The China Study by Thomas Campbell

40 THE CHINA STUDY

cancer. A classic illustration of this difficulty is that countries with more
telephone poles often have a higher incidence of heart disease, and
many other diseases. Therefore, telephone poles and heart disease are
positively correlated. But this does not prove that telephone poles cause
heart disease. In effect, correlation does not equal causation.
This does not mean that correlations are useless. When they are
properly interpreted, correlations can be effectively used to study nu-
trition and health relationships. The China Study, for example, has
over 8,000 statistically significant correlations, and this is of immense
value. When so many correlations like this are available, researchers
can begin to identify patterns of relationships between diet, lifestyle
and disease. These patterns, in turn, are representative of how diet and
health processes, which are unusually complex, truly operate. However,
if someone wants proof that a single factor causes a single outcome, a
correlation is not good enough.

STATISTICAL SIGNIFICANCE

You might think that deciding whether or not two factors are correlated
is obvious-either they are or they aren't. But that isn't the case. When
you are looking at a large quantity of data, you have to undertake a sta-
tistical analysis to determine if two factors are correlated. The answer
isn't yes or no. It's a probability, which we call statistical significance. Sta-
tistical significance is a measure of whether an observed experimental
effect is truly reliable or whether it is merely due to the play of chance.
If you flip a coin three times and it lands on heads each time, it's prob-
ably chance. If you flip it a hundred times and it lands on heads each
time, you can be pretty sure the coin has heads on both sides. That's the
concept behind statistical Significance-it's the odds that the correlation
(or other finding) is real, that it isn't just random chance.
A finding is said to be statistically Significant when there is less than
5% probability that it is due to chance. This means, for example, that
there is a 95% chance that we will get the same result if the study is
repeated. This 95% cutoff point is arbitrary, but it is the standard, none-
theless. Another arbitrary cutoff point is 99%. In this case, when the
result meets this test, it is said to be highly statistically significant. In the
discussions of diet and disease research in this book, statistical signifi-
cance pops up from time to time, and it can be used to help judge the
reliability, or "weight," of the evidence.