LC Ordinary Maths

Statistics and probability · Chapter 43

Standard deviation, normal data and margin of error

Clear notes, 7 worked examples and 3 questions to try.

01 · Understand the idea

The essentials

Key terms

Standard deviation
A measure of how spread out data are around their mean.
Margin of error
An allowance for uncertainty in a sample estimate.
Null hypothesis
The population claim being tested.

Key ideas

  • Standard deviation describes spread around the mean. Use the population standard-deviation setting when that is the quantity requested.
  • For an approximately normal distribution, about 68%\displaystyle 68 \% is within one standard deviation of the mean,95%\displaystyle , 95 \% within two, and 99.7%\displaystyle 99.7 \% within three.
  • The approximate 95%\displaystyle 95 \% margin of error for a population proportion is 1n\displaystyle \frac1{\sqrt n}. This is a proportion; multiply by 100\displaystyle 100 for percentage points.
  • An approximate interval is p^±1n\displaystyle \hat p\pm\frac1{\sqrt n}. State the population and sampling assumptions.
  • For the syllabus margin-of-error test, a hypothesised proportion outside the sample interval gives evidence against it at the approximate 5%\displaystyle 5 \% level. Inside means insufficient evidence to reject it.
  • Failure to reject a claim does not prove it is true. Random sampling and representativeness matter.
For approximately normal data, about 95% lie between mean minus two standard deviations and mean plus two standard deviations. This is an approximation, not a guarantee for every sample.

03 · Follow the working

Worked examples

Read the question first. Cover the working and decide what you would do next.

01Start here: a bigger sample gives a smaller margin

Estimate the 95%\displaystyle 95 \% margin of error for a random sample of n=100\displaystyle n=100.

Why this method? Use the syllabus approximation 1\displaystyle 1 divided by the square root of the sample size. Convert the proportion to percentage points at the end.

  1. n=100=10\displaystyle \sqrt n=\sqrt{100}=10
    Why?Take the square root before dividing.
  2. Margin of error≈110=0.10\displaystyle \text{Margin of error}\approx\frac1{10}=0.10
    Why?This is a proportion.
  3. 0.10×100=10 percentage points\displaystyle 0.10\times100=10\text{ percentage points}
    Why?For example, an estimate of 60%\displaystyle 60 \% would give about 50%\displaystyle 50 \% to 70%\displaystyle 70 \% .
02A simple standard deviation

Find the population standard deviation of 2,4,6\displaystyle 2 , 4 , 6.

Why this method? Identify the sample information first, calculate the uncertainty, then interpret the result in context.

  1. xˉ=2+4+63=4\displaystyle \bar x=\frac{2+4+6}{3}=4
    Why?Find the mean first.
  2. (2−4)2+(4−4)2+(6−4)2=8\displaystyle (2-4)^2+(4-4)^2+(6-4)^2=8
    Why?Find each distance from the mean, square it and add.
  3. σ=83≈1.63\displaystyle \sigma=\sqrt{\frac83}\approx1.63
    Why?Divide by the number of observations, then take the square root.
03Use the empirical rule

Approximately normal scores have mean 50\displaystyle 50 and standard deviation 5\displaystyle 5. Find the interval containing about 95%\displaystyle 95 \% .

Why this method? Identify the sample information first, calculate the uncertainty, then interpret the result in context.

  1. 50−2(5)=40\displaystyle 50-2(5)=40
    Why?Two standard deviations below the mean.
  2. 50+2(5)=60\displaystyle 50+2(5)=60
    Why?Two standard deviations above the mean.
  3. 40≤x≤60\displaystyle 40\leq x\leq60
    Why?About 95%\displaystyle 95 \% lie in this interval.
04A margin-of-error interval

A random sample of 400\displaystyle 400 gives a proportion 0.60\displaystyle 0.60. Estimate a95%\displaystyle a 95 \% interval.

Why this method? Identify the sample information first, calculate the uncertainty, then interpret the result in context.

  1. MOE=1400=0.05\displaystyle \mathrm{MOE}=\frac1{\sqrt{400}}=0.05
    Why?Use the number of people sampled,400\displaystyle , 400.
  2. 0.60−0.05=0.55,0.60+0.05=0.65\displaystyle 0.60-0.05=0.55,\quad0.60+0.05=0.65
    Why?Subtract and add the margin to the sample proportion.
  3. 55% to 65%\displaystyle 55\%\text{ to }65\%
    Why?Convert the proportion endpoints to percentages.
05Test a population claim

For the previous sample, test the claim that the population proportion is 0.50\displaystyle 0.50.

Why this method? Identify the sample information first, calculate the uncertainty, then interpret the result in context.

  1. 0.50<0.55\displaystyle 0.50<0.55
    Why?The claimed value lies below the sample interval 0.55\displaystyle 0.55 to 0.65\displaystyle 0.65.
  2. Reject the claim at the approximate 5% level.\displaystyle \text{Reject the claim at the approximate 5\% level.}
    Why?Under this margin-of-error method, the sample gives evidence against the claim.
06A claim inside the sample interval

A sample estimate is 55%,\displaystyle 55 \% , with an approximate 95%\displaystyle 95 \% margin of error of 5\displaystyle 5 percentage points. Test a population claim of 52%\displaystyle 52 \% .

Why this method? Build the interval first. Then check whether the claimed population percentage falls inside it.

  1. 55%−5%=50%,55%+5%=60%\displaystyle 55\%-5\%=50\%,\quad55\%+5\%=60\%
    Why?The interval runs from 50%\displaystyle 50 \% to 60%\displaystyle 60 \% .
  2. 50%≤52%≤60%\displaystyle 50\%\leq52\%\leq60\%
    Why?The claimed 52%\displaystyle 52 \% is inside the interval.
  3. Insufficient evidence to reject the claim.\displaystyle \text{Insufficient evidence to reject the claim.}
    Why?This does not prove that 52%\displaystyle 52 \% is the true population percentage.
07Test a population proportion using the OL margin of error

In a random sample of 400,54%\displaystyle 400 , 54 \% agree. Does this challenge a claim that 60%\displaystyle 60 \% agree?

Why this method?

  1. MOE=1400=0.05\displaystyle \mathrm{MOE}=\frac{1}{\sqrt{400}}=0.05
    Why?Use the Ordinary Level rule for a population proportion.
  2. From 49\displaystyle 49 per cent to 59\displaystyle 59 per cent
    Why?Add and subtract five percentage points from 54%\displaystyle 54 \% .
  3. 60%\displaystyle 60 \% is outside the interval.
    Final stepThe result gives evidence against the claim by this margin-of-error method.

Common mistake

Writing a margin of error of 0.05\displaystyle 0.05 as 0.05%\displaystyle 0.05 \% . A proportion of 0.05\displaystyle 0.05 is 5\displaystyle 5 percentage points. A sample estimate of 60%\displaystyle 60 \% with this margin gives approximately 55%\displaystyle 55 \% to 65%\displaystyle 65 \% .

Exam technique

Explain the conclusion in the context of the population claim.

Show your method, use units where appropriate and check whether the question asks for an exact or rounded answer.

04 · Try it yourself

Self-check questions

Work on paper first, then open the answer. These are practice questions, not submitted homework.

Question 1

Find the margin of error for n=100\displaystyle n = 100.

Show answer & check
0.10\displaystyle 0.10, or 10\displaystyle 10 percentage points.

Compare your result with the answer. If it differs, revisit a worked example and check each step.

Question 2

Estimate the margin of error for a random sample of n=400\displaystyle n=400.

Show answer & check
1400=0.05\displaystyle \frac1{\sqrt{400}}=0.05, or 5\displaystyle 5 percentage points.

A proportion of 0.05\displaystyle 0.05 is five percentage points.

Question 3

A sample estimate is 70%\displaystyle 70 \% with margin of error 4\displaystyle 4 percentage points. State the interval.

Show answer & check
Approximately 66%\displaystyle 66 \% to 74%\displaystyle 74 \% .

Subtract and add four percentage points.

Classroom resources

Open slides & animated PowerPoint →
Open full-size image ↗
Enlarged maths diagram