Lesson 1 of 8 · 13 min

Sample means vary

See the sample mean as a random variable whose value changes from sample to sample, tell statistics from parameters, and find the exact distribution of X̄ for a small population.

What you will be able to do

  • Distinguish a parameter (μ\mu, σ\sigma) from a statistic (xˉ\bar{x}, ss).
  • Explain why the sample mean Xˉ\bar{X} is a random variable, and the difference between Xˉ\bar{X} and xˉ\bar{x}.
  • Find the distribution of Xˉ\bar{X} for small samples from a small discrete population.

A transport researcher wants the mean one-way commute time μ\mu of every worker in Brisbane. That is far too many people to ask, so she surveys a random sample of 5 workers and calculates their mean, xˉ\bar{x}. A colleague surveys 5 different workers and gets a different xˉ\bar{x}. Neither of them is wrong. Sample means vary, and this whole topic is about how much.

Definition

Parameters and statistics

A parameter describes the whole population and is fixed: the population mean μ\mu and population standard deviation σ\sigma. Usually you don't know it.

A statistic is calculated from a sample, so it changes from sample to sample: the sample mean xˉ\bar{x} and the sample standard deviation ss. You use statistics to estimate parameters.

Interactive

Draw samples of commute times

Draw one sample at a time first and watch xˉ\bar{x} change. Then draw hundreds. Where is the histogram of xˉ\bar{x} centred, and is it as spread out as the population?

Before a sample is taken, its mean is unknown and could be any of many values, each with some probability. So the sample mean is a random variable, written Xˉ\bar{X}. Once you have a particular sample, its mean is a number, written xˉ\bar{x}.

If XX is the commute time of one randomly chosen worker, then for a sample of nn workers

Xˉ=X1+X2+⋯+Xnn\bar{X} = \frac{X_1 + X_2 + \cdots + X_n}{n}

where X1,X2,…,XnX_1, X_2, \ldots, X_n are the times of the nn workers, each with the same distribution as XX (mean μ\mu, standard deviation σ\sigma).

A random sample of 40 bags of rice has mean mass 1.012 kg. The machine is set to give a mean of 1.000 kg for all bags it fills. Which statement is correct?

Choose one answer

For a small population you can find the distribution of Xˉ\bar{X} exactly, by listing every possible sample. Roll a fair die twice, so n=2n = 2. A single roll XX has μ=3.5\mu = 3.5 and σ2=3512\sigma^2 = \frac{35}{12}. There are 36 equally likely ordered pairs of rolls, and each gives a sample mean.

The distribution of Xˉ\bar{X}, the mean of two rolls of a fair die.
Total of the two rollsxˉ\bar{x}P(Xˉ=xˉ)P(\bar{X} = \bar{x})
21136\frac{1}{36}
31.5236\frac{2}{36}
42336\frac{3}{36}
52.5436\frac{4}{36}
63536\frac{5}{36}
73.5636\frac{6}{36}
84536\frac{5}{36}
94.5436\frac{4}{36}
105336\frac{3}{36}
115.5236\frac{2}{36}
126136\frac{1}{36}

One roll is uniform on 1,…,61, \ldots, 6. The mean of two rolls already piles up in the middle: extreme means need both rolls to be extreme.

A fair die is rolled twice and Xˉ\bar{X} is the mean of the two rolls. Using the table, find P(Xˉ>4.5)P(\bar{X} > 4.5). Give an exact answer.

Worked example

The exact distribution of Xˉ\bar{X} for a small population

  • Simple familiar
  • Technology free

A large bag holds equal numbers of tokens marked 22, 44 and 66. Let XX be the number on a randomly drawn token. Two tokens are drawn at random, with replacement, and Xˉ\bar{X} is the mean of the two numbers.

(a) Find the mean μ\mu and variance σ2\sigma^2 of XX.

(b) Find the probability distribution of Xˉ\bar{X}.

(c) Find E(Xˉ)E(\bar{X}) and Var(Xˉ)\mathrm{Var}(\bar{X}), and compare them with μ\mu and σ2\sigma^2.

6 steps

Try it yourself first, then check the working one step at a time.

Method

  1. List every equally likely ordered sample and its mean.

  2. Group equal means to get the distribution of Xˉ\bar{X}.

  3. Use E(Xˉ)=∑xˉ pE(\bar{X}) = \sum \bar{x}\,p and Var(Xˉ)=E(Xˉ2)−[E(Xˉ)]2\mathrm{Var}(\bar{X}) = E(\bar{X}^2) - [E(\bar{X})]^2.

  4. Compare with μ\mu and σ2n\frac{\sigma^2}{n}.

Watch out for

  • Treating {2,4}\{2, 4\} and {2,2}\{2, 2\} as equally likely. Count ordered samples.

  • Forgetting to subtract [E(Xˉ)]2[E(\bar{X})]^2 when finding the variance.

Key idea

What the table already shows

The distribution of Xˉ\bar{X} is centred at the population mean (3.53.5), it is less spread out than one roll (Var(Xˉ)=3524\mathrm{Var}(\bar{X}) = \frac{35}{24}, exactly half of 3512\frac{35}{12}), and it is more bell-shaped than the population. The next two lessons make each of these precise.

Common mistake

x̄ is not μ

Your sample mean is an estimate of μ\mu, not μ\mu itself. Writing "μ=1.012\mu = 1.012" from a sample of 40 bags loses marks. Write "xˉ=1.012\bar{x} = 1.012, which estimates μ\mu".

Summary

  • Parameters (μ\mu, σ\sigma) describe the population and are fixed; statistics (xˉ\bar{x}, ss) come from a sample and vary.

  • The sample mean Xˉ=X1+⋯+Xnn\bar{X} = \frac{X_1 + \cdots + X_n}{n} is a random variable; xˉ\bar{x} is its value for one sample.

  • For a small population, list every equally likely sample to find the exact distribution of Xˉ\bar{X}.

  • The distribution of Xˉ\bar{X} is centred at μ\mu, narrower than the population, and more bell-shaped.

Finished the lesson?

Try the 2 quick checks above and the lesson completes itself, or mark it complete now.

Lesson 1 of 80% read

Save your progress with a free account