Russell McMahon wrote:

> IPCC speak table (courtesy some PICList poster) which the above 
> rapporteur was (presumably) using.
> 
>> Virtually certain > 99% probability of occurrence,
>> Extremely likely> 95%,
>> Very likely > 90%,
>> Likely > 66%,
>> More likely than not > 50%,
>> Unlikely < 33%,
>> Very unlikely < 10%,
>> Extremely unlikely < 5%.
> 
> This is utter travesty.
> P=0.5 is so close to the mean it matters not a whit.
> P=0.33 is one standard deviation out.

Russell, I think you may be confusing probabilities with confidence levels 
-- if I understand you correctly (and if I understand confidence levels 
correctly :). Note that they also have a suggested wording for confidence 
levels, which is different from the above.

For probabilities (not confidence levels), 50% does /not/ mean it's close 
to the mean. If I can give a probability of 50% for rain this afternoon, 
this is a meaningful data; the mean of rain or not (here at least) is quite 
different, and even if it were not different, the prediction that this 
afternoon the probability is close to the mean does add information.

There's nothing that says how the events these probabilities describe are 
distributed and what the mean is. (In fact, these number are exactly what 
the models are supposed to calculate.)

However, this probability doesn't yet say what the confidence level of such 
a prediction is. And I think that's where things get interesting, as the 
IPCC doesn't state a single confidence level with any of its probability 
predictions.

Gerhard

-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist