Re: Are there any non-smokers here in favor of allowing smoking in bars and restaurants?

Jim Carr <newsgroups@xxxxxxxxxxxxxx> wrote in
news:8ODaj.31053$1C4.13198@xxxxxxxxxxxx:

JoeSpareBedroom wrote:

If that's as good as the data can get, then what's the best thing to
do? Ignore them?

I don't mean any offense, Joe, but you ought to read up on statistics,
specifically confidence intervals. We've strayed away from the whole
smoking ban subject and are really discussing how the data is
analyzed.

Once you understand how it's done, you'll realize that people like
Aaron, Rich, Les, and myself really aren't trying to say that the
theory that smoking causes cancer is bunk. Far from it. Most of us
would like to engage in a reasonable discussion where we all
understand what the data means beyond the mantra, "it must be bad for
you!"

I think people are confusing discussions about the "how do we know" as
arguments against the idea the smoking causes disease.

Here's an example of dealing with statistics. The chances of having
identical twins is about 1 in 250 live births and is pretty consistent
worldwide. Let's accept that number as fact.

Suppose there's an isolated set of villages in Bumphuck, Nowhere. They
have about 1,000 births per year. For 10 years before and ten years
after our study, they had exactly a 1 in 250 rate of identical twins.

At the start of our three year study we expose half the child-bearing
age women with the virus that causes Runnington's Disease (getting
pissy every month). At the end of three years, we give them the
antidote (a pack of Camel Lights).

During the three years the exposed women had 10 sets of identical
twins instead of the expected 6. The other group had the expected 6.
How sure can we be that Runnnigton's Disease had an effect? What if
there were 20 in the virus group? 30?

When you know how to approach and form a response, you'll get at least
what I know I've been driving at and what I think others have been
driving at.


It's been a long time since I took a course in statistics and learned all
about confidence intervals. It's also been a long time since I took a
course in English grammar. Nonetheless, as I recall, it's "data are
analyzed" not "data is analyzed." While it's not particularly important,
it's something that an arrogant, uneducated, anal-retentive jackass like
yourself should not have missed,. It's also something you may want to
correct if/when you get around to publishing your dissertaion on the
subject.
.

Relevant Pages

  • Re: Are there any non-smokers here in favor of allowing smoking in bars and restaurants?
    ... I don't mean any offense, Joe, but you ought to read up on statistics, specifically confidence intervals. ... Once you understand how it's done, you'll realize that people like Aaron, Rich, Les, and myself really aren't trying to say that the theory that smoking causes cancer is bunk. ... The chances of having identical twins is about 1 in 250 live births and is pretty consistent worldwide. ... At the start of our three year study we expose half the child-bearing age women with the virus that causes Runnington's Disease. ...
    (alt.guitar.bass)
  • Re: TO COMPLETE
    ... Quantiles of test statistics "confidence intervals" ... if repeated daily only shows what Jack Tomsky ... Afonso again repeats his ignorance that the sample space has to be an interval. ...
    (sci.stat.math)
  • Re: Confidence Intervals for Bridge Simulations (long)
    ... range are impossible -- that's cookbook statistics. ... Grant me some sort of prior probability distribution assumption as to ... that's consistent with a population frequency as low as ... kind of prior distribution information to make confidence intervals ...
    (rec.games.bridge)
  • Re: Determining Confidence Interval
    ... > occurrences), how large a sample of total occurrences will be required ... > confidence that this result reflects a consistent trend or would I ... Confidence intervals are appropriate when you have ... are way off the mark for what statistics can tell you. ...
    (sci.stat.math)
  • Re: resampling methods are serious procedure?
    ... mentioned it right before you described sampling. ... However, you are teaching an approach to statistics, however ... I apologize for my hostile tone. ... I'm still not lined up with you re: confidence intervals. ...
    (sci.stat.math)