Listen to the podcast, or read the transcript.
http://fivethirtyeight.com/features/a-data-detective-story-did-a-british-nurse-kill-his-patients/

“I seem to have a jinx,” says Benjamin Geen, a former nurse who is in prison serving 17 life terms after being convicted of two murders and of causing grievous bodily harm to 15 other people. In the winter of 2003-04, more than a dozen patients admitted to Horton General Hospital in Banbury, England, suffered severe respiratory failure. Geen was a nurse on duty during all of the incidents, and that statistical anomaly set off an investigation that ended in his imprisonment. Geen maintains his innocence, and some statisticians who’ve investigated the case claim that, while unlikely, the pattern of deaths does not inherently point to criminal behavior.

—-

dv said:

Listen to the podcast, or read the transcript.
http://fivethirtyeight.com/features/a-data-detective-story-did-a-british-nurse-kill-his-patients/

“I seem to have a jinx,” says Benjamin Geen, a former nurse who is in prison serving 17 life terms after being convicted of two murders and of causing grievous bodily harm to 15 other people. In the winter of 2003-04, more than a dozen patients admitted to Horton General Hospital in Banbury, England, suffered severe respiratory failure. Geen was a nurse on duty during all of the incidents, and that statistical anomaly set off an investigation that ended in his imprisonment. Geen maintains his innocence, and some statisticians who’ve investigated the case claim that, while unlikely, the pattern of deaths does not inherently point to criminal behavior.

—-

If in doubt convict anyway…

The transcript is truncated, which makes it rather useless.

sibeen said:

The transcript is truncated, which makes it rather useless.

Well then listen to the podcast.

dv said:

sibeen said:

The transcript is truncated, which makes it rather useless.

Well then listen to the podcast.

Ain’t no one got time for dat.

What’s your verdict dv?

academic statistical opinion, however distinguished, is divorced from the actual facts

SCIENCE said:

academic statistical opinion, however distinguished, is divorced from the actual facts

What p value do you give that statement?

SCIENCE said:

academic statistical opinion, however distinguished, is divorced from the actual facts

That must make compiling statistics easy.

Bubblecar said:

What’s your verdict dv?

Without access to the raw data, I could not possibly offer independent opinion on this particular case. It would be odd if so many respected statisticians were all wrong about the fact that the use of statistics in this case was not appropriate: but of course, as they would admit, very odd things happen sometimes.

It does, however, seem anomalous that someone could be found guilty of murder without firmer evidence than this.

jiciwnc, it wasn’t not a quote from the transcript

I didn’t fail to not unknow that

dv said:

It would be odd if so many respected statisticians were all wrong about the fact that the use of statistics in this case was not appropriate: but of course, as they would admit, very odd things happen sometimes.

If the statistics include the application an assumed standard distribution to the tail of a real distribution, then that alone should be sufficient to have that evidence discarded.

Apart from which, if you have a huge number of samples, surely you must expect some very unlikely happenings to be found amongst them.

The Rev Dodgson said:

dv said:

It would be odd if so many respected statisticians were all wrong about the fact that the use of statistics in this case was not appropriate: but of course, as they would admit, very odd things happen sometimes.

If the statistics include the application an assumed standard distribution to the tail of a real distribution, then that alone should be sufficient to have that evidence discarded.

Apart from which, if you have a huge number of samples, surely you must expect some very unlikely happenings to be found amongst them.

Well their analysis took that into account.

SCIENCE said:

jiciwnc, it wasn’t not a quote from the transcript

What wasn’t not a quote from the transcript?

dv said:

The Rev Dodgson said:

dv said:

It would be odd if so many respected statisticians were all wrong about the fact that the use of statistics in this case was not appropriate: but of course, as they would admit, very odd things happen sometimes.

If the statistics include the application an assumed standard distribution to the tail of a real distribution, then that alone should be sufficient to have that evidence discarded.

Apart from which, if you have a huge number of samples, surely you must expect some very unlikely happenings to be found amongst them.

Well their analysis took that into account.

Part 2?

OK, ignore Part 2. What about the application of assumed standard distributions?

Are you telling me the statisticians didn’t do that?

The Rev Dodgson said:

dv said:

The Rev Dodgson said:

If the statistics include the application an assumed standard distribution to the tail of a real distribution, then that alone should be sufficient to have that evidence discarded.

Apart from which, if you have a huge number of samples, surely you must expect some very unlikely happenings to be found amongst them.

Well their analysis took that into account.

Part 2?

OK, ignore Part 2. What about the application of assumed standard distributions?

Are you telling me the statisticians didn’t do that?

No they did that.

I mean I’m not disagreeing with you here…