Many years ago I was chatting to an accountant who had been given the job of costing a range of standard analytical techniques used to assess pollution in natural waters.
I still recall the look on his face when I told him that the true value of an ammonia result of say 1.0 milligrams per litre (mg/l) might lie somewhere between 0.9 and 1.1 mg/l and occasionally would lie outside that range.
He was astounded. At first he couldn’t believe that a value of 1.0 mg/l of ammonia might not be exactly that. Even worse, it was unlikely to be exactly 1.0 mg/l even if that was the result reported and paid for. I had to draw a bell curve for him and explain the role of statistics in such analyses.
In the end he realised he’d learned something about chemical analysis and we moved on. What I didn’t mention was another word I could have introduced him to :-
Assuming the sample was taken from the right place.
Assuming the sampler used the standard protocol.
Assuming the analyst didn’t mix up the samples on the analyser.
Assuming there is nothing unusual to affect the analysis.
The natural world is exceedingly complex and the environmental sciences are riddled with measurement uncertainty and analytical pitfalls. Inevitably we always have to deal with that thoroughly human aspect of real life – we have to assume certain conditions which could affect our analytical results, measurements and our conclusions.
Suppose you are to conduct a limited survey of lead in a stretch of trout stream lying between two bridges. You have a sampling protocol borrowed from the Environment Agency and a contract with an accredited analytical laboratory where your samples are to be analysed.
Everything goes well, your samples are collected by a student and the analytical results are as expected for a trout stream. All seems hunky dory.
Apart from one result.
This single result shows an extremely high lead concentration in a single sample taken from the downstream bridge. So you have that particular sample reanalysed. After reanalysis the result stands – one high lead result is confirmed.
What do you do?
Report it and the entire stretch of trout stream comes under intense suspicion for intermittent contamination by lead. Yet the result appears hopelessly anomalous. You check with the student who took the samples. No problems there – all sampling protocols were followed.
After some soul-searching you leave the high result out of the final report. It’s too anomalous and too contentious. Frankly you don’t believe it because human error is hardly unknown in such cases, from contamination to mixed-up samples.
This is key – you don’t believe the result. You are convinced it is due to human error.
After your report is published, you discover that a field adjoining the trout stream was spread with sewage sludge in the nineteen seventies. The sewage sludge may have had a high lead content from lead in petrol and road runoff into the sewers. The anomalous lead result occurred shortly after a heavy storm so there may have been runoff from that field.
This little story is pure fiction and the high lead result could still have been due to human error. The problem it highlights is common in all environmental analysis - even satellite temperature measurements of the atmosphere.
When studying the natural world, we have our expectations and very often anomalous findings are due to human error, process malfunction or instrument failure. So one way or another, anomalous findings tend to be left out of the picture and the picture itself migrates towards a consensus.
There are often political pressures behind those expectations too. Good scientists know this and cope with it, but the potential for deceiving ourselves and others is considerable and insidious.
A few decades ago, climate scientists had a far more complex problem to deal with and many flunked it. They failed to cope with climate complexity and the pressures their assumptions eventually brought on their incautious heads. We should not be particularly surprised.
The much trumpeted climate consensus means all our assumptions are correct. Oh dear – have our energy policies been bent to breaking point over something so naive?
All original material is copyright of its author. Fair use permitted. Contact via comment. Nothing here should be taken as personal advice, financial or otherwise. No liability is accepted for third-party content, whether incorporated in or linked to this blog; or for unintentional error and inaccuracy. The blog author may have, or intend to change, a personal position in any stock or other kind of investment mentioned.