ok, so I'll admit I didn't look up what he said or any of the data about warming not happening until now. I'm not an expert in the field of climate change but did take a healthy number of related classes over my university career. I do, however, know a thing or two about statistics. Now that I've actually looked at the data and what he said I can point out the massive flaw. (besides what I've already said about not being able to use a short term trend as a substitute for long term results)
So. I really have nothing better to do here so I'll make this take an embarrassing amount of time.
Here's a fictional data sequence to work with:
View attachment 763378
random sequence of numbers between 0 and 30 sorted from lowest to highest. there's no disagreeing about the trend.
Now let's put an anomaly in the data.
View attachment 763379
If one sill looks at the data as a whole a single outlier won't really make much difference to a large data set, the r2 value will be lower, but that's about it. You can still safely assume the trend you see to be correct.
The danger is then when you focus in on one small section of a data set and say that the trend you see in that area is correct without looking at any of the data to either side. So if you focus in on the the section starting from the outlier and onwards it will look something like this:
View attachment 763380
Look the data shows there is no more upward trend. The slope of the regression is pretty much zero.
Does this mean the trend seen over the whole of the data has stopped? Of course not. 1 anomaly just makes it appear that way when looking at the data from that point forward.
A better way to represent the data (which is used in climate sciences) might be to do a moving average. So for each point along the x-axis an average will be represented for the past x amount of y values. For our data let's use 5. This is what you'd use to more or less smooth out data, primarily data over time, so it gets rid of the noise of short term trends caused by anomalies and such.
Then you get something like this:
View attachment 763381
So we can zoom in on any particular range we want and still get a far more accurate representation of the trend at that point. For instance we can now focus on the last portion of the data and get a good idea of what the trend is actually doing at this point.
So how does all this matter at all? Well turns out 1998 is an anomaly year. An El Nino year that just doesn't fit in since it was far warmer than one would expect. So, like in our example, you can start looking from 1998 onwards and make it seem like there is no warming trend.
To rectify this you can just look at the moving average (10 years is pretty common in climate science I just found out), and you can get something like this:
look from 1998 onwards.
It also helps explain why 9 of the last 10 hottest years on record have been since 1998. 1998 filling out the 10 at number 4. Also that the 2000's has been the hottest decade on record.
It's really a matter of data interpretation. It's super easy to lie with statistics, and saying warming has stopped because you're looking at data starting from 1998 is a good example of how to do that.
Anyways, take this as you will. Just wasting my time on the internet, it was between this or porn.