OK, as promised, here is the Thursday update. It deals with the use of the word "significant" and introduces the methodology paper and the headings we will explore over the coming weeks. So, let's begin.
WHY DID WE USE THE WORD "SIGNIFICANT"?
======================================
It's a legitimate question. The use of the word here is plainly not related to a p-value or a p-value threshold, so why use it? In January I considered adding a discussion of the point to the accompanying methodology paper, but after some agonising, I decided not to. The reason I didn't is because the point always boiled down to the same sentence, namely "we used the word 'significant' because Holmes and Haggett used it in the original paper". I tried to expand the point ("...Holmes and Haggett apply their method as a test to a series of numbers, and ascribe the word "significant" to those that pass the test...") but every time I just circled back to the same thing: we used it because they did.
THE METHODOLOGY PAPER
=====================
In its original incarnation, the story and the methodology were intertwined ("...We reproduced Holmes and Haggett's findings by applying their method as follows...we apply them to this series of numbers and obtained these results in the table below...") As it went through each iteration, the methodology was reduced to footnotes, allowing the story to predominate: "We reproduced Holmes and Haggett's findings[1] and found that Plymouth contributes significant movers[2] to Cornwall"). Tables became graphs, formulae were moved to the appendices, the appendices were hived off. This made the article intelligible to laymen but ran the risk of reducing it to gossip: I still had to be able to prove every point made. How to handle this?
To handle this, the methodology paper was born: the careful minutiae could be moved there and accessed via a link, leaving the broad sweep to the article. It was a simple solution and if that link had worked, a lot of the questions would never have arisen.
But it didn't and they did. So, what to do?
I've asked the editor to repair the link and no doubt he will do so when he can. Until then I thought it might be interesting to go through the methodology paper. It'll take me about two weeks to cover it from beginning to end so I'll just give the chapter headings tonight, then pick it up again on Monday.
THE CHAPTER HEADINGS
====================
The methodology paper is divided into sections as follows:
* What are the six groupings?
* Which tables did we use?
* EU15 (significant years, areas, countries of birth)
* EU8 (significant years, areas, countries of birth)
* EU-other (significant years, areas, countries of birth)
* Old Commonwealth (significant years, areas, countries of birth)
* New Commonwealth (significant years, areas, countries of birth)
* Overall (significant years, areas, countries of birth)
* Appendix 1: the Pearson algorithm (Javascript version)
* Appendix 2: Worked example of the Pearson’s algorithm
* Appendix 3: Assigned groupings (which countries/regions/etc belong to which groupings?)
* Appendix 4: Differences between LTIM and IPS (why did we use International Passenger Survey instead of Long Term International Migration figures?)
* Appendix 5: Superseded tables (what new figures have been released since the article was written?)
Regards, Martyn
******************************************************
Please note that if you press the 'Reply' button your
message will go only to the sender of this message.
If you want to reply to the whole list, use your mailer's
'Reply-to-All' button to send your message automatically
to [log in to unmask]
Disclaimer: The messages sent to this list are the views of the sender and cannot be assumed to be representative of the range of views held by subscribers to the Radical Statistics Group. To find out more about Radical Statistics and its aims and activities and read current and past issues of our newsletter you are invited to visit our web site www.radstats.org.uk.
*******************************************************
|