Camborne · UK
Cornwall; south-west England and close to the coast.
9MA0 study path
Large Data Set
Edexcel 9MA0
Paper 3
Pearson's weather data support Statistics assessment in Paper 3. Questions may assume you already know its terminology and contexts, or give you summaries, diagrams and samples based on it. You do not need to memorise individual readings or take the spreadsheet into the exam.
The data use daily observations for May–October in 1987 and 2015 across five UK and three overseas locations. That date window matters: it cannot automatically represent a whole year.
Locations and context
Heathrow · UK
London airport; inland relative to the coastal stations.
Hurn · UK
Near Bournemouth, a little inland of the south coast of England.
Leeming · UK
North Yorkshire in northern England.
Leuchars · UK
Fife in eastern Scotland; the most northerly UK station in the set.
Beijing · China
Northern-hemisphere, inland overseas location.
Jacksonville · USA
Florida; lower latitude and coastal.
Perth · Australia
Southern hemisphere, so May–October runs from autumn into spring.
Variable glossary
| Variable | Unit | Type | How it is used |
|---|---|---|---|
| Daily Mean Air Temperature | Unit°C | Typecontinuous quantitative | How it is usedCompare centre and spread; recognise seasonal and location effects. |
| Daily Total Rainfall | Unitmm | Typecontinuous quantitative | How it is usedTreat zero, trace rainfall and missing values differently; expect skew. |
| Daily Total Sunshine | Unithours | Typecontinuous quantitative | How it is usedCompare distributions or investigate association with cloud cover. |
| Daily Mean Windspeed | Unitknots | Typecontinuous quantitative | How it is usedCalculate summaries and compare stations or months. |
| Daily Mean Pressure | UnithPa | Typecontinuous quantitative | How it is usedUse summaries, diagrams and correlation in a weather context. |
| Daily Maximum Relative Humidity | Unit% | Typecontinuous quantitative | How it is usedInterpret a percentage measurement without treating it as a probability. |
| Daily Mean Visibility | Unitdecametres | Typecontinuous quantitative | How it is usedCompare distributions; watch the unit, which is not metres. |
| Daily Mean Total Cloud | Unitoktas (0–8) | Typediscrete quantitative | How it is usedPublished as whole oktas, each okta being an eighth of the sky covered; zero and eight have contextual meanings. |
| Daily Mean Wind Direction | Unitcardinal direction | Typecategorical | How it is usedRecorded as a compass direction, so use a categorical display rather than a mean. |
| Daily Maximum Gust | Unitknots | Typecontinuous quantitative | How it is usedA distinct variable from mean windspeed — a mark scheme may credit one and not the other. |
| Daily Mean Windspeed (Beaufort conversion) | UnitBeaufort scale | Typeordered categorical (Edexcel calls it qualitative) | How it is usedThis is the mean windspeed converted to a scale, not a separate measurement; do not treat it as a value in knots. |
| Date | Unitcalendar date | Typetemporal | How it is usedBuild samples by day, month or year and identify seasonal limitations. |
Missing and special values
n/a
Data not available for that day; it is missing data, not zero.
Keep it missing or omit it from the calculation, then state the valid sample size. Never silently enter 0.
tr
A trace of rainfall: less than 0.05 mm.
Pearson prescribes no conversion, so state your own convention before calculating and stay consistent. Any value from 0 to 0.05 mm is defensible, and 0 is accepted — so quote the convention alongside the answer, because a different convention gives a different mean.
0
A recorded zero: the measured quantity was zero at the stated precision.
Retain it as a genuine observation. It is different from both n/a and tr.
Sampling checklist
- Define the population and sampling frame: which locations, dates, months and years?
- For a simple random sample, number every eligible row and select unique random numbers.
- For a systematic sample, calculate the interval and choose a random start within the first interval.
- For a stratified sample, allocate in proportion to each relevant group and round consistently.
- Remove or preserve missing observations explicitly; report the valid sample size.
- Check whether May–October, the chosen stations or the two years can represent the claim being made.
LDS-style questions
Original practice
1. Rainfall values are 1.4, n/a, 0, tr and 2.1 mm. Explain how you would prepare them for a numerical mean, then state your mean and the convention it depends on.
Exclude n/a, leaving four valid observations; retain 0 as a genuine reading; and replace tr with a stated value, since Pearson prescribes none. State the convention with the answer. Taking tr = 0 gives (1.4 + 0 + 0 + 2.1) ÷ 4 = 0.875 mm; taking tr = 0.025 mm gives 0.88125 mm, which rounds to 0.88 mm at the precision of the data. Both are correct when the convention is stated; a mean quoted without it is not.
2. A station has 184 daily rows from May to October. Describe a systematic sample of 23 days.
Use interval 184 ÷ 23 = 8. Choose a random starting position from 1 to 8, then select that row and every eighth row after it, which gives exactly 23 rows for any of those starts. Choosing a random start anywhere from 1 to 184 and wrapping around at the end is equally valid.
3. A sample from Perth is warmer in October than in June. Give one contextual reason this is plausible.
Perth is in the southern hemisphere: June is winter, while October is spring.
4. A student uses the LDS to estimate a location's mean temperature for the whole of 2015. Criticise the claim.
The data cover May to October only, so the omitted months bias the estimate in a known direction — warm for the UK stations, since their winter is missing, and cool for Perth, whose summer is missing. Any of these is also a valid criticism: the sample covers one location rather than the whole country; only sampled days are used, not every date; and the figures are daily means. Credit a stated criticism with its consequence for the estimate.
5. Which LDS variable is recorded from 0 to 8, and why is a bar chart reasonable for it?
Daily Mean Total Cloud, measured in oktas. Although it is a daily mean, the data set publishes it as whole oktas from 0 to 8, so in practice it takes only nine discrete values and separated bars are appropriate. Treating the oktas as ordered categories is an equally acceptable justification.
6. A sampling frame contains 90 Heathrow rows, 60 Hurn rows and 30 Leeming rows. Allocate a stratified sample of 24 rows by location.
There are 180 rows, so allocate 24 × 90/180 = 12 Heathrow rows, 24 × 60/180 = 8 Hurn rows and 24 × 30/180 = 4 Leeming rows.