard deviation away) but more than 54 because otherwise the probability that is left on our curve would be much less than 10%.

Exercises

The operational lifespan of a specific model of marathon running shoes is normally distributed with a mean (\mu) of 700\text{ km} and a standard deviation (\sigma) of 50\text{ km}. Use the empirical rule benchmarks to determine the following proportions without a calculator:

  • 1. Find the probability that a pair of shoes lasts more than 750\text{ km}: \mathbb{P}(X > 750).
  • 2. Find the probability that a pair of shoes lasts less than 600\text{ km}: \mathbb{P}(X < 600).
  • 3. Find the probability that a pair of shoes lasts between 650\text{ km} and 800\text{ km}: \mathbb{P}(650 \le X \le 800).
  • 4. Find the probability that a pair of shoes lasts between 700\text{ km} and 750\text{ km}: \mathbb{P}(700 \le X \le 750).

The active battery life of a new generation smartwatch on a single charge under typical usage is normally distributed with a mean (\mu) of 48\text{ hours} and a standard deviation (\sigma) of 4\text{ hours}. Apply the precise normal distribution percentages to evaluate these metrics mentally:

  • 1. What proportion of smartwatches will run out of charge in under 44\text{ hours}? Evaluate \mathbb{P}(X < 44).
  • 2. What proportion of smartwatches will maintain a charge for over 60\text{ hours}? Evaluate \mathbb{P}(X > 60).
  • 3. What proportion of smartwatches will last between 40\text{ hours} and 52\text{ hours}? Evaluate \mathbb{P}(40 \le X \le 52).
  • 4. What proportion of smartwatches will last between 44\text{ hours} and 56\text{ hours}? Evaluate \mathbb{P}(44 \le X \le 56).

The maximum daily flight radius of worker honeybees from an apiary during prime foraging season is normally distributed with a mean (\mu) of 2.5\text{ km} and a standard deviation (\sigma) of 0.5\text{ km}. Use the empirical rule to deduce the following outcomes:

  • 1. Find the probability that a worker bee stays within 2.0\text{ km} of the apiary: \mathbb{P}(X < 2.0).
  • 2. Find the probability that a worker bee journeys further than 3.5\text{ km}: \mathbb{P}(X > 3.5).
  • 3. Find the probability that a worker bee travels between 1.0\text{ km} and 4.0\text{ km}: \mathbb{P}(1.0 \le X \le 4.0).
  • 4. Find the probability that a worker bee travels between 2.0\text{ km} and 3.0\text{ km}: \mathbb{P}(2.0 \le X \le 3.0).

For each scenario below, the boundaries do not fall on exact standard deviation benchmarks. Use the calculator to evaluate the probabilities directly.

  • 1. The drying time of a specific industrial eco-paint is normally distributed with a mean (\mu) of 110\text{ minutes} and a standard deviation (\sigma) of 12\text{ minutes}. Find the probability that a randomly selected wall patch takes between 95 and 122\text{ minutes} to dry completely: \mathbb{P}(95 \le X \le 122).
  • 2. The daily water consumption of a hydroponic vertical farm house is normally distributed with a mean (\mu) of 320\text{ liters} and a standard deviation (\sigma) of 25\text{ liters}. Find the probability that the farm house consumes more than 358\text{ liters} on a randomly chosen day: \mathbb{P}(X > 358).

A manufacturing process produces mechanical components whose individual weights are normally distributed with a mean (\mu) of 80\text{ grams} and a standard deviation (\sigma) of 5\text{ grams}. Draw the distribution graph for this random variable, marking the values at 1, 2, and 3 standard deviations away from the center line alongside their precise interval percentages.

The operational runtime of an commercial LED light bulb model before burning out is normally distributed with a mean (\mu) of 1200\text{ hours} and a standard deviation (\sigma) of 150\text{ hours}. Construct the symmetrical bell curve for this distribution system, clearly labeling the horizontal axis axis increments for \mu \pm \sigma, \mu \pm 2\sigma, and \mu \pm 3\sigma.

For each scenario below, use the given parameters to compute the specified interval probabilities and find the requested normal quantile threshold value using a statistical calculator or distribution table.

  • 1. The score on a national professional licensing exam is normally distributed with a mean (\mu) of 75 points and a standard deviation (\sigma) of 8 points.
        a. Find the probability that a randomly chosen candidate scores between 65 and 85 points: \mathbb{P}(65 \le X \le 85).
        b. Find the minimum passing score required to be in the top 15\% of candidates (the 85\text{th} percentile threshold value x such that \mathbb{P}(X \le x) = 0.85).
  • 2. The volume of juice bottled by an automated manufacturing machine is normally distributed with a mean (\mu) of 355\text{ mL} and a standard deviation (\sigma) of 5\text{ mL}.
        a. Find the probability that a randomly selected bottle contains less than 348\text{ mL}: \mathbb{P}(X < 348).
        b. Determine the maximum volume threshold for the bottom 5\% under-filled bottles (the 5\text{th} percentile threshold value x such that \mathbb{P}(X \le x) = 0.05).

Let X be a normally distributed random variable modeling a manufacturing or natural scenario. Use your knowledge of the empirical rule benchmark percentages to deduce the standard deviation \sigma for the following problem:

  • A normally distributed random variable X has an expectation (\mu) of 30. Knowing that the probability of X falling between 25 and 35 is exactly 68.26\% (\mathbb{P}(25 < X < 35) = 0.6826), determine what the standard deviation \sigma must be.
error: Content is protected!