tcarter2 at elp.rr.com
Wed Aug 13 13:48:06 EST 2003
Foremost allow me to apologize if I upset you. I know how
easily we humans can be irritated by opposing opinions, and I try to
avoid it. But I guess I'll have to try harder. I'd like to continue
our discussion in a dispassionate, scientific manner. If you can't,
feel free to cuss away. I have a thick skin, and a short memory for
such things. :>) And I believe I understand the emotions involved. I
also invite others to weigh in. After all we are discussing a study
which shows a rare life extension of about a year and a half by a
common food, in a cohort of reasonable size, (290) as well as
attenuation of brain aging after age 71, but is suspected of
accelerating brain aging. Or in other words a matter of some import. I
will snip freely, repost anything you wish.
> >At the very bottom of the table they give scores of
> >the intelligence test. (CASI) It turns out that the failures of the
> >low consumption group are only 2.0%. This is an impossible number
> >since the denominator of the calculation is 38, the group size. If
> >unadjusted it must clearly be zero, or some multiple of 2.6%,
> >corresponding to failures of 1,2, 3 individuals etc. divided by 38.
> >The paper says these values were adjusted for age, and education, (why
> >only these factors among the many available?) but such an adjustment
> >would leave the values at zero or raise them to some value higher than
> >2.6%, probably, judging from similar adjustments in the study to about
> >four or five, which would take away almost all of the impact of tofu
> >consumption. "Mistakes" like this are almost always indicative of data
> >doctoring IMO.
> IAN: It says, "CASI score adjusted for age (single year) and
> years of schooling completed." As such, it need NOT be 2.6%.
> So you base the charge of "data doctoring" on your assumption
> that the figure "probably" should have adjusted to some number
> other than 2%.
No, I did not say improbable, I said impossible. This is matter that
can be resolved by simple math and need not be disputed among
intelligent people. 1/38=2.6%. The low cohort of 38 was younger, and
had more education, so the stated adjustments could only have raised
the 2.6%, as I stated.
What a flimsy basis for such a harsh accusation!
If you call the accusation harsh, what you call the actual deed?
> > The hypothesis of the authors is not that the tofu consumption
> >in the cohort is associated with lifelong lower intelligence, but with
> >"accelerated brain aging". Figure two clearly shows them wrong on both
> >counts. The median age at testing was 80, putting the first two bar
> >graphs below the medium, and the last three above. A quick glance
> >shows that the younger members of the hi-hi tofu group failed at about
> >four times the rate of the low-low tofu group. As the members get
> >older this failure rate is gradually reduced to near equality.
> >Comparisons of the other two groups show the same trend but not as
> >significantly. ACCORDING TO THIS STUDY TOFU ATTENUATES BRAIN AGING
> >BETWEEN AGE 71 AND 93.
> IAN: That's a cross-sectional, not longitudinal, analysis,
> so you can't really say, "As the members get older this
> failure rate is gradually reduced," since each stratified
> group represents a different set of people. The important
> point is the sixth grouping showing the overall relation:
Are you quibbling about my syntax? I will restate it for you more
formally, but the conclusion is the same. The low-low group members
who died just after age 71 failed at one fourth the rate of the hi-hi
group members who died just after age 71. BUT the low-low group
members who died just after age 90 failed at nearly the same rate as
the hi-hi members who died just after age 90. The obvious presumption
is that at age 71 the low-low group members were more intelligent, but
those low-low members who survived to age 90 were only of equal
intelligence with the hi-hi group members. Thus between the ages of 71
and 90 the intelligence of the longer lived low-low members declined
relative to the intelligence of the hi-hi members.
This suggests that tofu attenuated brain aging.
> >Why were the hi tofu eaters so much less
> >intelligent at age 71-75 than the low group? They went to grade school
> >in Japan. People always (groups, not individuals) demonstrate a
> >relative lack of intelligence when dealing with written or oral tests
> >in a second language.
> IAN: But the study states: "The CASI has been validated as
> a screening instrument for dementia in the United States
> and Japan, in both English and Japanese languages. [...]
> participants selected either Japanese or English for testing,
> depending on the language with which they were most comfortable."
> So your conclusion that scores were biased by native education
> stands in direct opposition to information provided in the study.
No, people who switch from one language to another are less
comfortable in either language, but both my point and yours are
irrelevant to the question. The relevant point is that table 4 shows
that the wives who went to grade school in Japan failed at a rate of
270% higher than those who didn't. The fact that they had their choice
of languages for the test can not negate (and in fact doesn't even
relate to) this OVERWHELMINGLY large factor which was not applied to
> >A most pertinent example of this is the wives of
> >the cohort. Those who went to grade school in Japan failed the CASI
> >test at a rate of 2.7 times those who didn't (table 4).
> IAN: That's irrelevant due to your oversight noted above.
No, Ian, the relevance is that if the wives failed at a rate of 270%
because of their schooling, we should expect the husbands who went to
grade school in Japan to also fail at somewhere near the same
percentage because of their own schooling.
> >THIS HUGE
> >MULTIVARIENT FACTOR WAS NOT APPLIED TO HUSBANDS. Had it been, it would
> >have eliminated the difference in the CASI scores of the husbands at
> >age 71, and would have destroyed the thesis of the paper.
> IAN: That's a false assumption based on your oversight above.
About 20% more of the hi-hi group compared to the low-low group went
to grade school in Japan. If these men failed at just 200% the rate,
this would have tripled one fifth of the failings due to schooling in
Japan which would have added 60% to the total number of failings and
all but wiped out the 62% advantage of the low-low group in table 2
relative to the hi-hi group.
> Frankly, I just don't have time to respond point-by-point.
> Given that you're asking us to believe your personal analysis
> and thereby doubt eight expert epidemiologists and their peer-
> reviewers, it's fair to ask: exactly what are your credentials?
> I don't care for appeals to authority, but regarding complex
> statistical data analyses, expertise really means something.
No, I'm asking you and others to read, and follow my math and logic. I
don't think it's too complex for many of the group with a bit of
> One last comment:
> > Contrary to the CASI data the relationship with brain size was
> >most pronounced in the hi-hi group, being totally insignificant
> >between the other three groups. I would guess that this was caused by
> >lower growth due to the fact that soy protein is lacking in
> >methionine, a deficiency of which is known to slow growth rates, but
> >only at hi levels of deficiency. The high-high group, which show
> >lowered growth would have been deficient to at least some extent in
> >this amino acid, the restriction of which is also shown to extend
> IAN: So there IS a relation to smaller brain size and tofu
> (which was the primary focus of my post), but alas you've
> determined that it is a result of methionine deficiency.
But not brain atrophy, nor accelerated brain aging. At least not shown
by this paper. Some of the animal and in vitro data is quite
> http://www.jacn.org/cgi/content/full/19/2/242/F3 You know,
> you could broach your comments as hypotheses and questions,
> rather than, as you have, accusations of "data doctoring" and
> firm conclusions of "fatal mistakes" and such by the authors
> that you use to caste sweeping aspersion on peer-review per se.
> Moreover, you should locate and ask the authors directly before
> posting public accusations against their expertise and honor.
Actually I've always emailed a copy to any author who has a currently
valid address in their Pubmed listing. Most don't, and none have
replied. All humans, including myself, have a difficult time accepting
and dealing logically with unsettling facts. We can all profit by
recognizing and dealing with this "design flaw" to the best of our
More information about the Neur-sci