CO_{2} & Temperature correlation: Denial debunked

Patrick Moore posted a chart on social media that he states shows that there is no correlation between CO_{2} and temperature.

The chart is based on a rather bizarre variable, which is presumably one chosen by the CO_{2} Coalition - or somebody else - to misrepresent the known data.

On this page I will show how this trickery has been put together.

The chart on the right shows his plot of a temperature related entity to CO_{2}.

The entity is "% of days above 100F". Why? That means we're looking at "outliers" to assess the relationship between two continuous variables. From a statistical viewpoint that is a bizarre thing to do.

The valid way to assess such a relationship is to actually plot Temperature against CO_{2}.

Three other elements of the "analysis" also need to be dealt with in a valid way.

Firstly, we only have reliably recorded CO_{2} values back to 1959. Earlier values have been deduced using ice core data, so I will split the analysis into 2 charts, reflecting the difference in how the data were obtained.

Secondly, when it comes to correlation, whilst a careful choice of vertical scales can be used to hide high (or even low) correlation this is only a visual effect. Calculation of the level of correlation is independent of the scales used in a graphical represenation. So I'll use vertical scales that reflect the relationship inferred by the data, not that try to hide the relationship.

Thirdly, some of the data only relate to the United States. Clearly any analysis needs to be consistent in geographical scope.

Annual average global temperature anomaly values were obtained from Our World in Data.

Atmospheric CO_{2} values were obtained from NASA.

Plotting these produces the chart shown.

The coefficient of correlation over the period 1959-2023 is over 96%.

No correlation?

Really, Patrick?

Annual average global temperature anomaly values were obtained from Our World in Data.

CO_{2} values were obtained from NASA.

Plotting these produces the chart shown.

The coefficient of correlation over the period 1880-2023 is over 95%.

No correlation?

Really, Patrick?

Establishing correlation does not - of course - prove causation. However, my aim here was to show that there is correlation, in direct contradiction of the strange analysis presented by Patrick Moore.

On the subject of causation, however, this is an interesting paper.

© 2024 TobyArnott.com - Let me know of any errors in the data or of any broken links.