Another example of climate journalists and researchers not looking carefully at the data they are reporting.
It is on the face of it it is an interesting and significant story showing the tragic effects on humans as a result of a warming world.
More people are falling into the water and drowning due to walking on unstable ice or having their snowmobiles and other vehicles break the ice at times of the year when it had previously been stable. It seems, as the headlines proclaim, that warmer winters are linked to increased deaths according to a study just published in Plos One. It is reported that deaths from drowning were five times higher when warmer weather made the ice thinner and weaker.
My first thought was to wonder if those experienced in ice travel would judge safe conditions by the calendar using what the state of the ice would be on average at a certain time of the year rather that looking at the actual ice conditions. It made me want to look at the actual data in the research paper.
It comes from Canadian researchers who looked at data on 4,000 drowning events in 10 countries. “We can confidently say that there is quite a strong correlation between warmer winter air temperatures and more winter drownings,” Sapna Sharma from Your University in Toronto told the BBC.
It seems a simple story until one looks at the data which, as I suspected, paints a different picture than the headlines.
Look at their Figure 2 which shows deaths (per million people) plotted against December-March air temperature. An initial concern would be taking the average temperature over 4 months – this will include some variation in air temperature before and after the coldest temperatures of the year. But never mind, it is not the main concern there is with the data.
The authors of the report consider that Fig 2 shows that drowning deaths “increased exponentially” in regions with warmer winters when air temperatures exceeded 0°C. This is in my view is not a robust conclusion.
Clearly Estonia and Latvia are outliers: they have a far lower population that other countries in the data, yet have as much an ice culture as the others. Because of this I suspect that the data is in part a result of cultural and administrative factors in those countries.
The variance of Latvia’s data is huge, and Estonia’s data is sparse and is also widespread. Remove these two countries from the graph and the increasing exponentially effect disappears, as does the fundamental conclusion of this study.
In conclusion, this report has some interesting things to say about Latvia and Estonia but nothing to say about a worldwide effect on whether winter ice deaths are increasing as the world warms. It’s another example of specialist reporters of climate change only reading top lines, and not looking at the data of the stories they are reporting.