Jonathan Auerbach writes:
How well do you really know your federal data sources?
Life expectancy, gross domestic product, consumer price index—we reference statistics like these all the time. But how well do you really know the details?—the careful considerations that go into the official statistics we rely on every day?
Now’s your chance to find out. Take our quiz, share with that special data user in your life, and support the American Statistical Association’s Count on Stats initiative. (Ten questions randomly selected every hour.)
I just tried out the quiz. It was fun! I got 8 out of 10 correct.
I got 6 of 10, a result which I feel compelled to acknowledge is consistent with me being pretty sure about maybe 3 of 10.
Jeff:
Yeah, I guessed on a few of them too. But I feel like they were good guesses!
Thanks for posting — this was neat! I got 7/10, which is better than I expected given that I’m not a social scientist or anything related! I made only a few wild guesses — there are several questions for which one can make pretty sensible estimates, but I won’t describe this so no one’s experience is spoiled…
8/10. I guessed to some degree on every one except consecutive years of increasing life expectancy, which I knew was zero. I know women live longer than men. I didn’t know it was five years but i knew it wasn’t zero or ten years. Even $1B seems a bit low for a massive undertaking like the US census. The total length of roads question was an interesting one but I was pretty sure it wasn’t over 1%. I took it to mean just the road length not total lane length, if I had understood it that way I think it would have been harder to guess it.
I was 7/7 then I missed #s 8 and 10.
7/10 by blind luck and logic. I am Canadian. I may have used some FBI crime stats once.
StatCan is another matter.
5/10 let’s goooo
5/10 for me too. I didn’t think the quiz was very good, though. Some questions were just sort of definitional, while others seem more substantive, with practical implications. For example, the number of alternative measures of unemployment collected seems to matter little, while the issue of who is not included in unemployment measurement is of real importance. So, while it is a nice educational tool, a little more consistency in the questions would have put more focus on some important measurement issues and a little less like a guessing game.
8/10
Questions asked seem to vary, as some of the questions others here discuss (measures of unemployment) were not on the survey I took.
I did not appreciate the political pitch buried within a mundane statistical question.
That road mileage, by their measure, is up 0.3% per year while miles traveled is up 1.3% does NOT, by any stretch, mean that a comprehensive infrastructure bill is needed. The latter is a far more complicated question than a simple comparison of two numbers. Yeah, small point, but it seems to go to the general hubris of *SCIEEEEENCE!!!!!* – that data and the scientists who produce it necessarily imply policy, rather than that the latter is a complicated ball of judgements, tradeoffs, and the like…
Phil S.:
The Commissar for Traffic presents the latest Five-Year Plan.
I like Clark Williams-Derry’s article from the Sightline Institute’s on the Columbia River Bridge between Portland and Vancouver: https://www.sightline.org/2012/04/09/how-not-to-forecast-traffic/
He claims that the linear regression of increasing traffic growth is a “meaningless, cherry-picked line”. But his own plot shows a pattern of declining traffic growth prior to the opening of the I-205 bridge, after which traffic growth returned to it’s previous growth rate and possibly slightly faster! :) So, if “meaningless” means “exactly the way it happened in the past” then, yes, in this case, the linear projection is “meaningless”
He argues for a garbled model including all manor of unpredictable factors as being a more scientific model of traffic growth. It’s not clear why this would be better when the previous method was proven to work quite well in the past.
More likely Williams-Derry’s mistake is that he doesn’t understand the fundamental economics of transportation: more is almost always better, and – at least in a strong economy – traffic will almost always grow to use the available capacity, making his sophisticated models interesting in a scientific sense but wasteful for decision making.
(I also didn’t get the consecutive years of increasing longevity, tho I *did* get a question on female longevity advantage vs. males.)
And, as a further criticism, the fact that this survey apparently randomizes questions is kinda weird for folks who presumably are advocates for statistics and data. Because perhaps one of their best hopes was that the link would be picked up by sites like this, folks would take the quiz and compare scores. (As happened!)
But if the scores are based on different versions of the test (varying Qs), then, well, it’s kinda bad statistics for us test takers, and a (minor) strike against “statistics experts”, for those with only a casual interest (like me).
“does NOT, by any stretch, mean that a comprehensive infrastructure bill is needed. ”
Good point Phil! I saw the pitch and thought it was out of place but nodded absent mindedly. But you’re point is excellent: it’s these little factoids – hiding numerous mistaken assumptions and false inferences – that cause so much trouble.