Originally posted by

**caz**View PostI believe you mentioned in the past that he was a very accomplished chap, that's why I was confident that he could guide you through the obvious logical errors of Ike's statistical analysis.

Of course, I appreciate why you are hesitant to flip over this particular rock--you are worried that you will be bitterly disappointed and shocked by what he tells you.

Understood. But, if I may, let me just give you a small preview of the sort of thing he might say.

You see, when Ike was pontificating last week, and giving a wonderful demonstration of how

*not*to do statistics, I was sent a private message by a prominent member of these forums, who assured me that my concerns were justified.

But, being curious and fair-minded, he/she also decided to run Ike's analysis past Jeff Hamm. This was my correspondent's doing; I did not ask for it, but I was given permission to reprint the commentary if needed, with the caveat that Jeff had no interest in directly contributing to the 'Diary' debate.

Being a patient fellow, I decided not to reprint the comments, hoping that you and Ike would eventually see the error of your ways, and apologize to those who were abused & mocked for merely pointing out that Ike's methods were entirely invalid.

I know you respect Jeff's logic and abilities from your comments on the geographical profiling threads, so perhaps you'll believe it if you hear it from him instead of me...that is...if you can withhold your laughter at my expense long enough to try to understand what he is saying.

Good luck! If you still don't get it...I suggest that you try your brother or someone else you trust, perhaps down at the local university. I'm sure they can help.

Warmest wishes as always,

R P

- "
**Hi Roger, with best regards from Mr Hamm……**

“Ok, RJ's concern that the issue is about the probability of already known events is valid. It skews one's idea of what chance is. Ike's calculation of probabilities with coin flips (his 5/10 heads), is, while a correct calculation, isn't the probability one uses for inferential decisions (which this is). Given 5/10 heads is the most common outcome, for example, the fact that it only occurs 1/32 times, doesn't mean squat, it's still more probable than any other outcome, individually.

For coin flips, a typical example in statistics, one would look for things like "What is the probability of getting some outcome at least, or more extreme, than the target?", so what's the probability of getting at least 5, or more, of a kind out of 10 flips? And of course, that would be 100% (5:5 H:T, 4:6 H:T or 6:4 H:T, etc, it has to come out somehow).

Now, with the floor boards and the publisher, that would be a case of what Ike calls a "named sequence". (i.e. what's the probability of getting the exact sequence TTHHHHHTTH? It's really low, but that's what I just got as I typed this. But, it's a 6H 4T sequence, and the probability of getting 6:4 or more extreme, like 4H7T, or 7H3T or 7T3H, etc, is well over 90%, because it's everything other than 5:5 exactly).

To even begin to try and equate them, one has to consider all of the possible events one would consider as "acceptable" in terms of finding the Diary at Battlecrease (i.e. pulling up floor boards; or a house inspection of any sort; a visitor to the house who may have checked the library; a garage sale; etc). Also, how long before the "publisher event" would we accept? What if the floor boards had come up the day before? Or the week before? and so forth.

If one could decide upon all the possible events, and timelines, that would be viewed as acceptable (like we can for coin flips), then one start to approach a suitable calculation. And, given the fact that almost anything could be turned into a story that is acceptable, the probability of there being "an acceptable story by chance" is near 100%. The fact that this particular story might appear to have some low chance of occurring, is just like my above coin flip sequence.

Had I predicted that very sequence was going to occur, then I might be able to claim something, but the fact I flipped it first, and know the outcome, changes what I have to deal with - not only the currently known outcome, but all of the possible outcomes that didn't happen but that I would have deemed acceptable.

Hmmm, but long for a quote. Anyway, you can pass that on to RJ. I'm sure he'll be able to extract from that suitable points to deal with the conversation in the thread appropriately. I have no problem with him mentioning my name, but I also have no hesitancy in making the prediction that Ike will dismiss it and insist on his post-hoc calculation of specific events”

All the best

[Note: perhaps the best place to start would be to ask your brother to explain why a “post-hoc calculation of specific events” is not appropriate. And let me just add a final note: unless you and Ike (and Erobitha) eventually grasp this, I have precisely 37,577 reasons not to engage in this discussion further. RP]

## Comment