Friday, April 09, 2010

DR Chan’s findings distorted by media massage

Dr Yin Chan is always careful to be scientifically correct whenever he discusses his results. Not so the ‘communications’ experts massaging the media. The Doctor says his investigation, no statistical difference was found for all options except pastures improved with phosphate fertiliser. He doesn’t discount the possibility that other options could have an effect. He says the methodology – paired paddock comparisons - and the ‘field variability’ in samples could easily have ‘masked’ any differences.
But in the Booklet which reports the results, the headline reads “Pasture types do not affect SOC” which is very different to the text beneath: “We found no significant differences in SOC stocks between introduced or native pastures, and between annual and perennial pastures.”
This study had two “findings” reports – an interim report released in July 2009 and the elaborate Booklet “A Farmer’s Guide…” last month.
To judge by the tone of the ABC Rural report, the direction of the spin: “The latest science has debunked theories that rotational grazing methods can dramatically improve carbon storage in your soil. Dr Yin Chan of the NSW Department of Primary Industries says they compared carbon sequestration under set stock grazing compared to rotational grazing, native versus introduced pastures, and perennial versus annual plants. Dr Yin says none of those methods increased the amount of carbon taken up by soil in paddocks. He says adding phosphorous fertiliser was the only significant way to improve pasture production and soil carbon.”
He said no such thing. He was caerfulgt to say: "Actually after we got all the data and compared them statistically there's only one treatment we can find a statistical difference and that's due to P fertilisers," says Dr Yin Chan.
Dr Chan was not ‘debunking’ anything.

1 comment:

willem said...

with all due respect. I think you are misintepreting the figure. It is still possible that is a mock-up.
What you are looking at are the values on the contour lines, there are actually no real data points in the plot, although they are suggested by the "bomb craters". The contour lines might suggest the same values, but they are the result of the spatial integration that was used to make the plot and thus represent some sort of average between points.

I think in general the person who made the plot needs to learn a bit more about spatial statistics to make the plot more believable.