Day 034 - 11 Oct 94 - Page 11
1 positive association is right, a sceptic about association
2 of fat just cannot win, can he, because wherever there is
3 no association you say that does not help at all, and
4 wherever there is association you say: "Look, that
5 supports what I am saying"?
6 A. A sceptic, I think, could win if the great bulk of
7 studies showed no association whatsoever. If the animal
8 studies showed no association and if there were no
9 plausible biological mechanisms that had been established,
10 then I would say the sceptics had the weight of evidence
11 on their side.
12
13 Q. But only, in your mind, if that were so?
14 A. Yes.
15
16 MR. MORRIS: If, presumably, it is a question of the positive
17 associations that have been found were somehow by chance,
18 presumably, there would be as many studies that maybe
19 showed a negative association, if there was really no
20 association?
21 A. It is impossible to have this many positive findings
22 by chance. It is possible to have one in 20 studies could
23 be found by chance to be positive or to be negative. The
24 reason I use the No. 1 in 20 is that when statistical
25 significance is reported, for example, any one of these
26 positive marks, the first one Lea 1966, to report a
27 positive association between total fat and breast cancer
28 and to report that is statistically significant; what they
29 mean is they have analysed the data, they have done
30 statistical tests on them to show that 19 times out of 20,
31 or 95 per cent of the time, that could not have occurred
32 by chance alone; but that one time in 20 it is just
33 possible that that could have happened by chance; that
34 just that many more people happened by pure chance to be
35 consuming lots of fat and to end up with cancer; but in 19
36 out of 20 cases that would not occur by chance alone.
37
38 The reason I mention that is that if you do enough studies
39 you will find some that by chance alone are positive or
40 are negative. As we have seen, there are a couple of
41 negatives on this chart. The more studies one does the
42 more of those there might be, even if, in fact, that is
43 simply a fluke. One would expect about one in 20 studies
44 to show nothing other than a chance relationship.
45
46 MR. MORRIS: Those two -- well, yes, okay.
47 A. I am not saying that was the case in those studies.
48 I am saying that about one in 20 studies you can write off
49 to pure chance. However, it is impossible to write off
50 all of these studies. This is overwhelming, very
51 convincing evidence of a relationship.
52
53 MR. JUSTICE BELL: What I would like to do, Mr. Morris, unless
54 it is inconvenient to you, if you go through the cohort
55 studies and ask whatever questions you want to ask
56 Dr. Barnard of that; then you have been through the whole
57 table. Then, might I suggest, that you ask Dr. Barnard if
58 this were an accurate bird's eye view, what it meant to
59 him. He may have already expressed his view in the last
60 answer, I do not know. Then I can compare it with the
