Do-It-Yourself China Study

Using the 1989 China Study data, see the association of food choices with mortality from a variety of causes. For example does increased consumption of red meat (D050) appear to be associated with increased death by cancer (M023)? Observe the green trendline. If the trendline slopes upward, then death due to cancer is more frequent among those that consume more red meat. If it slopes downward, then those consuming more red meat experienced less death.

Select a dietary (Dxxx), blood test (Pxxx), or lifestyle (Qxxx) item:

Select a mortality cause:

Ignore Null Data (recommended)

1) Notice the use of the word “associated with” rather than “causes.” Using this data, one can establish that red meat is associated with greater longevity. This doesn’t mean red meat was the cause. It could be that people who could afford red meat got better medical care. There could be other factors too. So nothing like “this causes that” is proven here. However, to make a statement like “The China study proves that animal product is harmful,” as was claimed by the book of the same name, is ludicrous. The raw data, as the reader can see, indicates strongly that the opposite is far more likely to be the case.
To arrive at a conclusive answer requires more elaborate statistics, and such elaboration is subject to mischief as well.
We take it as a tenet that if the indications of the raw data are to be contradicted by some elaborate statistics, the argument needs to be very convincing.

2) Checking the “Discard Null Data” checkbox is recommended. It is often the case that data that is simply unavailable is recorded as zero, and using such data creates a false trend. Try it both ways.

3) Common sense. Any plot that has only a few number of points, or has the bulk of them lying on the horizontal or vertical axis is probably suspect. If the data forms s narrow vertical structure (eg. D020 & M006) the trend line is probably not meaningful.