Posts by Some Stats

  • Hard News: Change for the Better, in reply to JessicaRose,

    What do you do at those times? I’m ok admitting, I was frightened of that driver.

    When you encounter dangerous aggression from cars, call the police immediately. Use *555 which is for urgent traffic issues and goes through to the central control room. In my experience, the operators there are very nice and professional. Police in the area may be alerted to the jerk and pull them over. At the very least, the incident is recorded and the stats will add up...

    Auckland • Since Aug 2014 • 9 posts Report Reply

  • Field Theory: Our greatest fear,

    So how are you going compared to David Scott's predictions on Stats Chat?

    Auckland • Since Aug 2014 • 9 posts Report Reply

  • Legal Beagle: The Greg King Memorial…,

    Is it possible for you to write shorter posts? It's not a legal argument, it's a blog! I'd say you are at least 1000 words over here...

    Otherwise, great!

    Auckland • Since Aug 2014 • 9 posts Report Reply

  • Speaker: Why we need to stop talking…,

    A better quote from the follow-up paper is:

    In this paper, we have reported four studies that comprise a follow-up study to Thibodeau and Boroditsky. In contrast to the original studies, we consistently found no effects of metaphorical frames on policy preference. Additionally, there was no difference between the two metaphorical frames on the one hand and the non-metaphorical, neutral frame on the other hand, either. All three frames worked in the same way, consistently guiding all participants to a preference for enforcement-oriented policies. Our prediction that there might be an effect of metaphorical support for the metaphorical framing effects reported by Thibodeau and Boroditsky [1] was not supported either.

    Auckland • Since Aug 2014 • 9 posts Report Reply

  • Speaker: Why we need to stop talking…,

    The framing effect sounds too much like a psychological just-so story, especially as it is one small study purporting to demonstrate a large effect. A follow-up suggests alternative explanations:

    In this article, we offer a critical view of Thibodeau and Boroditsky.... We argue that the design of the study has left room for alternative explanations. We report four experiments comprising a follow-up study, remedying several shortcomings in the original design .... We do not find a metaphorical framing effect but instead show that there is another process at play across the board which presumably has to do with simple exposure to textual information. Reading about crime increases people's preference for enforcement irrespective of metaphorical frame or metaphorical support of the frame. These findings suggest the existence of boundary conditions under which metaphors can have differential effects on reasoning. Thus, our four experiments provide converging evidence raising questions about when metaphors do and do not influence reasoning.

    Auckland • Since Aug 2014 • 9 posts Report Reply

  • Speaker: Market failure in the research world, in reply to 81stcolumn,

    I miss your point...the argument is not that green open access replaces publishing to a restricted access venue, but augments it. Publish where you like but put a preprint somewhere it can be accessed by all.

    Auckland • Since Aug 2014 • 9 posts Report Reply

  • Speaker: Market failure in the research world, in reply to 81stcolumn,

    The cost of open access is not high in any field as there is always the "green" open access solution which is, generally, free.

    See Guardian blog Open access: six myths to put to rest that Mark linked to above.

    Auckland • Since Aug 2014 • 9 posts Report Reply

  • Hard News: Earning Confidence, in reply to Russell Brown,

    The Herald poll is up already. Labour slumps further.

    Of course, the results show no significant change in Labour's support from last week. Indeed, only the change in the Internet Mana support is near being statistically significant.

    Auckland • Since Aug 2014 • 9 posts Report Reply

  • Hard News: Earning Confidence, in reply to Myles Thomas,

    If p is the proportion and n is the sample size, the margin of error is 2*sqrt(p*(1-p)/n). So at 3% (that is p = 0.03) in a poll of 1000 people (n = 1000), you get a margin of error of about 1.1%.

    Stats Chat has a table giving margin of error at various levels of support. Notice that the confidence interval is not symmetric. On that page there is also a table that gives confidence intervals taking into account the lack of independent sampling (basically, they get a bit wider).

    A useful rule of thumb to decide if different results in different polls show a real difference or simply noise is multiply the margin of error by sqrt(2). So around 50%, a shift in poll numbers has to be about 4.5% before you would say it is a real shift.

    Auckland • Since Aug 2014 • 9 posts Report Reply