1. A 'values' voter speaks her mind on Obama

Comment #267188 by jt512 on October 20, 2008 at 10:26 am

These are the morons who got Bush elected twice, and who have pushed the United States to the brink of theocracy; and, if it weren't for the current financial crisis, would probably get McCain/Palin elected in the upcoming election. If it weren't for this financial crisis, it is not clear whether secular democracy in the US would survive the next decade.

2. Hail, ceaseless complexity: Review of 'Reinventing the Sacred'

Comment #256058 by jt512 on September 28, 2008 at 5:23 pm

The bit about a quantum Santa being everywhere at once reminds of this joke:

Q. What's the difference between a quantum mechanic and an auto mechanic?

A. A quantum mechanic can get his car into the garage without opening the door.

/nerd-humor

3. Look Who's Irrational Now

Comment #250526 by jt512 on September 19, 2008 at 8:47 pm

...traditional Christian religion greatly decreases belief in everything from the efficacy of palm readers to the usefulness of astrology.

Viewed differently, the same statistics suggest that belief in astrology and palm reading greatly reduces traditional Christian religion. If it's a choice, I'll take the astrologers. At least they're not trying to take over the world.

4. Autism and Vaccines: Why Bad Logic Trumps Science

Comment #245299 by jt512 on September 10, 2008 at 1:42 pm

"Apathy personified" wrote:

'Why don't we ever see quoted the incidence level of autism since the MMR jab incidence level declined?'

Answer -
Because it would just be a number, a statistic, from which we can't infer anything useful to serve our research purposes.

(Please correct me if i'm wrong)

Well, that's why a knowledgeable epidemiologist would not quote such a statistic. Unfortunately, those kinds of crude statistics get reported by the press and others who don't understand the limitations of them.

I am now very curious as to how an epidemiologist would look for a link between say, a certain vaccine and autism - Any info on that would be much appreciated.

One way is to conduct a cohort study, along the lines of the following:

1. Identify a relatively homogeneous population of children that includes a sufficient number of vaccinated and non-vaccinated subjects for statistical analysis.
   
   
2. For each child in the cohort, obtain their vaccination status, whether they developed autism, and information on other variables, like age at vaccination, mother's age at child's birth, etc., which could confound the relation between vaccination and autism.
   
   
3. Calculate the crude relative risk (RR) by dividing the autism rate in the the vaccinated group by the autism rate in the non-vaccinated group (so, an RR > 1 would indicate that autism incidence was higher in the vaccinated group).
   
   
4. Calculate an adjusted relative risk by using statistical techniques to adjust the crude relative risk for the potential confounding variables on which data were collected.
   
   
5. Test the adjusted RR for statistical significance.

Madsen et al (2003) conducted such a study among 537,000 children in Denmark. They found the incidence of autism and autistic-spectrum disorder was non-significantly lower in vaccinated children than in non-vaccinated children (RR = .92 and .83 for autism and autistic-spectrum disorder, respectively).

The following is a link to the abstract to the paper, from which you can link to the full text if you're interested.

http://tinyurl.com/madsen-2003

5. Autism and Vaccines: Why Bad Logic Trumps Science

Comment #244921 by jt512 on September 9, 2008 at 7:53 pm

ljirving wrote:

As a PhD biologist myself, I think Apathy's idea is a very good place to start

As an epidemiologist, I wouldn't expect a biologist (or someone in any of the natural sciences) to understand the practice of epidemiology. However, I would hope that you would be as open to a brief introduction to my field of study as I would be to yours.

Actually, as you say, a study such as that which AP has suggested might be a good place to "start," but we are no longer at the start of investigating the vaccination--autism hypothesis: over a dozen studies have already been published, some with far more rigorous methodologies than the study that AP is suggesting. As I said before, it doesn't make sense to follow up stronger studies with weaker ones; there is little to gain, and potentially a lot to lose. I'll get to the latter point in a bit. First we need to understand why the study that AP suggests is so poor.

What AP is suggesting is a type of study known in epidemiology is an ecologic study. The design is fraught with intractable problems. Like any observational study, raw associations observed in an ecologic study could be attributed to nuisance variables, or confounders, rather than to the exposure under investigation. However, in ordinary epidemiologic studies, confounders can potentially be controlled using statistical techniques. In ecologic studies they cannot because, by definition, measurements in an ecologic study are taken on groups of people, rather than on individuals. The consequences of this are subtle, but profound. Say, for example, that the study finds that the incidence of autism decreased when the incidence of vaccination decreased. And say, further, that in fact that this was true; that is, that the result was unconfounded at the group, or ecologic, level. At first it would seem that we could say, then, that children who were not vaccinated had a lower incidence of autism than those who were vaccinated. But oddly enough you cannot validly infer that from the data. For all we know, the incidence of autism actually declined in the vaccinated children. Mistakenly believing that an association between two variables that holds for grouped data implies the same association at the individual level is known as the ecologic fallacy. I briefly mentioned it in a previous post.

There are many examples of erroneous findings in the scientific literature, owing to this very misinterpretation of ecologic studies. Often they become widely reported in the news and on the internet because they seem so startling. One notorious example was a study that found that the greater the amount of dairy products a population consumed the greater was their incidence of osteoporosis. This suggested that milk, among the richest sources of calcium in our diet, might actually cause osteoporosis. However, the majority of studies that have examined the relation between dairy consumption and osteoporosis at the individual level have found the opposite to be true. Nonetheless, the erroneous results from the ecologic study continue to be widely cited on the Web and in popular books by vegan activists and wacko anti-milk fanatics.

Let's get back to AP's proposed study. Because of the inherent weaknesses of the methodology, no matter what the results of the study might be, we really wouldn't know how to interpret them. If the results, say, showed a decline in the incidence of autism, we would have no way of knowing whether that decline was actually due to the reduction in the vaccination rate. Nonetheless, the results would get reported in the popular press and would convince some parents who otherwise would have vaccinated their children not to. This is why it is important to avoid conducting bad epidemiologic studies. Bad studies by their very nature have negative public health ramifications.

Thanks for listening.

6. Autism and Vaccines: Why Bad Logic Trumps Science

Comment #243801 by jt512 on September 7, 2008 at 11:50 am

AP wrote:

An article mentions a debate about whether there's a correlation between two things; one of them has changed, i ask if the other one has also changed (although in this case i expect it hasn't) - that's all it was, just a simple question.

No, the article does not mention a debate about whether there is a correlation between vaccination and autism. It says that correlation does not prove causation; that well-controlled studies have shown that vaccination does not cause MMR.

You then ask why the author did not present statistics on whether there has been a change in the incidence of autism following a decrease in the incidence of vaccination. I think that my second sentence in the preceding paragraph answers that question. Do you see that such a statistic would be exactly the type of datum that the article is arguing against? Presenting such a statistic would be inconsistent with the author's argument.

7. Autism and Vaccines: Why Bad Logic Trumps Science

Comment #243792 by jt512 on September 7, 2008 at 11:18 am

Bonzai,

You have grossly misread AP's post. He was not suggesting that controlled experiments be run, but rather, that the observed incidence of autism be compared with the observed incidence of MMR vaccination. That is not only correlational data, it is worse: it is correlation between summary statistics, which compounds the temporal fallacy with the ecologic fallacy (that correlation between summary statistics is due to correlation at the individual level).

8. Autism and Vaccines: Why Bad Logic Trumps Science

Comment #243775 by jt512 on September 7, 2008 at 10:07 am

"Apathy personified" wrote:

[I wrote:]

It seems to me, that you have just fallen for the same logical fallacy discussed in the article.

How? In no way have i said i agree with the morons who think that the MMR jab causes autism.... I was merely asking why nobody mentions the incident rate of autism since the number of people taking the jab has decreased... i think it would be interesting to know.

It might be "interesting," but originally you called it a "crucial fact." In fact, it is not crucial because, as the article explains, "[T]wo events that appear to be causally related may not be; there are other possibilities. The appearance of causation may simply be coincidence.... Only carefully controlled scientific studies can conclusively tell the difference."

That is the crucial fact, and a major point of the article. It would require carefully controlled studies to determine the relevance of any observed relation (or lack of relation) between the incidence of autism and the incidence of MMR injection, so what would be the point in reporting it?

On the other hand, epidemiologists have already conducted carefully controlled studies that have shown that the vaccine does not cause autism. But you are suggesting that weak correlational evidence be presented to support a conclusion arrived at by strong, controlled studies. That contradicts a major point that the article is trying to convey.

Jay

9. Autism and Vaccines: Why Bad Logic Trumps Science

Comment #243755 by jt512 on September 7, 2008 at 9:16 am

Apathy personified wrote:

Out of pure curiousity,
Has the incidence level of autism varied with the reduction in the number of MMR jabs given?

That seems like a crucial fact, but it doesn't seem to be mentioned.

It seems to me, that you have just fallen for the same logical fallacy discussed in the article.

10. Imagine No Religion' signs to go up around town

Comment #239031 by jt512 on August 28, 2008 at 10:40 pm

Cool. Are you still in California? What kind of work do you do?

I have escaped the IE, but still live in the greater Los Angeles area. I'm a biostatistician.

11. Imagine No Religion' signs to go up around town

Comment #238992 by jt512 on August 28, 2008 at 9:12 pm

168. Comment #238239 by CocoCantare on August 27, 2008 at 7:52 pm
jt512,

Fancy that. . . it's a very small world! What did you study at CSU Fullerton? Where did you go to high school?

I studied biology and biochemistry at CSUF in preparation for grad school. I went to high school in Michigan.

12. Imagine No Religion' signs to go up around town

Comment #238196 by jt512 on August 27, 2008 at 6:21 pm

Coming late to the SoCal Inland Empire reunion from Page 1. I lived in Loma Linda and Riverside for six (long) years, and got a masters degree in nutritional epidemiology from Loma Linda Univ. I still do some consulting work for them. And for Coco: I also attended Cal State Fullerton for 2 years.

13. The Boundaries of Belief

Comment #205142 by jt512 on July 6, 2008 at 4:42 pm

Sam Harris wrote:

Pew's sample of 35,556 Americans included 515 respondents who identified themselves as "atheists" (1.6 percent). The margin of error for this subgroup appears to be around 5 percent �" which clearly makes a hash of many of the above findings.

Whatever is responsible for the survey indicating that a small percentage of self-described "atheists" have strong religious beliefs, it isn't the so-called "margin of error." The "margin of error" is an oversimplified way that public opinion pollsters have of expressing statistical uncertainty. It is actually the 95% confidence interval of 50%. That is, if 50% of respondents give a particular answer to a question, then the 95% confidence interval is 50% ± 5%. However, as results diverge from 50%, the actual 95% confidence interval narrows and becomes increasingly asymmetrical. Applying the pollsters' reported "margin of error" to extreme percentages results in gross errors, especially when the (sub)sample of respondents to which the percentage applies is small. For example, for the 3% of self-defined "atheists" who reported being "absolutely certain" that a personal god exists, the actual 95% confidence interval is 1.6%-4.8%.

My first reaction upon reading this article was that the seemingly contradictory beliefs among a small percentage of "atheists" are likely due to respondents making errors or deliberately giving misleading answers. However, the contradictions are even more numerous among self-defined "agnostics," 17% of whom report being "absolutely certain" about believing in God. Thus there seems to be real heterogeneity in religious beliefs among people who describe themselves as atheists or agnostics. It would be interesting to see more in-depth research into what these ranges of beliefs are. In fact, until we have a better understanding of this, interpretation of research on, for instance, trends in religiosity might be difficult: if surveys show increasing numbers of atheists, does that mean that fewer people are believing in god, or more believers are calling themselves atheists?