I spent a good portion of my life, some thirty years I guess, studying the distribution of rock-wallabies. Cute critters, go google them. The ones we have along the ranges (mostly) in NSW and Victoria are brush-tailed rock-wallabies. They’re the cutest, but OK, I’m biased. My studies involved information, maps and data – a lot of all of that. I was looking for reasons why they occur where they do and looking back into the past (few tens of millions of years) and into the future (considering land use changes as well as potential climate change effects).

The holy grail for any such research is cause and effect, i.e. this thing is happening/did happen (cause) and it made that other thing happen as a result (effect). I could, and did (mad man me) put all my data into a spreadsheet which resulted in some hundred odd columns and some 2,000 rows of data. So yes, that’s around 200,000 individual bits of data. I’m not even going to try to calculate how many possible inter-relationships, or correlations, that is, but I suspect it’s up there with the odds of winning lotto.

So, finding which of those correlations were ‘meaningful’ was time consuming. Trust me, I have two unfinished PhD studies to prove it. Both terminations were mutual – my problem being I couldn’t stop myself running down every rabbit hole I could find.

Anyway, “Yah me!”, because the total number of cause and effect relationships I found out of all this data and inter-relating was... zero. Yep, none, nada, zilch. Correlations though – oh my goodness, let me tell you about them... but I won’t as we both have lives to live. The really juicy ones I could even demonstrate via statistical analyses that the likelihood of those small handful of most promising correlations being random was acceptably small. In other words, these were meaningful correlations, i.e. something looks like it's going on for real here. Well, this is cause and effect, yes? No. It’s not.

OK, leaving rock-wallabies, let’s look at buttercups.

There is an old bit of trivial folklore that says that you can tell if a child likes butter by holding a buttercup under his or her chin. Oh my goodness, the stuff people believe, right?

Now, what if I tell you that I can prove with 100% accuracy that a child likes butter by doing just this, holding a buttercup under their chin? Yeah, right, pull the other one. Actually, no, what I’ve said is accurate. The trick is in the wording, and here beginneth the true lesson, one I learned through an awful lot of square eyes staring at my spreadsheet. What I didn’t say is ‘I can prove IF a child likes OR DOES NOT LIKE butter’. Give me a paddock full of kids, and if I run around with a buttercup it’s a fair bet that every little chin shines yellow. I end up with a big number of correct identifications – I mean how many kids don’t like butter – and a quite small number of incorrects. But of all those children I have identified as ‘yes, likes butter’ my mark is 100%, i.e. I didn’t miss a one who does like butter. Even if I account for those I got wrong, as the great majority of kids like butter anyway, then my ‘accuracy’ still looks good. At I guess I’d say 90% or better. Go ask a handful of kids and see how that goes.

You’d be surprised, or maybe not, how often this sort of stuff is reported as ‘recent study shows...’. Or maybe you wouldn’t. It’s actually easy to make this mistake, and that’s just humans erring, but deliberately making this sort of statement, knowing what the problem with it is, is not acceptable behaviour. Taking data like this, knowing it’s ‘falsity’ even if technically it’s correct as per the terminology used, is an example of what’s called cherry-picking. Basically you only present the information, or a particular way it can be presented, to bolster or support an argument without noting its limitations or inherent inaccuracies.

So, let's look at a 'real life' example, related to bushfires and that 2019/20 ‘Black Summer’.

A certain Australian politician appeared on an international news service during the height of that horrible bush fire season. In attempting to deflect criticism of the apparent lack of action on climate change in Australia, especially given it’s potential relevance to the then current raging bush fires, the politician said (words to the effect – as close as I can faithfully remember them) that ‘our scientists have told us that there has been no decrease in annual rainfall in Australia over the past decade'. His point being that if rainfall hadn’t decreased, how can there be a relationship to climate change (if that even exists) and to the wildfires? His view being there wasn’t one, and that the fires were the result of arsonists and not enough fuel reduction burning. Warning – here be buttercups!

What he said was true. In Australia (the whole continent) the annual rainfall had been more or less the same over the past decade prior to the 2019/20 summer. Now, given that it took me less than a minute to get to the Bureau of Meteorology site and find the annual precipitation maps, I find it hard to believe this politician, and/or his advisers, didn’t know exactly what he was doing. Yep, there are decade by decade annual rainfall data for Australia and they pretty much say just what he said; but there’s also maps. Blues of increasing darkness showed higher than average rainfall, reds in darkening shades showed lower than average rainfall. The map for the past decade, which summarised geographically the data he had been quoting, showed a continent split in half. You could pretty much take a ruler and draw a straight line from the north-east (mid-coast Queensland) to the south-west (a bit north of Perth) and you had a blue half a continent above that line and a red half below it. You could also go grab the fire spread maps from the Rural Fire Service fire mapping for the then current fires, toss it onto the decade annual rainfall map and the correlation was as clear as it was telling, i.e. to all intents of purposes the fire areas all fell within the lower than average and up to extremely lower than average rainfall areas. As far as cherry picking, or ‘buttercupping’, goes, his take on the data available was at best negligent and at worst bordering on criminal, considering the loss of properties and life that was occurring.

Back to finish on rock-wallabies, unknown numbers of which would also have been lost in those fires.

I never did establish a clear cause and effect in all that distribution data I was studying, but I did find a lot of interesting correlations; a lot more work (and I mean a LOT) may have lead to some clearer relationship of ‘this therefore that’. Perhaps though most importantly I also learned a lesson that has stood me in very good stead well beyond my studies, especially in these seemingly and increasingly conspiracy theorising times. Don’t believe anything you hear or read on face value. Go chase up the original source yourself and be brave enough to actively seek out views and information counter to what you want to hear and read. We’re all entitled to our opinions, but unless you’ve taken due diligence to prove what you’re espousing from unbiased, original sources, then that’s all they are – opinions. Not fact.

And while you’re online, for heaven’s sake do yourself a favour and go check out some rock-wallabies!