Making sense of data: our tips
We’ve all heard the one about how there are lies, damned lies, and statistics. With the increasing focus on ‘big data’ across the private, government and non-profit sectors, we need to be ever mindful of the perils of ‘bad data’.
One of the best pieces of advice I’ve received when it comes to using any data to tell a story is that data does not equal insight, but you can’t have insight without data.
So what does that mean? Well, any data is a narrow slice of information influenced by sampling, time, indicators used, cleansing and categorising, bias, etc. etc. When data is used in isolation, the potential to misuse or misinterpret data exists, either deliberately or not. This is where ‘cherrypicking’ information to suit a desired narrative becomes quite prevalent and something politicians have turned into an art form. Insight is derived from the entire process of gathering relevant data then making decisions about what it is saying as part of a bigger story.
“The best vision is insight”
— Malcolm Forbes
We’ve been knee deep in data over the last couple of months supporting the Gold Coast Primary Health Network with their needs assessment process. We have developed data-informed summaries across a range of population health areas, such as immunisation, cancer screening, chronic disease, Aboriginal and Torres Strait Islander health, and aged care.
Whether it be describing the health needs of a community or measuring the effect of your program or service, here are a few traps and tips to increase the rigour of your data analysis based on our experience:
Scan widely for all potential data sources—the open data movement means more data is becoming publicly available
Understand that external data sources are often not designed for the same purpose as yours — reading the data definition is important
Be a discerning consumer of data—don’t be afraid to dispute the quality of a data source
Be transparent if data is conflicting and explain your rationale to opt for one source over another
Use visuals to display data, but also provide commentary in simple terms explaining what is shown
Reference your data source—your audience will expect to be able to trace down the original material
Use a logical approach to telling the story of the findings—it could be chronologically or using a ‘funnel’ approach (start broad then narrow)
Sense-check your findings and assumptions with a broad range of stakeholders, such as focus groups with client/consumer representatives
Present the findings in a visually appealing way—invest in graphic design if you want people to remain engaged.
As part of our current work, we have put in place a number of tactics to avoid these common pitfalls and provide a level of assurance around the quality of the data we use.
The level of rigour you require will depend on the scale of the project and the intended application. The most important thing is being confident to justify the statements you’re making by using a wide range of data sources and understanding where the data is coming from.