Lesson 1 of 7 · 12 min

From a point estimate to an interval estimate

See why one sample proportion is almost never exactly right, and why a range of plausible values for p is a more honest and more useful estimate.

What you will be able to do

  • Distinguish a parameter (pp) from a statistic (p^\hat{p}).
  • Calculate a sample proportion as a point estimate of pp.
  • Explain what an interval estimate is and why it is more useful than a point estimate alone.

A Cancer Council survey asks a random sample of 1200 Queensland adults whether they wore sunscreen the last time they spent more than an hour outdoors. 744 say yes. What proportion of all Queensland adults wore sunscreen?

The honest answer is "we don't know exactly, but probably about 62%". This topic is about making "probably about" precise.

Definition

Parameters and statistics

  • The population proportion pp is a parameter: a fixed number describing the whole population. It is usually unknown.
  • The sample proportion p^=Xn\hat{p} = \dfrac{X}{n} is a statistic: calculated from a sample of size nn, where XX is the number of successes. It changes from sample to sample.

Using p^\hat{p} as our single best guess for pp makes it a point estimate.

In the survey, 744 of the 1200 adults wore sunscreen. Calculate the point estimate p^\hat{p} of the proportion of all Queensland adults who wore sunscreen.

A different random sample of 1200 adults would give a slightly different p^\hat{p}: perhaps 0.60, perhaps 0.64. In Topic 4 you saw that p^\hat{p} is a random variable with mean pp and standard deviation p(1−p)n\sqrt{\frac{p(1-p)}{n}}, approximately normal for large samples. Draw some samples below and watch p^\hat{p} scatter around pp.

Interactive

Samples scatter around p

Draw many samples. Almost no sample gives p^\hat{p} exactly equal to p=0.62p = 0.62, but nearly all land within about 0.03 of it. Try n=100n = 100: the spread grows.

Key idea

Why a single number is not enough

A point estimate is almost certainly a little wrong, and on its own it doesn't say how wrong it might be. An interval estimate gives a range of plausible values for the parameter, such as

0.593<p<0.647,0.593 < p < 0.647,

and attaches a level of confidence to the method that produced it. The width of the range tells you how precise the estimate is.

The point estimate p^=0.62\hat{p} = 0.62 and an interval estimate (0.593,0.647)(0.593, 0.647) for pp. You will learn how to calculate this interval in lesson 3.

A news report says: "Between 31% and 37% of Australian teenagers spend more than 4 hours a day on screens." What kind of estimate is this?

Choose one answer

Worked example

Point estimate or interval estimate?

  • Foundation

In a random sample of 1500 Australian adults, 465 had a flu vaccination this year. A report based on the survey states that "a 95% confidence interval for the proportion of Australian adults vaccinated against flu this year is (0.287,0.333)(0.287, 0.333)".

(a) Calculate the point estimate of the proportion.

(b) Identify the parameter being estimated, and explain what the interval adds to the point estimate.

4 steps

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

Method

  1. p^=Xn\hat{p} = \frac{X}{n} is the point estimate.

  2. The parameter is the population proportion pp, described in context.

  3. An interval estimate shows the precision of the estimate.

Watch out for

  • Saying the interval is for the sample proportion: we know p^=0.31\hat{p} = 0.31 exactly.

Common mistake

Mixing up p and p̂

pp belongs to the population and never changes; p^\hat{p} belongs to a sample and changes every time. A sentence like "the sample proportion is between 0.593 and 0.647" is wrong: we know p^=0.62\hat{p} = 0.62 exactly. The interval is a statement about pp.

A random sample of 400 Brisbane households finds that 132 have rooftop solar panels. Which statement is correct?

Choose one answer

Summary

  • A parameter such as pp describes a population and is fixed but unknown; a statistic such as p^=Xn\hat{p} = \frac{X}{n} describes a sample and varies from sample to sample.

  • p^\hat{p} is the point estimate of pp. It is almost never exactly right.

  • An interval estimate gives a range of plausible values for pp; its width shows how precise the estimate is.

  • An interval estimate is a statement about pp, not about p^\hat{p}.

Finished the lesson?

Try the 3 quick checks above and the lesson completes itself, or mark it complete now.

Lesson 1 of 70% read

Save your progress with a free account