Many of us understand the concept of the odds of a horse winning a race. Or perhaps we don’t and that’s why the gambling industry is so profitable. When it comes to weather forecasts, climate projections and health risks, is the concept of probability any more complex?
Regards from the @EconnectTeam
Perceptions versus facts
By Jenni Metcalfe
When reporting the odds in science, it is important to remember that people don’t always react rationally to the facts presented. This is due to how we perceive risk.
When people voluntarily choose an action, such as getting into a car or skiing down a steep slope, they see it as being less risky than something they don’t choose to do, such as eat food which has been genetically modified without their knowledge.
This is true even when the odds of being hurt in a car accident are far greater than the minimal risk that genetically modified food might pose.
In another example, people are more likely to refuse a vaccine that has a 1 in 10,000 chance of causing death, even when without the vaccine there is a 1 in 1000 chance of catching the disease and dying.
So what does this mean for communicating scientific odds or probabilities?
- Avoid using statistics to persuade people to make decisions about risk, especially when they feel they have no or little control over that risk.
- Acknowledge people’s concerns as being valid, even when they appear irrational; after all, it’s perfectly sensible to worry about the quality of the food you and your family eat.
- When explaining the level of risk, beware of analogies that compare an involuntary risk with a voluntary risk. For example, don’t say: “You are more likely to be run over by a car than to suffer any negative side effects from this medication”.
- Seek to understand the misperceptions that people may have, and why they have them, so that you can respond to and empathise with their values and points of view.
- Design messages that stress the correct information about a scientific issue and that reflect your understanding of the misperceptions people may have.
Using numbers, words and colours to communicate chance
By Sarah Cole
Are you a numbers, colours or words person?
How would you convey the chance of rain this afternoon?
Meteorologists make this decision whenever they communicate the odds of weather events.
If they communicate the odds well, it can help us decide when to have that family picnic, go sailing or plant crops.
Last year, in our newsletter on infographics, we wrote about using visual tools to make the intangible tangible. There are many ways to represent chance.
For example, the CliMate app gives farmers answers to questions such as: ‘How likely is it that Longreach will get more than normal rainfall in the next 3 months?’
The answer is presented in multiple formats on one screen:
- percentages (67.5%)
- likelihoods (‘7 out of 10 years’ had…)
- words (‘good’, ‘very high’)
- pie charts which represent percentages
- colours which represent percentages.
In the national temperature outlook (Dec 2013 – Feb 2014), Australian meteorologists have used:
- colour to represent percentages
- phrases such as “for every ten summer outlooks with similar odds to these, about six or seven of them would be expected to be warmer than average over these areas, while about three or four years would be cooler” and “chances of receiving a cooler or warmer than normal summer is roughly equal (i.e., close to 50%)”.
Warmer summer days more likely for eastern Australia
Bureau of Meteorology national temperature outlook for December to February
Seeing chance displayed in a range of formats can help us understand the odds of that storm arriving in the middle of our picnic.
Risks and ratios – making sense of statistics
By Jenni Metcalfe
Relative and absolute risk
If you’re communicating risk, understand the sort of risk you are talking about.
Absolute risk is the likelihood something will happen.
Relative risk is the difference in the likelihood that something will happen under different conditions. It allows you to compare different risk levels but it tells you nothing about the actual risk.
Buying a second lottery ticket doubles your chance of winning, but the chance was so small to start with that it’s hardly a meaningful increase.
Here’s another example of relative risk: A smoker’s risk of developing coronary heart disease is two to four times that of a non-smoker’s.
Always talk in terms of absolute risk, if possible.
If, for example, a substance increases the risk of cancer, describe how many extra cases it would produce in 100 people, or 1000 people, rather than the percentage risk increase.
Odds ratios – beware
Odds ratios are often reported in scientific studies. They can be hard to translate into everyday language and are best avoided.
In statistics, the ‘odds’ are the chance that something WILL happen divided by the chance that it will NOT happen. So, if there’s an 80% chance of rain tomorrow, the odds of rain are 4 (i.e. 80/20).
An ‘odds ratio’ is used when you want to compare the odds of something occurring to two different groups. It is the ratio of the odds for the first group and the odds for the second group.
An example: A study examined the link between acupuncture and successful IVF. In this study the odds ratio—i.e. the odds of pregnancy through IVF and acupuncture divided by the odds of pregnancy through IVF without acupuncture—was 1.65.
You might think this means that acupuncture increases the success rate of IVF by 65%—but the improvement in this study was more like 10%. In absolute terms, 10 patients would need to be treated with acupuncture to bring about one additional pregnancy.
[Ed: This article is an edited extract from ‘Understanding statistics and numbers’, a module in SciJourno, an online course for journalists, which we launched in October with AusSMC (the Australian Science Media Centre) and the University of Queensland. This module was written by AusSMC.]