"There are three kinds of lies: lies, damned lies, and statistics."

-- Benjamin Disraeli


--- In [email protected], "authfriend" <[EMAIL PROTECTED]> wrote:
>
> --- In [email protected], TurquoiseB <no_reply@> wrote:
> >
> > > > The real problem with the study is the design itself.
> > > > If it had a better design than a simple pre-post
> > > > (which  makes no sense for research of this sort) non
> > > > of these question would be discussed.
> > > 
> > > Are you saying the 36-hour hike in the murder rate
> > > *was* an anomaly and that it *was* legitimate not
> > > to take it into account?
> > 
> > Isn't discounting a large surge in the murder rate during 
> > the period that crime was being measured
> 
> It was not "a large surge in the murder rate during
> the period that crime was being measured."  It was a
> spike occurring during a 36-hour segment of that
> period (as I said).  The immediately following week,
> while the course was still going on, there were far
> *fewer* murders than normal, so the average number of 
> murders per week over the duration of the course
> remained the same as usual.
> 
> > a lot like saying:
> >  
> > "The IA course has successfully created a lasting state of
> > peace, worldwide. We have not counted Iraq, Afghanistan, 
> > Darfur, Chad, Sudan, Western Sahara, Somalia, Nigeria, and 
> > Chechnya because they are anomalies."
> 
> Even overlooking the fact that certain kinds of
> anomalies are, indeed, statistically insignificant
> (as the TM researcher new morning cited who was
> defending the study pointed out, this was such
> a case, given the small total number of murders
> in proportion to the *much* larger total number
> of violent crimes whose rate was being studied),
> no, the spike in the murder rate isn't at all like
> what you say.
> 
> The D.C. study did not claim to have successfully
> eliminated violent crime in D.C. on a permanent
> basis; it claimed to have been responsible for a
> temporary overall decline in the total number of
> incidents of violent crime compared to what would
> have been expected for that period if the course
> had not taken place, and it *did* count the spike
> in the number of murders per week.
> 
> The claim is that the spike, given the very
> small percentage of incidents of violent crime
> that murders always represent (there are over
> 10 times fewer murders than there are assaults),
> was not significant with regard to the overall
> decline in the number of violent crimes during
> the study period.
> 
> This was in response to the ignorant claim by a
> critic that the spike completely invalidated the
> results of the study.




Reply via email to