Algebraic expressions, formulae and probability experiments
Week 22 · Units 18–19: Algebraic expressions and formulae; probability experiments · 4 lessons
4
Is it fair? Testing for bias
Student’s Book 8: Units 18–19, pp. 175–183
GoalUse the results of an experiment to decide whether a dice, coin or spinner is biased, and justify your decision.
New words
biased · not fair: some outcomes are more likely than othersfair · every outcome is equally likelyexpected frequency · the number of times an outcome should happen, found by multiplying the probability by the number of trialshypothesis · an idea that you test with an experiment
Explanation
A dice is fair if each number has probability 1/6. To test it, roll it many times and write the results in a frequency table. If there were 120 rolls, each number is expected about 120 × 1/6 = 20 times. The results will not match exactly. If all the results are fairly close to 20, there is no evidence of bias. If one result is very far from 20, the dice is probably biased. Always use a large number of trials, because a small number can be misleading. A good experiment states a hypothesis, the method, the results, an analysis and a conclusion. If the experiment shows bias, you can change the object (for example, balance the spinner) and test again.
Number
1
2
3
4
5
6
Expected in 120 rolls
20
20
20
20
20
20
Worked examples
Two dice are each rolled 120 times. We expect each number about 20 times. The first dice gave 18, 21, 19, 22, 17, 23: all close to 20, so there is no sign of bias. The second dice gave 12, 15, 13, 14, 16 and 50 sixes. The six came up far more often than the expected 20, so the second dice is probably biased towards 6.
A spinner has three sectors: red takes half of the circle, and blue and green a quarter each. Theory: P(red) = 1/2, P(blue) = P(green) = 1/4. In 100 spins the results are 47, 28 and 25. The relative frequencies 0.47, 0.28 and 0.25 are close to the theory, so the spinner behaves as designed. It is not ‘fair’ in the sense of equal chances, but it is not faulty either.
Number of flips
Heads
Relative frequency
40
29
0.725
400
214
0.535
A coin gives 29 heads in 40 flips (relative frequency 0.725). That looks biased. After 400 flips there are 214 heads: 0.535. This is close to 0.5. The early result was due to chance, and with more trials the coin looks fair. The larger experiment is more reliable.
Class activity
Is my dice fair? Roll a dice 60 times and tally the results. Write the expected frequency (10 each) beside your tally. Does any number differ from 10 by a lot? Now combine your results with two other people’s, and compare the relative frequencies from 60 rolls and from 180 rolls.
Practice
1
Mo flips a coin 20 times and gets 15 heads. Ana flips the same coin 500 times and gets 270 heads. Find both relative frequencies. Whose result is more convincing? What do you conclude about the coin?
Flips
Heads
Mo
20
15
Ana
500
270
Mo: 15/20 = 0.75. Ana: 270/500 = 0.54. Ana’s result is more reliable because she did many more flips. Her 0.54 is close to 0.5, so there is no good evidence that the coin is biased; Mo’s 0.75 was probably due to chance with so few flips.
2
A dice is rolled 300 times. The results for 1 to 6 are 48, 52, 47, 51, 50 and 52. What is the expected frequency of each number? Is the dice fair?
Number
1
2
3
4
5
6
Frequency
48
52
47
51
50
52
Expected = 300 × 1/6 = 50 each. All results are within 3 of 50, so there is no evidence of bias; the dice looks fair.
3
A spinner with 5 equal sectors is spun 150 times. The results for 1 to 5 are 24, 31, 29, 32 and 34. What is expected for each? Is there convincing evidence of bias? Justify your answer.
Number
1
2
3
4
5
Frequency
24
31
29
32
34
Expected = 150 ÷ 5 = 30 each. The number 1 is 6 below expected and 5 is 4 above, but these are small differences for 150 spins. There is no convincing evidence of bias, but more spins would give a more reliable answer.
4
Plan an experiment to find out whether a bottle top lands ‘top up’ half the time. Write a hypothesis, a method and what you would do with the results.
Hypothesis: the bottle top lands top up with probability 1/2. Method: flip it 100 times from the same height onto a table and tally ‘up’ and ‘down’. Analysis: work out the relative frequency of ‘up’ and compare with 0.5. Conclusion: if it is far from 0.5 (for example 0.2 or 0.8), the top is not equally likely to land either way.