Hugely important insight here for analysts and data scientists. It isn't enough to know the right answer: you have to persuade people to pay attention to it...
This is a big problem since the group of people who have the insights (the analysts) are often the least effective at persuading decision makers to pay attention and actually make different decisions...
Analysts are often introverts and often focus on the art of analysis (getting the right answer) rather than the much wider job of persuading the people in power to make the right decisions...
The skill set required to persuade is very different to the skill set required to do analysis. But to achieve significant improvement, both are needed...
Nightingale's contemporary, the great John Snow (the father of epidemiology, not the GoT character) also knew this, luckily for the population of cities as he led to the elimination of cholera epidemics...
Current analysts need to learn the skills of persuasion if they are to have any impact on the world. Fixing the NHS isn't just about knowing the right answers, it is about persuading people to pay attention to them...
Interestingly, one of Nightingale's key skills was #dataviz. She knew that pictures were often far more persuasive than numbers and deployed very original visualisations in her campaigns to persuade the powerful to change policy...
Modern analysts frequently undervalue the importance of #dataviz . "I've got the right numbers, why waste effort on pictures" they seem to assume...
But, as Nightingale realised, a persuasive graphic is vital. In my experience designing that picture requires even more effort than getting the right numbers...
But the overall message is that learning #dataviz and the arts of persuasion should be more than half the job of all analysts if they really want to change the world...
Knowing the right answer is useless if nobody pays attention to it.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
I complained a few days ago about this story which reports A&E performance with "headline" numbers allowing obfuscation of bad performance in major A&Es. I thought I should follow up with some numbers... @NickTriggle
The major point is that using the headline number allows trusts with terrible performance in major A&Es to dilute their headline by including the (much better) performance in type 3 units (where poor performance is rare)...
It has been argued that this doesn't matter as type 3s divert minors from type 1 A&Es leaving them with worse performance. But this is only true if they are co-located, which is rare...
Here is an experimental chart that doesn't fit well on twitter: trust type 1 A&E performance vs the national performance, monthly since 2010...
Key message: for many trusts, relative performance persists over long periods...
This is a quick analysis not a thorough one and has some limitations. One is that a trust with middling but consistent performance will look like it is improving as national performance declines...
Let's say an NHS hospital could achieve only one of three things by taking action right now: 1) 3% cost improvement next year 2) 5% more elective activity next year 3) 30% higher productivity in 5 years time
Which would be better? Which would NHSE want? ...
The are mutually exclusive alternatives. You can only have one.
Which are they incentivised to do? ...
My suspicion is that in most years NHSE will pay far more attention to 1) and will actively intervene if that target is missed...
This story in today's Times by @eleanorhayward has the headline "GP shortage means busiest doctors look after 3,000 patients" which seriously understates the problem...
The analysis averages the GP staffing across ICBs which are large regions with hundreds of GPs each. But the data exists for individual practices and the variation inside ICBs is huge...
So even "good" ICBs might have some practices with much worse ratios. Nationally there are more than 1,600 practices (out of about 6,500 in England) with >3k patients per GP (april 2022 data):
It uses headline performance (which allows some trusts to include MIU and WIC) numbers which obscures how bad their major A&Es are...
So, for example, the Royal Cornwall goes from a bottom 5 performance in its major A&E of only 43% in 4hr to a top 10 performer with nearly 75% in 4hr including its other units...
Type 3 units (MIUs, WICs etc.) don't usually have many long waits. But hospitals can include those units in their headline numbers to dilute the bad performance of their major A&Es...
The Reform think tank have just released an extremely interesting analysis of why A&E performance in the NHS is so poor (with rather more analysis than the recent NHSE plan): reform.uk/publications/t… ...
What is interesting is how their conclusions echo the conclusions experts were talking about in 2013 and which were thoroughly explored in several analyses between 2014 and 2016 (notably by the Nuffield Trust and Monitor)...
Let's have a look at some of the big conclusions and their implications (long quotes as screenshots to avoid Twitter thread overload, sorry)...