Posts by 81stcolumn

  • Speaker: What I learned in Class: Should…, in reply to Joe Wylie,

    Heh they had real special way of getting round corners, self opening doors and choke that could be used as a cruice control.

    Nawthshaw • Since Nov 2006 • 753 posts Report Reply

  • Speaker: What I learned in Class: Should…, in reply to Sofie Bribiesca,

    Attachment

    Sofie, just for you -

    Nawthshaw • Since Nov 2006 • 753 posts Report Reply

  • Speaker: What I learned in Class: Should…, in reply to steven crawford,

    I should have added that my Nana sent us sheepskin seatcovers from good ole NZ :-)

    Nawthshaw • Since Nov 2006 • 753 posts Report Reply

  • Speaker: What I learned in Class: Should…,

    Attachment

    Bogan? You decide:

    On my top ten favourite songs is this - Eddie Cochran’s Somethin’ Else note the gorgeous moment when he out Elvis’s Elvis at 1:31-1:33. Was there ever a better song about cars and girls?

    I did a lot of growing up in a car town where they made Marinas, Maxis and errrm Princesses, I rebuilt one of my own ragtop Triumph in later years. But the car that needs to carry my ashes is one of these. The Vauxhall Viscount 3.3 Straight six, 3 speed GM transmission, electric windows, leather seats, walnut dash, just don’t go round any corners at speed. The old man had a very interesting taste in cars, our model was a ’72 just don’t ask the cost of the spares.

    Long hair? Check

    Leather jackets? Check

    I still have the above in my wardrobe……


    As for music

    There was of course Black Sabbath from Birmingham - where they also built cars.

    Ownership of Deep Purple’s Made in Japan was obligatory, containing one Highway Star.

    Forgotten upstarts? That would be Saxon and their wheels of steel.

    Whatever we were it wasn’t f***ing Chavs - they had yet to be born and Burberry had yet to sell out as a fashion brand.

    Nawthshaw • Since Nov 2006 • 753 posts Report Reply

  • Speaker: Identification strategy: Now…,

    81st promises to put the ghost of Ronald Fisher to sleep...

    Maybe over one of those missed cups of coffee eh?

    Nawthshaw • Since Nov 2006 • 753 posts Report Reply

  • Speaker: Identification strategy: Now…,

    Presumably you've occasionally done research that you weren't allowed to just release to the world without someone else's say so before?

    Yes indeed and I would see anonymity and redaction an important part of the process, especially with respect to qualitative data. The goal here is to act ethically and where possible, avoid harm to those who provide data. At the end of this I still have to decide on my own terms whether the remaining data is a fair reflection of the processes and question(s) involved. But that is not the same as obscuring methodology.

    The problem with IP on analytical procedures is that you cannot gain any sense of the robustness of that approach, consequently the worth of what you are seeing. Moreover it removes an important point of agreement regarding truthfulness of the conclusions drawn.

    nb. Stats programmes use IP to process data but the method is open and testable. I can test the tool using standard data sets and know what should come back.

    I guess because this is "sensitive IP for Labour" the credibility issue really sits with the Labour party (ahem - a group of politicians). Consequently my ire may have been mis-directed - but Rob did publish this in a post defending the method used.

    Consequently, for a political party that seeks an electoral mandate to obscure statistical method in pursuit of political goals......well my comments would still very much apply. Indeed I'd go further, the findings as they stand are not really credible, which leaves the inflammatory approach. Labour, Phil and others should not have released this analysis.

    Nawthshaw • Since Nov 2006 • 753 posts Report Reply

  • Speaker: Identification strategy: Now…,

    Okay while we are here:

    I like the idea of including priors it makes a lot of sense. But I think Bayesian analysis has to do its time on the wheel of epistemological rigour.

    i) The scope and impact of assumptions made in Bayesian analysis still need some testing, for example history effects on sampled ratios.

    ii) The tractability of the frequentist approach is a grand asset which is not trivial.

    It may take some time to develop a way of reporting and auditing assumptions made in Bayesian analysis that supersedes the evolution of the frequentist approach.

    @Tze Ming Mok - Cool poeple are still doing quants in our wonderfully post positivist world. Sure do miss you.

    Nawthshaw • Since Nov 2006 • 753 posts Report Reply

  • Speaker: Identification strategy: Now…, in reply to BenWilson,

    Ben - The problem remains that the method isn't public/transparent, which may satisfy Rob and it's fine for him to draw inferences as he chooses. But when you go public with stuff like this its not fair for anyone to have to take his word.

    Nawthshaw • Since Nov 2006 • 753 posts Report Reply

  • Speaker: Identification strategy: Now…,

    I won’t be describing that process further, as that is sensitive IP for Labour.

    Perhaps I am old fashioned in this respect but I don't hold researchers who do this in high regard. When it comes to politics especially on contetious issues with racial implications, this is writing a credibility cheque (trust us we know what we are doing coz we have teh statz). My point being, if you don't want to show it all then keep it to yourself, there can be no middle ground here.

    Nawthshaw • Since Nov 2006 • 753 posts Report Reply

  • Hard News: Friday Music: About Apple Music,

    And a gratuitous vaguely connected favourite of mine.

    I wonder what Dave Brock makes of it all.

    Nawthshaw • Since Nov 2006 • 753 posts Report Reply

Last ←Newer Page 1 2 3 4 5 76 Older→ First