Day 036 - 13 Oct 94 - Page 19
1 mentioned.
2
3 Q. In which table?
4 A. We were just looking at line 2 of table 4?
5
6 Q. Lines 1 and 2.
7 A. OK.
8
9 MR. JUSTICE BELL: Explain your point in relation to saturated
10 fat.
11 A. Without belabouring it, the P value refers to the
12 probability that the result is by chance alone. A high
13 number there is suggestive that it was simply chance.
14
15 MR. RAMPTON: A high number?
16 A. Yes.
17
18 MR. JUSTICE BELL: Yes, because it would be, for instance,
19 there would be a -- we have had lots of Ps of .05?
20 A. Yes, .05 or .01.
21
22 Q. So, there is a 5 per cent chance that it is just chance
23 rather than any factor at all?
24 A. That is exactly right.
25
26 Q. You see .71 you are saying there?
27 A. Yes, .71; that implies that this is probably a chance
28 association, or not necessarily chance but that simply the
29 study was not designed to answer that question in a
30 meaningful way. It does not mean it is a poorly designed
31 study. It simply means it does not respond to that
32 question.
33
34 MR. RAMPTON: Yes, but you get a P value of .29 for the total
35 fat residual in the same table, do you not?
36 A. Yes, you do.
37
38 Q. Can I ask you to look at ----?
39 A. But probably the only one of these -- just a quick
40 glancing over these, the only one than would be considered
41 statistically significant by which most authorities would
42 say a P less than or equal to .05 appears in table 3 from
43 polyunsaturate.
44
45 MR. RAMPTON: Yes, exactly. You will see, when we look at the
46 discussion, that the authors of this report recognise some
47 of these difficulties. The discussion begins on 1095,
48 which is two pages before the tables, those tables
49 anyway. It says: "Our results indicate a nonsignificant
50 positive association between dietary fat and risk of
51 postmenopausal breast cancer. In some analyses, the
52 strongest association was seen with polyunsaturated
53 fat".
54 A. I am with you.
55
56 Q. I will read it again. The discussion begins at the bottom
57 of the left-hand column: "Our results indicate a
58 nonsignificant positive association between dietary fat
59 and risk of postmenopausal breast cancer. In some
60 analyses, the strongest association was seen with
