February 1, 2017
Briefly
- Maps as a research communication tool: a research project into the relationship between rateable value and sales value for homes in Milwaukee, and who ends up overpaying their rates.
- From the ‘no way this could go wrong’ department. New York is to collect drop-off as well as pick-up times and locations for taxis, Ubers, Lyfts, etc. And the data appear to be subject to freedom of information requests.
- Cathy O’Neil (mathbabe.org, Weapons of Math Destruction) is now a Bloomsberg View columnist
- Petrol price margins are in the news again, so I’ll remind everyone again that MBIE collects data and draws pretty graphs. (They aren’t so good on regional variation or discounting, but it’s still useful reality-based information).
- No, toast isn’t going to kill you: Michelle Dickinson, David Spiegelhalter.
- Sherlock Holmes opined “It is a capital mistake to theorize before one has data”. As Kaiser Fung shows, this doesn’t apply to complex data-gathering tasks such as counting the number of heads on a movie poster.
- What happens when you make a major change to the definition of an important official statistic.
- One of the minor aspects of Donald Trump’s awful executive order is collection and reporting of crimes committed by immigrants. Whether this was a mistake for him depends on how good people are at denominators: immigrants commit less crime on average than people born in the US, and there are fewer immigrants than people think there are, so the number will be smaller than people should expect. But that takes maths. Or in the US, math.
Thomas Lumley (@tslumley) is Professor of Biostatistics at the University of Auckland. His research interests include semiparametric models, survey sampling, statistical computing, foundations of statistics, and whatever methodological problems his medical collaborators come up with. He also blogs at Biased and Inefficient See all posts by Thomas Lumley »
Those official crime stats, dont they make ‘back casts’ so that the new numbers have a more valid comparison with the previous years
7 years ago
They can’t make back-casts with the new definition, because it uses data that weren’t previously collected. They could present what the old definition would give, and they do something like that here
7 years ago