Are graphs more powerful than words?
The same information displayed on different graphs can have a very different influence on people’s actions. Dr Yasmina Okan talks to us about her research to find the best way to display information to help people follow health advice.
If a disease affects one in ten people, you are more likely to contract it than if it affects one in 100 people. Obvious? Maybe — but studies have shown that over a quarter of people in the USA and Germany can’t tell which presents the greater risk.
Given how often we are asked to make decisions on the basis of statistical information, particularly in healthcare, this lack of understanding is worrying. Whether it’s taking part in a screening programme or judging the likelihood of side effects, important choices are made on the basis of an understanding of risk.
Understanding risks
Presenting risk information using graphs can improve understanding — but it’s not as straightforward as it might sound, according to Dr Yasmina Okan of the Centre for Decision Research at Leeds University Business School.
“Risk information is often easier to understand in simple graphical formats, but complex graphs can sometime have the opposite effect” she says. “Many studies have shown that ‘icon arrays’ — that use human figures or circles rather than just numbers — can help people to understand health risks significantly better. For example, you could use black icons to show the number of people who experience a side effect, and white icons for the number who don’t. But some research has shown that icon arrays that include complex interactive features can in fact lead to worse understanding of the side effects associated with different cancer treatments.”
Small scale studies carried out by Dr Okan as part of an international collaboration between the Business School, the Max Planck Institute for Human Development in Berlin, the University of Oklahoma, and Wake Forest University in the USA found that different graph designs can also promote different responses.
For example, bar graphs representing the number of people affected by a heart attack (but not the total number of people at risk) resulted in people being more willing to take a treatment to reduce heart attack risk, but worse risk understanding. However, adding simple numerical labels to the top of bars promoted better understanding, while still positively affecting people’s willingness to take the treatment.
Unintended consequences
In another study, bar graphs were used to show people’s average blood glucose measurements over the previous week. Even though measurements had been ideal, so there was no need to modify blood sugar levels, the researchers found that different graph designs resulted in people wanting to take a treatment to either raise or lower their blood sugar levels.
“What we learned from these studies is that graphs can have unintended consequences,” says Dr Okan. “So to ensure they work to both promote risk understanding and have a good outcome for healthcare, we need to understand how different aspects of the graphs create certain biases that then impact on the decisions people take. We also need to make sure that graphs are designed in line with current best practices and recommendations.”
Dr Okan — with colleagues from Wake Forest and Oklahoma Universities in the USA — is now funded by Cancer Research UK to look at how presenting information in simple evidence-based graphs could improve women’s understanding of the benefits and risks of cervical cancer screening, with the aim of promoting informed uptake.
Although the risk of cervical cancer is relatively low compared to other types of cancer, the majority of cervical cancer cases could be prevented through screening, which aims to detect abnormal cells before they become cancerous. But in the UK, screening uptake has been declining for the last ten years and recently hit an all-time low.
While some of the reasons for non-participation are emotional or practical — fear about the procedure or problems finding time for an appointment — it may also be caused by lack of understanding as to how screening can reduce the risk of developing cervical cancer.
The team have been analysing the information that women will most commonly access on cervical cancer, including websites and leaflets or pamphlets that can also be accessed online. Their aim was to look at what kind of information was included and what excluded, what level of detail was provided, how often graphs and numbers were used and what formats were used to describe risk.
“Our initial findings are that graphs are used very little for cervical cancer information” says Dr Okan. “We also found that risk was regularly being quantified only in verbal terms, such as high or low likelihood. These are open to very different interpretations, which could be affecting how women respond to the information they receive.”
The team are also carrying out cognitive interviews with around 40 women, using the NHS leaflet on cervical cancer screening as a basis for the discussion. This contains information about cervical cancer and its causes, the benefits and risks of screening, what the procedure involves, possible screening results, and how to secure an appointment. Through these interviews, the team is building a picture of how women respond to the information in its current format, what they find hard to understand and what aspects motivate or prevent them from getting screened.
Improving understanding of risk
In the next stage of the project, the team will improve the leaflet and related web-based materials by adding graphical information and, potentially, other elements known to increase the effectiveness of risk communication. The new versions will be tested again through face-to-face interviews and an online survey that can randomly allocate the original or the updated versions of the information.
“We want to see if the inclusion of simple graphs can improve women’s understanding of information concerning screening,” says Dr Okan. “We also want to gauge their emotional reaction to the information and whether it will affect their willingness to be screened.”
Based on the outcome of these interviews, the materials will be further improved and then tested again in a longer study. This will assess women’s immediate reaction to the materials and then follow up after three and six months, to see if they did either get screened, talk to their GP about screening or look online for further information.
The project is set to run for three years and in total will involve over 1,700 women who are eligible for cervical screening (aged between 25 and 64), and with varying demographic characteristics, including different education levels. The findings will be useful for any organisation involved in communicating about cervical cancer or related health risks, says Dr Okan.
“We should be able to show how to present information to ensure people have the best understanding of benefits and risks of screening” she said. “Many charities and health organisations already work hard to try and present this kind of information in the best way possible, but they can still be ineffective, if the information is too complex for people to understand.”
“These kinds of leaflets are sent to everyone who is eligible of screening and are an important part of encouraging informed uptake, but in some cases, as with cervical cancer, they may not be working as intended. We hope to help overcome that problem.”
Find out more about the research taking place at the University of Leeds
