My experiment I’m doing this week on explaining my weight trends provides a very simple dataset to look at how “cherry-picking” data and extrapolating trends can show whatever bias the “analyst” wants. Below is the raw data and three different “trends”:
The measurements for each day show “noise” (variations in measured value, in this case as a digital scale reports). Any real world measurements will have noise, whether it’s my weight day-to-day or whether it’s temperature, over some area or span of time. Given all that noise how do you determine a trend?
The simplest technique is just linear regression (assuming you believe the trend to be linear, otherwise you might use some other function). And critically you use all the data – you don’t selectively include or throw out data because doing so would add a bias. Even if you do have a “bad” data point you’ve should have gotten enough data that even large random errors will be washed out. So in this graph (and dataset) that’s the trendline marked as “science“. This shows a nice steady downward trend which happens to translate to 3.5lbs/week weight loss (would it be so, but I’ve also “cherry-picked” data for a time period where that is what is happening). Nonetheless (assuming this is all the data I had) this is the unbiased or objective view, otherwise known as fact.
Now OTOH you can choose to cherry-pick the data in order to report whatever conclusion you want the data to show. In global warming there are people who would like to be called ‘skeptics’ or ‘contrarians’, but since these people usually end up being paid by some ’cause’ (like the Koch brothers) it’s dubious that they’re objective and so really it is entirely appropriate to call them ‘denialists’, that is people who are putting out propaganda that can be run through the rightwingnut echo chamber to feed their gullible “base” to believe what they want them to believe. So these people would selectively pick two points from the data, draw and trendline through that, and declare an outcome. That is shown on the line marked “denialist” which then shows that the data is increasing (in this case my weight is going up, in their case it might be polar ice or winter temperatures or whatever data they want to distort). They would claim a 1.2lbs/week increase, otherwise known as a lie.
The tagging of one analysis as fact and the other as lie is not an opinion or different points of view. There is no “controversy” here. There is just one answer, the truth, and then there is a lie. It is a lie, not a mistake, because it is a deliberate attempt to distort what the data is saying to produce a foregone conclusion. That’s lying, not controversy or skepticism. This is what the climate change denialists are doing, lying pure and simple.
Now the denialists also accuse the scientists of sky-is-falling alarmism which sometimes, regrettably, may be true (for instance, some stuff RFKJr puts out). In this case the true alarmist would also cherry-pick data and thus show the trendline marked “alarmist“. This does show the correct trend (unlike the lying denialist) but is too extreme in magnitude. In this case it shows 7.6lbs/week loss (a number that could be dismissed via other analysis). This is appropriately called a distortion since it does correctly indicate the direction of the trend but exaggerates the magnitude.
Now if all three groups will openly explain their methodology (and not just spout their “conclusion” without explaining how they got it) rational debate would quickly dismiss both the denialist and alarmist positions. This would be the normal process in science, discussion and debate until consensus is reached. The outliers, the denialist and the alarmist, would be rejected. Of course this is what has happened in the global warming debate and this is why 90%+ of the scientists agree to both the trend and approximately on the magnitude. People who are outside those bounds, as all the Repugs are, are at best ignorant, more likely just liars.
So there you have it, so you decide: am I losing weight or not?