← Map

Survey questions and fair samples

Method: Check the question and who is asked

A good survey question matches your goal, is fair, and lets people give a real answer. A leading question pushes people towards one answer. A yes/no question cannot tell you which choice is the favourite. The people you ask matter too: friends, or a group gathered in one place, often think alike, and a very small group can easily be unusual. Before believing a claim about everyone, ask who was surveyed.

Explore this idea with maths tools

Maths you can move

Build 8 + 5

Tap two gold counters to fill the ten-frame.

8 in the frame + 5 waiting. The total stays the same.

Words to know

survey
asking people questions to collect data
sample
the people who were actually asked
bias
something that pushes the results unfairly one way

Worked examples

Goal: the class's favourite sport. A: Do you like sport? B: What is your favourite sport? C: Don't you agree soccer is best?

  1. 1.A is yes/no and C pushes an answer.
  2. 2.B lets everyone name a favourite.

= B

How should you find how children at the school get there? A: ask children at the bus stop B: ask 5 children from every class C: ask 2 friends

  1. 1.A all catch the bus and C is too few friends.
  2. 2.B includes every class.

= B

9 of 12 chess club members chose chess, so 'most children love chess'. Problem? A: nothing B: only chess club members were asked C: 9 is not most

  1. 1.9 of 12 really is most of that group.
  2. 2.The problem is who was asked.

= B

For parents

Method: Check the question and who is asked

What to say: Does this question push people to one answer? Are these people like everyone else?

Common mistakes:

  • Chooses a question that pushes people towards one answer.
  • Chooses a yes/no question that cannot answer a 'which one?' goal.
  • Chooses a question about a different topic.
  • Surveys only friends, who tend to be alike.
  • Surveys a group chosen by where they are, so they share a habit or interest.
  • Accepts a sample that is far too small.
  • Accepts a claim about everyone from a biased sample.
  • Blames the arithmetic instead of the sample.
  • Focuses on how results are recorded or shown, not on the sample.