How to correctly approach statistical claims

It’s very easy to read a statistical claim that reflects your current opinion on that same matter, and agree with that claim without questioning it. We may even go as far as sharing it with others on social media, before double checking what we’ve read. The FT discussion between Tim Harford and Sarah O’Connor contains some useful tips on how we should be interacting with statistical statements and also things we read or hear in general. 

Firstly, Tim says we should be assessing how we feel about the subject, whether or not this opposes or affirms our original opinion. We should “notice our emotional reaction” and be aware of this. We’re often looking for things we desperately wish to be true. He then tells us to try and understand the statement being made, what is meant exactly by certain words, such as “nurse” and “inequality”. Look out for whats being left out is another one of his points – be careful of noting the datasets that could have existed prior to the data being used. For example, data collected in 2005 might well be relevant to a story that is only focused on data from 2006-2010. Do your own research about what you’re reading to get a broader picture. Any presentation of data is going to leave out some form of information, and it’s important to be sceptical.

A particular example of misleading data or claims of data is a one made my Donald Trump regarding America’s economical growth under his presidency.  It isn’t news that Donald Trump is a first place fabricator – he’s made 2,000 false claims since taking office as of January 2018, and when he spoke to the United Nations in September last year, he told them that “Countries are moving back, creating job growth the likes of which our country hasn’t seen in a very long time.” Politifact used data to show via an infographic that this level of growth was actually mimicked during the Obama administration.

Whilst reading an article on the independent about tanning beds, a statistic from cancer research UK is quoted, saying that if you use a tanning bed before the age of 35 you are 75% more likely to develop skin cancer. Instead of taking this at face value, it would be interesting to know the frequency of tanning sessions needed to increase this chance, or do cancer research literally mean single use before age 35? This links well to what Tim said about looking out for what could be missing.