One Incontrovertible, Unequivocal, Undeniable Fact Which Refutes the Diary

[erobitha]

Seriously, RJ, once again I pissed myself at your desperation to be right at all times.

Or is that he wants Orsam to be so right? Maybe it’s me, but sometimes it’s hard to tell who is who sometimes. Thankfully the odd Americanism alerts me to the fact they are not one and the same.

Otherwise I’d be wondering if we had a multiple personality disorder on our hands.

[Iconoclast]

No Ike, I'm not going to give you the answer, and I'll tell you why.

Because you don't know how to do it, I'd guess. I thought you'd maybe run it by your local newsagent or a farmer or the guy who trims your hedge every Sunday and he or she could furnish you with the answer?

It will leave the impression that this example is relevant, and somehow analogous to your own statistical analysis, and that the difficulty we face is one that can be resolved through simple mathematics.

I think we can all work out that - in reality - it would show that actually you don't understand how simple statistics work which - in fairness to me - is exactly what I know will happen.

That would be wrong, because the example you give involves calculating the odds of events that we know happened.

That is the strongest form of statistical analysis, yes - events which definitely happened taken from a known population of possibles.

By contrast, you are attempting to calculate the odds of a theoretical event, that you don't know happened, and that is what is leading you into the quagmire.

You aren't fooling me or our dear readers, RJ. There are no theoretical events in my calculation. Maybrick died on May 11, 1889 so - slightly arbitrarily perhaps but a real statistician would have understood the irrelevance of worrying about the specific start date once we got as far as the 1990s - we are simply calculating the probability of Maybrick's study floorboards being raised for the first time definitely on record AND someone approaching a literary agent with a potential diary of James Maybrick ON THE SAME DAY. If you know for a fact that one or more of these events did not happen, I think you should publish it as it rather flies in the face of the basic facts of the case.

Ike

[rjpalmer]

Eddie was working full-time hours, from Monday to Saturday, from December 1991 to 7th March 1992, over in Skem, so opportunities to bump into Mike in the Saddle during that time would have been limited, unless they both drank there outside of Eddie's working hours, for example on Saturday nights. But the landlord, speaking to Feldy in 1993, only said he remembered Mike - and Tony - and occasionally Mike's father, coming in for lunchtime drinks.

This is a little misleading, isn't it?

"The landlord only said..."

Here's Feldman's account below. It is far from clear whether Feldman even asked the landlord about Eddie's visits to the Saddle, and Feldman is not great about giving the details, or sticking with a coherent chronology. So we don't really know what--if anything--the landlord said about Eddie, or what Feldman asked.

[Iconoclast]

Or is that he wants Orsam to be so right?

Hi erobitha,

Yes, you hit the nail on the head. I think that RJ represents a number of posters on the Casebook who would rather the aristocratic artisan was accurate in his assessment that Anne and her awful other 'arf authored the ancient article because it would just be simpler, wouldn't it? No awkward details to work around and what have you.

Hey, by the way, have you ever noticed how remarkably similar Maybrick's signature in the back of the mooted Maybrick Watch is to his known signature on his wedding licence? Any idea how that was possible?

You know, that sort of awkward detail ...

Ike

[rjpalmer]

Do you think her years of training on double-entry accounting instinctively told her that statisticians had been calculating simple probabilities incorrectly for decades before she came along with her DIY Chi-Squared Test?

This is quite sad. You're pettiness is beneath you, Ike, and you should really try to do something about it.

Not that it is any of your business, but my bookkeeper friend also happens to have a degree in Psychology.

You probably don't know this, but, at the time she earned her college degree, psychology students were required to take two college level courses in statistics, so they would understand the pitfalls of statistical analysis, which is often used in psychology. And often wrongly used by psychologists.

The field of statistics is particularly rife with pitfalls and flawed logic and poorly drawn conclusions--the kind that you are committing in your own statistical analysis, where the data you are using is irrelevant to the questions you should be asking.

There's many articles and books on the subject. Feel free to consult them.

P.S. As for Lord Orsam, I have no vested or emotional stake in him being "right." I merely observe that he understands why your statistical analysis is moonshine.

But you won't ever listen to either him, nor me, which is why I suggest that you seek the services of a trained logician/statistician with no interest in the Mabyrick Diary who can patiently explain your error. Perhaps then, you'll listen and understand.

[Iconoclast]

You probably don't know this, but, at the time she earned her college degree, psychology students were required to take two college level courses in statistics, so they would understand the pitfalls of statistical analysis, which is often used in psychology. And often wrongly used by psychologists.

Oh, perfect! I completed three highly-intensive years of research methods and statistics as part of my degree!

Okay, RJ, let me release you from the purgatory of not knowing how to do simple statistics. You might want to run this one by your hairdresser, by the way.

If someone tosses a coin nine times (and we accept the assumption that none land on their side), what is the probability of getting eight heads and one tails?

Well, it's very simple. In a simple dichotomous outcome (heads or tails), all you need to do is calculate 0.5 (the probability of a heads or a tails each time) by itself as many times as the highest frequency of outcomes.

The lowest frequency in our example was tails and that was 1.
The highest frequency in our example was heads and that was 8.

Therefore, we can ignore one of the coin tosses - it doesn't matter which one - because it was guaranteed to give us an outcome we required (a head or a tail) which would have us multiple by 1 which is pointless, so we then multiple 0.5 only by the number of times we need a specific outcome, which in this example is eight.

So, 0.5 x 0.5 x 0.5 x 0.5 x 0.5 x 0.5 x 0.5 x 0.5 = 0.0039062 (or 1/256)

Therefore, if you tossed a coin nine times 256 times, you shouldn't be in the slightest bit surprised if you get eight heads and one tails at least once.

Doesn't matter if your lowest and highest frequencies are the same, by the way. So, if you tossed a coin eight times and got four heads and four tails, the probability of doing so is (and I'll add back in the 1s this time but remember they do not change the final outcome):

1 x 1 x 1 x 1 x 0.5 x 0.5 x 0.5 x 0.5 = 0.03125 (1/32)

Therefore, if you tossed a coin eight times 32 times, you shouldn't be in the slightest bit surprised if you get four heads and four tails at least once.

This throws up (pardon the pun) the interesting statistical anomaly that throwing a single coin and getting, say, a heads has the same probability as tossing two coins and getting a heads and a tails. The probability of throwing one coin and getting, say, a heads is obviously 0.5. The probability of tossing two coins and getting a heads and a tails (in no specific order, by the way) is also 0.5 because the first coin toss is guaranteed to give us one of our desired outcomes (which gives us a probability of 1) therefore the only probability which matters is the second toss which has to be the opposite of our first toss, therefore we're back to our original 0.5 chance, or 1 x 0.5 to be precise.

Therefore, if you tossed a coin twice 2 times, you shouldn't be in the slightest bit surprised if you get one head and one tail at least once.

Before you ask, this is not the same as the probability of throwing a coin twice and getting, say, a head first then a tail second. In this case, each throw has to have a specific outcome so it would be 0.5 x 0.5 = 0.25 (1/4).

Therefore, if you tossed a coin twice 4 times, you shouldn't be in the slightest bit surprised if you get a head followed by a tail at least once.

Now, RJ, run this one by your bookkeeper and let us all know why these calculations are wrong (obviously, we don't need probability theory to know that that's what you'll report because that's just what you do because you just always have to be right).

Ike

[Iconoclast]

So, dear readers, ahead of RJ informing us all of how awful I am at statistics, when I say that there was in 1-in-37,557 chance on March 9, 1992 that Maybrick's floorboards would be lifted and someone would bring Maybrick's mooted diary to the attention of a literary agent on the same day by sheer chance alone, please be assured that I am correct (given May 12, 1889 as the first day when either of these events could have happened previously).

These two events occurred and no other factors need to be massaged-in to change the analysis: two statistical miracles happened on March 9, 1992, and that was the first one ...

PS Obviously the second statistical miracle was that a member of the team drank in the same pub as the guy who contacted the literary agent. The first guy, by the way, obviously had to be in Liverpool, but there is no reason to assume that the second guy also had to be in Liverpool that day (just in case you don't realise how staggeringly unlikely it was that they drank in the same pub). Imagine if you knew that Maybrick's floorboards came up and also that a guy in Portsmouth or Inverness contacted Rupert Crew on that same day offering Maybrick's diary (even if he didn't realise it was Maybrick's on March 9, 1992). Which of the following would be your more natural reaction to this staggeringly-unlikely 'coincidence':

"Oh, that happens all the time in my experience", or
"There just has to be a link between these two events"?

In the case of when these two events actually did occur, the link was The Saddle Inn in Anfield. If you think that's fairly unremarkable, I would ask that you quickly Google Map 'Merseyside' and then Google '1992 population of Merseyside'.

Ike

[Scott Nelson]

Ike, your calculations in #6576 are meaningless. The probability is always going to be 50-50.

[erobitha]

Ike, your calculations in #6576 are meaningless. The probability is always going to be 50-50.

Only if the toss is once. Did you say you do engineering?

[Scott Nelson]

Yes, engineering. No matter how many coin tosses, the odds are always going to be 50-50. It's common sense. There's nothing more to calculate.

The point of all this being that you can't attempt to quantify possible outcomes in a nebulous historical context unless you have ironclad data to begin with. The coin toss was just an example of straight numerical probability.

[Yabs]

Only if the toss is once. Did you say you do engineering?

Do you mean it’s 50/50 if you toss a coin once in your whole lifetime?
If not, what does the gap have to be between coin tosses to make it 50/50 again?

[Scott Nelson]

Any gap doesn't matter.

It's always going to be 50-50, heads or tails, no matter how many times you flip the coin, no matter what day of the week it is, what year, etc.

[rjpalmer]

Any gap doesn't matter.

It's always going to be 50-50, heads or tails, no matter how many times you flip the coin, no matter what day of the week it is, what year, etc.

In Ike's defense (gasp!), re-read his excruciatingly dull post.

He's not calculating the odds of a single theoretical coin flip.

He's calculating the odds of a known event that has already happened (or would eventually happen if you kept flipping coins long enough). He really did flip a coin 9 times, and it really did come up heads 8 times. He's simply figuring out the odds of this having happened randomly.

Which is an entirely different problem than what he is facing with the bogus Maybrick Diary "double event," where he is confronted with a series of theoretical, unconfirmed events, and then being foolish enough to try and calculate the odds. He doesn't even know these events ever COULD have happened, because he doesn't have enough data to confirm that.

That's the root of the flawed logic that Ike is using. He is a veritable human calculator--he just can't grasp that he's using bogus numbers for his calculations.

From Post #5230 of this thread.

The chances of the floorboards coming up on the very day Barrett rang Montgomery is around 1-in-26,000 (the number of times it could have happened divided by the number of times it did happen).

Dang, he’s only off 11,000 or so.

This is 'incestuous' logic, of course. Ike's statement "the number of times it could have happened" is not in evidence, since it is entirely possible, even likely, that the diary didn't exist for the great majority of that span. Indeed, it may not even have existed on March 9th, 1992. We are back to "garbage in/garbage out” and the misuse of statistics.

One can also see that our resident statistician has now published 3 different sets of odds. No doubt it is the fault of Excel.

And really, using Ike's logic, why couldn't the diary have been found under the floorboards May 8, 1889? Or May 10th? Doesn't the last bogus entry date to May 6th? Do we need a fourth calculation?

Currently, I have more faith that my hairdresser, if I had one, would come up with a legitimate number.

Sorry, Ike.

RP

_________________

"Maybrick accusers use statistics in the same way that a drunk uses lamp-posts—for support rather than illumination" – (with apologies to Mr. Andrew Lang).

[Yabs]

Any gap doesn't matter.

It's always going to be 50-50, heads or tails, no matter how many times you flip the coin, no matter what day of the week it is, what year, etc.

Of course, I agree.
The odds of ten heads being spun in a row is just as likely or unlikely as any other combination of head and tails.

[Iconoclast]

In Ike's defense (gasp!), re-read his excruciatingly dull post.

He's not calculating the odds of a single theoretical coin flip.

He's calculating the odds of a known event that has already happened (or would eventually happen if you kept flipping coins long enough). He really did flip a coin 9 times, and it really did come up heads 8 times. He's simply figuring out the odds of this having happened randomly.

Thank you, RJ (gasp!). I was indeed answering the question I first posed to Scott last night and then to you today.

And, Scott, of course every single toss of a coin is a probability 're-set'. It's always 50-50 for that individual throw. Obviously! You just weren't paying attention to the question, mate.

Originally posted by Iconoclast View Post
The chances of the floorboards coming up on the very day Barrett rang Montgomery is around 1-in-26,000 (the number of times it could have happened divided by the number of times it did happen).

I think you'll find that 1/26,000 was generously factoring-out weekends and Bank Holidays and I think you'll find I stated that quite clearly either in the post you quote or previous posts.

Honestly, when the odds of something happening by chance alone start to get into four figures (never mind five figures), you know it's not random so a generous 1/26,000 or a hard-nosed 1/37,557 absolutely makes no material difference to the massively unlikely probability that an event has happened purely by chance.

A statistician would have known that, of course. You should have fact-checked this first with your gardener, RJ.