Posts by linger

  • Legal Beagle: Last call on the Electoral…,

    Perhaps specify "near unanimity" as "one or two dissenters" or "two-thirds", whichever is the maximum: consider the case of a party with fewer than 5 MPs.

    Tokyo • Since Apr 2007 • 1802 posts Report Reply

  • Hard News: On joining the international…, in reply to Dennis Frank,

    Problem is, phrases such as "political correctness", "social engineering", and "nanny state" are loaded terms that often reveal more about the user than about the intended referent. E.g. railing against "nanny-statism" has been used to excuse the blocking or dismantling of regulations around food-origin labelling, building safety, energy efficiency, environmental protection… meanwhile the same governments tend not to see any contradiction at all in enacting laws extending the powers of the state against personal privacy and peaceful protest.

    Tokyo • Since Apr 2007 • 1802 posts Report Reply

  • Hard News: On joining the international…,

    The danger even in a seemingly positive claim such as Brash’s is that an assertion “Group X is more successful” reliably feeds conspiracy theories attacking Group X on the assumption that their success has been obtained unfairly.

    (What is needed is more testable general hypotheses regarding possible causal mechanisms, rather than claiming exceptionism of any one group. E.g., we may hypothesize that – as one ingredient – a monotheistic culture favours a search for single-cause deterministic models, which may promote the advancement of mathematics, science and technology. As another ingredient, the culture’s traditions should be open to question and debate, allowing a faster adoption of improvements as they are discovered.)

    Tokyo • Since Apr 2007 • 1802 posts Report Reply

  • Hard News: On joining the international…,

    Oh, there may well be a correlation; but as always, that doesn’t prove causation. We should be wary of any assertion of g(u)ilt by association, whether cast in terms of “race” or “culture”.

    Tokyo • Since Apr 2007 • 1802 posts Report Reply

  • Hard News: On joining the international…,

    I rather liked the DomPost letter today to the effect that they should be on the Wellington comedy circuit.

    Tokyo • Since Apr 2007 • 1802 posts Report Reply

  • Hard News: Dilemmas: the drug-driving ad…, in reply to Dennis Frank,

    Ads are primarily designed to be short, which may or may not mean a holistic design. It does mean that, as far as possible, an ad draws on common experience (e.g. using instantly recognisable stereotypes, story schema, and default value judgements). All I’m saying is that in this case, if you haven’t had that experience, the ad doesn’t work. (It’s usually a bad sign when every referent leads you down the rabbit hole to a previous ad.)

    Tokyo • Since Apr 2007 • 1802 posts Report Reply

  • Hard News: Dilemmas: the drug-driving ad…,

    Really confusing. It looks like Ghost Chips, but drops any pretence of linking the images into a coherent narrative. (E.g., where does Nek Minit dude fit in?) Maybe makes more sense if you're actually high while watching?

    Tokyo • Since Apr 2007 • 1802 posts Report Reply

  • Hard News: On joining the international…,

    Except that Watkin’s article didn’t tell anyone to do anything:
    (i) it cited a colleague’s tweeted suggestion, and enthused on how it might have trapped them; and
    (ii) it was published after the snakes had got on a plane.

    But I agree that hosting abhorrent views isn’t something that anyone should be required or expected to take on.

    Tokyo • Since Apr 2007 • 1802 posts Report Reply

  • Hard News: Reassure Me: cannabis,…, in reply to BenWilson,

    A stratified sample is one in which there are fixed quotas for certain defined subgroups. But we need to distinguish two possible types of stratification, used for different purposes.

    The first has the aim of ensuring those groups are all represented, usually in order to facilitate subsequent comparisons of those groups. For such a purpose, minorities should be over-represented to reduce the uncertainty in estimates based on subsamples. Such a stratified sample is “representative” in the sense of ensuring diverse representation, rather than in the sense of accurately reflecting overall population statistics.

    The second type of stratified sample design is to ensure statistically fair representation of some groups that would otherwise be underrepresented by normal sampling methods, because they are harder to contact or less likely to respond to requests. Here the quotas are based on the proportions of each group in the population, so the final sample should accurately reflect the wider population. (In this case, there may be no intention to compare the subgroups.)
    Presumably the second use is more what you have in mind?

    N.B. Whether the sample is random is a separate issue. It is possible to fill each subsample quota by random selection from its target subgroup population, though in practice a networking approach may be necessary for group classifications for which membership is less easily externally verifiable/ publicly available.

    Tokyo • Since Apr 2007 • 1802 posts Report Reply

  • Hard News: Reassure Me: cannabis,…,

    Note Bridges' cynical attempt to head off any chance of a wider-scope "reeferendum" by trying to limit the discussion to high-end medical extracts.

    Tokyo • Since Apr 2007 • 1802 posts Report Reply

Last ←Newer Page 1 2 3 4 5 180 Older→ First