Studying health risk and perceptions of risk
Studying this course? Keep your progress in one place.
Key facts
- Morbidity is the rate of illness and mortality the rate of death; compare populations using a rate such as per 100 000.
- A correlation does not prove a causal relationship because a confounding variable may affect both.
- Studies conflict because sample size, duration, methods and outcomes differ.
- A sample must be representative; a larger sample is more reliable.
- People often underestimate voluntary, long-term risks and overestimate dramatic, rare ones.
Analysing illness and mortality data
Convert counts to a rate per 100 000 so that populations of different sizes can be compared fairly.
Morbidity is the rate of illness and mortality is the rate of death. To compare populations of different sizes, express data as a rate such as deaths per 100 000 people.
Quote figures from the table, use comparisons (ratios or percentage differences), and consider age, sex and lifestyle.
- Rate per 100 000
Worked example: rates
Town X has 48 CHD deaths in a population of 32 000. Town Y has 30 deaths in 12 500. Which town has the higher risk?
- 1
Town X: 48 ÷ 32 000 × 100 000 = 150 per 100 000.
- 2
Town Y: 30 ÷ 12 500 × 100 000 = 240 per 100 000.
- 3
Compare: 240 ÷ 150 = 1.6.
Why should deaths be converted to a rate before comparing populations?
Correlation and causation
A correlation shows a link between variables, but only a mechanism or a controlled experiment supports a causal relationship.
A correlation is a link between two variables. A causal relationship means a change in one directly produces the change in the other.
A confounding variable may affect both. For example, dog owners may have lower CHD death rates because they walk more, not because they own a dog.
Causation is more convincing with a plausible mechanism, a dose-response pattern, consistent results from several studies, and a trial in which only one variable is changed.
- 1
Observation
dog owners have lower CHD death rates
- 2
Possible confounding variable
dog owners also walk more, so take more exercise
- 3
Exercise lowers CHD risk
it may be the real cause of the lower rate
- 4
So the link may not be causal
a mechanism or controlled experiment is needed
Correlation
- A statistical link between two variables
- Does not show cause
Causal relationship
- A change in one directly produces a change in the other
- Needs a mechanism or a controlled experiment
A study finds that dog owners have fewer CHD deaths. What is the best explanation?
Conflicting evidence
Studies differ in sample size, population, duration, method and outcome, so one study is rarely conclusive.
Studies on health risk sometimes conflict. Possible reasons include different sample sizes and populations, different durations, different methods of measuring diet or lifestyle, and different outcomes (CHD events or a risk factor such as LDL cholesterol).
To judge conflicting evidence, compare sample size, duration, controls and the quality of the measurement.
- 1
Different samples and populations
size, age, country, lifestyle
- 2
Different methods
duration, how diet or lifestyle is measured
- 3
Different outcomes
CHD events or a risk factor such as LDL cholesterol
- 4
Conflicting results
so one study is rarely conclusive
- One studyrarely conclusive
- Several large, well-controlled studiesmore convincing weight of evidence
Why might two studies of the same risk factor give conflicting results?
Evaluating the design of risk studies
A good study uses a representative sample, a large size, controlled variables and reliable measurements.
Health risk studies include cohort studies (a group followed over time), case-control studies (people with a disease compared with people without) and controlled trials.
Control confounding variables or match groups for validity. Repeat readings, calculate means and use statistical tests for reliability. Self-reported data, such as diet questionnaires, may be inaccurate.
- 1
Recruit a representative group
large, without the disease at the start
- 2
Follow them over time
control or match confounding variables
- 3
Record who develops the disease
use reliable, repeated measurements
- 4
Compare rates by exposure
for example smokers and non-smokers
| Good design | Weak design | |
|---|---|---|
| Sample | Representative: random, covering all ages, sexes and lifestyles | Biased: volunteers from one group, such as a gym |
| Size | Large, so chance and anomalies matter less | Small, so chance has more effect |
| Measurement | Controlled and repeated | Self-reported, may be inaccurate |
Good design
- Sample:
- Representative: random, covering all ages, sexes and lifestyles
- Size:
- Large, so chance and anomalies matter less
- Measurement:
- Controlled and repeated
Weak design
- Sample:
- Biased: volunteers from one group, such as a gym
- Size:
- Small, so chance has more effect
- Measurement:
- Self-reported, may be inaccurate
Which sample would be biased?
Perceptions of risk
People often underestimate voluntary, long-term risks and overestimate dramatic, rare ones they cannot control.
A person's perception of risk often differs from the actual risk.
Underestimating (for example CHD): voluntary risks feel under personal control, effects are long-term, and optimism bias says 'it will not happen to me'.
Overestimating (for example flying): dramatic, rare events get heavy media coverage, and people fear risks they cannot control.
Poor perception affects lifestyle choices, so public health messages need clear data.
- 1
Actual risk
measured from data such as mortality rates
- 2
Perception is shaped by
control, timescale, drama and optimism bias
- 3
Common errors
underestimating CHD; overestimating dramatic, rare events
| Underestimated risk | Overestimated risk | |
|---|---|---|
| Example | CHD from smoking and diet | Violent crime or flying |
| Why | Voluntary, long-term, optimism bias | Dramatic, rare events with heavy media coverage; risks outside control |
Underestimated risk
- Example:
- CHD from smoking and diet
- Why:
- Voluntary, long-term, optimism bias
Overestimated risk
- Example:
- Violent crime or flying
- Why:
- Dramatic, rare events with heavy media coverage; risks outside control
What is optimism bias?
Try an exam question
A study of risk factors for coronary heart disease recruited only volunteers from a local gym. Evaluate the sample used.
[4 marks]
- [1]The sample is not representative of the whole population (ages, sexes and lifestyles).
- [1]It is biased, because gym members are likely to be more active than most people.
- [1]So the results cannot be applied to the whole population.
- [1]A larger random sample covering all ages and lifestyles would be more reliable.
That's the notes covered.
Carry on to the next subtopic.
Try a question
Which statement is best supported by these data?
1 mark, marked against the mark scheme
Which statement is best supported by these data?
1 mark, marked against the mark scheme