BI08-03 Biology Watch
Reading disease data: sampling, correlation and cause
In this lesson
Reading disease data for GCSE Biology, with worked examples and the mistakes examiners report. AQA 8461, Edexcel 1BI0 and OCR J247 GCSE. Health and disease 3 of 4. More at https://scholafly.com
By the end: Read disease data from frequency tables, bar charts, histograms and scatter diagrams, judge from how the sample was taken whether the data supports a conclusion, and tell a correlation apart from a proven cause.
What it covers
- 0:00 Intro: reading disease data
- 0:31 Four forms of data
- 2:24 Comparing rates
- 3:25 Reading the scale
- 4:49 How many [times higher] is E's rate
- 5:27 The scale trap
- 5:57 Reading disease data: correlation
- 7:29 A student's claim
- 8:19 People on a heart medicine have more
- 8:49 Causal mechanism
- 9:36 Grey hair and the heart
- 10:27 The lighter survey: what is the [third
- 11:30 Which sample?
- 12:06 Judging a headline
- 13:17 Closing a data answer
Key words
About this video
Reading disease data: sampling, correlation and cause | GCSE Biology | BI08-03
Reading disease data for GCSE Biology, with worked examples and the mistakes examiners report.
AQA 8461, Edexcel 1BI0 and OCR J247 GCSE. Health and disease 3 of 4.
More at https://scholafly.com
Watch first: BI08-02 - https://youtu.be/fuqCn7ykhJU
Video code: BI08-03 (search "ScholaFly BI08-03")
#GCSEBiology #HealthAndDisease #ReadingDiseaseData
For teachers
This GCSE Biology lesson teaches reading disease data: sampling, correlation and cause. By the end, students should be able to read disease data from frequency tables, bar charts, histograms and scatter diagrams, judge from how the sample was taken whether the data supports a conclusion, and tell a correlation apart from a proven cause. It works through four worked examples and the mistakes examiners report, and suits Foundation and Higher tier students.
Exam board specification references:
AQA 8461
- 4.2.2.5
- 4.2.2.6
Edexcel 1BI0
- 5.23
- 5.24
- 5.25
OCR J247
- B6.3t
- BM6.3i
- BM6.3ii
- BM6.3iii
- BM6.3iv
Read the transcript
Imagine a survey where the people who carry a lighter turn out to be far more likely to get lung cancer. Would banning lighters bring lung cancer down? The numbers are honest, and nobody miscounted. Yet that conclusion feels wrong, and by the end you will be able to say why. That starts with reading the data properly.
Disease data reaches you in four forms, and each suits a different kind of data. A frequency table lists how many cases fall in each group. A bar chart shows separate categories, like areas of a county, with gaps between the bars. A histogram shows grouped data that runs on continuously, like age in ten-year bands, so its bars touch. A scatter diagram plots one point for each person or place, against two variables. A question may give you a table and ask for a graph, or the other way round. The numbers stay the same, and only the form changes.
Now match a form to some data. New cases of a disease, sorted into age bands of ten years: which form fits, a bar chart, a histogram or a scatter diagram?
A histogram, because age is continuous data put into groups, so the bars touch.
One examiner's report, looking back over a whole paper, says questions asking students to pull information from diagrams, tables of data and graphs were answered well. So this is a skill you can score on. Incidence means the number of new cases of a disease. A city will always have more cases than a village, so incidence is given as a rate, the number of cases per hundred thousand people. A rate lets you compare a town, a country and the whole world fairly, whatever their size.
Now compare three places. A town's rate is four hundred and twenty per hundred thousand, its country's is three hundred, and the world's is one hundred and fifty.
Which comparison is fair? A, the town has more cases than the world. B, the town's rate is higher than the country's, and nearly three times the world's. C, the country has more cases than the town.
B is the fair one. Four hundred and twenty divided by one hundred and fifty is two point eight, so nearly three times. A rate says nothing about the total number of cases, so A and C cannot be told from these figures.
Notice what B does. When a question says compare, say which value is higher or lower, because two numbers quoted side by side are not a comparison. Before you read any value from a graph, find what one small square is worth, and write it down as its own line.
Here is a bar chart of new cases of type two diabetes per hundred thousand adults last year, in five areas of one county, A to E. The vertical axis runs from nought to six hundred. Each big square is a hundred cases, and each big square is split into five small squares. So the first line is the scale. A hundred divided by five is twenty, so one small square is twenty cases per hundred thousand.
Now read area C. Its bar ends three small squares above the three hundred line. What is its rate?
Three small squares is three times twenty, which is sixty. Three hundred plus sixty is three hundred and sixty cases per hundred thousand.
Reading each small square as ten would give three hundred and thirty. Reading it as one would give three hundred and three. Both come from skipping the scale line.
Now a second reading. Area E's rate is five hundred and forty, and area A's is ninety. How many times higher is E's rate than A's?
Five hundred and forty divided by ninety is six. Area E's rate is six times area A's.
One more line belongs on every chart answer. This chart says where the rate is higher, but not why it is higher.
One examiner's report, on a graph of crop yield, says many students did not look carefully at the scale on the y axis. The line sat at three point seven five, and students read it as three point seven or three point eight, so their final answers came out wrong. The context there was crops, not disease, and the habit is the same. Write one small square equals, and its value, before you read anything.
On a scatter diagram, the pattern the points make is called a correlation. If one variable goes up as the other goes up, that is a positive correlation. If one goes down as the other goes up, it is a negative correlation. If the points are scattered with no pattern at all, there is no correlation.
Here is a scatter diagram with one point for each of twelve towns. Across is the percentage of adults who smoke, from ten to thirty per cent. Up is deaths from lung cancer per hundred thousand people per year. The points rise from bottom left to top right. One sits at twelve per cent and thirty-five, and another at twenty-eight per cent and eighty.
How would you describe the relationship, in one sentence with two values?
It's a positive correlation. As the percentage of adults who smoke goes up, lung cancer deaths go up, from thirty-five per hundred thousand at twelve per cent to eighty at twenty-eight per cent.
That sentence has three parts every time. Name the correlation, say what happens to one variable as the other changes, and quote two values. Saying it goes up does none of that.
Now a student writes, this graph proves that smoking causes lung cancer. Is that right, wrong, or half right?
It's half right. The graph shows a correlation, and a correlation on its own cannot prove a cause, because something else could differ between the towns.
A correlation shows that two things change together. It does not show that one causes the other. Before calling it a cause, check three other explanations. The pattern could be chance. A third factor could be behind both. Or the link could run the other way.
Now test the checklist on a case. People who take a heart medicine turn out to have more heart disease than people who do not. Which explanation fits best: chance, a third factor, or the link running the other way?
The link runs the other way. Having heart disease is the reason people take the medicine, not the result of taking it.
To call something a cause, scientists need a causal mechanism, which is a proven biological route from the factor to the disease. Smoking is accepted as a cause of lung cancer because a causal mechanism has been found. Chemicals in tobacco smoke have been shown to damage lung cells, and that proof comes from outside this graph. Some risk factors have a causal mechanism like that. For others, only a correlation has been found. And one heavy smoker who never gets lung cancer does not break the correlation. A correlation is a trend across many points, and some points always sit off it.
Now a different study. In a survey of two thousand adults, those with grey hair had three times the rate of coronary heart disease.
Does grey hair cause it, and if not, what hidden factor explains the pattern?
No, because no causal mechanism links the colour of your hair to your arteries. The hidden factor is age, because older people are more likely to have grey hair, and more likely to have that illness too.
To check, compare people of the same age, with and without grey hair. If the difference disappears, age was behind the pattern.
Now back to the lighter in that survey. What is the third factor there?
Smoking. People who smoke carry lighters, and smoking has a causal mechanism for lung cancer, so banning lighters would change nothing.
Nobody can study everybody, so every disease study uses a sample, a smaller group chosen to stand for everyone. A sample supports a conclusion only if it is large enough, chosen at random, and representative of the group the conclusion is about. Representative means it matches that group in the ways that matter, such as age, sex, where people live and how they live. A sample that does not match is called biased. A biased sample can make a pattern appear that is not there, or hide one that is.
Now two samples of a hundred people, to study exercise and type two diabetes in one town. A is picked at random from the whole town. B is picked at the gym. Which can support a conclusion about the town?
Sample A. Sample B is the same size, but gym members exercise more than the town as a whole, so it is biased, and adding more gym members would not fix that.
Now a newspaper headline says porridge every morning prevents type two diabetes. The study behind it asked thirty volunteers from one running club to eat porridge every morning for a year, and none of them developed type two diabetes. Find three reasons the data do not support the headline.
Pause it there and work it out for yourself.
First, the sample is small, only thirty people, which is not enough to trust. Second, it is not random or representative. Runners exercise a lot, and exercise lowers the rate of type two diabetes anyway, so the study cannot separate porridge from running. Running is a third factor. Third, there is no comparison group who did not eat porridge, so the result has nothing to be measured against. If you found a fourth, a year is short, and the headline reaches far beyond thirty runners.
Every data question can close the same way, in three steps. First, what the data show, with numbers. Second, what they do not show. Third, a one-sentence conclusion that stays inside the people who were studied.
One examiner's report names pulling information from bar charts to make judgements and conclusions as a key area for teachers to revisit.
Those three steps are the fix, and the conclusion step is what turns a description of the bars into a judgement.
Now the method again, turned into questions.
An axis goes up in fifties, with ten small squares to each fifty. What is one small square worth?
Fifty divided by ten is five, so one small square is worth five.
Now, ice cream sales and cases of sunburn rise together through the year. What third factor explains it?
Hot, sunny weather, which makes people buy ice cream and also burns their skin.
Now, what is a causal mechanism?
A causal mechanism is a proven biological route from a factor to a disease.
Now a survey of a thousand people at one vegan food festival. Why is it a poor sample for a whole country's diet?
It is not random or representative. Everyone in it eats one kind of diet, so however large it is, it cannot stand for the whole country.
Now the survey with the lighters. Would a ban on lighters lower the rate of lung cancer?
No. Carrying a lighter is only correlated with lung cancer. Smoking is the cause, with a causal mechanism behind it, and the lighter just travels with the smoker.
If that's in the bag, give it a thumbs up, so you know this one is safe to skip. If it's still shaky, save it. Reading data comes with the questions, and every data question you try is practice.
Next in the chapter is Cancer: benign and malignant tumours. Hopefully that's next in your feed, but if not, search for ScholaFly B I oh eight, oh four.
Every other video is on scholafly.com too, easy to search by video code or by keyword. Whatever you're revising, go there and have a look, or, watch the next video.
Related terms
For: AQA GCSE 8461, Edexcel GCSE 1BI0, OCR GCSE J247
On the specification
| Board | Spec | Statement |
|---|---|---|
| AQA GCSE 8461 | 4.2.2.5 | Health issues |
| AQA GCSE 8461 | 4.2.2.6 | The effect of lifestyle on some non-communicable diseases |
| Edexcel GCSE 1BI0 | 5.23 | Describe that many non-communicable human diseases are caused by the interaction of a number of factors, including cardiovascular diseases, many forms of cancer, some lung and liver diseases and diseases influenced by nutrition |
| Edexcel GCSE 1BI0 | 5.24 | Explain the effect of lifestyle factors on non-communicable diseases at local, national and global levels, including: a exercise and diet on obesity and malnutrition, including BMI and waist : hip calculations, using the BMI equation: mass (kg) BMI = height (m) b alcohol on liver diseases c smoking on cardiovascular diseases |
| Edexcel GCSE 1BI0 | 5.25 | Evaluate some different treatments for cardiovascular disease, including: a life-long medication b surgical procedures c lifestyle changes |
| OCR GCSE J247 | B6.3t | Analyse the effect of lifestyle factors on the incidence of non- communicable diseases at local, national and global levels |
| OCR GCSE J247 | BM6.3i | Translate information between graphical and numerical forms |
| OCR GCSE J247 | BM6.3ii | Construct and interpret frequency tables and diagrams, bar charts and histograms |
| OCR GCSE J247 | BM6.3iii | Understand the principles of sampling as applied to scientific data |
| OCR GCSE J247 | BM6.3iv | Use a scatter diagram to identify a correlation between two variables |
For teachers
This GCSE Biology lesson teaches reading disease data: sampling, correlation and cause. By the end, students should be able to read disease data from frequency tables, bar charts, histograms and scatter diagrams, judge from how the sample was taken whether the data supports a conclusion, and tell a correlation apart from a proven cause. It works through four worked examples and the mistakes examiners report, and suits Foundation and Higher tier students.