More Data, More Openness, More Swing-o-Meters

0

El Universal’s Eugenio Martínez seems to be the Go To guy for Election Projections in the Veneco Tweetosphere these days (@puzkas) – and the guy doesn’t think much about our Swing-o-Meter. As he sees it, circuit-by-circuit polling shows a much bleaker picture for the opposition.

Thing is, Puzkas doesn’t actually cite any polls, or any pollsters, or any dates, or tell us which circuits he’s talking about, or even pause to explain why he has to hold back such information. He just sort of waves vaguely in the direction of some quasi-mythical set of 21 circuit-by-circuit polls and says that the opposition is likely to top-out at 55 seats, with another handful going to PPT.

Maybe he’s right, but we have no way to check. 

Personally, this way of doing journalism drives me up a wall. It’s the Wikileaks era, people: if you have data, you link to it. If you don’t, we’re entitled to assume you’re making it up as you go along.

Of course, part of this is that it eats at me that somebody somewhere has a poll that I don’t. So, if you have circuit-by-circuit polling, by all means leak it here: I’ll keep your name out of it. But I will publish the results in detail.

Because openness is crucial on an issue like this. If we’re not up-front about our data, our models, and our methods, there’s no way we can look over one another’s shoulders and check if work is being done competently and in good faith.

Because, lets be clear: Puzkas could well be right. It’s just that, if he won’t show us his data, how could we possibly know? Guy could well be spinning us…or he could well be getting spun. Certainly, there’s a powerful incentive for the opposition to keep expectations low ahead of the vote – motivate the base, be able to claim a "bigger-than-expected victory" down the road, etc. 

One way or another, over the next few months we’ll be working hard to refine the Swing-o-Meter: both the model itself, and the data behind it. I’ve already made some cosmetic adjustment that make it easier to use. But more substantive improvements are coming.

For now, though, what we need is just more data. More polling data. More stuff we can regress. Remember, we’re all part-timers and DIYers here, so if you have some Data or some modelling experience, don’t just sit back and tell us what to do: make a projection and send it in!

Crowdsourcing is the way to solve a problem like this.

Download the new look Swing-o-Meter here.

Send it to all your friends. 

(La versión en español está aquí.)

 

Leave a Reply