The simulator is new to 270toWin in 2008, so it is still a work-in-progress for us. We like the statistics behind it, but need to point out an important issue.
Most national polls of late show the race to be very tight with Obama ever so slightly up; one poll out today (Reuters/Zogby) shows McCain has pulled ahead by 5 points. Let’s call it statistically tied for argument’s sake. The important point is that the change has happened fairly quickly.
The simulator is based on state-level polls. Ultimately, the election will be decided at the state level, so this is the right approach. However, at this point, most states are only being polled once or twice a month. Until this polling gets more frequent, be aware that the simulator results will lag behind what the national polls are saying.
We hope that as Election Day nears, everything will sync up nicely and the tool will be quite useful, but until then, use caution in reading too much into the simulator outcomes.
10 responses so far ↓
1 mark // Aug 20, 2008 at 7:04 pm
Thank you for the explanation……Im sure I wasnt the only person getting impatient with the simulator……Your timely response is very appreciated…..
2 B Caldwell // Aug 20, 2008 at 9:21 pm
I like the simulator, but I always run it 10-15 times. That reveals the various combinations and permutations that could occur, and reveals what states each candidate must win. Then you go back and play with the 2008 map. The simulator can’t predict the election, but it can show you the effect of a candidate winning/losing a particular state or states.
3 mark // Aug 21, 2008 at 12:00 pm
perhaps the answer to the problem of lag time…………….a national tracker that allows for the curve…along side the simulator……….you could show the correlation between the nationals.and the state polls……….showing how they converge will at least give the appearance of context to….what state polls vs. national………..perhaps some new shiny colors …and a new graph line.lol
4 quincyconscience // Aug 25, 2008 at 6:29 am
how long does it take for pollsters to add running mates to the polling questions?
5 Oaza // Sep 4, 2008 at 2:34 am
Congratulations for developing this tool! This is one of the most educational pieces of on-line software I have ever seen. It teaches geography, politics, statistics, probability, electoral law, and most of all, teases the brain. Excellent puzzle! Great fun! Thank you on behalf of everyone who enjoys this beautiful tool!
6 Anonymous // Sep 4, 2008 at 2:21 pm
So when are you going to update the polling data for the simulator? The simulator page says it was last updated August 27. We should see the polling pick up in earnest by next week.
270toWin: There aren’t really any polls coming out during the convention. There are a few from CNN/Time but they are of questionable value.
7 Jesse // Sep 7, 2008 at 12:32 pm
Do you have any evidence that your methodology is at all accurate? If you had historical polling data from 1-2 months before the each of the last 4 elections and you ran a simulation to see how they matched up with the results, that would be good to know.
270toWin: We believe the methodology is statistically valid. However, nobody will vote based on today’s polls so there is no way of knowing if they are accurate. Even if the election ends up completely different in two months, it doesn’t mean the poll inputs weren’t an accurate reflection when they were taken. We’re not trying to guess the future here.
8 Jesse // Sep 7, 2008 at 1:33 pm
Thanks for the quick reply.
Another thought: national and state polls can be divergent, i.e. state polls can still show Obama winning by 30 EVs whereas national polls show him even with McCain. This could be the case if, suppose, red states were getting redder yet McCain was not making any inroads into blue or purple states.
Am I right?
270toWin: That is certainly one possibility mathematically, although a pollster would probably be better able to answer. One interesting aside is that this is the final election before the 2010 census and its reallocation of electoral votes. As a result, one could argue that states that have grown this decade will be slightly under-represented in the 2008 election and vice-versa for those that have shrunk. This could distort a popular vs. electoral analysis, although we have zero ability to quantify that.
9 Christopher // Sep 12, 2008 at 2:09 pm
How often do you update the simulator?
I enjoy a website: http://www.electoral-vote.com/ which updates the national map daily whenever a new poll is out. By scrolling over the state, I can see how recent the poll is.
It would be nice if your simulator had some type of similar scroll-over feature. I have no way to know if the numbers you are using are the most recent or not.
270toWin: Unfortunately, we didn’t build that in to the simulator, nor into our other maps. Would be nice to have, but our resources are such that it isn’t going to happen this election cycle. We update the polls as we find them, which automatically updates the simulator. You can see roughly how each state polls for simulator purposes by looking at the polling map (http://www.270towin.com/2008_polls/mccain_obama/).
10 Barton // Oct 17, 2008 at 6:07 pm
I really like using the simulator, one of the things that I think would be nice would be to show the effect of the polls closing in various time zones… have the eastern time zone display results first, followed by central, mountain and pacific. Perhaps seeing the states where the margins are thinner display after the ones where one party or the other has a large margin, simulating states where one contender went over the top, or, where the counts are close, put in a delay.
This would add a little drama to each run of the simulator.
270toWin: This has been requested by a number of people. Good idea, probably not something that we’re going to be able to do for this election, however.
Leave a Comment