General Discussion
Related: Editorials & Other Articles, Issue Forums, Alliance Forums, Region ForumsTarnished Silver (Nate Silver) by Paul Krugman
http://krugman.blogs.nytimes.com/2014/03/23/tarnished-silver/?_php=true&_type=blogs&_php=true&_type=blogs&_r=1&Timothy Egan joins the chorus of those dismayed by Nate Silvers new FiveThirtyEight. I sorry, but I have to agree: so far it looks like something between a disappointment and a disaster.
But Id argue that many of the critics are getting the problem wrong. Its not the reliance on data; numbers can be good, and can even be revelatory. But data never tell a story on their own. They need to be viewed through the lens of some kind of model, and its very important to do your best to get a good model. And that usually means turning to experts in whatever field youre addressing.
Unfortunately, Silver seems to have taken the wrong lesson from his election-forecasting success. In that case, he pitted his statistical approach against campaign-narrative pundits, who turned out to know approximately nothing. What he seems to have concluded is that there are no experts anywhere, that a smart data analyst can and should ignore all that.
But not all fields are like that in fact, even political analysis isnt like that, if you talk to political scientists instead of political reporters. So, for example, before glancing at some correlation and asserting causation, you really should talk to the researchers.
snip
BumRushDaShow
(127,281 posts)Cirque du So-What
(25,811 posts)Which of DU's erstwhile heroes will be thrown under the bus - Nate Silver or Paul Krugman? Epic battles will ensue.
randome
(34,845 posts)Like some other poll agencies have done.
I agree, though, it's getting tough to see the truth with such diametrically opposed opinions. The impetus is to trust Silver since he was right the first time. Does that mean he will be right a second time? A third? That remains to be seen.
[hr][font color="blue"][center]A 90% chance of rain means the same as a 10% chance:
It might rain and it might not.[/center][/font][hr]
Cirque du So-What
(25,811 posts)to verify Krugman's claim - that Nate Silver has heavily based current predictions on the word of pundits? If so, it gives me hope that his predictions do not take into account the variable of voter turnout, over which Democrats have some degree of control...if only they exercise it.
randome
(34,845 posts)Numbers tell a story but not necessarily the story. And TPM is saying that based on Silver's data, the GOP is a 'slight favorite'. And it's still early in the year.
[hr][font color="blue"][center]It made sense when we were children. Not so much now.
Talk to a stranger today. You might learn something. You might help someone.[/center][/font][hr]
Hekate
(90,189 posts)If I read your reference to "some other poll agencies" wrong, I apologize. But Krugman is meticulous.
WhaTHellsgoingonhere
(5,252 posts)Nate says GOP 60% likely to win Senate and 30% likely to have a landslide victory.
Buh-bye, Nate, but you've outlived your usefulness at DU
Nate Silver revealed a new forecast on Sunday via his new venture, FiveThirtyEight.com, projecting that Republicans have about a 60 percent chance of wining at least 6 Senate seats to take control of the upper chamber in November.
http://talkingpointsmemo.com/livewire/nate-silver-gop-senate
Chathamization
(1,638 posts)That's not to say that I think he's terrible or anything, and I enjoyed reading FiveThirtyEight when it was at the Times. Having said that, the climate change article over at FiverThirtyEight right now is pretty bad, and doesn't give me much of a reason to look at the rest of the site.
Krugman's insistence on treating the word "data" as if it's plural (well, trying to at least) is annoying, though.
cthulu2016
(10,960 posts)One piece of data is datum.
There is no word "datums"... two or more datum = data.
Chathamization
(1,638 posts)and it's used by people that actually use the word datum on a regular basis (people involved in topography), as supposed to those who only bring up datum when they want to defend treating datum as if it's plural. This should tell us something.
muriel_volestrangler
(101,149 posts)The technically correct singular is 'datum', but many people will treat 'data' as singular or plural, so that, by now, using it as singular is no longer incorrect. But you shouldn't get annoyed by someone being correct.
Chathamization
(1,638 posts)Here's an example from Krugman's article on Silver:
But you cant be an effective fox just by letting the data speak for itself because it never does. You use data to inform your analysis, you let it tell you that your pet hypothesis is wrong, but data are never a substitute for hard thinking. If you think the data are speaking for themselves, what youre really doing is implicit theorizing, which is a really bad idea (because you cant test your assumptions if you dont even know what youre assuming.)
If when they're able to is it somewhat consistently, they seem to accept terms like "piece of data" and "data scientist" which aren't consistent with treating it like a plural. And this is on top of a somewhat shaky theoretical ground (as I noted in the other post, people who actually use the word "datum" frequently use the plural "datums" .
I wouldn't call it correct.
muriel_volestrangler
(101,149 posts)though there's nothing in 'data scientist' that implies it's singular. Substitute 'numbers' for data, which is obviously plural (and has a similar meaning here), and you get 'numbers scientist', which is fine (or, if the concept jars a little, consider 'data expert/numbers expert').
Clearly, it is correct to treat it as a plural noun; and, even if you think it can be a singular noun as well, then the plural of 'data' is 'data' ("here is the data for Monday; the data for Tuesday and Wednesday are also available" .
The use of 'datums' in surveying is a bit different; the Oxford English Dictionary explicitly says that the plural of 'datum' is 'data' except for the surveying use (and 'datum' to mean an item of information, with plural 'data', dates back to 1630, while the surveying usage only dates from 1795).
Chathamization
(1,638 posts)As for "data scientist," I doubt most would pick "plants scientist" over "plant scientist" or "cars mechanic" over "car mechanic." However, those common data terms work if it's treated as a singular mass noun (and people who treat it as such don't tend to switch between singular and plural in one sentence the way pluralists do).
As for datums - the thing is, topography is where you actually see the word datum used. To say that datums is only correct for surveying use is like saying that datums is the correct plural of datum when you actually use the word datum, and that data is the plural of datum for in those instances when datum would never be used.
AngryAmish
(25,704 posts)Nate Silver has not released his model. Fine. But it may or may not be accurate. So all of his predictions are suspect.
As for Krugman. Krugman is an economist. For the really useful stuff (debt load of a nation, interest rates, if a multiplier exists etc) there is profound disagreements. There is no predictive value in the process that calls itself economics.
The future is inherently unpredictable. We should be more modest in our claims.
pa28
(6,145 posts)I enjoy his new site and let's not forget it's "new". He's got the time and money to find his feet and his voice and I think he'll do just fine.