Judging a Political Claim
Campaigns use numbers to persuade. Official statistics can help check those claims. Examine the source, baseline, denominator and time window, and ask what assumptions sit behind a costed promise.
What a learner can do afterwards
- Take a political statistic and identify its baseline, denominator and time window
- Explain how the same policy can be reported as a cut and as an increase
- Say what you would need to know before believing a costed promise
1 · Read
You have compared policies and their costs. A political claim needs an evidence check too. A statistic can support a claim when its source and meaning fit the question. If bus use rises after fares are removed, that shows changed use. It does not establish that free fares caused the rise. Changes to routes or working hours could also matter.
Find the baseline, the comparison value, and the dates covered. A rate's denominator is the group or total at the bottom of the fraction. For unemployment, divide unemployed people by the labour force. Percent means out of 100. Multiply the ratio by 100 to find the number before the % sign. Here, 12 ÷ 200 = 0.06 = 6%, since 0.06 × 100 = 6. Later, 10 ÷ 125 = 0.08 = 8%, since 0.08 × 100 = 8. The count fell, but the rate rose. Check that the definitions match.
The rates are percentages. A smaller unemployed count can accompany a higher rate when the labour force also changes.
A fictional service receives £40 million in Year 1, £55 million in Year 2 and £50 million in Year 3. Year 3 is £10 million above Year 1, but £5 million below Year 2. It is an increase against one baseline and a cut against the other. Both comparisons need their baseline stated. These cash amounts do not by themselves show what the money buys after prices change.
For a poll, identify whom it aims to describe and how people were recruited. Asking passers-by at one shopping centre is convenience sampling; it may miss other groups. More interviews there do not remove that selection problem. A poll of eligible voters can describe that group. To forecast an election, researchers estimate who is likely to vote. They cannot know future turnout for certain.
Trace a campaign figure to its original release or dataset. Check the dates, population, units and definitions. Official statistics agencies publish methods; other sources need checking too. Two unemployment figures may use different measures. Read the notes before comparing them. Agencies can revise figures or methods openly. A change does not, by itself, show deception.
For a costed promise, multiply the number served by the cost per person, then check for omitted expenses. Ask whether the total is annual or one-off, whether the inputs are realistic, how prices may change and where funding comes from. For example, 300 places at £400 each cost £120,000 before extra expenses. Correct multiplication does not prove that the estimate or funding is sound.
Check the source, baseline, denominator and dates. For a promise, check costs and funding. Separate what a number shows from what a speaker says caused it.
Lesson, questions and flashcards written for this dataset. The numerical budgets and training-place estimates are fictional teaching examples.
These four readings were written for this dataset about evidence, polling, budgets and statistical releases. Supplied textbook records were studied for facts and design, without reproducing their prose.
More to read
What was measured, and what explains it?
Split a political claim into an observation and an explanation. They may need different evidence.
Imagine a town records more bus journeys after removing fares. The records support a claim about increased journeys. To judge whether free fares caused that increase, investigate other changes, such as new routes or changed working hours. The order of events alone does not settle the cause.
Check what the measure counts. Journeys and individual passengers are different. One person can make several journeys, so a journey total cannot tell you how many different residents benefited.
Write a conclusion that matches the evidence. You might accept that recorded journeys increased while leaving the cause open. Explain what further information would help. These examples are fictional.
Who does the poll describe?
A percentage needs a population. Residents, eligible voters and likely voters are different groups. First identify which group the poll aims to describe.
Then check recruitment. Interviews outside one shopping centre depend on who visits that place at that time. Increasing the number interviewed there does not remove that selection problem. Ask how other groups were reached and how nonresponse was handled.
Read the question too. Supporting a proposal and supporting it if taxes rise are different questions. A headline should describe the question people actually answered.
Election forecasts add uncertainty about turnout. Researchers estimate likely voters; they cannot know the final voting population in advance. A poll of eligible voters can still be useful, but it describes a different group from the people who eventually cast ballots.
Check a budget headline and a promise
Suppose fictional spending moves from £70 million to £82 million and then to £76 million. The last amount is above the first but below the second. A headline calling it a cut or an increase needs to name its baseline.
Check whether the figures are cash totals or spending per person. Population changes can affect that comparison. Price changes also affect what a cash budget can buy.
For a promise, reconstruct the calculation. Identify the number of places or people served and the assumed cost of each. Then ask what that price includes. Buildings, staffing or other expenses might be counted separately. Distinguish annual expenses from one-off costs.
A correct total does not establish that the assumptions are realistic. Nor does it establish that a proposed tax, saving or borrowing plan will provide the money. Check the cost estimate and the funding claim separately.
Trace a figure to its release
Follow the source named by a campaign to the release or dataset behind its figure. If no source is named, record that limitation.
Check the dates, population, unit and definition. Two unemployment figures for the same month may describe different groups or measures. Read the notes before comparing them.
In the United States, the Bureau of Labor Statistics is part of the Department of Labor. It publishes labour-market and price statistics. It is one agency within the wider federal statistical system, not the leader of that whole system.
Revisions need checking too. An agency may receive more information or change a method and explain the change publicly. A changed figure does not by itself establish dishonesty. Ask whether the campaign quoted the release accurately and whether its interpretation follows from the measure.
2 · Watch
Take it off screen
Where it sits
Learn first
Optional connections
This opens up
Nothing builds on it yet.
Where this leads
Jobs that lean on this skill. Follow one to see everything it is built on.
8 questions wait behind this lesson, each with its answer explained. Every answer feeds the sky: stars light as they are learned, and dim when it is time to come back.